# MENTAL DISORDERS ASSOCIATED WITH NEUROLOGICAL DISEASES

EDITED BY : Yi Yang, Chunxue Wang, Yu-Tao Xiang, Thomas Penzel and Jun Lu PUBLISHED IN : Frontiers in Psychiatry and Frontiers in Neurology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-719-5 DOI 10.3389/978-2-88963-719-5

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# MENTAL DISORDERS ASSOCIATED WITH NEUROLOGICAL DISEASES

Topic Editors: Yi Yang, The First Hospital of Jilin University, China Chunxue Wang, Beijing Tiantan Hospital, China Yu-Tao Xiang, University of Macau, China Thomas Penzel, Charité – Universitätsmedizin Berlin, Germany Jun Lu, Harvard Medical School, United States

Citation: Yang, Y., Wang, C., Xiang, Y.-T., Penzel, T., Lu, J., eds. (2020). Mental Disorders Associated With Neurological Diseases. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-719-5

# Table of Contents


Yiming Deng, Luyao Wang, Xuan Sun, Lian Liu, Meifang Zhu, Chunxue Wang, Binbin Sui, Mi Shen, Weibin Gu, Dapeng Mo, Ning Ma, Ligang Song, Xiaoqing Li, Xiaochuan Huo, Zhongrong Miao, Duanduan Chen and Feng Gao

*22 The Orbitofrontal Cortex Gray Matter is Associated With the Interaction Between Insomnia and Depression*

Siyi Yu, Zhifu Shen, Rui Lai, Fen Feng, Baojun Guo, Zhengyan Wang, Jie Yang, Youping Hu and Liang Gong

*30 Mechanisms and Therapeutic Targets of Depression After Intracerebral Hemorrhage*

Yinan Wu, Liangliang Wang, Kaimin Hu, Chengcheng Yu, Yuanhan Zhu, Suzhan Zhang and Anwen Shao


Xinzhen Yin, Ying Zhou, Shenqiang Yan and Min Lou

*66 An Investigation on the Clinical Features and Neurochemical Changes in Parkinson's Disease With Depression* Teng-Hong Lian, Peng Guo, Li-Jun Zuo, Yang Hu, Shu-Yang Yu, Li Liu, Zhao Jin, Qiu-Jin Yu, Rui-Dan Wang, Li-Xia Li, Ying-Shan Piao and Wei Zhang


Meidan Fang, Lili Zhong, Xin Jin, Ranji Cui, Wei Yang, Shuohui Gao, Jing Lv, Bingjin Li and Tongjun Liu

*112 Anxiety in Patients With Acute Ischemic Stroke: Risk Factors and Effects on Functional Status*

Wei Li, Wei-Min Xiao, Yang-Kun Chen, Jian-Feng Qu, Yong-Lin Liu, Xue-Wen Fang, Han-Yu Weng and Gen-Pei Luo

*120 Does the Use of Antidepressants Accelerate the Disease Progress in Creutzfeldt–Jakob Disease Patients With Depression? A Case Report and A Systematic Review*

Yifan Liang, Yan Li, Huibin Wang, Xi Cheng, Meiting Guan, Shanshan Zhong and Chuansheng Zhao


Ming-Ya Luo, Zhen-Ni Guo, Yang Qu, Peng Zhang, Zan Wang, Hang Jin, Hong-Yin Ma, Shan Lv, Xin Sun and Yi Yang

*142 Brain Iron Deposits in Thalamus is an Independent Factor for Depressive Symptoms Based on Quantitative Susceptibility Mapping in an Older Adults Community Population*

Wenhua Zhang, Ying Zhou, Qingqing Li, Jinjin Xu, Shenqiang Yan, Jinsong Cai, Yeerfan Jiaerken and Min Lou

# Editorial: Mental Disorders Associated With Neurological Diseases

#### Yi Yang1,2 \*, Chunxue Wang<sup>3</sup> , Yutao Xiang<sup>4</sup> , Jun Lu<sup>5</sup> and Thomas Penzel 6,7

*<sup>1</sup> Department of Neurology, The First Hospital of Jilin University, Changchun, China, <sup>2</sup> Department of Neurology, Clinical Trial and Research Center for Stroke, The First Hospital of Jilin University, Changchun, China, <sup>3</sup> Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>4</sup> Faculty of Health Sciences, University of Macau, Macao, China, <sup>5</sup> Department of Neurology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, United States, <sup>6</sup> Sleep Medicine Center, Charite University Hospital, Berlin, Germany, <sup>7</sup> Saratov State University, Saratov, Russia*

Keywords: mental disorders, post-stroke depression, cognitive impairment, depression, post-stroke anxiety, Parkinson's disease

#### **Editorial on the Research Topic**

#### **Mental Disorders Associated With Neurological Diseases**

Mental disorders are important comorbidities of nervous system diseases and they have a lot in common in risk factors and pathogenesis. However, mental disorders can be easily neglected by neurologists. The mechanisms underlying the association of mental disorders with neurological disease are largely unclear. This Research Topic provides a collection of research into post-stroke depression and anxiety, cognitive impairment, and depression in Creutzfeldt-Jakob and Parkinson's disease. Although the present research collection cannot cover the whole range of advancements in the field, it highlights certain key findings regarding mental disorders associated with neurological diseases and we hope that this will inspire further interest and new research efforts in this exciting area.

Depression is a global chronic medical illness that leads to low mood, loss of interest, change in appetite, insomnia, and neurocognitive dysfunction. Despite the prevalence of depression, there are still many aspects to be explored and understood. An increasing prevalence of late-life depression has been identified, the mechanisms of which remain unclear. Previous studies demonstrated that iron deposition was related to the severity of symptoms in patients with depression. Zhang et al. investigated the role of iron deposits in depression among older adults and found that iron deposits in the thalamus was an independent factor relating to depressive symptoms. This new finding may inform future studies into the underlying pathophysiological mechanisms of depression.

Cerebral autoregulation was initially considered as an intrinsic protective mechanism of the brain, which ensures relatively constant cerebral blood flow despite fluctuations in arterial blood pressure or cerebral perfusion pressure. The impairment of cerebral autoregulation has been reported to be a feature of several diseases, including cerebral stroke, and Alzheimer's disease. Luo et al. observed that cerebral autoregulation was compromised in patients with depression and negatively correlated with the depression score. Though the mechanism is still unknown, improving cerebral autoregulation could be a potential therapeutic approach to treating the neurological symptoms of depression.

Depressive disturbances are common in patients with Parkinson's disease, but the neurochemical changes that occur in these cases are still unknown. In order to address this, Lian et al. investigated clinical features and neurochemical changes in patients with Parkinson's disease. The authors report that a high proportion of patients with Parkinson's disease had depression. Motor symptoms, postural instability, gait difficulty, anxiety, and fatigue are the significant influencing factors in cases of Parkinson's disease with

#### Edited and reviewed by:

*Paul Stokes, King's College London, United Kingdom*

\*Correspondence: *Yi Yang doctoryangyi@163.com*

#### Specialty section:

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

Received: *13 December 2019* Accepted: *28 February 2020* Published: *24 March 2020*

#### Citation:

*Yang Y, Wang C, Xiang Y, Lu J and Penzel T (2020) Editorial: Mental Disorders Associated With Neurological Diseases. Front. Psychiatry 11:196. doi: 10.3389/fpsyt.2020.00196*

**5**

depression. Moreover, dopamine may play a more important role in Parkinson's disease with depression compared to 5-HT. In another study, Zhu et al., observed that high concentrations of dopamine may cause the high incidence of restless leg syndrome (RLS) in Parkinson's disease patients, which was accompanied by anxiety, depression, insomnia, and other mental health symptoms. This finding highlights the importance of monitoring such symptoms in the clinical management of patients with Parkinson's disease.

Post-stroke depression, the most common psychiatric implication of stroke, negatively impacts patients' rehabilitation results, cognitive function, and quality of life. In this Research Topic, Huang, Zhao et al. explore the potential interaction between depressive symptoms and cognitive impairment after stroke and found remitters of post-stroke depression had more significant cognitive improvements than non-remitters. Therefore, early recognition and intervention for potential depression after stroke is of great importance. This study also identified predictors of remission in patients with earlyonset post-stroke depression, which included neurological impairment, major life events, major medical comorbidities, and frontal lobe lesion at baseline. A review in this Research Topic by Wang, Shi et al. provides a comprehensive overview of etiologies of post-stroke depression. In their article, the authors identify several factors related to the pathogenesis of post-stroke depression, including monoamine neurotransmitter change, inflammation, the hypothalamus-pituitary-adrenal axis and the hypothalamus-pituitary-thyroid axis, glial cells (astrocytes and microglia), vitamin D levels, homocysteine levels, neural network dysfunction, genetic background, and social psychological mechanisms.

Focusing on depression after intracerebral hemorrhage, Wu et al. provide a detailed review about the pathophysiology of depression after intracerebral hemorrhage, including inflammation, oxidative stress, apoptosis, and autophagy. This overview is accompanied by an in-depth exploration of the associated signaling pathways.

Recently, there has been an international focus on research into inflammation and post-stroke depression. Fang et al. produced a comprehensive review summarizing how neuroinflammation affects stroke rehabilitation and poststroke depression, potentially offering new therapeutic targets for stroke and post-stroke depression. Contributing also to the topic of post-stroke depression, Wang, Wang et al. share their research into the association between post-stroke depression, aphasia and physical independence, in stroke patients in China at a 3-months follow-up. The authors found the incidence of post-stroke depression was independently associated with physical dependence.

Another prevalent mental disorder after a stroke is anxiety. However, studies investigating the effects of post-stroke anxiety on functional status are very limited. The article by Li et al. describes that severity of post-stroke anxiety in the acute stage was a significant indicator for daily living and stroke-specific quality of life.

In this Research Topic, Liang et al. report on a case of sporadic Creutzfeldt-Jakob disease with depression. They found that the patient's condition worsened after using antidepressants. This finding was followed by a systematic survey, which showed that survival period was associated with the type of antidepressant used (especially serotonin and noradrenaline reuptake inhibitors).

Insomnia is a highly prevalent symptom in patients with mental disorders. Exploring the common and different brain mechanisms underlying such symptoms may help refine existing treatments. Yu et al. found that the interaction of depression and insomnia was associated with decreased gray matter volume in the right orbitofrontal cortex. This finding provides new insights into the mechanisms underlying the comorbidity of insomnia and depression. The comorbidity of insomnia and anxiety disorders is also worthy of further exploration. Another study into insomnia in this Research Topic is from Huang, Zhan et al., who reveal that cortical excitability in patients with generalized anxiety disorder comorbid with insomnia is modulated by insomnia. The authors examined the recovery functions of median nerve somatosensory evoked potentials, thus shedding light on the underlying neurobiological correlates of the effects of insomnia on generalized anxiety disorder.

Cognitive impairment seems to mark a high-risk population for developing dementia and plays a crucial role in the course of mental disability. The pathophysiology of it is a field where much work is yet to be done. The contribution from Wei et al. addresses a surrogate marker (the peak width of skeletonized mean diffusivity) for cognitive impairment in cerebral white matter lesions patients, which provides new insights into the pathophysiology of cognitive impairment in these patients. A study in this Research Topic by Deng et al. found that patients with severe vertebra-basilar stenosis showed a decline in cognitive ability, and that chronic posterior circulation hypoperfusion was an independent risk factor for cognitive impairment. The cerebral venous system also plays an important part in the progress of cognitive impairment. Compression and stenosis of the draining veins have been reported to be linked with transient global amnesia (a specific kind of cognitive impairment) via magnetic resonance imaging studies. Using ultrasound examination, Han et al. further confirmed a decrease in the total flow volume of the vertebral and internal jugular veins in patients with transient global amnesia. In addition, internal jugular vein drainage was relatively compromised during the Valsalva maneuver (an activity that can trigger transient global amnesia). To clarify the mechanisms involved in dementia, Zhou et al. performed a meta-analysis of the association between cortical superficial siderosis and dementia. The authors found that pre-existing cortical superficial siderosis could be a candidate imaging indicator for Alzheimer's disease. Another interesting study in this Research Topic is from Yin et al., who found that cerebral blood flow damage in white matter is associated with global cognitive dysfunction in CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy). Moreover, this exploratory study found cerebral blood flow was more strongly associated with global cognitive function than mean diffusivity and that this could be a biomarker used to monitor alterations of global cognitive function in CADASIL.

Finally, the editors would like to acknowledge the authors who contributed to this Research Topic. Their honest efforts and hard work are truly admirable. The editors hope that papers comprising this Research Topic will inspire significant progress in the field of mental disorders associated with neurological diseases.

# AUTHOR CONTRIBUTIONS

YY: drafted the manuscript. CW, YX, JL, and TP: revised the manuscript. All authors read and approved the final manuscript.

# FUNDING

This article was supported by the National Key R&D Program of China (2016YFC1301600) to YY, and the RF Government grant No 075-15-2019-1885 to TP.

**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Yang, Wang, Xiang, Lu and Penzel. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Association Between Post-stroke Depression, Aphasia, and Physical Independence in Stroke Patients at 3-Month Follow-Up

Shuo Wang1,2,3,4,5, Chun-Xue Wang1,2,3,4,5,6 \*, Ning Zhang1,2,3,4,5, Yu-Tao Xiang<sup>7</sup> , Yang Yang1,2,3,4,5, Yu-Zhi Shi 1,3, Yi-Ming Deng3,4,8, Mei-Fang Zhu1,2,3,4,5, Fei Liu1,2,3,4,5 , Ping Yu1,2,3,4,5, Gabor S. Ungvari <sup>9</sup> and Chee H. Ng<sup>10</sup>

*<sup>1</sup> Department of Neurology, Beijing Tian Tan Hospital, Capital Medical University, Beijing, China, <sup>2</sup> Department of Neuropsychiatry and Behavioral Neurology and Clinical Psychology, Beijing Tian Tan Hospital, Capital Medical University, Beijing, China, <sup>3</sup> China National Clinical Research Center for Neurological Diseases, Beijing, China, <sup>4</sup> Center of Stroke, Beijing Institute for Brain Disorders, Beijing, China, <sup>5</sup> Beijing Key Laboratory of Translational Medicine for Cerebrovascular Disease, Beijing, China, <sup>6</sup> Department of Clinical Psychology, Capital Medical University, Beijing, China, <sup>7</sup> Unit of Psychiatry, Faculty of Health Sciences, University of Macau, Taipa, Macau, <sup>8</sup> Department of Interventional Neuroradiology, Beijing Tian Tan Hospital, Capital Medical University, Beijing, China, <sup>9</sup> University of Notre Dame Australia, Marian Centre and Graylands Hospital, Perth, WA, Australia, <sup>10</sup> Department of Psychiatry, University of Melbourne, Melbourne, VIC, Australia*

#### Edited by:

*Gianluca Serafini, Ospedale San Martino (IRCCS), Italy*

#### Reviewed by:

*Jeffrey Guina, Wright State University, United States Paul David Barrows, University of Nottingham, United Kingdom*

> \*Correspondence: *Chun-Xue Wang snowsen@126.com*

#### Specialty section:

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

Received: *08 May 2018* Accepted: *25 July 2018* Published: *20 August 2018*

#### Citation:

*Wang S, Wang C-X, Zhang N, Xiang Y-T, Yang Y, Shi Y-Z, Deng Y-M, Zhu M-F, Liu F, Yu P, Ungvari GS and Ng CH (2018) The Association Between Post-stroke Depression, Aphasia, and Physical Independence in Stroke Patients at 3-Month Follow-Up. Front. Psychiatry 9:374. doi: 10.3389/fpsyt.2018.00374* Objective: Few studies have examined the association between post-stroke depression (PSD), aphasia, and physical independence in Chinese patients. This study investigated the above association in stroke patients in China at 3-month follow-up.

Methods: Altogether 270 patients within 14 days after ischemic stroke were recruited and followed up at 3 months. PSD, aphasia, and physical functional status were measured using the Stroke Aphasia Depression Questionnaire (SADQ), Western Aphasia Battery (WAB), and modified Rankin Scale (mRS), respectively. Patients with mRS total score >2 were considered as having "physical dependence."

Results: Out of 248 patients at 3-month follow up, 119 (48%) were rated as having physical dependence. Multiple logistic regression analyses revealed that female (*p* = 0.04; OR = 2.2; 95% CI: 1.0–5.1), more severe stroke at admission (*p* < 0.01; OR = 1.4; 95% CI: 1.3–1.5), and more severe PSD at 3 months (*p* = 0.01; OR = 1.05; 95% CI: 1.01–1.1) were independently associated with physical dependence at 3 months.

Conclusions: Greater PSD and stroke severity were independently associated with physical dependence at 3 months after stroke. Aphasia was also associated with physical dependence but the relationship was not significant. Early and effective depression screening, treatment and stroke rehabilitation appear to be important to improve the physical outcome and reduce the burden of stroke survivors.

Keywords: aphasia, ischemic stroke, depression, PSD, physical independence

# INTRODUCTION

Post-stroke depression (PSD) is one of the most common psychiatric comorbidities in stroke survivors, with a prevalence ranging from 20 to 65% (1–4). PSD is significantly associated with poor treatment adherence and increased risk of disability, mortality, stroke recurrence, and poor quality of life (5, 6). Aphasia occurs in about a third of ischemic stroke patients and is associated with impaired activity of daily life (7, 8) and higher risk of PSD (9, 10). Poor functional outcome after stroke could result in significant personal distress and family burden (11). Understanding the association between PSD, aphasia and physical independence is hence important to develop comprehensive treatment strategies for stroke survivors.

In China, the lifetime prevalence of stroke was 2.08% (95% CI, 2.02–2.13%) (12) in 2017, which translates to ∼2.9 million stroke patients. An Italy study found PSD was the only significant factor related to functional recovery from discharge to 3-month followup after stroke but not aphasia (13). To the best of our knowledge, there are no studies that have investigated the independent association between PSD, aphasia, and physical independence in stroke survivors in China. This study thus aimed to examine the association between PSD, aphasia, and physical independence in Chinese stroke patients at 3-month follow-up.

# METHODS

# Participants and Study Setting

This prospective cohort study was conducted between April 2014 and October 2015 in the Stroke Centre of Beijing Tiantan Hospital. A total of 320 patients were consecutively screened if they fulfilled the following criteria: (1) aged 18 years or older; (2) had an acute ischemic stroke within 14 days according to the WHO diagnostic criteria (14) confirmed by computed tomography (CT) or magnetic resonance imaging (MRI); and (3) had the ability to provide informed consent and complete the assessment. Exclusion criteria included: (1) history of language impairment; (2) drug and alcohol abuse and severe psychiatric disorders; (3) other major medical conditions, such as Parkinson's disease; and (4) severe cognitive deficit defined by the Mini Mental State Examination (MMSE) total score <18 (15, 16). The first three were excluded based on the self-reported prestroke histories by unstructured interviews and the last was made at the start of the study. The study protocol was approved by the ethics committee of Beijing Tiantan Hospital, Capital Medical University. All participants provided written informed consent.

### Measurement Instruments and Evaluation

Assessment was conducted at baseline and 3 months after index ischemic stroke. Patients' socio-demographic and clinical characteristics at baseline were recorded via a review of electronic medical records and confirmed by a clinical interview conducted by trained research neurologists. Severity and type of stroke was assessed with the National Institutes of Health Stroke Scale (NIHSS) (17–19) and the Trial of Org 10172 Acute Stroke Treatment (TOAST) (20, 21).

Physical independence and degree of handicap was evaluated using the modified Ranking Scale (mRS) at 3 months, with mRS total score >2 indicating physical dependence (22, 23). As most depression scales cannot be used in stroke patients with aphasia due to their impaired communication ability, the severity of depressive symptoms at 14 ± 2 days and 3 months after index stroke was measured using the 21-item Stroke Aphasic Depression Questionnaire (SADQ) (24, 25) with total score ranges from 0 to 63, ≥19 indicating the presence of PSD, ≥ 22 indicating moderate depressive symptoms and ≥ 26 indicating major depressive symptoms. The SADQ relies on external observation of emotional behavior by nursing staff and family members in recent time. The hospital version (SADQ-H) focus on the recent week. The severity and type of aphasia was evaluated with the Aphasia Quotient (AQ) derived from the Western Aphasia Battery (WAB), with AQ < 93.8 indicating the presence of aphasia (26). The language assessment was also conducted at 3-month follow-up with much missing data on AQ score, so the data was not analyzed.

#### Statistical Analysis

Data were analyzed with SPSS 21.0 (SPSS Inc., Chicago, IL, USA). Socio-demographic and clinical variables were compared between physical independence and dependence groups using Chi-square test, independent sample t-test and Mann-Whitney U-test, as appropriate. Independent correlates of the physical dependence at 3 months were examined using multivariate logistic regression analysis with the "enter" method. The outcome at 3 months was the dependent variable, while variables that significantly differed between both groups in the univariate analyses were entered as independent variables. Significance was set at 0.05 (two-tailed).

# RESULTS

Out of 320 patients with ischemic stroke who were consecutively screened, 270 fulfilled the study entry criteria and participated in the study, giving a participation rate of 84.4%. At baseline, 160 (59.3%) patients had aphasic symptom with the incidence of PSD was 47.5% compared with 29.1% in non-aphasiac patients (p < 0.01). At the 3-month assessment, 22 patients dropped out due to lack of interest, moving house or other unknown reasons. Of the 248 patients who completed the 3-month assessment, 119 (48.0%) had physical dependence.

**Table 1** shows the socio-demographic and clinical characteristics of the whole sample and separately by outcome. Patients with physical dependence were less likely to be married, and have large-artery atherosclerosis (LAA) TOAST type and pulmonary infection, but more likely to be female, have aphasia at baseline and more severe depressive symptoms at both baseline and 3 months after stroke. In addition, they had higher NIHSS scores at baseline, and higher SADQ scores at baseline and 3 months (all p-values < 0.05).

The independent correlates of physical dependence are shown in **Tables 2**, **3**. Due to collinearity between SADQ-H (Hospital version using during hospitalization) and SADQ score, two multivariate logistic regression analyses were performed with

#### TABLE 1 | Comparison of demographic and clinical variables between physical independence and dependence groups.


*Values p* < *0.05 are bolded; a, Mann-Whitney U-test; LAA, Large-Artery Atherosclerosis; NIHSS, National Institutes of Health Stroke Scale; SADQ-H, Stroke Aphasic Depression Questionnaire (Hospital version); SADQ, Stroke Aphasic Depression Questionnaire; TOAST, Trial of Org 10172 in Acute Stroke Treatment.*

SADQ-H and SADQ separately. Finally, female, NIHSS score at admission and SADQ score at 3 months were independently associated with physical dependence at 3-month follow up (adjusted R <sup>2</sup> = 0.410 in **Table 2** and adjusted R <sup>2</sup> = 0.423 in **Table 3**).

#### DISCUSSION

Generally about a third of patients with ischemic stroke have poor functional outcome at different follow-up time points (27, 28). In this study, the prevalence of physical dependence was 48% using the mRS cut-off score, which is similar to the 5-year prevalence of poor functional outcome (45%) in another study with the same measure (29). However, any direct comparison should be done with caution due to the different assessment measures, follow-up time points, and demographic characteristics.

A report of the American Heart Association/American Stroke Association in 2017 and meta-analyses found that approximately a third of stroke patients develop PSD at any point after stroke (30–32). A large-scale prospective cohort study in China involving over 2,000 stroke patients found that the cumulative incidence of PSD at 1 year after the index stroke was 42% (4). Depression was found predictive of worse functional outcome in an updated meta-analysis about the impact of depression on stroke outcome (33). While up to 62–70% in aphasic patients with stroke were diagnosed as major depression according to DSM-III-R criteria at 3 months and 1 year after stroke (9, 10, 34) which was higher than our results. The reason may due to the different scales, population and time points. Aphasia have been considered for the potential association with PSD with inconsistent conclusion (35, 36). We found that the incidence of PSD in stroke patients with aphasia and non-aphasia was significantly different indicating aphasia may be a risk factor of the development of PSD.

In this study, the incidence of PSD was 34.7% at 3 months after stroke, which was independently associated with physical dependence. Similar findings were also reported previously (37– 39). Depressive symptoms after stroke could result in behavioral and biological abnormalities, such as poor treatment adherence and dysregulation in autonomic system activation, which in turn, could lead to physical dependence (11, 37). However,



*Values p* < *0.05 are bolded; a, Mann-Whitney U-test; LAA, Large-Artery Atherosclerosis; NIHSS, National Institutes of Health Stroke Scale; SADQ-H, Stroke Aphasic Depression Questionnaire (Hospital version); TOAST, Trial of Org 10172 in Acute Stroke Treatment.* #*The severity of depressive symptoms was tested by SADQ-H at baseline and SADQ at 3 months. There was collinearity between SADQ-H at baseline and SADQ at 3 months, therefore they were not entered in the analysis concurrently.*

TABLE 3 | Independent correlates of physical dependence at 3 months# .


*Values p* < *0.05 are bolded; a, Mann-Whitney U test; LAA, Large-Artery Atherosclerosis; NIHSS, National Institutes of Health Stroke Scale; SADQ, Stroke Aphasic Depression Questionnaire; TOAST, Trial of Org 10172 in Acute Stroke Treatment.* #*The severity of depressive symptoms was tested by SADQ-H at baseline and SADQ at 3 months. There was collinearity between SADQ-H at baseline and SADQ at 3 months, therefore they were not entered in the analysis concurrently.*

since more than half of the patients suffered from aphasia in this sample, the use of the SADQ required the input of nursing staff and/or family members. Therefore, we could not exclude the possibility that nursing staff and family members were unable to distinguish between insomnia, irritability, poor appetite, anxiety, and depressive symptoms, which would cause bias in the incidence of PSD to an uncertain extent.

In this study, aphasia was measured using the AQ that covered complete aphasia, motor aphasia, sensory aphasia, transcortical mixed aphasia, transcortical motor aphasia, transcortical sensory aphasia, anomic aphasia, and conduction aphasia. Aphasia may lead to communication or comprehension difficulties, social avoidance and decreased attention, which is associated with physical dependence in stroke survivors (40). However, this finding was only confirmed in the univariate, but not in the multivariate analyses in this study. It is speculated that the association between aphasia and poor physical independence was moderated by other variables, such as depressive symptoms and severity of stroke. In addition, traditional Confucian culture favors family support and inter-dependence, particularly for family members with illness. For example, 94.8% of patients in this study were living with others. Thus, the strong family support may have offset the association between aphasia and physical dependence.

The association between demographic characteristics and physical dependence in stroke survivors have been inconsistent (41, 42). In this study, only female was independently associated with physical dependence, which is supported by previous findings (43, 44). The gender difference in physical independence in stroke survivors may be related to menstrual cycles, neuroendocrine regulation and more frequent physical comorbidities, such as diabetes, atrial fibrillation, and coronary heart disease in women with stroke (45). As expected, stroke severity as measured by the NIHSS was positively associated with physical dependence, which is consistent with previous findings (27, 29).

There are several methodological limitations to this study. First, this was a single-center study with relatively small sample size, therefore the findings could not be generalized to all stroke patients in China. Second, depressive symptoms were measured using the SADQ based on the observation by nursing staff or family members. There may be a gap between observerrated and self-reported measures of depression, although there are a number of self-reported measures specific for aphasia such as the Visual Analogue Mood Scales (VAMS) (46), Visual Analogue Self-Esteem Scales (VASES) (47), Disc Intensity Scale Circles (DISCS) (48), and Dynamic Visual Analogue Mood Scales (D-VAMS) (49). We chose SADQ from the perspective of relatively short items, easy operation and short time. Third, some important variables related to physical dependence, such as the use of medication, treatment adherence, the size and lesions of infarcts and the missing data about aphasia at 3-month, were not evaluated in the 3 months assessment.

In conclusion, physical dependence at 3-month follow up was common in Chinese stroke patients, which was associated with gender, greater PSD and stroke severity. Aphasia was also associated with physical dependence but the relationship was not significant. Our findings call for early and effective depression screening, treatment, and stroke rehabilitation to improve physical outcome and reduce the burden of stroke survivors in China.

#### AUTHOR CONTRIBUTIONS

SW and C-XW: study design. SW, NZ, YY, Y-ZS, Y-MD, M-FZ, FL, PY, and Y-TX: collection, analysis and interpretation of data. SW, C-XW, and Y-TX: drafting of the manuscript. GU, CN, and Y-TX: critical revision of the manuscript. All coauthors approval of the final version for publication.

#### FUNDING

This study was funded by the Ministry of Science and Technology and the Ministry of Health of the People's Republic of China. Individual grants include the National Key Research & Development Program of China (No. 2016YFC1301720), Beijing Brain Research (Z161100000216131), the Beijing Municipal Science and Technology Commission (Z151100004015127), and the Build High Level Technology Talents of Health System in Beijing (No.2015-3-038).

## REFERENCES


### ACKNOWLEDGMENTS

The authors would like to thank all of the participating colleagues, patients, and their families.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wang, Wang, Zhang, Xiang, Yang, Shi, Deng, Zhu, Liu, Yu, Ungvari and Ng. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Association Between Cerebral Hypoperfusion and Cognitive Impairment in Patients With Chronic Vertebra-Basilar Stenosis

Yiming Deng1,2,3†, Luyao Wang4†, Xuan Sun1,2,3, Lian Liu1,2,3, Meifang Zhu2,3,5 , Chunxue Wang2,3,5, Binbin Sui 2,3,6, Mi Shen2,3,6, Weibin Gu2,3,6, Dapeng Mo1,2,3 , Ning Ma1,2,3, Ligang Song1,2,3, Xiaoqing Li 1,2,3, Xiaochuan Huo1,2,3, Zhongrong Miao1,2,3 , Duanduan Chen7,8 \* and Feng Gao1,2,3 \*

<sup>1</sup> Department of Interventional Neuroradiology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>2</sup> China National Clinical Research Center for Neurological Diseases, Beijing, China, <sup>3</sup> Center of Stroke, Beijing Institute for Brain Disorders, Beijing, China, <sup>4</sup> Intelligent Robotics Institute, School of Mechatronical Engineering, Beijing Institute of Technology, Beijing, China, <sup>5</sup> Departments of Neuropsychiatry and Clinical Psychology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>6</sup> Department of Radiology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>7</sup> School of Life Science, Beijing Institute of Technology, Beijing, China, <sup>8</sup> Key Laboratory of Convergence Medical Engineering System and Healthcare Technology, The Ministry of Industry and Information Technology, Beijing Institute of Technology, Beijing, China

#### Edited by:

Yuan-Pang Wang, Universidade de São Paulo, Brazil

#### Reviewed by:

Rong Chen, University of Maryland, Baltimore, United States Bingkun Kevin Chen, Mayo Clinic, United States

#### \*Correspondence:

Duanduan Chen duanduan@bit.edu.cn Feng Gao gaofengttyy@yeah.net

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry

Received: 13 June 2018 Accepted: 31 August 2018 Published: 26 September 2018

#### Citation:

Deng Y, Wang L, Sun X, Liu L, Zhu M, Wang C, Sui B, Shen M, Gu W, Mo D, Ma N, Song L, Li X, Huo X, Miao Z, Chen D and Gao F (2018) Association Between Cerebral Hypoperfusion and Cognitive Impairment in Patients With Chronic Vertebra-Basilar Stenosis. Front. Psychiatry 9:455. doi: 10.3389/fpsyt.2018.00455 Objective: This study aimed to investigate the association between cognitive impairment and cerebral haemodynamic changes in patients with chronic vertebra-basilar (VB) stenosis.

Methods: Patients with severe posterior circulation VB stenosis and infarction or a history of infarction for more than 2 weeks from January 2014 to January 2015 were enrolled (n = 96). They were divided into three groups, namely, the computed tomography perfusion (CTP) normal group, the CTP compensated group, and the CTP decompensated group. Cognitive function was assessed using a validated Chinese version of the Mini-Mental State Examination (MMSE), the Frontal Assessment Battery (FAB), and the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS). Regression models were used to identify independent risk factors for cognitive impairment.

Results: The MMSE and FAB scores of patients in the CTP decompensated group were significantly lower than those of patients in the CTP normal and CTP compensated groups (all p < 0.05). The RBANS total and its domain scores, including immediate memory, visual acuity, and delayed memory, in the CTP compensated and CTP decompensated groups were significantly lower than those in the CTP normal group (all p < 0.05). Multiple regression analyses showed that CTP compensation, CTP decompensation, severe VB tandem stenosis, and multiple infarctions were independent risk factors for cognitive impairment.

Conclusions: Low perfusion caused by severe VB stenosis can lead to extensive cognitive impairments in areas such as immediate memory, visual span, and delayed memory.

Keywords: vertebra-basilar stenosis, cognitive impairment, cerebral hypoperfusion, cerebral infarction, stroke

#### INTRODUCTION

Neurocognitive function changes with age (1) and disease progression (2–4), which is related to pathologic mechanisms and is easily examined clinically. Carotid artery stenosis is closely related to vascular cognitive impairment (VCI) (5). Carotid artery stenosis can not only directly lead to the occurrence and rapid progression of VCI but also accelerate the development of degenerative diseases, such as Alzheimer's disease (6). Because of the collateral circulation in cerebral arteries, stenosis at the same site may cause different levels of cerebral blood flow perfusion. Studies have found that changes in cerebral flow perfusion were related to VCI in patients with carotid artery stenosis. Hypoperfusion caused by carotid artery stenosis can lead to frontal lobe damage, which in turn reduces the attention, language fluency, spatial structure, short-term memory, and executive function of patients (7). Compared with studies of VCI induced by carotid artery stenosis, few studies have examined the contribution of the posterior circulation or vertebra-basilar (VB) artery stenosis to cognitive impairment. Additionally, the correlation between cerebral blood flow perfusion and VCI in patients with VB artery stenosis remains unclear.

The stroke recurrence rate of the VB artery is reported to be relatively high (8, 9). For strokes in the posterior circulation or VB artery, transient ischaemic attack (TIA) accounts for ∼20% of ischaemic stroke cases (10). The clinical presentation of posterior circulation ischaemic strokes is unapparent and differs from those of anterior circulation or carotid artery strokes. Consequently, this type of stroke is often hidden (11). Manifestations such as vertigo, diplopia, and coughing while drinking water are generally ignored by patients. In contrast, anterior circulation symptoms, such as facial or limb paralysis, are often more likely to be noted (12).

Basilar artery stenosis may lead to poor attention, poor executive function, and long-term memory impairment in patients (13). In this study, computed tomography perfusion (CTP) was used to analyse the relationship between cognitive impairment and cerebral haemodynamic changes. We aimed to investigate the cognitive status of patients with chronic posterior circulation hypoperfusion, which, to our best knowledge, has received little systemic investigation.

#### MATERIALS AND METHODS

#### Subjects

This study was a prospective cohort study (Clinical Trial Registration URL: http://www.clinicaltrials.gov. Unique identifier: NCT01968122.). All methods were performed in accordance with the relevant guidelines and regulations. A total of 96 patients who were diagnosed with severe posterior circulation VB stenosis and had infarction or a history of infarction for more than 2 weeks from January 2014 to January 2015 were enrolled in the current study. The inclusion criteria were as follows: (1) Patients who had vertebral artery or basilar artery stenosis confirmed by CT angiography (CTA) or digital subtraction angiography (DSA) examination, with a stenosis area equal to or greater than 70% of the vascular area (14–18). In this study, 70% of patients underwent CTA examination, 50% underwent DSA examination, and 20% underwent CTA and DSA examination. (2) Cranial magnetic resonance imaging (MRI) showed that the area of nonlacunar infarction [multiple infarctions, high signal greater than or equal to two diffusion-weighted imaging (DWI) images] was <1/3 of the hemisphere area. The exclusion criteria were as follows: (1) patients who failed to complete the scale evaluation due to aphasia, apraxia, and dysphonia; (2) patients who had cognitive impairment caused by Alzheimer's disease and other related nervous system degeneration or nonvascular factors; (3) patients who had nervous system diseases (such as central nervous system hereditary diseases, tumors, encephalitis, demyelinating disease, Parkinson's disease, craniocerebral injury, and epilepsy) that could lead to cognitive impairment; (4) patients who had anxiety, depression, or other mental disorders; (5) patients who had severe diseases of the liver, kidney, heart, or blood; (6) patients who had hypothyroidism, chronic alcoholism, infection, or other cognitive function-related diseases; (7) patients who had a history of substance abuse, drug addiction, carbon monoxide, pesticide, and other chemical poisoning, brain parasites, etc.; and (8) patients whose first-degree relatives had dementia and psychosis, cerebral lacuna infarct, or leukodystrophy revealed by brain MRI examination. The recruitment diagram is shown in **Figure 1**. After a complete description of the study, all subjects gave their written informed consent to participate in the study. This study was approved by the Regional Committee for Ethics of Beijing Tiantan Hospital.

#### Baseline Data Assessment

The patients were divided into three groups: the CTP normal group, CTP compensated group, and CTP decompensated group (19). Baseline information, including gender, age, length of education, left- or right-handedness, high blood pressure, diabetes, atrial fibrillation, and smoking, was collected. Information on patient history of hypertension, diabetes, and hyperlipidaemia was recorded. Briefly, blood pressure ≥140/90 mm Hg (1 mm Hg = 0.133 kPa) was defined as hypertension, and fasting blood glucose ≥7.0 mmol/L, 2 h postprandial blood glucose ≥11.1 mmol/L or random blood glucose ≥11.1 mmol/L were defined as diabetes. Atrial fibrillation was diagnosed according to the 1979 World Health Organization (WHO) diagnostic criteria. Hyperlipidaemia was diagnosed based on the "Chinese Adult Dyslipidaemia Prevention and Control Guidelines" from 2007. Hyperlipidaemia was diagnosed when the patients met one of the following criteria: blood cholesterol concentration >5.17 mmol/L; blood concentration

**Abbreviations:** VB, Vertebra-Basilar; CTP, Computed Tomography Perfusion; MMSE, Mini-Mental State Examination; FAB, Frontal Assessment Battery; RBANS, Repeatable Battery For The Assessment Of Neuropsychological Status; VCI, Vascular Cognitive Impairment; TIA, Transient Ischaemic Attack; CTA, Ct Angiography; DSA, Digital Subtraction Angiography; DWI, Diffusion-Weighted Imaging; MRI, Magnetic Resonance Imaging; TTP, Time To Peak; MTT, Transit Time; CBF, Cerebral Blood Flow; CBV, Cerebral Blood Volume; PCA, Posterior Cerebral Arteries; PCI, Percutaneous Coronary Intervention.

of triglycerides >1.7 mmol/L; or blood concentration of lowdensity lipoproteins >3.1 mmol/L.

#### Cognitive Function Evaluation

Cognitive function was assessed using the validated Chinese version of the Mini-Mental State Examination (C-MMSE) (20), the Frontal Assessment Battery (FAB) (21), and the Repeatable Battery for the Assessment of Neuropsychological Status


(RBANS) (22). Three researchers participated in a cognitive function training course before the study started. Repeated evaluations showed that the overall correlation coefficient of the MMSE, FAB, and RBANS for the three researchers was >0.8 after training. Finally, an RBANS total score >77.5 was defined as cognitively normal, and an RBANS total score ≤77.5 was defined as cognitively impaired (23).

#### Imaging Evaluation

Posterior circulation acute ischaemic infarction (including cerebral infarction) was diagnosed in patients with clinical manifestations. Additionally, such patients had high-density lesions on magnetic resonance DWI or TIA in the posterior circulation. CTP was performed using a Siemens dual-source spiral CT machine with 128 layers (Germany). Briefly, a volume of 60 mL of contrast agent (iohexol, 370 mg I/mL) was injected into the elbow middle vein at a rate of 8 mL/s using a doubletube high-pressure syringe (Ulrich Missouvi XD2501-C), and a volume shuttle scan with a scanning range of ∼110 mm was started after a delay of 4 s. Intravenous injection of iohexol was performed using an EZEM high-pressure syringe (America) at a rate of 5 mL/s. The base section plane was selected, and two layers were continuously scanned 40 times with the parameters of 80 kV, 200 mA, layer thickness 12 mm, and pitch 0.75. Forty images in each layer were scanned, and a total of 80 images were obtained. Four images of the temporal lobe and 4 images of the occipital lobe were selected from each layer image as the region of interest (ROI). The original CTP image was introduced into a dedicated postprocessing workstation (Neusoft Medical Co., Shenyang, China) and analyzed with CT perfusion software. Time to peak (TTP), transit time (MTT), cerebral blood


<sup>a</sup>Group I, CTP normal group; <sup>b</sup>Group II, CTP compensated group; <sup>c</sup>Group III, CTP decompensated group. \*P < 0.05, compared with group I.

(d), and DSA indicated severe proximal stenosis of the basilar artery (e). (C) Stage I3: TTP (a) and MTT (b) were prolonged, CBF was decreased (c), CBV was slightly decreased (d), and DSA indicated severe stenosis in the middle part of the basilar artery (e). (D) Stage I4: TTP (a) and MTT (b) were prolonged, CBF was decreased (c), CBV was decreased (d), and CTA demonstrated occlusion in the middle part of the basilar artery (e).

flow (CBF), and cerebral blood volume (CBV) were calculated. The qualitative assessment of perfusion in the ROI, which was used in a previous study (14), was grouped as follows. The patients in the CTP normal group had complete perfusion. The patients in the CTP compensated group had hypoperfusion and preserved cerebral vascular reactivity (a lower peak, delayed TTP, increased MTT, decreased CBF, and normal or elevated CBV). In addition, the patients in the CTP decompensated group had hypoperfusion without adequate cerebral vascular reactivity.

### Statistical Analysis

All statistical analyses were performed using SPSS software version 23.0 (SPSS Inc., Chicago, IL, USA). Demographic and clinical variables of the multiple groups were compared using one-way ANOVA for continuous variables and X<sup>2</sup> (chi-square test) or Fisher's exact test for categorical variables. Where there was significance in the ANOVA, we used the Fisher minimum significant difference (LSD) test for post hoc comparisons between groups.


<sup>a</sup>Group I, CTP normal group; <sup>b</sup>Group II, CTP compensated group; <sup>c</sup>Group III, CTP decompensated group. \*P < 0.05, compared with group I. <sup>1</sup>P < 0.05, compared with group II. MMSE, Mini-Mental State Examination; FAB, Frontal Assessment Battery; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status.

TABLE 3 | Regression models of independent risk factors for cognitive impairment.


<sup>a</sup>Adjusted for age, sex, hypertension, diabetes, atrial fibrillation, cigarette smoking, hyperlipidaemia, lesion site, and infarct pattern; <sup>b</sup>Adjusted for age, sex, hypertension, diabetes, atrial fibrillation, cigarette smoking, hyperlipidaemia, perfusion type, and infarct pattern; <sup>c</sup>Adjusted for age, sex, hypertension, diabetes, atrial fibrillation cigarette smoking, hyperlipidaemia, perfusion type, and lesion site.

A linear regression model was used to identify risk factors for cognitive impairment in patients with VB artery stenosis. P < 0.05 was considered statistically significant.

# RESULTS

#### Sociodemographic Data and Clinical Background Characteristics

A total of 96 patients were ultimately included in this study. Among them, 46 patients had severe basilar artery stenosis, 38 patients had severe intracranial artery stenosis, 32 patients had vertebral artery extracranial stenosis, 20 patients had tandem lesions, 12 patients had no new infarct (TIA), 56 patients had a single infarct, and 28 patients had multiple infarctions. The number of patients in the CTP normal group, CTP compensated group, and CTP decompensated group was 22, 42, and 32, respectively. There were no differences in the sociodemographic characteristics between the three groups (all p > 0.05). The rate of intracranial artery stenosis in the CTP compensated group was lower than that in the CTP normal group, (p < 0.05); however, the rate of intracranial artery stenosis in the CTP decompensated group was higher than that of the normal group (**Table 1**).

### Association Between MMSE, FAB, and RBANS Scores and CT Perfusion

The stages of posterior circulation perfusion are summarized in **Figure 2** in the order of cognitive decline.

As presented in **Table 2**, the MMSE, FAB, and RBANS scores of the CTP decompensated group were significantly lower than those of the CTP normal and CTP compensated groups (all p < 0.05). The RBANS total, immediate memory, visual acuity, and delayed memory scores in the CTP compensated and CTP decompensated groups were significantly lower than those in the CTP normal group (p < 0.05). CTP compensated patients had reduced attention compared to that of CTP normal patients (p < 0.05).

### Regression Models of Independent Risk Factors for Cognitive Impairment

Based on the RBANS total score, 68 patients were included in the cognitive impairment group (RBANS score ≤77.5), and 28 patients were considered to have no cognitive impairment (RBANS score >77.5 points). After adjusting for other relevant factors, CTP compensation (p = 0.30), CTP decompensation (p < 0.01), severe VB tandem stenosis (p = 0.021), and multiple infarctions (p = 0.023) were found to be independent risk factors for cognitive impairment (**Table 3**).

# DISCUSSION

The primary findings of this study could be summarized as follows: (A) patients who have chronic posterior circulation hypoperfusion showed a decline in cognitive ability; (B) medial temporal lobe perfusion was associated with serious cognitive impairment; (C) in addition to language ability, there were other dimensions of cognitive impairment; and (D) in patients with chronic posterior circulation hypoperfusion, multiple stenosis and multiple infarcts were independent risk factors for cognitive impairment.

The basilar artery branches into two posterior cerebral arteries (PCA), which supply the majority of blood to the temporal lobe and thalamus. Previous studies found that cognitive impairment existed in patients with infarcts in these regions (24–26). In our study, the cognitive ability of patients with low-perfusion percutaneous coronary intervention (PCI) generally decreased, which might be associated with chronic ischaemia and hypoxia of the brain structures mentioned above. Studies showed that the state of ischaemia and hypoxia was associated with damage to the neural network between the brainstem or cerebellar regions and the anterior circulation (25–27). Low perfusion leads to a decrease in thrombus clearance; additionally, the formation of microemboli that result from lesions caused by cerebral vascular stenosis also leads to VCI (28). In animal studies, microemboli were found to decrease the number of brain-derived neurotrophic factors in the hippocampus and lead to impaired memory in mice (29).

In our study, the executive function, immediate memory, delayed memory, and visual range of patients with PCI accompanied by hypoperfusion were impaired, which is in agreement with previous findings (30, 31). However, the language function of these patients was retained in our study, which is inconsistent with previous studies (31, 32). In these patients, the memory function, including short-term memory and delayed memory, was severely damaged, which might be related to long-term ischaemia and hypoxia of the medial temporal lobe structures. The efferent fibers and afferent fibers of the temporal lobe have a wide range of links with the frontal lobe, parietal lobe, occipital lobe, and hippocampus (33). The hippocampus plays an important role in mood, neuropsychological activities, memory, execution, language (including fluency and repetition), and other cognitive activities (34). Memory impairment may occur before stroke, which might be associated with the chronic ischaemia and hypoxia caused by the hypoperfusion of the medial temporal lobe (34). Executive function impairment may be caused by damage in part of the tissues of the VB artery, whose function is linked to the thalamus, parietal lobe, and frontal lobes (25, 26, 35, 36). Visual span impairment might be associated with chronic ischaemia and hypoxia in the occipital lobe and temporal lobe (19, 37). Further multifactor logistic regression analysis revealed that low perfusion of blood supply areas, tandem, or multiple stenosis, and multiple PCI were independent risk factors for cognitive impairment in patients with PCI. Both CTP compensated and CTP decompensated patients had cognitive impairment. The incidence of cognitive impairment in CTP decompensated patients was 6.8 times higher than that observed in the normal metabolic patients. A previous study found that the prognostic MRS score of patients with PCI was significantly higher than that of patients with anterior circulation infarction (38). Although the neurological function of patients with PCI recovers well, their cognitive function is likely to suffer sustained damage if the collateral circulation is

#### REFERENCES


not sufficient or chronic hypoperfusion is persistent. Tandem lesions or multiple stenosis can lead to a further decrease in perfusion in the posterior circulation area (39). The presence of chronic persistent hypoperfusion can lead to multiple infarcts in the brain, which also aggravates the cognitive impairment of patients (40). It was reported that in first-onset mild stroke patients, the occurrence of multiple infarcts and decreased hippocampal volume were positively correlated with cognitive impairment (41). Even in patients with asymptomatic stroke, multiple infarcts caused by hypoperfusion or microemboli also led to reduced hippocampal volume, resulting in decreased memory. In addition, multiple cerebral infarctions led to declines in language function, processing speed, and visual spatial competence (42).

The present study has some limitations. First, the RBANS was performed by only a single independent reviewer; therefore, there was some subjectivity in the judgement of graphic memory. Second, a small number of patients had carotid artery stenosis. As a result, these patients may be affected by cognitive effects due to anterior circulation cerebral hypoperfusion. Third, the sample size of the study is relatively small, which limits the generalizability of the results. Hence, the conclusions must be further confirmed with a larger sample size. In the future, we will design different experiments related to neurocognitive function (43, 44) and use various analytical methods to explore the pathologic mechanisms of neurocognitive deficits in patients.

# AUTHOR CONTRIBUTIONS

YD and LW: analyzed and interpreted the data, wrote the paper. XS, LL, MZ, and CW: contributed to the conception or design of the work, interpreted the data. BS, MS, WG, and DM: conceived and designed the experiments, performed the experiments. NM, LS, XL, ZM, and XH: performed the experiments, drafted and revised the work. DC and FG: revised the paper, approved the final version.

# ACKNOWLEDGMENTS

We acknowledge and thank the subjects involved in the study. This study was financially supported by the National Key Research and Development Program of China under grant 2018YFC0115400, the National Natural Science Foundation of China (grant number 81471752); the Beijing Municipal Science & Technology Commission (grant number Z161100001116122); and the Beijing Nova Program (grant number Z171100001117057).

and transient ischemic attack. CNS Neurosci Therap. (2018) 24:154–61. doi: 10.1111/cns.12787


to Alzheimer's disease. Theranostics (2018) 8:3237–55. doi: 10.7150/thno. 23772


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Deng, Wang, Sun, Liu, Zhu, Wang, Sui, Shen, Gu, Mo, Ma, Song, Li, Huo, Miao, Chen and Gao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Orbitofrontal Cortex Gray Matter Is Associated With the Interaction Between Insomnia and Depression

Siyi Yu1†, Zhifu Shen1†, Rui Lai <sup>2</sup> , Fen Feng<sup>3</sup> , Baojun Guo<sup>1</sup> , Zhengyan Wang<sup>4</sup> , Jie Yang<sup>1</sup> , Youping Hu<sup>1</sup> \* and Liang Gong5,6 \*

*<sup>1</sup> Department of Acupuncture & Tuina, Chengdu University of Traditional Chinese Medicine, Chengdu, China, <sup>2</sup> Department of Anesthesiology, People's Hospital of Deyang, Deyang, China, <sup>3</sup> Affiliated Hospital of Chengdu University of Traditional Chinese Medicine, Chengdu, China, <sup>4</sup> Department of Pain Management, Sichuan Integrative Medicine Hospital, Chengdu, China, <sup>5</sup> Department of Neurology, Chengdu Second People's Hospital, Chengdu, China, <sup>6</sup> Department of Neurology, Affiliated ZhongDa Hospital, School of Medicine, Southeast University, Nanjing, China*

#### Edited by:

*Yu-Tao Xiang, University of Macau, China*

#### Reviewed by:

*Yang-Kun Chen, Dongguan People's Hospital, China Huajun Liang, University of Maryland, United States*

#### \*Correspondence:

*Youping Hu hypcdutcm@yeah.net Liang Gong seugongliang@hotmail.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

Received: *02 June 2018* Accepted: *16 November 2018* Published: *04 December 2018*

#### Citation:

*Yu S, Shen Z, Lai R, Feng F, Guo B, Wang Z, Yang J, Hu Y and Gong L (2018) The Orbitofrontal Cortex Gray Matter Is Associated With the Interaction Between Insomnia and Depression. Front. Psychiatry 9:651. doi: 10.3389/fpsyt.2018.00651* Insomnia and depression are highly comorbid symptoms in both primary insomnia (PI) and major depressive disorder (MDD). In the current study, we aimed at exploring both the homogeneous and heterogeneous brain structure alteration in PI and MDD patients. Sixty-five MDD patients and 67 matched PI patients were recruited and underwent a structural MRI scan. The subjects were sub-divided into four groups, namely MDD patients with higher or lower insomnia, and PI patients with higher or lower severe depression. A general linear model was employed to explore the changes in cortical thickness and volume as a result of depression or insomnia, and their interaction. In addition, partial correlation analysis was conducted to detect the clinical significance of the altered brain structural regions. A main effect of depression on cortical thickness was seen in the superior parietal lobe, middle cingulate cortex, and parahippocampal gyrus, while a main effect of insomnia on cortical thickness was found in the posterior cingulate cortex. Importantly, the interaction between depression and insomnia was associated with decreased gray matter volume in the right orbitofrontal cortex, i.e., patients with co-occurring depression and insomnia showed smaller brain volume in the right orbitofrontal cortex when compared to patients with lower insomnia/depression. These findings highlighted the role of the orbitofrontal cortex in the neuropathology of the comorbidity of insomnia and depression. Our findings provide new insights into the understanding of the brain mechanism underlying comorbidity of insomnia and depression.

Keywords: insomnia, depression, comorbidity, structural MRI, orbitofrontal cortex

# INTRODUCTION

Insomnia represents a common symptom seen in the world population (about 30%), with about 6–10% of the adult population reaching the diagnostic criteria for Primary insomnia (PI) (1). In contrast, major depressive disorder (MDD) is the second cause of disability worldwide, which is to become the world's most frequent and economically burdensome illness by 2030 (2, 3). Sleep complaints, especially the symptom of insomnia, are reported in up to 90% of MDD patients and can profoundly impact both the severity of depression and the course of the illness (4, 5). In addition, about 20% of patients with insomnia suffer from depression (6, 7). Furthermore,

**22**

insomnia was found to be a predictor of depression given that non-depressed people with insomnia have a 2-fold risk to develop depression according to a recent meta-analysis of longitudinal epidemiological studies (8). These findings suggest that the link between insomnia and depression is bidirectional. In addition, both antidepressants and hypnotic medication are commonly prescribed to patients with the combined condition of depression and insomnia (9, 10). Therefore, due to the heterogeneity and homogeneity between insomnia and depression, exploring the common and different brain mechanisms underlying such symptoms may help refine existing depression and insomnia treatments and develop personalized treatment for PI and MDD.

In the last decades, accumulating neuroimaging studies suggested PI and MDD to be associated with some functional and structural alterations in the brain of patients (11, 12). For example, Winkelman et al. found that patients with chronic PI reported an increased cortical volume in the rostral anterior cingulate cortex (rACC) when compared to normal sleepers, which was an indication of clinical severity (13). Other studies also reported volumetric differences in the frontal cortex, OFC, parietal cortex, precuneus and hippocampus (14, 15). In contrast, neuroimaging studies in patients with MDD described smaller volumes of the hippocampus, thalamus, insula, frontal lobe, orbitofrontal cortex and rACC (16, 17). However, due to the heterogeneity between the two diseases, such results regarding the structural brain alteration were not always consistent (11, 18). More recently, an increasing number of researchers pay attention on both the common and different brain mechanisms underlying insomnia and depression. For example, Cheng et al. conducted a study on a bit sample of healthy individuals and found an increased functional connectivity in the lateral orbitofrontal cortex (OFC), dorsolateral prefrontal cortex, anterior and posterior cingulate cortex and insula, which was associated with both sleep and depressive scores (19). Furthermore, Liu et al. reported increased amplitude of low-frequency fluctuations (ALFF) during the resting state in the right inferior frontal gyrus and anterior insula in MDD patient with insomnia when compared to MDD patients without insomnia. In addition, they suggested that the abnormal ALFF was associated with sleep disturbance scores (20). Moreover, Li et al. found that patients with PI had reduced gray matter volume in the middle cingulate cortex, which was significantly associated with self-rating for depression score (21). Yang et al. suggested that decreased gray matter volume both in the left lingual gyrus and cerebellum predicts insomnia in female MDD patients (22). Considering the high comorbidity of insomnia and depression in the two neuropsychiatric disorders, only a few study have investigated their interaction effect on brain structure in both PI and MDD patients.

In the present study, we aimed at exploring the potential brain mechanism underlying the comorbidity of insomnia and depression using structural magnetic resonance imaging (MRI). First, we detected the main and the interaction effects of insomnia and depression on brain cortical thickness and volume in four heterogeneous subgroups of patients, i.e., MDD patients with higher or lower insomnia (MDD-HI or MDD-LI), and PI patients with higher or lower depression (PI-HD or PI-LD). Second, we explored the clinical association between the influenced brain regions in each group. Based on previous neuroimaging finding on PI and MDD (21, 23, 24), we hypothesized that the prefrontal cortex, especially the OFC and the anterior cingulate cortex, would be influenced by the interaction between insomnia and depression.

# METHODS AND MATERIALS

### Participants

The present study is a preliminary and retrospective research, the enrollment is separately for MDD and PI group. All participants were recruited from the outpatient of department of neurology and psychiatry of the Chengdu University of Traditional Chinese Medicine (CDUTCM). We selected sixty-five MDD patients and 67 age-, gender-, and education-matched PI patients (**Table 1**). This study was approved by the Research Ethics Committee of CDUTCM and all participants gave written informed consent. The following eligibility criteria were considered for MDD patients: (1) Met the diagnostic criteria for MDD according to the Diagnostic Statistical Manual of Mental Disorder, fourth Edition (DSM-IV); (2) the Hamilton Rating Scale for Depression-17 (HAMD) score was equal or above 17; (3) Naïve to antidepressant medications or a washout period of at least five half-lives of the previously prescribed medicine was undergone (25, 26); (4) Age between 18 and 55; and (5) Age at onset was <50 years. The following inclusion criteria were considered for PI patients: (1) Met the diagnostic criteria for PI according to the DSM-IV; (2) complaints of difficulty of falling asleep, maintaining sleep or early awakening for at least 3 months; (3) Age between 18 and 55; and (4) Age at onset was <50 years. The exclusion criteria for all patients included: (1) a history of other major psychiatric disorders or a neurological illness history, except for anxiety in the current state; (2) substance abuse, including caffeine, nicotine, and alcohol (27); (3) any brain lesions found by a T2 MRI scan.

# Behavior Assessment and Subgroup Division

All participants underwent both a clinical and a behavioral assessment, while a neuropsychiatric examination was performed by two experienced neurologists (SY and ZS) who reached a consensus diagnosis. The HAMD for depression severity, the Hamilton Rating Scale for Anxiety (HAMA) for anxiety evaluation, and the HAMD sleep subscale (HAMD-S) for insomnia evaluation were used for assessing MDD patients. Following, according to the HAMD-S score, the MDD group was divided into the MDD subgroup with higher insomnia (MDD-HI, HAMD-S score >3) and the MDD subgroup with lower insomnia (MDD-LD, HAMD-S score <3) (20, 28, 29). Given that 5 MDD patients reported a HAMD-S score equal to 3, they were excluded from the statistical analysis. In contrast, the Pittsburgh Sleep Quality Index (PSQI) for evaluating the insomnia severity (30), the self-rating depression scale (SDS) for depression severity, and the self-rating anxiety scale (SAS) for anxiety severity evaluation were used for assessing PI patients. Following, the PI group was also divided into two subgroups

Yu et al. OFC Abnormal in Insomnia-Depression


TABLE 1 | Demographic and clinical characteristics for all participants.

*† The p-value was obtained by chi-square test; other p-values were obtained by a two-way T-test or one way analysis of variance. MDD-HI, major depressive disorder with higher insomnia; MDD-LI, major depressive disorder with lower insomnia; PI-HD, primary insomnia with higher depression; PI-LD, primary insomnia with lower depression. PSQI, Pittsburgh Sleep Quality Index; HAMD, Hamilton Rating Scale for Depression; HAMD-S, HAMD sleep subscale; SDS, self-rating depression scale; SAS, self-rating anxiety scale; HAMA, Hamilton Rating Scale for Anxiety.*

according to the SDS scores, namely the PI subgroup with higher depression (PI-HD, SDS score >55) and the PI subgroup with lower depression (PI-LD, SDS score <50). Given that 10 patients reported to have a SDS score between 50 and 55, they were not included in the statistical analysis (31).

#### Image Data Acquisition and Processing

All participants underwent MRI scanning on the same 3.0-Tesla magnetic resonance scanner (Discovery MR750, General Electric, Milwaukee, WI, USA) equipped with a standard head coil. All participants were instructed not to consume caffeine, alcohol, or any other psychoactive substance in the 48 h prior to the scan. Tight however comfortable foam padding was used to minimize head motion, and earplugs were employed to reduce the scanner noise. Sagittal 3D T1-weighted images were acquired using a brain volume sequence with the following parameters: repetition time (TR) = 8.16 ms, echo time (TE) = 3.18 ms, flip angle (FA) = 7 ◦ , field of view (FOV) =256 × 256 mm<sup>2</sup> , matrix =256 × 256; slice thickness = 1 mm, no gap; and 188 sagittal slices. During the MRI examination, all subjects were instructed to relax with their eyes closed without falling asleep or thinking about something. All participants were checked for their wakening status following the scan and they all claimed to be awake during the course of the study.

The images were processed using the standard surfacebased workflows in the FreeSurfer version 6.0 (http://surfer. nmr.mgh.harvard.edu/fswiki/recon-all/), including motion correction, non-parametric non-uniform intensity correction, intensity normalization, skull strip, automatic subcortical segmentation, white matter segmentation, tessellation, original surface smoothing, inflation, automatic topology fixer, surface registration, and cortical parcellation (32). The resulting surface reconstruction was visually inspected and manually edited in the following trouble shooting step (https://surfer.nmr.mgh.harvard. edu/fswiki/FsTutorial/TroubleshootingDataV6.0) by a single rater, blind to the diagnostic status. The troubleshooting included skull strip errors, segmentation errors, intensity normalization error, pial surface misplacement, and topological defect. The estimated Total Intracranial Volume (eTIV) was extracted as a covariate in a later analysis (33).

# Statistical Analysis

#### Demographic and Behavioral Data

One-way analysis of variance (ANOVA), independent samples T-tests and chi-square tests were conducted using the Statistical Package for the Social Sciences version 20.0 (SPSS, Inc., Chicago, IL, USA) to compare the demographic and behavioral data among groups. The Pearson correlation was employed to examine the relationships between HAMD/SDS scores and HAMD-S/PSQI scores. Considering the difference in both the disease duration between the two PI subgroups and in the HAMD score between the two MDD subgroups, the Pearson Correlation analysis was also conducted to investigate the relationships between both disease duration and SDS score in the PI groups and HAMD-S scores in the MDD groups. The significance was set at p < 0.05.

#### Structural Imaging

A general linear model (GLM) analysis was employed to explore the effect of depression, insomnia, and their interaction on the reconstructed cortical surface image (mri\_glmfit, FreeSurfer), regressed out the effect of gender, age, years of education, and the eTIV. The design matrices were created by a FreeSurfer Group Descriptor File (http://surfer.nmr. mgh.harvard.edu/fswiki/Fsgdf4G1V). For the main effect of depression, we set the MDD-HI, MDD-LI, and PI-HD as depressed patients, PI-LD as non-depressed patients; For the main effect of insomnia, we set the MDD-HI, PI-LD, PI-HD as insomnia patients, MDD-LI as non-insomnia patients; For the interaction of depression and insomnia effect, we set the MDD-HI, PI-HD as the interaction of depression and insomnia. The the contrast matrix is set as the equation below:

$$\begin{aligned} Y\_i &= \ \beta\_0 + \ \beta\_1 \ast Gender + \ \beta\_2 \ast Age + \beta\_3 \ast Education + \beta\_4 \ast eTIV \\ &+ \ \beta\_5 \ast Duration + \beta\_6 \ast Ins + \beta\_7 \ast Dep + \beta\_8 \ast (Ins \ast Dep) + \varepsilon \end{aligned}$$

where Y<sup>i</sup> represents the thickness or volume of each vertex in cortex of the ith participant; β<sup>0</sup> is the intercept of the straightline fitting in the model; β1, β2, β3, β4, and β<sup>5</sup> stand for the main effect of gender, age, education, eTIV volume, and disease duration, respectively, which were discarded as covariates of no interest in the GLM models. β<sup>5</sup> represents the main effect of depression, β<sup>6</sup> is the main effect of insomnia, β<sup>7</sup> describes the interaction effect of depression and insomnia. The error term ε was assumed to have a Gaussian distribution so that no correlation across participants was shown.

After the GLM analyses, a cluster-wise correction for multiple comparison was processed using the Monte Carlo Simulation (mri\_glmfit-sim, FreeSurfer, the vertex-wise threshold is p < 0.0001, the cluster-wise p < 0.05, iteration is 10,000, adjust p-values for two hemispheres) (34).

#### The Relationship Between Structural Imaging Features and Behavior

To further detect whether depression and insomnia influenced the brain structural regions associated with clinical characteristics in both the MDD and PI groups, a Partial correlation analysis was conducted, after controlling for the effect of age, gender, years of education, disease duration and eTIV. The significance was set at p < 0.05, and the false discovery rate (FDR) correction was used for multiple comparison correction.

# RESULTS

# Demographic Information and Clinical Characteristics

Significant differences in age, gender, years of education and eTIV were not observed between the four groups. As **Table 1** illustrates, the disease duration in the PI-LD subgroup is shorter when compared to other groups. In addition, the PSQI score in the PI-LD subgroup is lower than the one in the PI-HD subgroup (t = 2.25, p = 0.02), while the HAMD score in the MDD-HL subgroup is higher when compared to the one in the MDD-LI subgroup (t = 5.24, p < 0.001). In contrast, a significant difference in the anxiety symptom was not found between the two PI and MDD groups. In the MDD group, the HAMD score was positively correlated with both the HAMA score (r = 0.46, p < 0.001), and the HAMD-S score (r = 0.57, p < 0.001). In concordance, the PSQI score was positively correlated with the SDS score in the PI group (r = 0.27, p = 0.027). Significant relationships between disease duration, PSQI and SAS scores were not seen in PI groups. An absence of other significant relationships between the clinical characteristics was present.

# Main Effect of Depression on Brain Structure

As illustrated in **Figure 1** and **Table 2**, a main effect of depression on cortical thickness was seen in the bilateral superior parietal lobule (left SPL, 2.39 ± 0.16 vs. 2.45 ± 0.10; right SPL, 2.37 ± 0.12 vs. 2.43 ± 0.11), the right middle cingulate cortex (MCC, 2.38 ± 0.15 vs. 2.46 ± 0.12), and the right parahippocampal gyrus (2.42 ± 0.18 vs. 2.45 ± 0.16), after adjusting for the effects of gender, age, and years of education and TIV. The cortical regions affected by depression were thinner in patients with depression when compared to patients with an absence of severe depression symptom. In contrast, a significant effect of depression on cortical volume was not found.

#### Main Effect of Insomnia on Brain Structure

After adjusting for the effects of gender, age, and years of education, a main effect of insomnia on cortical thickness was observed in the right PCC (2.39 ± 0.16 vs. 2.43 ± 0.14). As showed in **Figure 2** and **Table 2**, the right PCC was thinner in patients with insomnia when compared to patients with an absence of severe insomnia symptom. In contrast, a significant effect of insomnia on cortical volume was not seen.

## Interaction Effect of Depression and Insomnia on Brain Structure

A significant interaction effect of depression and insomnia on the right orbital frontal cortex (OFC) volume was detected, after both a flexible multiple comparison correction (p < 0.05) and controlling for the effects of gender, age, and years of education. As **Figure 3** displays, patients with co-occurring depression and insomnia showed smaller brain volume in the right OFC when compared to patients without severe insomnia/depression, in both the MDD and the PI groups. A significant interaction effect of depression and insomnia on cortical thickness was not present after multiple comparison correction.

## Correlation Analysis

An absence of significant FDR-adjusted correlations between influenced brain regions and relative clinical traits in the MDD and PI groups was reported.

# DISCUSSION

To our knowledge, the present study was the first attempt to explore the interaction of insomnia and depression on brain structure in both PI and MDD patients. Three major findings were reported. First, insomnia and depression symptoms were closely associated in both the PI and MDD patients groups. Second, depression influences the brain structure in the SPL, MCC, and parahippocampal gyrus, while insomnia mainly influences the PCC thickness. Third, depression and insomnia interaction contributes to cortical volume loss in the right OFC. These findings indicated that the OFC might be a core region for the neuropathological alteration in comorbidity of insomnia and depression. Therefore, our findings provide new insights into the understanding of the brain mechanism underlying comorbidity of insomnia and depression.

The clinical features of the four subgroups verified the close association and the comorbidity between MDD and PI. In fact, patients with insomnia in the MDD group showed more severe depression, which is consistent with a previous

FIGURE 1 | Main effect of depression in cortical thickness across all patients. The color bar indicates the –log10(p) value after clusterwise correction for multiple comparisons using Monte Carlo simulations (vertex *p* < 0.0001, cluster *p* < 0.05). L, left; R, right; MCC, middle cingulate cortex; SPL, superior parietal lobule; PHG, parahippocampal gyrus.


TABLE 2 | The depression and insomnia and their interactive effects on cortical thickness and volume.

\**The threshold of multiple comparison correction was set as vertex p at 0.05, and cluster p at 0.05. TalX,Y,Z, the Talairach coordinate of peak vertex; VtxMax, number of peak vertex of the significant cluster; CWP, cluster-wise probability and the nominal p-value; NVtxs, number of vertices in cluster.*

epidemiologic study (35). In concordance, PI patients with depression have a longer disease duration and more severe insomnia, which is positively associated with depression but not with the anxiety symptom. These results were proved also consistent with previous epidemiologic results on the long-term consequences of insomnia (1). The clinical results presented here support the notion that the relationship between insomnia and depression is bidirectional, and this relationship can change over time following disease progression or release (7).

In the present study, we investigated the insomnia and depression effects on brain structure in a pooled patient group. This approach enables, in fact, the detection of the insomnia/depression effect in two patient groups and might reveal the common neurobiology underlying such symptoms. Only cortical thickness was significantly influenced by insomnia and depression in the pooled population. Patients with depression showed thinning of the SPL, MCC, and parahippocampal gyrus. The SPL is involved in spatial cognition

and sensorimotor integration (36, 37); previous studies also reported decreased gray matter volume and reduced functional connectivity in the SPL in MDD patients (38, 39). Although the alteration of the MCC is less commonly reported in depression studies, the posterior MCC is involved in attentional control during emotion regulation and showed a functional abnormality in MDD patients (40, 41). The present result provides a new insight into emotion regulation abnormality in patient with depression. Last, the parahippocampal gyrus is important for episodic memory and emotion cues processing (42); the thinning parahippocampal gyrus here found is consistent with previous structural imaging results in MDD patients (43), and indicated a common structural basis underlying both emotion and episodic memory processing in the patients with depression. Regarding patients with insomnia, the common structural alteration was seen in PCC, which is the core region of the default mode network and has been considered to be involved in emotion, memory and intrinsic control, and to be represent a neural substrate for human awareness (44, 45). The metabolism, functional connectivity abnormality and structural connectivity of the anterior default mode network was also reported in PI patients (46, 47). Generally, our findings on the main effect of insomnia and depression highlighted the morphological alteration in the thickness of the SPL, MCC, and parahippocampal gyrus in depression and of the PCC in insomnia.

Our most important finding was the localization of the insomnia/depression interaction effect on brain structure in the right OFC. The orbitofrontal area is involved in emotion, especially in reward processing, decision-making, and problemsolving abilities (48, 49). Previously, gray matter decrements in the OFC were reported in insomnia and depression patients, separately (50, 51). The decreased PFC volume may explain the insufficient decision-making and problem-solving abilities, and the attenuated recognition of environmental temperature in insomnia patients (24, 52). In addition, the reduced OFC volume may account for the distorted emotional stimuli evaluation, abnormal emotional and visceral regulation, anhedonia symptoms as well (23, 50). Interestingly, we found that co-occurrence of insomnia and depression is related to the worst volume decrease seen in the orbitofrontal cortex, especially, in MDD patients with insomnia. Baglioni and Riemann proposed a hypothesis stating that the link between persistent insomnia and depression is the alteration of the arousal system and its subsequent impact on affective and cognitive systems (53). Given that OFC receives widespread projections from arousal systems, including thalamus and amygdala (54), the decreased OFC volume may indicate an abnormal top-down control mechanism for arousal systems and dysfunctional emotional regulation and reward processing in patients with comorbidity of insomnia and depression. Overall, our results suggested that OFC is an essential area underlying the neuropathological mechanism of the comorbidity of insomnia and depression.

Two factors are likely to account for the absence in clinical significance of the alteration in cortical morphology of regions. First, considering that the behavioral assessments of depression and insomnia are not conformed between groups, we can only conduct the correlation analysis separately. Second, given that the clinical characteristics were robustly associated with demographic factors and relative stability of brain structure, some recent studies found that cortical structure feature is a relatively weak discriminant factor for current episode clinical status (55, 56). Further studies should include detailed and conformed neuropsychological tests, and a larger sample to verify this.

The limitations of the current study will now be highlighted. First, the present study is a preliminary and retrospective research, so the present study did not use the same estimating tool for depression and insomnia, we cannot explore the relationship between the influenced structural region and the clinical traits in a pooled group. In addition, the HAMD subscale of insomnia scores only has three items to estimate the insomnia, the HAMD total score is positively correlated with HAMD-S score, so it's difficult to completely remove the depression severity effect in the data analysis. These limitations regarding the clinical evaluation restricted the elaboration of our findings on brain structure. Second, not only insomnia, but also hypersomnia, is present in MDD patients (5); therefore, future studies should explore the brain mechanism underlying hypersomnia in MDD patients. Third, significant differences were found between the PI-HD and PI-LD groups in the PSQI total score, and between the MDD-HI and MDD-LI groups in the HAMD total score, which might bias the findings to uncertain extent. Fourth, the present study strictly focused on the structural change of cortical regions in depression and insomnia, whereas the subcortical and cerebellar regions were unexplored. Future studies should explore these issues further. Fifth, the present study did not include a healthy control group, which again limits the discussion of our findings. Lastly, considering that our study had a cross-sectional design with relatively small sample sizes, future research using larger sample sizes, data pooling, and longitudinal designs is deemed necessary.

In conclusion, our present results indicated that the OFC may be a core region for the neuropathological mechanism in comorbidity of insomnia and depression. In addition, our findings provide new insights into the understanding of the brain mechanism underlying comorbidity of insomnia and depression and the development of rational treatment strategies for these patients.

### AUTHOR CONTRIBUTIONS

SY, YH, and LG contributed to the conception and design of study. RL, BG, ZW, and FF contributed to the clinical

#### REFERENCES


estimate acquisition of imaging data. SY and LG performed imaging data analysis. SY and ZS wrote the manuscript. ZS and JY contributed revising the manuscript logically for important theoretical content. All authors contributed to the manuscript and have approved the final manuscript.

### FUNDING

This work was supported by the programs of the National Natural Science Foundation of China (81303057, 81503670, and 81373560); The Science and Technology Department of Sichuan Province (2018JY0249).

# ACKNOWLEDGMENTS

We thank all the participants involved in the study. We thank Jiaojian Wang for the help of statistical analysis of GLM.

sleeper controls: an MRI volumetric study at 3 Tesla. Sleep Med. (2010) 11:576–82. doi: 10.1016/j.sleep.2010.03.009


people entering treatment for substance addictions. Am J Occup Ther. (2015) 69:6903250010p1-9. doi: 10.5014/ajot.2015.014050


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Yu, Shen, Lai, Feng, Guo, Wang, Yang, Hu and Gong. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Mechanisms and Therapeutic Targets of Depression After Intracerebral Hemorrhage

Yinan Wu1†, Liangliang Wang2†, Kaimin Hu<sup>3</sup> , Chengcheng Yu<sup>4</sup> , Yuanhan Zhu<sup>5</sup> , Suzhan Zhang<sup>3</sup> \* and Anwen Shao<sup>6</sup> \*

*<sup>1</sup> Cancer Institute, Key Laboratory of Cancer Prevention and Intervention, China National Ministry of Education, The Second Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China, <sup>2</sup> Interdisciplinary Institute of Neuroscience and Technology, Qiushi Academy for Advanced Studies, Zhejiang University, Hangzhou, China, <sup>3</sup> Department of Surgical Oncology, The Second Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China, <sup>4</sup> Department of Orthopedics, The Second Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China, <sup>5</sup> Department of Neurosurgery, Rongjun Hospital, Jiaxing, China, <sup>6</sup> Department of Neurosurgery, The Second Affiliated Hospital, School of Medicine, Zhejiang University, Hangzhou, China*

#### Edited by:

*Yi Yang, Jilin University, China*

#### Reviewed by:

*Jianwei Lei, Second Hospital of Nanchang, China Zhiyuan Zhu, University of Hong Kong, China*

#### \*Correspondence:

*Suzhan Zhang zuci@zju.edu.cn Anwen Shao 21118116@zju.edu.cn; anwenshao@sina.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

Received: *15 October 2018* Accepted: *23 November 2018* Published: *17 December 2018*

#### Citation:

*Wu Y, Wang L, Hu K, Yu C, Zhu Y, Zhang S and Shao A (2018) Mechanisms and Therapeutic Targets of Depression After Intracerebral Hemorrhage. Front. Psychiatry 9:682. doi: 10.3389/fpsyt.2018.00682* The relationship between depression and intracerebral hemorrhage (ICH) is complicated. One of the most common neuropsychiatric comorbidities of hemorrhagic stroke is Post-ICH depression. Depression, as a neuropsychiatric symptom, also negatively impacts the outcome of ICH by enhancing morbidity, disability, and mortality. However, the ICH outcome can be improved by antidepressants such as the frequently-used selective serotonin reuptake inhibitors. This review therefore presents the mechanisms of post-ICH depression, we grouped the mechanisms according to inflammation, oxidative stress (OS), apoptosis and autophagy, and explained them through their several associated signaling pathways. Inflammation is mainly related to Toll-like receptors (TLRs), the NF-kB mediated signal pathway, the PPAR-γ-dependent pathway, as well as other signaling pathways. OS is associated to nuclear factor erythroid-2 related factor 2 (Nrf2), the PI3K/Akt pathway and the MAPK/P38 pathway. Moreover, autophagy is associated with the mTOR signaling cascade and the NF-kB mediated signal pathway, while apoptosis is correlated with the death receptor-mediated apoptosis pathway, mitochondrial apoptosis pathway, caspase-independent pathways and others. Furthermore, we found that neuroinflammation, oxidative stress, autophagy, and apoptosis experience interactions with one another. Additionally, it may provide several potential therapeutic targets for patients that might suffer from depression after ICH.

Keywords: intracerebral hemorrhage, depression, inflammation, oxidative stress, apoptosis, review, therapeutic target, autophagy

# INTRODUCTION

Each year, about 795,000 individuals suffer from a new or recurrent stroke. Nearly 610,000 among these patients experience first time attacks in their entire life; the remaining cases are reported as recurrent strokes. All stroke cases, 87% are ischemic, while intracerebral hemorrhage (ICH) strokes account for 10%, and subarachnoid hemorrhage (SAH) strokes only make up 3% of the total (1).

**30**

Post-stroke depression (PSD), the most common and frequent mental disorder after stroke, has a strong association with exacerbate deterioration of functional recovery, physical, and cognitive outcome, as well as quality of life. Moreover, PSD is even independently associated with enhanced morbidity, disability, and mortality (2–4).

Intracerebral hemorrhage (ICH) is a dangerous type of stroke, which is severer. Evidence have shown that 20% of ICH survivors existed with explicit signs of depression (5, 6). Numerous studies of PSD have been revealed, especially ischemic strokes; studies were based on medical examinations in researches. Nevertheless, the etiological factors that cause post-ICH depression are far from being elucidated. Hence, it is vital to understand the specific etiopathology of depression after ICH that can thus help people to develop effective therapeutic strategies aimed at etiological factors.

The present review will address the mechanisms, especially involved signaling pathways, and will introduce several potential therapeutic agents for the therapy of post-ICH depression. Finally, we will provide suggestions, that we hope can guide future research.

# ICH TYPES

ICH is divided into primary and secondary types, depending on the response to the fundamental cause of hemorrhage (7). Primary ICH develops without any underlying vascular malformation or coagulopathy. However, some cases like tumors, trauma, as well as coagulation could lead to the formation of secondary ICH, as well as the use of thrombolytic agents (7). In any ICH case, primary brain damage will occur because of the hemorrhage. And with the development of a hematoma, secondary brain injury will gradually appear on account of a pathological and physiological reaction.

Intracerebral hemorrhage is a lethal type of stroke, in the United States, every year there are 30,000 people who die from a stroke. If the patient is lucky enough to survive, then the growth of the hematoma in the brain parenchyma could trigger multiple of reactions that lead to another insult or even more severe neurological impairments. We will discuss several aspects of secondary cerebral injury following ICH and underline the key mechanisms correlated with post ICH depression (8).

### SECONDARY BRAIN INJURY-INDUCED DEPRESSION

Secondary injury is a key factor in the deterioration of the nervous system in ICH patients (9, 10). Secondary brain injury after ICH is caused by intraparenchymal hemorrhage, which then activates several signaling pathways such as inflammatory, oxidative, autophagic, and apoptotic pathways. These pathways, in vivo, become the bridge that links intracerebral hemorrhage and depression (8, 11–13).

#### Inflammation

Inflammation is a significant host defense response to cerebral damage following ICH. Once ICH occurs, components in the blood such as leukocytes, RBCs, and macrophages immediately migrate into the brain parenchyma. There is growing evidence that inflammation-induced impairment plays a crucial role in the mechanism underlying secondary brain injury after ICH (8, 14–17).

#### Toll-Like Receptors in Inflammation

Toll-like receptors (TLRs) are an important component of inflammatory responses and innate immunity (18, 19). TLR4 on leukocytes are important for the infiltration of both neutrophils and monocytes out of circulation (20). Recently, several clinical studies have suggested that increased levels of TLR2 and TLR4 expression in peripheral monocytes is related to a poor prognosis in patients with ICH (21). Furthermore, some studies have found an improved neurological function in TLR4-knockout ICH animal models (20, 22). Moreover, TLR4 signaling, especially those on resident microglia and on blood-derived inflammatory cells, is specific to inflammatory damage induced by ICH (20, 22, 23).

Recently, more attention has been placed on understanding the underlying mechanisms of inflammation-induced depression. Kéri et al. found patients diagnosed as major depressive disorder (MDD) for the first time, usually had an accompanied upregulation of TLR-4 signaling. It is thought to be related to bacterial translocation or various molecular patterns that correlate with the type of injury (24). Strekalova et al. first showed that C57BL/6J mice models appeared to show depression- and anxiety-like behaviors when they were fed high amounts of cholesterol. Moreover, they reported an unexpected elevation in the level of TLR4 expression, which indicated that TLR4 may play a critical role in the central neuronal system (25). Habib and his colleagues clarified, in an experiment using diabetic/depressed rats, when dysfunctions occurred to blood vessels as well as the metabolic system, the expression of TLR-4 in the aorta increased rapidly, in addition to a rise in pro-inflammatory cytokines (26). Later, Cheng et al. found that stress-induced neuroinflammatory responses are regulated by the GSK3-dependent TLR4 pathway. This signaling pathway is involved in development of depressive-like symptoms (27). García et al. then concluded that the activation of TLR-4 in the brain and peripheral area leads to sickness symptoms, and its expression level is also a risk factor that contributes to depression (28). These results confirmed the correlation between an elevated level of the TLRs and the risk of developing depression.

An increasing body of evidence suggests that microglia are the main mediators of inflammation-related disorders, including depression. Slusarczyk et al. suggested that tianeptine, an antidepressant drug, could attenuate the level of inflammatory mediators related to TLR4 signaling and the NLRP3 inflammasome (29). In addition, chronic restraint stress (CRS)-induced depressive-like animal models were found to show inflammatory responses in the hippocampus via the tolllike receptor type 4 (TLR4)/p38 mitogen-activated protein kinase (MAPK) pathway, which could be treated by ketamine (30). Past studies demonstrated that the TLR4 signaling pathway in the CNS and the periphery are associated with activated glycogen synthase kinase-3 (GSK3), a kinase shown to be involved in depression (31, 32). GSK3 inhibition has been indicated to reduce the production of pro-inflammatory cytokines with the stimulation of TLR4 in several different immune cells, both in clinical and basic experiments (33–35). Moreover, antidepressants like fluoxetine or the GSK3 inhibitor, TDZD-8, could improve stress-induced depressive-like behaviors via the TLR4 signaling pathway (27).

#### NF-kB Mediated Signal Pathway

Recently, many studies have proven the instrumental role of proinflammatory cytokines in the development of ICH. For instance, the activation of NF-kB in microglia/macrophages, which contributes to brain damage after ICH, results in the upregulation of proinflammatory cytokines (36, 37). Moreover, inhibited NF-kB activity is also related to alleviated neurological deficits (22, 38).

Furthermore, plenty of research suggests that neuroinflammation may play a significant role in the pathogenesis of depressive disorders. Koo et al. first reported that NF-κB signaling may act as a key mediator in anti-neurogenic and stress-induced behavioral actions; it may provide therapeutic targets of depression, which have never been described before (39). A few years after, evidence was provided that MDD is characterized by up-regulation of redox-sensitive transcriptional factors (Nrf2 and NF-κB), which indicated the pro-oxidative state that exists in MDD patients' peripheral blood mononuclear cells (PBMC) (40). A review concerned with adult hippocampal neurogenesis similarly supported the finding that NF-κB signaling modulates neurogenesis in adult patients, as well as expressing antidepressant actions (41). Recently, Nadeem et al. discovered that IL-17A seems to participate in comorbid depression with those who have psoriatic inflammation; this was linked to NF-κB and p38MAPK pathways that function through the up-regulated inflammatory cytokines in the brain (42). What is more, chronic stress in the basolateral amygdala (BLA) would induce the upregulation of neuropeptides and subsequently cause depressive-like behaviors. The siRNA could mediate the inhibition of NF-kB signaling in the BLA and downregulate the expression of neuropeptides, which lead to the alleviation of depressive symptoms (43). Moreover, Su et al. (44) proved that chronic unpredictable mild stress (CUMS)—induced depression-like action could be mediated through the NLRP3 inflammasome. Furthermore, the depression rat model indicated that the CUMS-induced MAPK pathway could be regulated by NLRP3 inflammasome by activating the NF-κB protein complex (44). Depression is one of the upmost psychological illness that is closely tied with inflammation. Crocin could act as a promising therapeutic target for depressive-like behaviors and neuro-inflammation caused by lipopolysaccharide (LPS). Researchers found such a phenomenon was an outcome of inhibited NLRP3 inflammasomes as well as inhibited NFκB signaling in microglia (45). Pro-anthocyanidin, having potential anti-inflammatory and antioxidative activity efficacy, functions as an effective therapeutic candidate for depressionlike behaviors induced by LPS by regulating the NF-κB signal in many cerebral regions and inhibiting the LPS-induced iNOS and the increased expression of COX-2 (46). Senegenin (SEN) is a main bioactive component of Polygala tenuifolia Willd, which has sturdy effects including anti-inflammatory actions as well as neuroprotection. At the same time, it has been used to lessen the depressive behavior in CUMS-induced rat models by inhibiting NLRP3-regulated NF-κB signaling (47). Icariin (ICA), which could be extracted from a certain traditional Chinese herb, is able to freely transverse the blood-brain barrier. It reduces neuroinflammation and OS-induced brain damage to prevent depressive-like behaviors by inhibiting the activation of NF-κB signaling in addition partially inhibition of the NLRP3 inflammasome/caspase-1/IL-1β axis, which would increase the antioxidant and anti-inflammatory ability of the cerebrum (48). With associated neuroprotection and anti-inflammatory activities, Geraniol (GE) has the potential to treat antidepressantlike behaviors in CUMS-induced depression mouse models, possibly by inhibiting the NF-κB pathway activation. Likewise, it seems that the regulation of nucleotide binding and NLRP3 inflammasome expression are both involved in this process (49). On the other hand, Chrysophanol (Chr) was also reported to function as anti-depression treatment by influencing the P2X7/NF-κB signaling pathway (50).

#### PPAR-γ-Dependent Pathway

CD36, belonging to the class B scavenger receptor family, is usually expressed in macrophages or microglia. It is involved in phagocytosis of many pathogens such as bacteria, apoptotic cells and oxidized low-density lipoproteins (51– 53). Peroxisome proliferator-activated receptor (PPAR) -γ, which is a part of the nuclear hormone receptor superfamily, can transcriptionally regulate the expression of CD36 and participate in inflammation (54, 55). In addition, treated with PPARγ activator, the hematoma in the brain of ICH mouse model would regress quicker and neurological damage following ICH in adults would decline. Flores and his colleagues confirmed that PPARγ agonists (15d-PGJ2) raised short-term PPARγ levels, accompanied with enhanced CD36 expression and accelerated hematoma resolution. Furthermore, it improved neurological function results. Moreover, both long term ventricular dilatation after ICH and white matter loss were decreased (56).

In several clinical and basic experiments, PPAR-γ agonists have exerted anti-depressive behavioral effects. Nevertheless, no one explained these mechanisms clearly. Gold and his colleagues proposed that PPAR-γ may exhibit as a conceptually new remedial target that improves the affective, cognitive and systemic manifestations of MDD (57). Later, Agudelo et al. opened a novel therapeutic avenue for treating depression through the PGC-1α1-PPAR axis, which was usually expressed in skeletal muscles, rather than by crossing the blood-brain barrier (58). Colle et al. found that PPAR-γ agonists have antidepressant effects in 3 out of 4 RCTs and in 4 open-label studies. Consequent studies concluded that PPAR-γ agonists may have antidepressant effects (59). Recently, several studies suggested that PPAR-γ agonist exhibit their antidepressant-like effects through various pathways: Liao and his colleagues firstly revealed the regulation of the CREB/BDNF and NF-κB/IL-6/STAT3 pathways, together with the potential effects on central 5-HT neurotransmission may be implicated in depressive-like behaviors via PPAR-γ-related effects (60). Through the upregulation of PPARγ expression, Song and his colleagues proposed that neuroinflammation could be inhibited and even play a role in its antidepressant effects (61).

Selective agonists of the nuclear transcription factor PPAR-γ are used to treat type 2 diabetes. Several studies also seem to suggest their contribution to improvement of depressive symptoms. PPAR-γ agonist pioglitazone (60, 62, 63), rosiglitazone (64–66), Troglitazone (67), atorvastatin (68), folic acid (69), Astragaloside IV (61), all of which have been reported to ameliorate depressive-like behaviors in mice via the PPAR-γ inflammasome axis.

#### Other Pathways

Inhibiting transient receptor potential Classic 3 (TRPC3), a member of the calcium-permeable cation channels, significantly reduced the perihematomal accumulation of reactive astrocytes, indicating that TRPC3 plays an important part in activating astrocytes following ICH. Accumulating findings indicate that neurological functions improve with reduced cerebral edema by inhibiting activated astrocytes via the TRPC3 inhibitor Pyr3.

In recent years, several studies have reported that the alterations of intracellular Ca2<sup>+</sup> signaling are the basis for the pathophysiology of psychiatric disorders, including depression (70, 71). Qin and his colleagues showed a complete difference between the depression animal model group and the control group related to the expression level of TRPC3/5 and the morphology in neurons, located in the hippocampus (72). Moreover, Buran et al. found that TRPC3/6 inhibitors might play a critical part in the etiopathogenesis of depressive disorders with enhanced levels of miR-9-5p and miR-128-1-5p (73).

## Oxidative Stress (OS)

#### Nuclear Factor Erythroid-2 Related Factor 2 (Nrf2) Pathway

Nrf2 comprises of a basic leucine zipper (bZIP) domain, which plays an important part in regulating the cellular antioxidant defense system. This includes heme oxygenase (HO) and superoxide dismutase (SOD) (74). The functions of Reactive oxygen species (ROS) are to trigger the Keap1/Nrf2/ARE pathways so as to compromise oxidative stress (OS) following ICH, which is known as an adaptive response (75–77). Keap1 is an OS sensor that negatively regulates Nrf2. Upon exposure to ROS, Nrf2 decouples from Keap1 and relocates to the nucleus before activating the antioxidant response element (ARE) dependent cytoprotective gene that mediates cell survival (78). The neuroprotective effect of Nrf2 suggests that a greater brain damage in Nrf2 knockout mice is correlated with increased ROS and apoptosis (76, 77). Therefore, Nrf2 activation of pharmaceutical preparations is a promising target to alleviate OS-induced brain damage following ICH.

Some researchers have indicated that Nrf2 is a major redox-sensitive transcription factor, which gets involved in the process of cellular self-protection from oxidative damage and increases vulnerability to depression-like actions. As part of a review, depression was characterized by distortion in six interwoven pathways; Maes et al. proposed that inhibitors of the Nrf2 activator target the above six pathways and may produce antidepressant effects (79). Djordjevic et al. revealed the maladaptive characteristics of chronic stress at the Nrf2/Keap1 level, resulted in the production of pro-inflammatory symptoms, suggesting that these changes may take part in the pathogenesis of depression/anxiety disorders (80).

For the past few years, several drugs have been found to have an antidepressant effect by influencing the Nrf2 signaling pathway. Furthermore, their target proteins are expressed in the brain. Mendez-David and colleagues showed that the Nrf2 signaling pathway is necessary for fluoxetineinduced neuroprotection associated with SERT blockade of 5- HT transporters, rather than for enhancing BDNF expression (81). Martín-Hernández et al. confirmed that the Nrf2 pathway is involved in the oxidation/nitrosation damage detected in the prefrontal cortex (PFC); moreover, the antidepressant drug has a therapeutic effect through this route (82). By stimulating PFC, CA3, and TrkB in dentate gyrus in Nrf2-knockout animal experiments, the TrkB agonist, 7,8-dihydroxyflavone, has shown a significant antidepressant functionality (83). Mice pretreated with Nrf2 activator sulforaphane (SFN) revealed reduced depression symptoms, which resulted from frequent social defeat stress. This suggests that the Keap1-Nrf2 interaction has a critical role in the pathophysiology of depression (83). Other Nrf2-activating drugs like TBE-31 and MCE-1 have also been proven as effective for treatment of depression associated to inflammation (84). Agmatine, an endogenous neuromodulator, also promises to serve as adjuvant/monotherapy for depression. This reinforces the importance of antidepressant Nrf2 activators (85). Recently, another drug, cilostazol, manifests promising prophylactic antidepressant-like effect by activating the Nrf2 pathway as well as by recovering mitochondrial malfunction, which interrupts OS (86).

#### PI3K/Akt Pathway

Plenty of brain stroke studies have revealed that ROS/RNS not only directly oxidize cellular macromolecules, such as proteins, lipids, and nucleic acids, which are associated with oxidative damage, but are also involved in the cell apoptosis signaling pathways. The PI3K/Akt, MAPK/P38, and NF-κB pathways are three major OS-mediated pathway activators. Apoptosis mediated by cytochrome c is another important pathway that is mitochondria-dependent (87). In addition, there is growing evidence that the phosphatidylinositol 3-kinase (PI3K)/AKT pathway is associated with the pathophysiology of depression and the antidepressant-like effect of different compounds (88–90).

In recent years, numerous findings have been derived from both basic and clinical researches that suggest erythropoietin has the ability to fight the depressive-like symptoms. Through JAK2, erythropoietin and its receptor signaling activates plenty of downstream signaling pathways such as NF-κB, PI3K/Akt, MAPK, and STAT5, they are able to have a significant role in neuro-progression and inflammation in the CNS (91). Recently, Wu et al. concluded that following the activation and release of neuroinflammatory factor induced by stress, the probable mechanism relates to the idea that the AKT/GSK3β/CRMP-2 pathway changes the normal structure and function of the central nervous cell scaffold microtubule system, and subsequently leads to depression (92). Moreover, Tao et al. proposed that liquiritigenin may reverse depression-like behavior in UCMS-induced animal models by modulating PI3K/Akt/mTOR mediated BDNF/TrkB signaling pathway (93). Several studies have shown that fluoxetine, creatine, atorvastatin, valproic acid as well as IGF-1 can all counteract depression-like behaviors via the PI3K/Akt Pathway (94–99).

#### MAPK/P38 Pathway

Earlier studies have shown that p38 MAPK, which can be stimulated by cytokine, can influence the neuroendocrine function, monoamine neurotransmission as well as other behaviorally-associated pathophysiological pathways (100). Felger et al. indicated that during chronic IFN-α treatment, depressive symptoms are highly related to the sensitivity of the p38 MAPK pathway to immune-stimuli (101). The MAO-A enzyme and p38 MAPK cascade are both involved in OS. These data and in vitro experiments demonstrate that the function of MAO-A is strongly inhibited by the p38 MAPK cascade. Thus, these published data indicate that the endogenous approach could be adopted to deal with OS and disorders like depression (102).

Recent research on neuroscience indicates that neurodegenerative pathways and OS pathways are both involved in depression. Bruchas et al. (103) found that the serotonin transporter can translocate to the plasma cell membrane, and that neurotransmitter-uptake is enhanced at the serotonergic nerve terminals when stress induces the activation of p38α MAPK. This finding strongly suggests that a cascade of molecular and cellular events is initiated by stress, and consequently the activation of p38α MAPK leads to a change in the hyposerotonergic state, which underlies drug-seeking and depression-like behaviors (103). MAPK and its phosphatase MKP are found to be implicated in depression and drug-addiction. Findings by Jia et al. supported the idea that there is a direct link between the phosphorylation of MAPK and depression induced by prolonged morphine withdrawal (104). In addition, Park et al. demonstrated that p38 MAPK inhibits the hypoxia response pathway (105). Moreover, Martín-Hernández et al. (106) showed that CMS increased the expression of activated MAPK p38 in addition to decreasing antioxidant transcriptional factor Nrf2. These results suggested that the translocated bacteria played a role through p38 MAPK, which aggravated oxidative injury and neuro-inflammation. This is possibly strongly linked to the pathophysiology of depression (106). These studies indicated an indirect relationship with depression, which requires further research.

Regarding drug therapy, the acute MAPK pathway was blocked, which resulted in depression-like symptoms and prevented the positive effects of ketamine. This fact suggests that the antidepressant response of ketamine is probably regulated by the MAPK pathway in some brain regions (107). Yang et al. reported that fluoxetine (FLX) is able to reduce NFκB and p38 MAPK phosphorylation levels and may improve the anti-inflammatory consequence (108). Moreover, Moretti et al. extended the data relating to the anti-depressive-like effect of ascorbic acid, which distinctly decreased hippocampal phosphorylation of p38MAPK (109).

#### Autophagy

Increased autophagy has now been reported in the central nervous system after several different kinds of diseases, such as ICH. Autophagy is an essential intracellular pathway, which includes degradation and recycling of aged proteins and entire organelles (110, 111). The mTOR pathway, NF-κB pathway and PI3K pathway are major pathways involved in regulating autophagy.

#### The mTOR Signaling Cascade

Mammalian target of rapamycin (mTOR) is a serine/threonine protein kinase that belongs to the phosphoinositide kinaserelated kinase family (PIKK family) and is a downstream effector of the PI3K/PKB (protein kinase B) signaling pathway. When its signaling pathway is activated, it has an important presence in regulating protein development, synthesis, proliferation and cell survival. Wang et al. conducted an experiment on mice to try to understand the negative effects of mTOR signaling (and its downstream products) on brain damage that results from ICH. It was found that if mTOR is blocked with rapamycin, PICs, scilicet, TNF-α, IL-6, IL-1β, and Caspase-3 all were upregulated indicating that apoptotic cell death could be reduced remarkably (112).

Several studies have found that ICH upregulated the expression level of miRNA-144 but decreased mTOR expression, which lead to increased inflammation and microglial autophagy. Their findings suggested that miRNA-144 was a critical regulator of autophagy by modulating mTOR (113). Later, another study published by Wang et al. indicated that miRNA-144 contributed to activated autophagy of microglia through the mTOR signaling pathway, which might be mediated by hemoglobin (114). More recently, Shi et al. suggested that IL-17A is a mediator who promoted the activation of inflammation and autophagy in microglial cells (115).

Structural and neurochemical changes in the limbic system are related to depression. The limbic structures include the hippocampus, which plays an important part in controlling emotion and mood. How the mTOR signaling pathway is relating to depression is discussed in many studies. Severe damages in mTOR signaling are revealed in postmortem studies, especially the mTOR signaling that exists in the prefrontal cortex of MDD patients (116). Feng et al. proposed that the depressive disorder is related to PLD-mTOR signaling (117). Lately, studies suggest microRNAs (miRNAs or miRs) such as miR-124-3p are implicated in certain signaling pathways, which might be associated to the pathophysiological mechanism of MDD. It was also suggested that DNA damage-inducible transcript 4 (DDIT4) is an inhibitor of the mammalian target of rapamycin (mTOR) signaling pathway and positively correlates with the expression of miR-124-3p (118).

Recent investigations found that mTOR signaling is related to several types of antidepressant drugs. Yu et al. indicated that the antidepressant effects of ketamine, in patients who have depression, could be reversed by the mTOR signaling inhibitor rapamycin (119). Cui et al. confirmed that by improving plasticity and neurogenesis, the mammalian target of rapamycin (mTOR) signaling pathway has an important role involved in mediating the antidepressant effect of ketamine (120). Nonetheless, drugs such as imipramine are not the same as ketamine, which could inhibit the PI3K/Akt/mTOR signaling to exert its antidepressant effect (116). Moreover, Liu et al. indicated that Resveratrol expresses antidepressant effects in CUMS-induced depressive-like animal models, which was partly mediated by its up-regulation of phosphor-Akt and mTOR expressions in the PFC and hippocampus playing a part in its antioxidant effects (121). Zhang et al. presented a new insight into the role of the dopaminergic system located in mesocortical region, which revealed antidepressant actions during the l-SPD mediated antidepressant process via the D1R/PKA/mTOR signaling cascade in the mPFC (122).

#### NF-κB Pathway

Numerous studies have demonstrated that in several disorders, autophagy is associated with inflammation. Moreover, as a critical controller in inflammation, NF-κB is either mediated by autophagy-related proteins or regulates autophagy directly. Shen et al. indicated that autophagy is activated after ICH, which may exacerbate ICH-induced cerebral damage in animal models. Furthermore, the regulation of the NF-κB pathway maybe a key reason that results in neuro-damage via its promotion of apoptosis and inflammation (123).

As we have described in the former part of this manuscript, the NF-κB pathway is an important pathway that links ICH and ICH-induced depression. Drugs such as Crocin, Icariin, Proanthocyanidin, Senegenin, and others all have antidepressant effects via the NF-κB pathway in ICH-induced depression patients.

# Apoptosis

Study findings suggest that both necrosis and apoptosis following ICH causes cell death. Some experiments revealed that apoptotic cell death existed in brain tissues from both animal models and ICH patients (124, 125). DNA fragmentation and apoptotic cell death are a consequence of activated caspases that are a part of a series of overwhelmingly complicated apoptotic mechanisms. It has been reported that cell death in ICH-induced animal models results from apoptosis mediated by caspases (126, 127). Intrinsic and extrinsic pathways are mainly responsible for apoptosis.

### Apoptotic Pathways

1. Death receptor-mediated apoptosis pathway—cell death signals are likely initiated by different stimuli, such as tumors, trauma or others. Subsequently, upstream signals bind to Fas-associated proteins that have associated death domains (FADD) and receptors. Then, caspase- 8 is activated via the p53, BCL-2, FAS, NF-κB, and others, which would ultimately lead to the activation of the executioner caspase. The effector caspases then activate endonucleases, resulting in DNA fragmentation, which subsequently orchestrates the dismantling of the whole cell structure.


Finally, together, procaspase-8,−9, cytochrome c and other signal proteins constitute the "apoptosome," which activates the initiator caspases such as caspase 8 and−9. After that, either the extrinsic or intrinsic apoptosis pathway delivers the cell death signal to the final executioner caspase (caspase-3,−6,−7) and subsequently initiates enzymes that degrade DNA, RNA and ribose. After the process of activating procaspase to caspase, programmed cell death is initiated (128).

Depression is a condition related to abnormal brain energy metabolism that is also marked with increased apoptosis in specific cerebral areas. Bay 60-7550 (Phosphodiesterase 2 inhibitors) has been shown to be a mediator in the apoptotic process, possibly via the SOD-cGMP/PKG-antiapoptosis signaling pathway in neuronal cells, and by inhibiting PDE2; it may be a significant novel antidepressant therapy (129). Moreover, water extracted from Panax ginseng (WEG) has been used as a treatment of several CNS disorders. Ding et al. suggested previously that WEG performed antidepressantlike effects in animal models of depression that was treated for both chronic and acute stress. Its neuroprotective effect relies on corticosterone-induced apoptosis via the downregulation of cytochrome C, ICAD, caspase-3, caspase-9 and so on (130). Moreover, both risperidone, at medium dose, and paroxetine were reported to improve modified stress re-stress (SRS)-induced depressive- like behaviors with associated down-regulated levels of cytochrome-C and caspase-9 in several regions of the brain (131). In addition, a novel antidepressant drug, Agomelatine (AG), might play an important part in the pathophysiology of depression with the amelioration of the apoptotic cells and the increase of neurogenesis in the hippocampus (132). Moreover, Apocynum venetum leaf extract (AVLE) was also reported to exert antidepressant-like activities in CUMS-induced rat models, which possibly suppressed neuronal apoptosis by regulating the Bcl-2/Bax signaling pathways, and improved the BDNF expressions in the hippocampus (133). Overload of Ca2<sup>+</sup> entry as well as excessive OS in neurons are the **two** main causes of depression. Recently, Demirda¸s et al. reported that with the treatment of Duloxetine (DULOX), TRPM2 and TRPV1 channels (associated with Ca2<sup>+</sup> entry-induced neuronal death), were regulated to reduce apoptosis in depression-like rats models (134).


TABLE 1 | The relationship among pathophysiology of ICH-induced depression, underlying signal pathways and its potential antidepressant drugs.

#### Other Pathways

Pannexins serves a significant role in the regulation of cellular signal transduction of glial cells and extracellular neuronal regenerative currents. Nevertheless, there have been no reported findings regarding the effects of pannexins in various cerebrovascular diseases. Zhou et al. first suggested that the upregulation of Pannexin-1 (Panx1) expression may be correlated with degeneration and apoptotic cell death of neurons in the rat cerebrum after ICH. Furthermore, he speculated that this may lead to subsequent cognitive dysfunction (135). Recently, Ni et al. used a broad-spectrum Panx1 inhibitor called Mefloquine (MFQ), demonstrating that the Panx1 channel played an important role in chronic stress and MFQ-induced depression and anxiety behaviors (136).

Recently, NIX was elucidated as a novel p75 neurotrophin receptor (p75NTR) binding protein as well as a member of the pro-apoptotic BH3-only group of proteins. When exposed to glutamate, the connection between NIX and p75NTR, there was marked increase in the apoptosis of neurons and activation of the JNK-p53-Bax pathway (137). Fujii et al. previously offered verification for the connection between the Ser205Leu polymorphism of the p75(NTR) gene as well as MDD, which indicated that the Leu205 allele provides a protective influence to fight the development of MDD (138).

In addition, Zhou et al. suggested that the Wnt/β-catenin signaling pathway is related to the level of proliferating cell nuclear antigen (PCNA) that is present in the cerebrum of the ICH rat, in addition to the rate of cell apoptosis, it could even regulate the balance of cell proliferation and apoptosis (139). Furthermore, Ni et al. (136) found that lncRNA TCONS\_00019174 exerts an antidepressant effect in rats by activating the Wnt/β-catenin pathway (139).

These new signaling pathways are proposed as potential clinical therapeutic targets for depressive disorders. This may require further research in order to explore further the relationships between ICH and depression.

# CONCLUSIONS

The pathophysiology of PSD is extremely complex; A multitude of ischemia-induced neurobiological mal-function as well as psychosocial distress are involved. The symptom for alterations of monoaminergic neurotransmitter systems has been well presented due to the injury of frontal-basal ganglia brainstem pathway. It has also been proved that there is a strong relationship between neuroinflammation and acute ischemic stroke: stress-induced activation of the hypothalamic-pituitaryadrenal (HPA) axis and the deficit of adaptive response (140).

In this review, we addressed the mechanisms and therapeutic targets of post-ICH depression. We divided the mechanisms into inflammation, oxidative stress, autophagy, and apoptosis, and clarified them through several signaling pathways. Inflammation is mainly related to TLRs, NF-κB mediated signal pathway, the PPAR-γ-dependent pathway and other signaling pathways. OS is related to Nrf2, the PI3K/Akt pathway and the MAPK/P38 pathway. Autophagy is associated with the mTOR signaling cascade and NF-kB mediated signal pathway. Meanwhile, apoptosis is related to the death receptor-mediated apoptosis pathway, mitochondrial apoptosis pathway, caspase-independent pathways as well as other pathways. Based on the evidence listed above, we found that neuroinflammation, OS, autophagy and apoptosis interacted with each other. OS-related brain injury is part of the pathogenic mechanism of neutrophil infiltration that follows ICH (16). Inducible NOS (iNOS) is synthesized through the induction of proinflammatory cytokines, and the molecular mechanisms for NOS activation after ICH are primarily NFkB dependent (141, 142). If NF-κB and antioxidative defense components can be inhibited, then OS and inflammation can be reduced via PPARγ; in the meantime, the cerebral damage caused by ICH would be improved. Proinflammatory cytokines, namely TNF-α and IL-1, could induce iNOS expression in microglial cells via the KC/p38MAFP/NF-kB pathway (143). Free radicals can also induce apoptosis, and antioxidant therapy could alleviate neuronal apoptosis after ICH (144, 145). The NF-kB pathway has also been detected to mediate Hb-induced apoptosis and autophagy (146). mTOR, as a downstream effector of the PI3K/PKB signaling pathway, also plays a significant part in CNS apoptosis and autophagy. Interactions of TLRs with pathogenassociated molecular patterns (PAMP) and damage-associated molecular patterns (DAMP) initiate signaling through myeloid differentiation primary response-88 (MyD88) and produce cytokines through the activation of the transcription factor nuclear factor kappa beta (NF-kB) (147). Furthermore, PPARγ could stimulate hematoma regression mediated by phagocytosis, and facilitate the cleanup of the hematomas, which may reduce the generation of inflammation and toxicity. Overall, from the assessed antidepressant drugs for ICH-induced depression, we found that several drugs exerted their antidepressant-like

# REFERENCES


effects via different signaling pathways and may have different pathophysiological origins (e.g., ketamine could treat depression through mTOR signaling cascade, the MAPK/P38 pathway or TLR-related signal pathways). This could provide us with evidence that some underlying correlations may exist between different signaling pathways. However, this still requires more research.

In summary, our review presents the signaling pathways relevant to post-ICH depression. Additionally, it may provide several potential therapeutic targets for the treatment of patients who show depressive behavior after ICH (**Table 1**).

Depression has a complex relevance with enhanced mortality and morbidity in ICH patients. In spite of its great clinical evidence, the underlying etiological mechanisms are still worthy to be explored. To better understand its pathophysiology and to pursue a more promising outcome of post-ICH depression, therapeutic interventions have become progressively more important for future studies.

# AUTHOR CONTRIBUTIONS

All authors participated in designing the concept of this manuscript. YW, LW, CY, and KH reviewed the literature and drafted the article. YZ, SZ, and AS finalized the paper and provided suggestions to improve it.

# FUNDING

This work was funded by China Postdoctoral Science Foundation (2017M612010) and National Natural Science Foundation of China (81701144).


antidepressant-like effect of folic acid in the forced swimming test in mice. J Psychopharmacol. (2012) 26:714–23. doi: 10.1177/0269881111424456


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wu, Wang, Hu, Yu, Zhu, Zhang and Shao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Predictors of Remission of Early-Onset Poststroke Depression and the Interaction Between Depression and Cognition During Follow-Up

Jing Huang1,2,3,4†, Fu-Chun Zhou5,6,7†, Boyuan Guan1,2,3,4, Ning Zhang1,2,3,4 , Anxin Wang1,2,3,4, Ping Yu1,2,3,4, Lei Zhou1,2,3,4, Chuan-Yue Wang5,6,7 \* and Chunxue Wang1,2,3,4 \*

#### Edited by:

Renerio Fraguas, University of São Paulo, Brazil

#### Reviewed by:

Guido Gainotti, Università Cattolica del Sacro Cuore, Italy Ranji Cui, Second Affiliated Hospital of Jilin University, China

#### \*Correspondence:

Chuan-Yue Wang wang.cy@163.net Chunxue Wang snowsen@126.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry

Received: 17 May 2018 Accepted: 13 December 2018 Published: 08 January 2019

#### Citation:

Huang J, Zhou F-C, Guan B, Zhang N, Wang A, Yu P, Zhou L, Wang C-Y and Wang C (2019) Predictors of Remission of Early-Onset Poststroke Depression and the Interaction Between Depression and Cognition During Follow-Up. Front. Psychiatry 9:738. doi: 10.3389/fpsyt.2018.00738 <sup>1</sup> Department of Neuropsychiatry and Behavioral Neurology and Clinical Psychology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>2</sup> China National Clinical Research Center of Neurological Diseases, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>3</sup> Center of Stroke, Beijing Institute for Brain Disorders, Beijing, China, <sup>4</sup> Beijing Key Laboratory of Translational Medicine for Cerebrovascular Disease, Beijing, China, <sup>5</sup> Beijing Key Laboratory of Mental Disorders, Beijing Anding Hospital, Capital Medical University, Beijing, China, <sup>6</sup> Beijing Institute for Brain Disorders Center of Schizophrenia, Beijing Anding Hospital, Capital Medical University, Beijing, China, <sup>7</sup> The National Clinical Research Center for Mental Disorders, Beijing Anding Hospital, Capital Medical University, Beijing, China

Objectives: This study aimed to examine the rate of remission in individuals experiencing early-onset poststroke depression (PSD) in China and to identify predictors of remission during a 3-month follow-up. This study also explored the interaction between cognitive impairment and depression.

Methods: A total of 820 patients with PSD from a massive multicenter prospective cohort project in China (PRIOD) were included in the present study. Depressive symptoms were measured with the Hamilton Depression Rating Scale (17 Items, HDRS-17) at 2 weeks and the endpoint of the 3-month follow-up. The cut-off score of HDRS-17 (<8) was used to define remission of depression at the endpoint. The Mini-Mental State Exam (MMSE) was used to evaluate the cognitive impairment of the patients (at the 2-week follow-up and 3-month endpoint). The National Institutes of Health Stroke Scale (NIHSS) was used to measure the severity of stroke.

Results: (1) Six hundred and forty-two patients completed the 3-month follow-up, and 332 (51.7%) patients remitted by the end of the study. Univariate analyses indicated that there was a higher proportion of patients who had hypertension, frontal lobe lesion, basal ganglia lesion, poor outcome at 2 weeks, high scores on the NIHSS at 2 weeks, major life events within 3 months, and major medical diseases within 3 months in the nonremission group. In stepwise multiple logistic regression analyses, remission was significantly predicted by lower NIHSS scores at 2 weeks (p = 0.001, OR = 1.086, 95% CI 1.035–1.139), fewer major life events (p = 0.036, OR = 5.195, 95% CI 1.111–27.283), fewer major medical comorbidities (p = 0.015, OR = 2.434, 95% CI 1.190–4.979), and fewer frontal lobe lesions (p = 0.042, OR = 1.717, 95% CI 1.019–2.891). (2) After controlling for confounding variables, repeated measures analysis of variance revealed

**42**

a significant interaction between time (2 weeks vs. 3 months) and group (remitters vs. nonremitters) on MMSE scores [F(1, 532) =20.2, p < 0.001].

Conclusions: Early-onset PSD patients with milder neurological impairment, fewer major life events, fewer major medical comorbidities and no frontal lobe lesion at baseline were more likely to achieve remission 3 months after stroke. Only remitters of PSD improved significantly in cognitive impairment after stroke.

The PRIOD trial is registered at http://www.isrctn.com/, number ISRCTN62169508.

Keywords: early-onset, poststroke depression, predictors of remission, cognitive impairment, follow-up

#### INTRODUCTION

Stroke is still the third leading cause of death, and the incidence rate of stroke has been increasing by 8.7% every year in China (1). However, a progressive decrease in stroke mortality and the subsequent increase of survivors with residual impairments and disabilities have been observed in recent years (2). Stroke can cause physical disability as well as essential emotional and cognitive complications, which can be seriously debilitating. Poststroke depression (PSD) is the most common psychiatric implication of stroke. The prevalence rate is estimated to be 41.8% in the first year following stroke (3), although the rate varies across studies due to methodological discrepancies. It is important to note that depression can result from physical disability (4), and vice versa. Emotional distress can also have a negative influence on the mortality, recovery, physical and cognitive functioning, and quality of life of stroke survivors (5– 7).

The mechanisms of PSD are complex and likely to involve multifactorial interactions (8). Previous studies have identified robust predictors of PSD, such as level of functional impairment (4) and stroke severity (9). Sociodemographic factors, such as young/old age, female sex, low education, living alone, a neurotic personality, and unemployment, were often found to be associated with PSD (10–12). Some studies have revealed that patients were more likely to develop PSD if their lesions were on the left side, on the frontal lobe or in the basal ganglia (13–15), but inconsistent findings were also reported (16–18).

To better understand PSD, a few studies have focused on the natural progression and explored trajectories of depressive symptoms following stroke, but the results were far from conclusive (19). PSD cases present in the initial poststroke period may differ from those who develop PSD later, in terms of mechanism and symptomatology. It has been reported that neuroanatomical factors, such as left hemisphere lesions involving the basal ganglia, are responsible for the initial poststroke period depression (20), whereas psychological factors could contribute to both the initial period and later PSD (21). The term "early-onset PSD" is often used to describe the depressive symptoms that appear in the acute phase (within 1–2 weeks after the stroke attack) of stroke (22). Limited attention has been paid to the predictive factors of the remission of early-onset depressive symptoms over time. Therefore, knowledge about the predictors of remission of early-onset PSD is warranted to facilitate clinicians to make an optimal treatment plan, given that administration of antidepressants and nonpharmacological interventions remains controversial (23, 24).

Cognitive impairment after stroke has often been reported with various prevalence rates, possibly due to methodological differences such as the tools and timing of the cognitive evaluation (25, 26). Notably, cognitive impairment after stroke has been associated with reduced functional recovery, increased risk of mortality, and the possibility of evolving to degenerative diseases (27–30). There seems to be a complicated interaction between depressive symptoms and cognitive functioning in poststroke patients over time (31). Cognitive impairments partly overlap with depressive symptoms, and the two syndromes may coexist in patients suffering from stroke (32, 33). In some cases, cognitive impairment caused by stroke may increase the risk of PSD (34, 35); whereas in other cases, cognitive impairment may also result from the depressive symptoms (36). The relationship between PSD and cognitive impairment after stroke has not been sufficiently elucidated in previous studies.

To date, there have been some studies investigating the natural course of PSD, but very few of them specifically focused on earlyonset PSD and its interaction with cognitive function over time. This is an important topic that is essential for improving both the long-term physical and psychological outcomes after stroke. The primary aim of this study was to describe the trajectory and outcome of early-onset PSD over a 3-month follow-up, with a focus on remission of depressive symptoms as well as its interaction with cognition. The hypothesis was that patients who remained depressed at the 3-month endpoint would be more likely to have had more severe functional impairments and more risk factors for developing PSD than patients who remitted from their depressive symptoms. The second hypothesis was that the remitters would have more significant cognitive improvements than nonremitters.

#### METHODS

#### Study Participants

The present study was part of a larger project: Incidence and Outcome of patients with poststroke Depression in China (PRIOD) (Project No. ISRCTN62169508, April 2008 to April 2010). PRIOD is a multicenter prospective cohort study with the participation of 56 hospitals (3). The project aimed to investigate the prevalence of PSD in China during a one-year follow-up period after first stroke onset and related risk factors for PSD.

The inclusion criteria of PRIOD were described in detail in a previous publication (3). In brief, patients who fulfilled the following criteria were enrolled in PRIOD: (1) a diagnosis of stroke according to the WHO diagnostic criteria, which was confirmed with CT or MR imaging; (2) onset of stroke within 14 days prior to recruitment; and (3) aged over 18 years old. The exclusion criteria of PRIOD were (1) patients with dementia or other neurological diseases that could affect cognitive functions; (2) patients with a history of or current major psychiatric disorders or alcohol or drug abuse; and (3) patients who did not appropriately communicate.

#### Data Collection and Scale Assessment

Eligible patients were consecutively enrolled in the present study. Patients' demographic information, medical history, personal history, family history, diagnostic information, and intervening measures were collected with a case report form at baseline. Major medical comorbidities were defined as cancer, severe cardiovascular disease (acute myocardial infarction, congestive heart failure, angina pectoris), severe kidney disease, and stroke relapse. Major life events were assessed using a self-designed form, and most of the items were adapted from "the list of threatening experiences," including the death of parents, spouses, or children, as well as serious family conflicts, family members suffering from serious illness, and divorce (37). All these events were commonly reported to cause moderate or marked longterm threat (37).

The stroke patients were screened for depression at 2 weeks after stroke, as was usually done in other studies in early-onset PSD (22, 38). The follow-up assessments were scheduled at 3 months after stroke, because this time point seemed to be a watershed for patients with PSD. Previous studies demonstrated that some biological features of early-onset PSD disappeared at or beyond 3 months after stroke (17, 39).

Experienced neurologists who implemented the scale assessment were blinded to the patients' clinical information. They all received standardized training for the assessments, and interrater reliability reached an acceptable level. Major or minor depression was determined by the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV). DSM-IV suggests that a person should be considered to have mild depressive symptoms if he/she experienced at least 2, but less than 5, of the depressive symptoms listed as the diagnostic criteria for at least 2 weeks, and at least one of the symptoms must be either depressed mood or loss of pleasure/interest (40). The Hamilton Rating Scale for Depression-17 (41) was applied to monitor the degree of depression at the 2-week and 3-month follow-up points. The validity and reliability of the Chinese version HRSD-17 had been proven in previous studies (42). In the present study, early-onset depression was defined as the presence of a depressive episode at 2 weeks after stroke. Remission of depression was defined by the cut-off score on the HRSD-17 (<8) at the endpoint.

The National Institutes of Health Stroke Scale (NIHSS) was used to measure the severity of Stroke (43). The modified RANKIN scale (mRS) was used to assess the neural functional recovery after stroke at the baseline and at the 3-month followup point. In the present study, mRS < 2 represented a favorable prognosis (benign outcome), while mRS ≥ 2 indicated an unfavorable prognosis (poor outcome), as in previous studies (44, 45). Cognitive impairment was evaluated with the Mini-Mental State Examination (MMSE) (46). The score for the MMSE scale ranged from 1-30 points. The higher the score, the better the cognitive function.

In the present study, hemorrhagic stroke and ischemic stroke were determined by MRI or CT scan results. The images of MRI (T1 and T2 weighted, fluid-attenuated inversion-recovery sequence, diffusion weighted imaging) or CT scans were retrieved from clinical routine exams. The lesions responsible for the stroke event were identified and reported by radiologists from each site who were blinded to the patients' psychiatric diagnoses. All the research radiologists among various sites have received standard training regarding image interpretation. When the stroke lesions were located in more than one brain region, every affected region was identified, recorded and used in the statistical analyses.

Telephone or face-to-face interviews were conducted at the follow-up point of 2 weeks and 3 months after the stroke attack. Information about the death, stroke relapse, medication regimen, life events, mRS scores, MMSE, and HRSD-17 scores was collected. Antidepressants were prescribed by the treating physicians according to the patients' clinical needs and clinical practice guidelines for depression. Participants were also allowed to take psychotherapy of various types, durations and number of sessions. The use of antidepressants was recorded at 3 months after stroke, which was discussed in detail in our previous article (47).

### Statistical Analysis

Data were analyzed with SPSS 23.0 (SPSS, Inc., IBM Company, USA). Comparisons between remitters and nonremitters with regard to sociodemographic characteristics, scores on the NIHSS and MMSE, functional outcomes as measured by the mRS, lesion locations, medical and personal history with respect to smoking, drinking, taking antidepressants, and receiving psychotherapy were performed using independent sample t-tests, Mann–Whitney U tests, Fisher's exact test, and chi-square tests, where appropriate. Multivariate logistic regression analyses with the backward Wald method were used to identify predictors of remission at the 3-month follow-up point. In the regression analyses, remission was entered as the dependent variable, and all variables that showed significant differences between the two groups in the aforementioned univariate analyses were entered as independent variables. Receiver operating characteristic (ROC) curves with the area under the curve values were calculated for remission, showing the predicted probabilities from the final model of logistic regression analysis.

Repeated measures analysis of variance (ANOVA) was performed for the Mini-Mental State Examination (MMSE) scores with group (remitters vs. nonremitters) as the betweengroup factor, time (2 weeks vs. 3 months) as the within-group factor, and variables that showed significant differences between the two groups in the aforementioned univariate analyses as covariates. The effects of time, group, and the interaction between time and group were examined. A two-tailed probability value of p < 0.05 was considered to indicate statistical significance.

#### RESULTS

Among 2828 patients who participated in PRIOD, 1992 patients were exclude because their HDRS scores were missing or they did not have depression (HDRS-17 total score = <7) 2 weeks after stroke. Among the remaining 836 patients with depression, 16 patients had a past history of mental disorders. Eight hundred and twenty PRIOD participants met the PSD criteria within 2 weeks after stroke and entered the present study for assessment. Therefore, the prevalence rate of PSD was 29.37% (820/2828). During the follow-up period from the beginning of the 2nd week to the end of the 3rd month, 10 patients died, 153 patients lost contact, and 15 patients lacked HRSD-17 scores. In the end, a total of 642 patients were included in the final analysis. They were divided into the nonremission group (HRSD-17 score ≥ 8, n = 310) and the remission group (HRSD-17 score < 8, n = 332) according to the HRSD-17 score at the 3-month follow-up (**Figure 1**).

#### Comparison Between Patients Included and Those Excluded From the Analyses

Among 820 patients who entered the study, 178 patients were excluded from the analyses. The excluded patients (n = 178) did not differ significantly from the included patients (n = 642) with regard to gender (male: 64.04% vs. 59.66%, p = 0.289), marital status (married: 89.89% vs. 93.45%, p = 0.107), education status (≥12 years of education: 42.24% vs. 35.78%, p = 0.183), stroke type (ischemic stroke: 77.97 vs. 77.88%, p = 0.795), first episode stroke (68.98 vs. 75.04%, p = 0.059), diabetes (28.32 vs.

22.84%, p = 0.135), hypertension (68.39 vs. 71.18%, p = 0.475), hyperlipidemia (23.57 vs. 19.44% p = 0.255), smoking (29.57 vs. 19.44%, p = 0.255), drinking (11.24 vs. 12.62%, p = 0.620), positive family history of stroke (24.56 vs. 18.50%, p = 0.078), stroke relapse (8.33 vs. 2.50%, p = 0.134), major life events (0.00 vs. 2.18%, p =1.000), and basal ganglia lesion (47.75 vs. 51.25%, p = 0.409). However, there were significant differences in age (64.51 ± 11.67 vs. 61.80 ± 11.54, p = 0.006), NIHSS score at 2 weeks (5.25 ± 3.95 vs. 4.08 ± 3.46, p < 0.001), MMSE at 2 weeks (23.06 ± 6.36 vs. 24.66 ± 5.26, p = 0.006), HDRS-17 at 2 weeks (13.37 ± 4.60 vs. 12.24 ± 3.90, p = 0.003), major medical comorbidities (20.83 vs. 6.54%, p = 0.022), poor outcome at 2 weeks (67.80 vs. 56.23%, p = 0.006), taking antidepressants (62.22 vs. 18.07%, p < 0.001), receiving psychotherapy (32.26 vs. 9.03%, p < 0.001), and frontal lobe lesion (20.22% vs. 11.06%, p < 0.001) compared to the patients included in the analysis. The two groups of patients were compared in order to determine whether the patients included in the following analyses were representative of the whole sample, and whether the conclusion could be generalized to other populations.

## Comparison of Demographic and Clinical Characteristics Between Remitters and Non-remitters at 3 Months After Stroke

At the end of the 3-month follow-up, there were 332 (51.7%) remitters and 310 (48.3%) nonremitters determined by the cutoff point of the HDRS-17 total score. The average total score on the HDRS-17 was 13.33 ± 4.63 in nonremitters and 3.89 ± 2.14 in the remitters. The baseline demographic and clinical characteristics of remitters and nonremitters at the 3-month follow-up are presented in **Table 1**. Univariate analyses indicated that there was a higher proportion of patients with hypertension (74.92 vs. 67.69%, p = 0.046), frontal lobe lesion (13.87 vs. 8.44%, p = 0.028), basal ganglia lesion (55.48 vs. 47.29%, p = 0.038), poor outcome at 2 weeks (61.29 vs. 51.51%, p = 0.013), high scores on NIHSS at 2 weeks (4.61 ± 3.67 vs. 3.59 ± 3.18, p < 0.001), major life events within 3 months (3.87 vs. 0.60%, p = 0.005), and major medical comorbidities at 3 months (9.68 vs. 3.61%, p = 0.002) in the nonremission group. On the other hand, remitters had a low HDRS-17 total score (11.42 ± 3.32 vs. 13.11 ± 4.27, p < 0.001) at 2 weeks.

#### Exploring the Independent Predictors of Remission at 3 Months After Stroke

Stepwise multivariate logistic regression analyses were used to identify predictors of remission of PSD at 3 months. In the regression analyses, remission was entered as the dependent variable, and all variables that showed significant differences between two groups in the aforementioned univariate analyses were entered as independent variables. The results are shown in **Table 2**. Remission was significantly predicted by lower NIHSS scores at 2 weeks (p = 0.001, OR = 1.086, 95% CI 1.035– 1.139), fewer major life events (p = 0.036, OR = 5.195, 95% CI 1.111–27.283), fewer major medical comorbidities (p = 0.015, OR = 2.434, 95% CI 1.190–4.979), and not having frontal lobe lesions (p = 0.042, OR = 1.717, 95% CI 1.019–2.891). **Figure 2** TABLE 1 | Comparison of demographic and clinical characteristics between remitters and nonremitters at 3 months.


SD, Standard Deviations; NIHSS, National Institutes of Health Stroke Scale; MMSE, Mini-Mental Exam; HRSD-17, Hamilton Depression Rating Scale (17 items); \*P < 0.05.

presents the ROC curve for the predicted probabilities from the final model of the multiple logistic regression analysis. The area under the ROC curve was estimated to be 0.637 (p < 0.001, 95% CI 0.593–0.680), indicating that the overall accuracy of the final model to predict patients' remission (with a predicted probability of 0.5 or greater) was acceptable.

### Comparison Between Remitters and Nonremitters Regarding Longitudinal Changes in MMSE

At 2 weeks after stroke, there was no significant difference in MMSE between the two groups. However, nonremitters performed significantly poorer on MMSE than remitters at 3 months after stroke (**Table 1**). After controlling for NIHSS, hypertension, major life events, major medical comorbidities, frontal lobe lesion and basal ganglia lesion, the results of a repeated measures ANOVA revealed significant time (2 weeks vs. 3 months) <sup>∗</sup> group (remitters vs. nonremitters) interaction on MMSE [F(1, 532) = 20.2, p < 0.001]. In the remitter group, MMSE scores changed toward better performance from 2 weeks (24.95 ± 4.89) to 3 months (26.63 ± 3.65). In the non-remitter group, MMSE scores did not change from 2 weeks (24.31 ± 5.67) to 3 months (24.72 ± 5.31) (**Figure 3**).

#### DISCUSSION

This study systematically examined predictors of clinical remission of early-onset PSD and explored the potential interaction between depressive symptoms and cognitive impairment after stroke. The two hypotheses were both


TABLE 2 | Predictors of nonremission at 3 months (multivariate stepwise logistic regression model) (n = 642).

confirmed by the results. Milder neurological impairment as indicated by NIHSS at 2 weeks, fewer major life events, fewer medical comorbidities and no frontal lobe lesion were significant predictors of the remission of PSD 3 months after stroke. These factors have been reported in previous studies as risk factors for the development of PSD (15, 48). For early-onset PSD patients, only remitters gained significant improvement in cognition over 3 months of follow-up.

Among the patients from PRIOD, 29.37% had depressive symptoms. This prevalence rate of PSD is close to most of the rates reported in previous studies, in which PSD was generally observed in approximately one-third of stroke cases, despite a large variety of criteria used to diagnose PSD (49–52). Among depressed patients who completed the 3-month followup, the remission rate was 51.7%. Some authors believe that the symptoms of PSD are self-limiting, and a longitudinal study reported that most patients diagnosed with acute PSD recovered (53). Previous studies demonstrated that approximately 50% of PSD patients would have symptom remission after 6 months, and the rate would increase to 89% at 12 months after stroke (54). Remission of PSD over the first a few months after stroke is vital, and symptom remission has been associated with higher recovery in the activity of daily living function for these patients (55).

The logistic regression analysis in the present study clearly confirmed that the course of PSD was a result of multifactorial interactions involving both biological and psychosocial determinants. Stroke severity and lesion locations served as the main biological factors in this study. Stroke severity, as indicated by NIHSS scores, has been one of the most consistent risk factors for PSD. A growing body of evidence has shown that the more severe the stroke is, the more likely a patient would develop PSD (26, 56). Higher NIHSS scores were associated with more severe depression after stroke. The results from Ilut et al.'s study suggested that the NIHSS score can predict the long-term prognosis of stroke, and a score over 11 could even bring a 9.4 fold higher probability of experiencing severe depression (15). This association may represent the biological mechanism of the pathophysiology of PSD. Severe stroke could lead to a series of biological changes in the brain and body, as well as psychological and functional alterations, all of which could contribute to the development of depression. Serious brain lesions may damage the function of some brain regions that are responsible for mood (57, 58). Much attention has been paid to the relationship between the onset of PSD and lesion locations, although this issue remains inconclusive (59, 60). Frontal lobe lesions were identified as a predictor of nonremission, which is in line with previous clinical and laboratory studies. Researchers have been trying to establish a connection between neuroimaging markers and the occurrence and development of depressive symptoms after stroke. Some neural circuits have been implicated in the development of both major depressive disorder and PSD (61– 65), often involving frontal areas (51, 66, 67). On the one hand, the frontal lobe plays a critical role in regulating emotion and cognitive functions (68). On the other hand, metabolic changes were discovered in the frontal lobe in PSD patients through MRI spectroscopy (69). Experimental rats with middle cerebral artery occlusion (MCAO) were found to be 14 times more likely to exhibit depressive-like behaviors than sham-operated control rats, and BDNF levels were downregulated in certain brain regions in the frontal and other cortical regions (70). The prefrontal cortex has also been implicated in the bilateral internal carotid artery occlusion (BICAO) model as one of the several vulnerable brain areas associated with depressive-like

behaviors after ischemia (71). PSD patients with frontal lobe lesions exhibited more persistent or recurrent symptoms than those without frontal lobe lesions in the first year after stroke onset (72).

Moreover, the stroke attack may also result in a decreased socioeconomic status, quality of life and general self-efficacy (GSE) (73), which contributed to a vicious circle between functional deficits and onset of PSD (74, 75). Other psychosocial factors include suffering from major medical comorbidity and exposure to major life events, which were also independent predictors of nonremission. Patients who underwent severe major medical diseases (including stroke relapse) after stroke attack may have an increased risk of depression onset or deterioration. Comorbid medical conditions may result in a patient's increased psychological burden, including a decline in the quality of life and rehabilitation faith, which may further cause depressed mood (76). Then, the depressed mood may in turn negatively affect the existing medical conditions, thus creating a vicious circle (77). Even in depressive patients who had no previous stroke, somatic symptoms have been reported to potentially worsen the outcome of depressive disorder. A study in patients with major depressive disorder indicated that remission rates in patients with more severe somatic symptoms were significantly lower than those in patients without somatic symptoms (78). Another 2-year follow-up study showed that "somatic symptoms" was an independent predictor of a worse prognosis of MDD (79). Therefore, major medical comorbidities and their companying somatic symptoms may have a tremendous negative impact on the probability of remission of depression. In the general population, exposure to major life events is a risk factor for the subsequent development of the depressive disorder (80). Likewise, studies demonstrated that patients with PSD had more major life events than nondepressed stroke patients before and after 6 months of stroke onset (48, 81, 82). The present study found that major life events not only served as a risk factor for developing depression but also prevented PSD patients from achieving remission. Therefore, more medical attention should be paid to PSD patients experiencing major life events.

Hypertension and poor outcomes at week 2 showed statistical significance in the univariate analysis, but they were eventually removed in the multivariate logistic regression. Concerning hypertension, the present result suggested that it did not contribute to the symptom resolution of depression, which is consistent with previous literature (83). The chi-square test indicated that mRS scores were significantly lower in the remission group than in the nonremission group. Many studies have shown that functional impairments might play a key role in the pathogenesis and development of PSD (16, 54). In a review in 2014, the researchers revealed that depression was negatively associated with functional outcome in stroke survivors and that the severity of stroke was the most significant contributor to PSD (84). However, a poor functional outcome may also be due to stroke severity and major medical comorbidities, and these are more robust independent predictors of remission in the present study.

The beneficial effects of antidepressant treatment in patients with PSD have been proved in a number of previous studies. These studies suggested noradrenaline reuptake inhibitors (NRIs), selective serotonin reuptake inhibitors (SSRIs), and tricyclic antidepressants (TCAs) all brought a considerable higher HAMD score reduction than the control treatments (85). There has also been a body of evidence indicating positive effects of cognitive behavioral psychotherapy in patients with PSD (86). However, neither antidepressants nor psychotherapy was associated with higher remission rate in the present study. PRIOD is a non-interventional study. Antidepressants and psychotherapies were used according patient's clinical needs with various doses, regimens, durations and number of sessions. These confounding factors could partly explain the inconsistent results with previous studies.

The present study also found a significant group (remitters vs. nonremitters) by time (from 2 weeks to 3 months) interaction with respect to MMSE scores, and only remitters had a significant improvement on the MMSE. Cognitive impairment and its relationship with depression in stroke survivors have long been of interest to the research community. Ischemic brain injury can cause both dementia (87) and depression (56). However, the causal relationship between cognitive impairment and depression after stroke remains debatable. In some studies, significant improvement had been achieved with regard to poststroke depressive symptoms during treatment with antidepressants, while cognitive function remained impaired (88–90). Therefore, some authors claimed that depressive symptoms might be secondary to cognitive impairment, which was caused by stroke and would follow its own course of recovery (88). However, cognitive functioning did not improve significantly in the aforementioned studies may be attributed partly to the inclusion of mixed cohorts of patients with various severities of depression. Patients with mild depressive disorder would not be expected to show cognitive improvement (32, 91). In Murata et al.'s study, only patients with major depressive disorder were enrolled, and patients with a significant reduction in depression severity also showed significant cognitive improvement over time (32). Although the present research also included both major and mild depression, as described in the method section, the outcome measure is symptom remission rather than reduction of depression severity. In addition, the sample size of the present study is much larger than that of previous studies. Therefore, the discrepancies with some previous studies could possibly be due to these methodological differences. The interaction between cognitive function and depressive symptoms in patients with PSD warrants further exploration.

The strengths of the present study include a large sample size, a wide range of sociodemographic and clinical variables, and the exclusion of patients with a history of mental disorders. However, the results should be explained with caution due to the following methodological limitations. First, 21.7% patients were excluded from the analyses because of incomplete followup information, and they were significantly different from the included patients in some factors. For this reason, the conclusion could not be simply applied to all studied populations. Second, patients with dementia and severe aphasia were excluded from the present study, and the subjects enrolled had relatively moderate deficits in neurological function and lower mortality. The exclusion of patients with severe cognitive and neurological deficits may potentially prevent generalization of findings to all stroke patients. Third, the MMSE has been found to overestimate impairments in persons over age 60 and in persons with less than 9 years of education (92); thus, it may not be sensitive to the cognitive changes in the present sample. A more sensitive and detailed neuropsychological battery is needed to monitor the longitudinal changes in cognitive function in patients with early-onset PSD. Fourth, psychological determinants, such as personality and coping styles, also play an important role in the course of PSD, but they were not collected in the present study.

# CONCLUSION

In conclusion, this study showed that approximately half of early-onset PSD patients remitted 3 months after stroke. Patients with less severe stroke, fewer major life events, fewer major medical comorbidities and frontal lobe lesions were more likely to have a favorable outcome regarding depression 3 months after stroke. Moreover, only remitters of PSD improved significantly in cognitive impairment after stroke. These results highlight the importance of early identification and intervention for patients with potentially persistent depression after stroke.

# DATA AVAILABILITY STATEMENT

The datasets analyzed for this study can be found in the website: http://www.tt.zhinanmed.com/.

# ETHICS STATEMENT

The PRIOD protocol was approved by the Medical Ethics Committee of Beijing TianTan Hospital, Capital Medical University. The project was carried out in accordance with the Declaration of Helsinki Guidelines, and all participants offered written consent form for the study.

# AUTHOR CONTRIBUTIONS

JH and F-CZ wrote the draft of the manuscript. JH, F-CZ, BG, PY, and LZ organized the database. AW and F-CZ performed the statistical analysis. C-YW and CW contributed the revision of the final version. CW contributed conception and design of the study. All authors contributed to manuscript revision, read and approved the submitted version.

### FUNDING

This study was supported by the National Key Research & Development Program of China [grant number 2016YFC1307200], the National Key Technology Research and Development Program of the Ministry of Science and Technology of China [grant number 2015BAI13B03], the Beijing Brain Research [grant number

### REFERENCES


Z161100000216131], the Beijing Municipal Science & Technology Commission [grant number Z151100004015127] and the National 11th 5-year Scientific and Technological Brainstorm Project [grant number 2006BA101A11], the Build High Level Technology Talents of Health System in Beijing (No.2015-3-038), and the Beijing Municipal Administration of Hospitals' Youth Programme (QML20161902).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Huang, Zhou, Guan, Zhang, Wang, Yu, Zhou, Wang and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Effect of Insomnia on Cortical Excitability in Patients With Generalized Anxiety Disorder

Zhaoyang Huang1,2, Shuqin Zhan1,2, Chao Chen<sup>3</sup> , Ning Li 1,2, Yan Ding1,2, Yue Hou1,2 , Li Wang1,2 and Yuping Wang1,2 \*

*<sup>1</sup> Department of Neurology, Xuanwu Hospital, Capital Medical University, Beijing, China, <sup>2</sup> Beijing Key Laboratory of Neuromodulation, Beijing, China, <sup>3</sup> Key Laboratory of Complex System Control Theory and Application, Tianjin University of Technology, Tianjin, China*

#### Edited by:

*Chunxue Wang, Beijing Tiantan Hospital, China*

#### Reviewed by:

*Yudan Lv, Jilin University, China Feng Liu, Tianjin Medical University General Hospital, China Ning Zhang, Beijing Tiantan Hospital, China Jianjun Meng, Carnegie Mellon University, United States*

> \*Correspondence: *Yuping Wang mdwangyp@sina.cn*

#### Specialty section:

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

Received: *15 September 2018* Accepted: *20 December 2018* Published: *10 January 2019*

#### Citation:

*Huang Z, Zhan S, Chen C, Li N, Ding Y, Hou Y, Wang L and Wang Y (2019) The Effect of Insomnia on Cortical Excitability in Patients With Generalized Anxiety Disorder. Front. Psychiatry 9:755. doi: 10.3389/fpsyt.2018.00755* The high rate of comorbidity between insomnia and anxiety disorders have been confirmed by previous studies. However, the underlying neurobiological correlates of the relationship between insomnia and anxiety disorders are largely unknown. The aim of the present study was to investigate the effect of insomnia on cortical excitability in patients with generalized anxiety disorder (GAD) by examining the recovery functions of median nerve somatosensory evoked potentials (SEPs) in patients with GAD without insomnia and patients with GAD comorbid with insomnia. We studied the recovery functions of median nerve SEPs in 12 medication-naive patients with GAD without insomnia, 15 medication-naive patients with GAD comorbid with insomnia, and 15 age and sex matched healthy controls. SEPs in response to single stimulus and paired stimuli at interstimulus intervals (ISIs) of 20, 60, 100, and 150 ms were recorded. The recovery function of the P25 component showed significantly reduced suppression in patients with GAD without insomnia as compared to patients with GAD comorbid with insomnia and healthy controls. There were no significant differences in the recovery functions of median nerve SEPs between patients with GAD comorbid with insomnia and healthy controls. The present study suggested that the cortical excitability of right parietal cortex increased in patients with GAD without insomnia, and cortical excitability in patients with GAD comorbid with insomnia was modulated by insomnia. Our findings provide new insights into the underlying neurobiological correlates of the effects of insomnia on GAD, which could ultimately be used to inform clinical intervention.

Keywords: insomnia, generalized anxiety disorder, somatosensory evoked potential, recovery function, right parietal cortex

# INTRODUCTION

Epidemiological studies have shown a high rate of comorbidity between insomnia and anxiety disorders. A number of longitudinal studies indicated that the relationship between insomnia and anxiety disorders is bidirectional. Insomnia contributes to the development of anxiety disorders, and anxiety disorders result in insomnia (1, 2). Treating chronic insomnia can often reduce the severity of anxiety symptoms, and similarly, treating anxiety can often improve insomnia (3). Generalized anxiety disorder (GAD) is the most common anxiety disorder, which is characterized by pervasive worry, difficulty concentrating, feeling restless, easily fatigued, muscle tension, and sleep disturbances (4). About 75% of patients with GAD have insomnia (5, 6). Sleep difficulties are included in the diagnostic criteria for GAD (4). Previous studies showed that GAD independently predicted insomnia, and higher levels of insomnia significantly predicted higher levels of GAD (7). Despite the well-documented association between insomnia and anxiety, the underlying neurobiological correlates remain unclear.

Patients with anxiety disorders characteristically show physical and psychological arousal (8). It is believed that anxiety is associated with alterations in brain excitability (9). Previous studies found that patients with anxiety disorders had significant high right parietal activity (10–13). These results suggested that the cortical excitability of right parietal lobe might be abnormal in patients with anxiety disorders. Previous studies also found that the functions of right parietal lobe in patients with GAD are abnormal. Etkin et al. (14) found an altered functional connectivity between bilateral posterior parietal cortex and the amygdala in patients with GAD using fMRI. Wu et al. (15) noted that patients with GAD had more activity in their right parietal lobes during vigilant tasks. Brambilla et al. (16) found that white-matter connectivity is impaired in the right parietal lobe in patients with GAD. These findings indicated that the cortical excitability of right parietal lobe might be abnormal in patients with GAD.

Several previous studies have revealed abnormal cortical excitability in anxiety disorders, including obsessive-compulsive disorder (17), social anxiety disorder (18), post-traumatic stress disorder (19), and generalized anxiety disorder (20). So far, there have been only one published study investigating cortical excitability in patients with GAD. Li CT et al. measured motor cortical excitability of patients with GAD using pairedpulse transcranial magnetic stimulation (20). They found that GAD patients had significantly lower intracortical facilitation (ICF). They concluded that GAD was associated with impaired intracortical facilitation, and such ICF deficits predicted the severity of anxiety. However, they did not consider the impact of insomnia on cortical excitability in patients with GAD.

Previous studies demonstrated that insomnia have a significant impact on GAD. Insomnia and sleep deprivation can significantly increase anxiety (21), and GAD patients with higher levels of insomnia have higher levels of anxiety symptoms (7). Thus, we assumed that insomnia could lead to changes in cortical excitability of patients with GAD, and cortical excitability might be different between patients with GAD without insomnia and patients with GAD comorbid with insomnia.

Paired-pulse stimulation techniques are used as common tools to investigate cortical excitability and cortical plastic changes. The recovery function of cortical somatosensory evoked potential (SEPs) component in the paired-pulse paradigm, which has been used in our previous study (22), has been applied to study the cortical excitability in patients with various psychiatric and neurological disorders (23, 24). When paired stimuli are delivered at different inter-stimulus intervals (ISIs), the amplitude of the SEP evoked by the second stimulus is suppressed depending on the interstimulus interval. The longer is the ISI, the higher is the amplitude of the SEP evoked by the second stimulus, until a complete amplitude recovery is observed (25–27). To improve the understanding of the neurophysiological effects of insomnia on GAD, the present study investigated alterations of cortical excitability of right parietal cortex in patients with GAD without insomnia and patients with GAD comorbid with insomnia by examining the recovery functions of median nerve SEPs. We hypothesized that the cortical excitability of right parietal cortex might be abnormal in patients with GAD without insomnia, and the cortical excitability of right parietal cortex in patients with GAD comorbid with insomnia might be modulated by insomnia.

# METHODS

#### Participants

We studied the recovery functions of median nerve SEPs in 12 medication-naive patients with GAD without insomnia, 15 medication-naive patients with GAD comorbid with insomnia, and 15 age and sex matched healthy controls. All participants were recruited from neurology outpatient clinics of Xuanwu Hospital. All procedures of this study were approved by the Institutional Review Board of Xuanwu Hospital and written informed consent was obtained from each participant.

All participants were interviewed and examined by two experienced neurologists. The interview included the administration of the Pittsburgh Sleep Quality Index (PSQI) (28), the Hamilton Anxiety Rating Scale (HAMA) (29) and the Hamilton Depression Rating Scale (HAMD, 24-item version) (30).

All participants aged from 18 to 60 years old. All patients met diagnostic criteria for GAD according to the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders-IV-TR (DSM-IV). Patients with GAD comorbid with insomnia also met diagnostic criteria for insomnia based on criteria for insomnia related to another mental disorder from the DSM-IV with duration of insomnia ≥ 3 months. Patients with GAD without insomnia were required to have the HAMA score of ≥ 14, the PSQI score of < 7, and the HAMD-24 score of < 20. Patients with GAD comorbid with insomnia were required to have the HAMA score of ≥ 14, the PSQI score of ≥ 7, and the HAMD-24 score of < 20. All patients were required to have no prior history of other psychiatric diseases, including all types of anxiety disorders other than GAD, depression, substance or alcohol abuse or dependence, and other sleep disorders.

Healthy controls were required to have no history of psychiatric diseases and sleep disorders, and have the PSQI score of < 7, the HAMA score of < 7, the HAMD-24 score of < 8.

Exclusion criteria for both groups were as follows: evidence of neurological or other physical diseases such as respiratory, cardiac, renal, hepatic, and endocrinal diseases as assessed by clinical history, physical examination or routine laboratory tests; any medication that might affect central nervous system within 14 days; irregular sleep patterns associated with shift work, frequent travel or personal preference (as indicated by a weekly variation > 3 h in bedtime or wake time, or time in bed duration < 5.5 or > 10 h per night); concurrent psychotherapy or counseling; pregnancy or breastfeeding women.

#### SEPs Recording Procedure

For SEPs recording, we used the same method and parameters as in our earlier study (22). Left median nerve was stimulated at the wrist at an intensity fixed at about 1.2 times the motor threshold (stimulus duration: 0.2 ms, stimulus rate: 1 Hz). SEPs were recorded (the Neuropack M1 MEB-9200 EP/EMG measuring system, Nihon Kohden Corporation, Japan) with the recording electrodes placed over the ipsilateral Erb point, the spinous process of the sixth cervical vertebra (Cv6), and the contralateral parietal area (C4', 2 cm posterior to the C4 placement of the international 10–20 system). All recording electrodes were referred to the right earlobe.

We recorded the N9 potential from the ipsilateral Erb's point, the N13 potential from Cv6, and the P14, N20, and P25 potentials from the parietal region contralateral to the stimulation side.

#### SEPs Recovery Functions

Recovery functions of SEPs were studied using the same method and parameters as in our earlier study (22). Paired stimuli of equal intensity were given at ISIs of 20, 60, 100, and 150 ms. Single trial SEPs were taken as control. The sequences of these trials were randomized among the subjects. At least three hundred sweeps were averaged for each condition. To ascertain reproducibility of results, SEPs of each condition (single stimulus, and paired stimuli at ISIs of 20, 60, 100, and 150 ms) were recorded at least twice, one trial after another trial. Then we obtained the average SEPs time series of each condition used for subtraction. To obtain SEPs evoked by the test stimulus (T-SEPs), we subtracted SEPs evoked by single stimulus alone (S-SEPs) from those elicited with paired stimuli.

We measured amplitudes of SEPs from the preceding peak (peak-to-peak) to prevent the impact of baseline shift on the results. For SEPs recovery functions, the amplitudes of each component in the subtracted SEPs waveform were measured. Then we calculated the relative amplitude ratios of T-SEPs to those of the corresponding S-SEPs at different ISIs. Finally, we obtained SEP recovery curves (SEP-Rs) by plotting the amplitude ratios of T-SEP/S-SEP against the interstimulus intervals. The value of ratio ≥ 1 means that there is no suppression.

#### Statistical Analysis

All statistical analysis was carried out with SPSS version 19.0 for Windows (SPSS Inc., Chicago, IL). The demographic and clinical characteristics at baseline were compared among the three groups using Pearson's chi-square test and One-way Analysis of Variance (ANOVA) followed by the LSD post-hoc test. For the amplitudes of SEPs obtained by single stimulus, we used one-way ANOVA test. For recovery functions obtained by paired stimuli, we employed a repeated measures ANOVA with ISI as the within-subjects factor and group as the betweensubjects factor. A p ≤ 0.05 was considered statistically significant results.

TABLE 1 | Demographic and clinical characteristics of the participants.


*PSQI, Pittsburgh Sleep Quality Index; HAMA, Hamilton Anxiety Rating Scale; HAMD, Hamilton Depression Rating Scale.*

*a Indicates significant differences between patients with GAD without insomnia and the controls (p* < *0.01).*

*b Indicates significant differences between patients with GAD comorbid with insomnia and the controls (p* < *0.01).*

*c Indicates significant differences between patients with GAD comorbid with insomnia and patients with GAD without insomnia (p* < *0.01).*

*d Indicates significant differences between patients with GAD comorbid with insomnia and patients with GAD without insomnia (p* < *0.05).*

### RESULTS

### Demographic and Clinical Characteristics

The demographic and clinical characteristics of the participants, including age, sex, and the scores of PSQI, HAMA, and HAMD, are summarized in **Table 1**.

There were no significant differences among the three groups with respect to age (F = 0.54, p = 0.59) and sex (χ <sup>2</sup> = 0.16, p = 0.92). The PSQI score was significantly higher in patients with GAD comorbid with insomnia than the other two groups (p < 0.01). There were no significant differences in PSQI scores between patients with GAD without insomnia and the controls (p > 0.05). The HAMA scores were significantly higher in patients with GAD comorbid with insomnia (p < 0.01) and patients with GAD without insomnia (p < 0.01) than the controls. The HAMA scores were significantly higher in patients with GAD comorbid with insomnia than patients with GAD without insomnia (p < 0.05). Similarly, the HAMD scores were significantly higher in patients with GAD comorbid with insomnia (p < 0.01) and patients with GAD without insomnia (p < 0.01) than the controls. The HAMD scores were significantly higher in patients with GAD comorbid with insomnia than patients with GAD without insomnia (p < 0.01).

### Single-Pulse SEPs

Mean values and standard deviations of the amplitudes of SEPs components in the single-pulse condition are shown in **Table 2**.

In the single stimulus condition, there were no significant differences in the amplitudes of SEPs components among the three groups (p > 0.05).



*Each value is expressed as mean (standard deviation).*

#### SEPs Recovery Functions

Mean values and standard deviations of the amplitude ratios of T-SEP/S-SEP at different ISIs are shown in **Table 3**.

The N9 component evoked by the test stimulus in healthy controls were suppressed (the amplitude ratios of T-SEP/S-SEP ratios < 1.0) at ISI of 20 ms, and recovered at ISI of 60 ms. In patients with GAD without insomnia and patients with GAD comorbid with insomnia, the N9 component behaved similarly to the normal controls. The repeated measures ANOVA showed that there were no significant differences among the three groups (F = 0.80, p = 0.46).

The N13 component evoked by the test stimulus in healthy controls were suppressed at all ISIs of 20, 60, 100, and 150 ms. In patients with GAD without insomnia and patients with GAD comorbid with insomnia, the N13 component behaved similarly to the normal controls. The repeated measures ANOVA showed that there were no significant differences among the three groups (F = 2.35, p = 0.11).

The N20 component evoked by the test stimulus in healthy controls were suppressed at all ISIs of 20, 60, 100, and 150 ms. In patients with GAD without insomnia, the N20 component evoked by the test stimulus recovered at ISI of 150 ms. In patients with GAD comorbid with insomnia, the N20 component behaved similarly to the controls. The repeated measures ANOVA showed that there were no significant differences among the three groups (F = 0.31, p = 0.74).

The P25 component evoked by the test stimulus in healthy controls were suppressed at all ISIs of 20, 60, 100, and 150 ms. In patients with GAD without insomnia, the P25 component evoked by the test stimulus were not suppressed at ISI of 20 ms, and suppressed at ISIs of 60, 100, and 150 ms. In patients with GAD comorbid with insomnia, the P25 component evoked by the test stimulus were suppressed at all ISIs. The repeated measures ANOVA showed that the recovery functions of the P25 component were significantly different among the three groups (F = 10.96, p < 0.01). Post-hoc tests showed significant differences between the patients with GAD without insomnia and the controls (p < 0.01), the patients with GAD without insomnia and the patients with GAD comorbid with insomnia (p < 0.01), but no significant differences between the patients with GAD comorbid with insomnia and the controls (p = 0.89).

**Figure 1** shows mean (± SD) recovery curves of the P25 component in the three groups.

#### DISCUSSION

In the present study, we investigated changes in cortical excitability in patients with GAD without insomnia and patients with GAD comorbid with insomnia by examining the recovery functions of median nerve SEPs. Our findings demonstrated that patients with GAD without insomnia exhibited reduced suppression of the cortical P25 component, but patients with GAD comorbid with insomnia showed no significant differences in the recovery functions of median nerve SEPs compared with the other two groups.

Previous studies suggested that the N20 component originates mainly in Brodmann's area 3b, and the P25 potential is generated by neurons in the Brodmann Areas 1and 2 of the parietal cortex (31, 32). It has been proposed that the N20 component reflects thalamocortical input to the primary somatosensory cortex, whereas the P25 component represents intracortical processing (33). The SEP recovery function of the cortical components is believed to reflect cortical excitability (24). Therefore, the normal recovery pattern of the N20 component and the disinhibited recovery pattern of the P25 component in patients with GAD without insomnia suggested an increased excitability of the parietal cortex.

The gamma-aminobutyric acid (GABA) system is believed to play a key role in the pathophysiology of GAD (34–36). Benzodiazepines, which act by enhancing inhibitory activity in the GABAergic receptor complex, are considered to be one of the most effective agents for GAD (37). Previous findings suggested that a GABA receptor-mediated mechanism in cerebral cortex might play a crucial role in the mechanism of pairedpulse inhibition (38–40). Therefore, these findings suggested that the dysfunction of inhibitory GABAergic interneurons in cerebral cortex might contribute to the disinhibited pattern of the cortical P25 component in patients with GAD without insomnia.

The level of excitability of cortical neurons depends on the balance between the GABA-related inhibitory and glutamaterelated excitatory systems (41). Glutamate neurotransmitter system has also been identified to be involved in anxiety disorders (42). Pregabalin has been shown to be effective in the treatment of GAD. It works in part by reducing the release of glutamate (43). Riluzole, a drug that reduces glutamate release and consequently increase the expression of glutamate receptors, may also be effective in the treatment of mood and anxiety disorders. Other compounds, which act on the glutamate system, have also been demonstrated to have the potential to treat GAD (44, 45). Thus, these findings suggested that the glutamate neurotransmitter system might also contribute to the disinhibited pattern of TABLE 3 | The amplitude ratios of T-SEP/S-SEP at different ISIs in the three groups.


*ISIs, Interstimulus intervals. Each value is expressed as mean (standard deviation).*

the cortical P25 component in patients with GAD without insomnia.

Interestingly, the present study showed that recovery pattern of the P25 components in patients with GAD comorbid with insomnia was not significantly different from the other two groups. We proposed that the most possible mechanism is the effects of insomnia on GAD. Previous studies suggested that insomnia and sleep deprivation can significantly increase anxiety (21), and higher levels of insomnia significantly predicted higher levels of GAD (7). In the present study, the HAMA scores were significantly higher in patients with GAD comorbid with insomnia than patients with GAD without insomnia. These results suggested that insomnia aggravates the severity of the disease. Thus, the cortical excitability in patients with GAD comorbid with insomnia might be modulated by insomnia. The present study found an increased excitability of the parietal cortex in patients with GAD without insomnia. Our previous study also found that the cortical excitability of the parietal cortex increased in patients with primary insomnia (22). Previous findings suggested that a decrease in GABA and a compensatory increase in glutamate might be involved in the mechanisms of the increased excitability of the parietal cortex in both patients with GAD without insomnia and patients with primary insomnia (46, 47). We proposed that because of the impact of insomnia, there might be not enough glutamate being released, that might result in decompensation of the GABA-related inhibitory and glutamaterelated excitatory systems in patients with GAD comorbid with insomnia.

Our study has several limitations. First, the present study used the recovery function of median nerve SEPs to investigate cortical excitability. This method can only reflect the regional cortical excitability of parietal lobe. Future studies could use task-related fMRI to explore differences in brain activation patterns between patients with GAD without insomnia and patients with GAD comorbid with insomnia. Then we can better understand the underlying neurobiological correlates of the relationship between insomnia and GAD. Second, we did not investigate the GABA and glutamate systems directly. Future studies could use magnetic resonance spectroscopy (MRS) to investigate differences in cortical GABAergic and glutamatergic neurotransmission between patients with GAD without insomnia and patients with GAD comorbid with insomnia. Third, the relatively small sample size is another limitation of the present study. Future studies with larger sample sizes could be conducted to confirm our conclusion.

In conclusion, the present study demonstrated that the cortical excitability of right parietal cortex increased in patients with GAD without insomnia. The cortical excitability in patients with GAD comorbid with insomnia was modulated by insomnia. The cortical GABA-related inhibitory and glutamate-related excitatory systems might play key roles in the mechanisms of the effects of insomnia on GAD. Our findings provide new insights into the underlying neurobiological correlates of the effects of

#### REFERENCES


insomnia on GAD, which could ultimately be used to inform clinical intervention.

# AUTHOR CONTRIBUTIONS

YW designed the research, supervised the project and revised the article. ZH and CC performed the research, drafted the article and analyzed data. YH, NL, YD, and LW collected data and interpreted the data. SZ interpreted data and revised the draft. All authors reviewed the paper and approved it to submit.

## ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China, Grant No. 81301138, 81571294, 61806146, the Beijing Municipal Administration of Hospitals, Grant No. QML20150802, Beijing Municipal Science & Technology Commission, Grant No. Z161100002616001, Natural Foundation of Capital Medical University, Grant No. PYZ2018069, and the National Key R&D Program of China, Grant No. 2015AA020514, 2018YFC1314500.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Huang, Zhan, Chen, Li, Ding, Hou, Wang and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effects of Cerebral Blood Flow and White Matter Integrity on Cognition in CADASIL Patients

#### Xinzhen Yin, Ying Zhou, Shenqiang Yan\* and Min Lou

*Department of Neurology, School of Medicine, The Second Affiliated Hospital of Zhejiang University, Hangzhou, China*

Background: It remains unclear whether the degree of white matter tract damage or cerebral hypoperfusion can better predict global cognitive impairment in CADASIL. We sought to determine the independent effects of cerebral perfusion status and white matter integrity on the cognition.

#### Edited by:

*Yi Yang, Jilin University, China*

#### Reviewed by:

*Hong Jiang, Central South University, China Yuhu Zhang, Guangdong Academy of Medical Sciences, China*

> \*Correspondence: *Shenqiang Yan shenqiangyan@zju.edu.cn*

#### Specialty section:

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

Received: *15 October 2018* Accepted: *14 December 2018* Published: *14 January 2019*

#### Citation:

*Yin X, Zhou Y, Yan S and Lou M (2019) Effects of Cerebral Blood Flow and White Matter Integrity on Cognition in CADASIL Patients. Front. Psychiatry 9:741. doi: 10.3389/fpsyt.2018.00741* Methods: We reviewed prospectively collected clinical and imaging data from genetically-confirmed CADASIL patients who underwent both arterial spin labeling (ASL) perfusion MRI and diffusion tensor imaging (DTI). We analyzed the cerebral blood flow (CBF), mean diffusion (MD), and fractional anisotropy (FA) by dividing the brain tissue into white matter hyperintensity (WMH) and normal-appearing white matter (NAWM). Global cognitive function was evaluated by using Mini-Mental State Examination (MMSE) and the Montreal Cognitive Assessment (MoCA).

Results: Of the included 29 CADASIL patients, the mean age was 48.4 ± 7.9 years, and 17 (58.6%) were women. MD was significantly correlated with CBF in both WMH (*r* = −0.407, *P* = 0.035) and NAWM (*r* = −0.437, *P* = 0.023) after adjusting for age and WMH volume. A MoCA score was obtained in 13 patients and was significantly correlated with CBF in both WMH (*r* = 0.742, *P* = 0.004) and NAWM (*r* = 0.659, *P* = 0.014). Both CBF in WMH (area under the curve, 0.767; 95% CI, 0.586-0.947, *P* = 0.015) and MD in WMH (area under the curve, 0.740; 95% CI, 0.557–0.924, *P* = 0.028) were good predictors for cognitive impairment (MMSE score < 27). However, multiple linear regression analysis revealed that global cognitive function was independently associated with CBF in WMH only (standardized β = 0.485, *P* = 0.015), after adjusting for age, gender, WMH volume, the presence of subcortical infarcts and DTI metrics.

Conclusions: Our findings suggested that cerebral hypoperfusion was more strongly associated with global cognitive dysfunction than the severity of brain microstructural damage, supporting that CBF assessed by ASL could serve as a candidate imaging indicator for monitoring alterations of global cognitive function in CADASIL.

Keywords: NOTCH3, CADASIL, diffusion tensor imaging, cognitive impairment, cerebral blood flow, arterial spin labeling

# INTRODUCTION

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is an early-onset monogenic variant of cerebral small vessel disease (CSVD) caused by mutations in the NOTCH3 gene (1), whose prevalence is at least 4.6 per 100,000 adults (2). Pathologically, there is deposition of granular osmiophilic material in the basal membrane of small arteries and capillaries in close association with progressive degeneration of smooth muscle cells (3, 4). Cognitive impairment is the second most frequent clinical manifestation of CADASIL, principally affecting processing speed, executive function, and attention from an early stage (5–7).

The mechanism of cognitive dysfunction in CADASIL remains uncertain. No strong correlation has been found between cognitive function and T2 lesion load on conventional MRI (8), since T2 high signal can be caused by severe neuronal loss or subtle damage to vascular tissues that leaves neural fibers intact. Therefore, diffusion tensor imaging (DTI), which is sensitive to the microstructural integrity of white matter, is used in the CADASIL population. Previous studies have shown that the degree of white matter tract damage may relate to global cognitive function (9, 10). On the other hand, cerebral hypoperfusion was also found to be correlated with cognitive impairment or dementia in CADASIL (11, 12). However, interaction may exist between cerebral hypoperfusion and disruption of brain microstructure. It's unclear whether white matter integrity or cerebral perfusion condition could better predict the global cognitive outcome.

To our knowledge, few previous studies have investigated the relationship between cerebral perfusion status and white matter integrity or their correlations with cognitive function in CADASIL. We thus performed both 3D arterial spin labeling (ASL) perfusion MRI, which provides measures of cerebral blood flow (CBF), and DTI in genetically-confirmed CADASIL patients, in order to determine the independent effects of CBF and DTI metrics on cognition.

# MATERIALS AND METHODS

#### Study Subjects

This was an investigator-initiated prospective single-center study. During the years 2007–2017, we performed NOTCH3 gene testing in patients with probable CADASIL. CADASIL suspicion arose when typical clinical features (migraine, stroke, cognitive deficits, or psychiatric symptoms), positive family history, or neuroimaging were suggestive of an inherited CSVD. Affected family members of index patients were not included in the current study. We then enrolled patients who (i) had a deleterious mutation of NOTCH3; (ii) underwent both ASL and DTI at the same time; and (iii) received a cognitive function assessment based on the Mini-Mental State Examination (MMSE) and/or the Montreal Cognitive Assessment (MoCA). We excluded patients whose image quality was poor due to motion artifacts. This study has been approved by our local human ethics committee. All clinical investigation has been conducted according to the principles expressed in the Declaration of Helsinki. Informed consent was obtained for all patients.

We retrieved demographic, clinical, and radiological data including age (disease onset and first visit) and gender; the vascular risk factors such as history of hypertension, diabetes mellitus, hyperlipidemia, and smoking; clinical features including ischemic events, migraine, family history, MMSE and MoCA score; and conventional neuroimaging findings such as intracranial arterial stenosis, the severity of white matter hyperintensities (WMHs), temporal poles hyperintensity, external capsule involvement, and subcortical infarcts (single and multiple) were recorded. Family history was collected by means of a structured interview that focused on the typical CADASIL disturbances referred to by all the proband's relatives. The family history was considered positive when at least one typical disturbance was present in at least one of the proband's first-degree relatives.

#### MRI Parameters

MRI was performed on a 3.0T system (MR750, GE Healthcare, United States) equipped with an 8-channel phased array head coil. MR sequences contained high-resolution 3D sagittal T1 weighted imaging (T1-WI), fluid attenuated inversion recovery (FLAIR), DTI and 3D ASL. A single shot, diffusion-weighted spin echo echo-planar imaging sequence was performed for DTI. Maximum b-value was 1,000 s/mm<sup>2</sup> in 30 non-collinear directions; one volume was acquired without diffusion weighting (b-value = 0 s/mm<sup>2</sup> ). Other parameters of DTI were as follows: TR = 8,000 ms; TE = 80.8 ms; flip angle = 90◦ ; FOV = 25.6 × 25.6 cm<sup>2</sup> ; matrix size = 128 × 128; slice thickness = 2.0 mm without interslice gap. 3D ASL was acquired using spin-echo pulse sequence with TR/TE = 4611/10.5 ms, TI = 1,525 ms, flip angle = 111◦ , slice thickness = 4 mm, matrix = 128 × 128, FOV = 24 × 24 cm<sup>2</sup> . High-resolution 3D sagittal T1WI was acquired using spoiled gradient echo sequence with TR/TE = 7.3/3.0 ms, TI = 450 ms, flip angle = 8 ◦ , slice thickness = 1 mm, matrix = 250 × 250, FOV = 25 × 25 cm<sup>2</sup> . Time-of-flight magnetic resonance angiography (TOF-MRA) consisted of 3 slabs with TR = 20 ms; TE = 3.2 ms; flip angle = 15◦ ; FOV = 24 × 24 cm<sup>2</sup> ; matrix size = 320 × 224; slice thickness = 1.4 mm. FLAIR parameters were TR = 9,000 ms; TE = 150 ms; TI = 2,250 ms; FOV = 24 × 24 cm<sup>2</sup> ; matrix size = 256 × 192; slice thickness = 5.0 mm. Axial FLAIR sequence was used to measure the lesion volume of WMHs with the following parameters: TR = 8,400 ms; TE = 152 ms; FOV = 24 × 24 cm<sup>2</sup> matrix size = 256 × 256; flip angle = 90◦ ; TI = 2,100 ms; slice thickness = 4.0 mm without interslice gap. The whole brain was imaged.

### NOTCH3 Gene Analysis

We used the diagnostic strategy established by Joutel et al. (13). We initially screened exons 3 and 4 for mutations, and if no mutations were present, we then analyzed the remaining exons, i.e., 2 and 5–23. Co-segregation was analyzed if any variant was found, and the presence of all identified, novel, disease-associated variants was examined in 100 controls by direct DNA sequencing.

FIGURE 1 | Flowchart demonstrating our imaging process for each subject. The 3DT1 image (A) was segmented to white mater mapping (B) automatically on Statistical Parametric Mapping (SPM). Then the T2-FLAIR image (C) was co-registered to 3DT1 image. White matter hyperintensity (WMH) lesion segmentation was based on co-registered T2FLAIR image (C) on Lesion Segmentation Tool in SPM. Normal-appearing white mater mapping (D) was obtained by image subtraction of white mater mapping and WMH lesion.

#### Imaging Analysis

DTI images were post-processed using FSL (www.fmrib.ox. ac.uk/fsl) to extract brain, remove bulk motion, and eddy current induced distortions. Then we calculated the parametric maps of mean diffusivity (MD) (a measure of the apparent diffusion coefficient averaged in all spatial directions), and fractional anisotropy (FA) (a measure of the directionality of diffusion) with DTIfit command in FSL. The raw data of ASL were transferred to a separate workstation (ADW, GE), where the quantitative CBF maps were generated by a custom-built program. The segmentation of normal-appearing white matter (NAWM) and WMH tissue masks was automatically processed in the native space using 3D T1WI and FLAIR images by the lesion segmentation tool (LST) toolbox in Statistical Parametric Mapping Version 8 (SPM8) (**Figure 1**) (14). The processed WMH and NAWM masks were further manually corrected by using ITK-SANP software (www.itksnap.org). The steps of manual correction included (i) removal of non-brain tissue, deep gray matter, brain stem, and cerebellum; and (ii) correction of false segmentation (positives or negatives). After co-registration, the masks of WMH and NAWM were used to obtain averaged MD, FA, and CBF of corresponding tissues in each subject.

#### Statistical Analysis

All numeric variables were expressed as mean ± SD. The difference between FA, MD, and CBF in WMH and their TABLE 1 | Main characteristics of CADASIL patients.


*CADASIL, cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy; SD, standard deviation; TIA, transient ischemic attack; MMSE, Mini-mental state examination; WMH, white matter hyperintensity.*

counterparts in NAWM were compared using paired t-tests. Pearson's correlation analysis was used for the continuous variables. In addition, we performed partial Pearson's correlation analysis to determine the correlation between CBF and the metrics of DTI (FA and MD) by adjusting for age and WMH volume. The association of CBF, MD, FA, age, gender, WMH volume, and the presence of subcortical infarcts was tested using the univariate linear regression model. The association of the variables, whose P < 0.1, with MMSE score was estimated using the multiple linear regression model. Receiver operating characteristic curve analysis was used to determine predictive value. All analyses were performed blinded to participant identifying information. Statistical significance was set at a P < 0.05. All statistical analyses were performed with SPSS package.

### RESULTS

A total of 29 genetically-confirmed CADASIL patients were included for the final analysis. The demographic, clinical and imaging characteristics were demonstrated in **Table 1**. The age at disease onset was 42.0 ± 9.4 years (ranged from 21 to 57 years), and the age at first neurological examination was 48.4 ± 7.9 years (ranged from 31 to 63 years), and 17 (58.6%) were women. The MMSE score was 23.8 ± 6.9 (ranged from 7 to 30). Seventeen (58.6%) of them suffered at least one transient ischemic attack or completed stroke, 10 (34.5%) had a migraine

with aura, and 25 (86.2%) had a positive family history. Only one patient had severe arterial stenosis (right middle cerebral artery), with a MMSE score of 27 and MoCA score of 15. No obvious intracranial arterial stenosis was shown in the other patients.

MD, FA, and CBF in WMH and NAWM were not associated with age (all P > 0.05). There existed a significant association between the WMH volume and FA in WMH (Pearson r = −0.543, P = 0.002), and in NAWM (Pearson r = −0.757, P < 0.001), while a tendency was detected with MD in WMH (Pearson r = 0.322, P = 0.088) and in NAWM (Pearson r = 0.363, P = 0.053). As illustrated in **Figure 2**, FA and CBF were significantly decreased in WMH compared to NAWM, while MD was significantly increased. After adjusting for age and WMH volume, MD was significantly correlated with CBF in both WMH (r = −0.407, P = 0.035) and NAWM (r = −0.437, P = 0.023), while there lacked of an association between FA and CBF in both WMH (r = 0.196, P = 0.328) and NAWM (r = 0.159, P = 0.427) (**Table 2**).

Univariate linear regression analysis demonstrated that MMSE score was significantly associated with CBF in both WMH and NAWM, and mean FA and MD in WMH, but not in NAWM (**Table 3**). Nevertheless, MMSE score was not significantly associated with WMH volume (standardized β = −0.189, P = 0.325). The cut-off point of CBF in WMH was 27 mL/100 g/min (area under the curve, 0.767; 95% CI, 0.586–0.947, P = 0.015), and this yielded a sensitivity of 73.3% and a specificity of 78.6% for prediction of cognitive impairment (MMSE score < 27), while CBF in NAWM was not a good predictor (area under the curve, 0.676; 95% CI, 0.480–0.872, P = 0.106). The cut-off point of MD in WMH was 1.250 10−<sup>9</sup> m<sup>2</sup> /s (area under the curve, 0.740; 95% CI, 0.557–0.924, P = 0.028), and this yielded a sensitivity of 71.4% and a specificity TABLE 2 | Partial correlation between cerebral blood flow and DTI-derived indices after adjusting for age and lesion load of WMH.


*DTI, diffusion-tensor imaging; WMH, white matter hyperintensity; CBF, cerebral blood flow; NAWM, normal-appearing white matter; FA, fractional anisotropy; MD, mean diffusivity. Bold indicates p* < *0.05.*

of 60.0% for prediction of cognitive impairment, while FA in WMH was not a good predictor (area under the curve, 0.629; 95% CI, 0.419–0.838, P = 0.239). MoCA score was obtained in 13 patients. After adjusting for age and WMH volume, MoCA score was significantly correlated with CBF in both WMH (r = 0.742, P = 0.004) and NAWM (r = 0.659, P = 0.014), while there was a trend toward significance between MoCA and MD in WMH (r = −0.519, P = 0.069). Multiple linear regression analysis revealed that global cognitive function was independently associated with CBF in WMH only (standardized β = 0.485, P = 0.015), after adjusting for age, gender, WMH volume, the presence of subcortical infarcts, and DTI metrics (**Table 3**).

#### DISCUSSION

In the current study, we investigated the effects of both cerebral perfusion status and white matter integrity on cognitive



*MMSE, Mini-mental state examination; FA, fractional anisotropy; WMH, white matter hyperintensity; NAWM, normal-appearing white matter; MD, mean diffusivity; CBF, cerebral blood flow. Age, gender, lesion load of WMH, and the presence of subcortical infarcts were adjusted in the multiple linear regression models. Bold indicates p* < *0.05.*

function in CADASIL patients. Our findings suggested that cerebral hypoperfusion was more strongly associated with global cognitive dysfunction than the severity of brain microstructural damage, which might differ from that in sporadic CSVD.

Although almost all demented CADASIL patients appeared to have confluent and diffuse WMH, some asymptomatic or mildly affected subjects had similar lesions (15). Therefore, the extent of WMH on T2-weighted images or FLAIR did not account for the phenotypic severity in CADASIL. Chabriat et al. first used DTI to detect the microstructural tissular alterations underlying T2 signal abnormalities and found that water diffusivity measured within WMH correlated with both MMSE and Rankin scale scores (9). Another DTI study also showed that the diffusion abnormalities of the thalamus correlated with cognitive function in CADASIL without dementia, especially for executive dysfunction (10). In addition, attentional network connectivity was proven to be associated with cognitive performance in CADASIL based on functional MRI (16), while the increased rapid-onset cortical plasticity might contribute to largely preserved cognitive function despite extensive ischemic changes (17).

Increased Notch3 activity mediates reduction in maximal dilator capacity of cerebral arteries in CADASIL and may contribute to reductions in CBF (18). Previous studies also reported the relationship between cerebral hypoperfusion and cognitive dysfunction. A SPECT study of a German CADASIL family showed that cognitive impairment was linked to hypoperfusion in the basal ganglia, and demented patients had a pattern of frontal, temporal, and basal ganglial hypoperfusion (19). A significant reduction in absolute and relative CBF was found within areas of WMH, and this reduction was more severe in demented than in non-demented CADASIL patients (12). Another perfusion metric of cerebral blood volume was also proved to be correlated with disability and cognitive impairment in CADASIL (11).

In the current study, we found that both CBF and MD in WMH were good predictors for cognitive impairment according to receiver operating characteristic curve analysis, while multiple linear regression analysis revealed that CBF in WMH was more strongly associated with global cognitive function. There is no doubt that both cerebral hypoperfusion and brain microstructural damage contribute to cognitive decline. However, brain microstructural changes appear secondary to cerebral perfusion changes, considering both the association between regional MD and CBF, and the results of multiple linear regression analysis. Cerebral hypoperfusion in NOTCH3 mutation carriers was supposed to precede the development of brain microstructural damage. CADASIL patients had both impaired cerebral and peripheral vasoreactivity at an early stage (20), and hemodynamic parameters were found to be abnormal in the superficial nerve fiber layer of the optic nerve head and retinal capillaries (21, 22). Chronic cerebral hypoperfusion reduced the activity of extracellular signal-regulated kinases, leading to neuronal adaptive responses, and impaired the function of microglial cells, which were implicated in amyloidβ elimination (23). Interestingly, both cognitive decline and cerebral hypoperfusion improved in a CADASIL patient during 2-year administration of lomerizine (24). Our findings suggested that CBF assessed by ASL could serve as a candidate imaging indicator for monitoring alterations of global cognitive function in CADASIL.

Our study had several limitations. First, although we prospectively collected data using a CADASIL registry and MRI protocol, our study design was cross-sectional. Longitudinal studies are needed to explore the causality between cerebral hypoperfusion, white matter integrity and cognitive outcome. Second, the number of CADASIL patients included in the current study was small, which reduced the power to detect significant effects and precluded comprehensive statistical analysis. Third, the MMSE is a crude measure of cognitive functioning that is insensitive to executive dysfunction and is not sensitive enough to detect mild cognitive impairment. Since only part of the enrolled subjects had a MoCA score, the multivariate analysis for MoCA was inapplicable due to the small sample size. More specific tests of executive function and neuropsychological assessment are required in future studies. Fourth, we did not focus on the CBF in subcortical gray matter nuclei, which might also contribute to the cognitive function. Moreover, the emerging technique of diffusion kurtosis imaging allows the measurement of mean kurtosis, which does not require tissue's directionality and hence it could provide more detailed information of microstructural integrity than DTI. The relationship between mean kurtosis and cognitive function should be further investigated in CADASIL.

In conclusion, our study demonstrated a significant association between cerebral hypoperfusion and the severity of brain microstructural damage, while the cerebral perfusion status was more strongly associated with global cognitive function than with white matter integrity.

### AUTHOR CONTRIBUTIONS

XY drafted and revised the manuscript, participated in study concept and design, conducted the statistical analyses, analyzed and interpreted the data. SY participated in study concept and design, data interpretation and made a major contribution in revising the manuscript. ML participated in the study design and made contribution in revising the manuscript. YZ assisted in designing the MRI sequences and imaging analysis, and assessed the cognitive function of the participants.

#### REFERENCES


#### FUNDING

This study was supported by a grant from the National Natural Science Foundation of China (81500993 & 81701150), the Zhejiang Provincial Natural Science Foundation (LY18H090003), and the Young Elite Scientists Sponsorship Program by CAST to SY (2017QNRC001).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Yin, Zhou, Yan and Lou. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# An Investigation on the Clinical Features and Neurochemical Changes in Parkinson's Disease With Depression

Teng-Hong Lian<sup>1</sup> , Peng Guo<sup>1</sup> , Li-Jun Zuo<sup>1</sup> , Yang Hu<sup>1</sup> , Shu-Yang Yu<sup>1</sup> , Li Liu<sup>2</sup> , Zhao Jin<sup>1</sup> , Qiu-Jin Yu<sup>1</sup> , Rui-Dan Wang<sup>1</sup> , Li-Xia Li <sup>2</sup> , Ying-Shan Piao<sup>3</sup> and Wei Zhang4,5,6,7,8 \*

*<sup>1</sup> Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>2</sup> Department of Internal Medicine, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>3</sup> Center for Movement Disorder, Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>4</sup> Center for Cognitive Neurology, Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>5</sup> China National Clinical Research Center for Neurological Diseases, Beijing, China, <sup>6</sup> Key Laboratory for Neurodegenerative Disorders of the Ministry of Education, Capital Medical University, Beijing, China, <sup>7</sup> Center of Parkinson's Disease, Beijing Institute for Brain Disorders, Beijing, China, <sup>8</sup> Beijing Key Laboratory on Parkinson's Disease, Beijing, China*

#### Edited by:

*Yi Yang, Jilin University, China*

#### Reviewed by:

*Fang Deng, Jilin University, China Junlei Chang, Chinese Academy of Sciences, China*

> \*Correspondence: *Wei Zhang ttyyzw@163.com*

#### Specialty section:

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

Received: *04 June 2018* Accepted: *07 December 2018* Published: *18 January 2019*

#### Citation:

*Lian T-H, Guo P, Zuo L-J, Hu Y, Yu S-Y, Liu L, Jin Z, Yu Q-J, Wang R-D, Li L-X, Piao Y-S and Zhang W (2019) An Investigation on the Clinical Features and Neurochemical Changes in Parkinson's Disease With Depression. Front. Psychiatry 9:723. doi: 10.3389/fpsyt.2018.00723* Objective: To investigate the clinical features and neurochemical changes in Parkinson's disease with depression (PD-D).

Methods: A total of 478 PD patients were divided into PD-D and PD patients without depression (PD-ND) groups according to the 24-item Hamilton Depression Rating Scale (HAMD) score. Demographic variables, motor and non-motor symptoms and activities of daily living were evaluated. The independent influencing factors of PD-D were investigated via binary logistic regression analysis. The levels of neurotransmitters in cerebrospinal fluid (CSF) were measured and their correlations with HAMD score were analyzed.

Results: The proportion of PD-D was 59.0%, of which 76.95, 20.92, and 2.13% had mild, moderate, and severe depression, respectively. Anxiety/somatization was the most prevalent sub-factor of HAMD in PD-D. The scores of UPDRS III, postural instability/gait difficulty (PIGD) type and the scores of 14-item Hamilton Anxiety Scale (HAMA) and 14 item Chalder Fatigue Scale (FS) were independently associated with PD-D. The levels of dopamine (DA) and 5-hydroxytryptamine (5-HT) were all significantly reduced in PD-D group compared with those in PD-ND group. HAMD scores were negatively correlated with the DA levels in CSF.

Conclusions : PD patients have a high proportion of depression, mainly of mild and moderate levels. The profile of depression in PD population is subtly different from that of the general population. Motor symptoms, PIGD type, anxiety and fatigue are the significant influencing factors of PD-D. Compared to 5-HT, DA may play a more important role in PD-D.

Keywords: depression, Parkinson's disease, risk factor, dopamine, serotonin

**66**

# INTRODUCTION

Parkinson's disease (PD) is a common neurodegenerative disorder with both the characteristic motor symptoms and a variety of non-motor symptoms. Depression, one of the most common non-motor symptoms of PD (1), used to be regarded as a psychological reaction to PD, and thus frequently be underestimated and undertreated. The failure to detect PD with depression (PD-D) and offer timely treatment would lead to worse outcomes for patients and caregivers (2).

Clinically, studies on PD-D have yielded conflicting results. In terms of demographic information, for example, some investigators showed that the frequency of PD-D increased with age (3), while others claimed that younger patients were more susceptible to PD-D (4). Some studies identified that being female and having a longer disease duration were the risk factors for PD-D (5), but some did not (6). From the aspect of symptoms, worse motor symptoms might be associated with PD-D (7). However, there were few studies that investigated the relationship between clinical type of PD and PD-D. For non-motor symptoms, most studies agreed that anxiety was the independent influencing factor (8), but other non-motor symptoms, for instance, fatigue, cognitive decline and sleep disturbances, had considerably varied reports (4, 9). Therefore, there is a lack of consensus on the influencing factors of PD-D. Studies consisting of large samples are necessary to explore both general and specific factors related to PD-D. In addition, there are few studies investigating the sub-factors of the 24-item Hamilton Depression Scale (HAMD) in PD-D.

The underlying mechanisms of PD-D have not been adequately clarified. Currently, PD-D is known to be of biological cause, and may be affected by multiple factors including genetic predisposition, biochemical disturbances, and psychological events (7, 10). The altered levels of neurotransmitters may serve as the biochemical basis of depression. 5-hydroxytryptamine (5-HT), dopamine (DA), and norepinephrine (NE), which were altered in the pathological process of PD, were known to play important roles in primary depression (11). Therefore, the pathophysiology of PD-D might also be related to the changes in the serotonergic, dopaminergic and noradrenergic systems (12). However, the changes in the above neurotransmitters for PD-D patients may not be the same as those in patients with idiopathic depression, which major biochemical compromise is the serotonergic system. To our best knowledge, there are limited studies on the neurotransmitters levels in cerebrospinal fluid (CSF) from PD-D patients.

In this study, PD-D was assessed using HAMD (13); motor symptoms and other non-motor symptoms were evaluated by a series of rating scales; the levels of 5- HT, DA, and NE in the CSF were measured by high performance liquid chromatography (HPLC), and their correlations with HAMD scores were analyzed. The objectives of this study were to investigate the clinical features, influencing factors and neurochemical mechanisms of PD-D.

## METHODS

#### Ethics Statement

This study met the guidelines of Helsinki Declaration on ethical principles for medical research involving human subjects, and the protocol was approved by the ethical review board of Beijing Tiantan Hospital. All participants signed written informed consents before they were recruited in the study.

#### Participants

Total 500 PD patients were consecutively recruited from the Department of Neurology and Geriatrics of the Beijing Tiantan Hospital from November 2014 to November 2017. Patients were diagnosed with idiopathic PD based on the criteria of the Parkinson's UK Brain Bank (14).

Exclusion criteria included severe systemic diseases, such as heart failure, pulmonary disease, gastrointestinal disorders, anemia, infectious disease and chronic inflammatory disease, deep brain stimulation, and conditions that might interfere with the reliable completion of clinical assessments.

Demographic variables including sex, age, age of onset, disease duration, side of onset, education level and anti-parkinson therapy, including the types of drugs, levodopa equivalent daily dose (LEDD) and the durantion of taking medicines, etc. LEDD was calculated as previously proposed (15).

### Clinical Assessment For PD Patients Depression

PD-D was diagnosed according to the diagnostic criteria for depression in PD (16). Patients with depression resulted from systemic diseases, organic mental disorders, psychoactive substances or non-addictive substances and patients on antidepressants were excluded from analyses.

Twenty-two patients were excluded due to ineligibility. A final total of 478 PD patients were included in the research. Patients with HAMD score of ≥8 points and fulfilled the above diagnostic criteria were assigned to the PD-D group. Within the PD-D group, patients with HAMD scores of 8–19 points, 20–34 points and ≥35 points were categorized as having mild, moderate and severe depression, respectively. PD patients with HAMD scores of <8 points were assigned to the group of PD patients without depression (PD-ND).

The 24 items of HAMD can be grouped into the following 7 factors (5): (1) anxiety/somatization (6 items: psychic anxiety, somatic anxiety, gastrointestinal symptoms, hypochondriasis, insight, and general symptoms); (2) weight loss (1 item); (3) cognitive disturbances (6 items: self-guilt, suicide, agitation, depersonalization and derealization, paranoid, and obsessivecompulsive symptom); (4) circadian fluctuations (1 item); (5) retardation symptoms (4 items: depression, work and interests, retardation, and sexual symptoms); (6) sleep disturbances (3 items: difficulty falling asleep, superficial sleep and early awakening); (7) hopelessness symptoms (3 items: helplessness, hopelessness and worthlessness).

#### Motor Function, Non-motor Symptoms, and Activities of Daily Living

The severity of PD was evaluated according to the Hoehn and Yahr (H-Y) stage. The motor symptoms were assessed using the Unified Parkinson's Disease Rating Scale (UPDRS) III. PD patients were divided into three phenotypes according to the clinical phenotype classification by Jankovic (17): tremordominant (TD) subtype, postural instability and gait difficulty (PIGD) subtype and mixed subtype. Items 16, 20, and 21 of the UPDRS were for tremor symptoms; items 13, 14, 15, 29, and 30 of UPDRS were for PIGD symptoms. The phenotype was defined by the ratio of the mean tremor score to PIGD score. TD subtype was defined as the ratio ≥1.5, PIGD subtype as the ratio ≤1.0 and mixed subtype as the ratio between 1.0 and 1.5.

A variety of non-motor symptoms were assessed using the following scales: the Montreal Cognitive Assessment Scale (MoCA) (18) for cognitive impairment, the 14-item Hamilton Anxiety Scale (HAMA) (19) for anxiety, the 14-item Chalder Fatigue Scale (FS) (20) for fatigue, the Rapid Eye Movement Sleep Behavior Disorder (RBD) Screening Questionnaire (RBDSQ) (21) for RBD, the Scale for Outcomes in PD for Autonomic Symptoms (SCOPA-AUT) (22) for autonomic dysfunction, and the Restless Legs Syndrome (RLS) Severity Rating Scale (RLSRS) (23) for RLS.

Activities of daily living (ADL) were evaluated by ADL Scale.

# Measurements of Neurotransmitters in CSF From PD Patients

Anti-parkinsonian drugs were withdrawn for 12–14 h if the patients' condition allowed and longer time was considered unethical by our ethical committee. For medical washout period should be three times of the t1/2 of the medicine, the subjects taking drugs with long t1/2, including controlled release Sinemet (Sinement CR), Pramipexole and controlled release Piribedil (Piribedil CR) were excluded in the analysis of neurotransmitters. Under fasting condition, 5 ml CSF was taken in a polypropylene tube through lumbar puncture, between 7 and 9 a.m. CSF samples were centrifuged immediately at 3,000 rpm at 4◦C. Then, approximately 0.5 ml volume of CSF was aliquoted into separate Nunc cryotubes and kept frozen at −80◦C until usage in the assays.

The levels of neurotransmitters, including DA, 5-HT, and NE in CSF from PD patients were measured by HPLC (24). LC-MS-MS 6410 chromatograph and Phenomenex 150<sup>∗</sup> 2 mm and 150<sup>∗</sup> 3 mm chromatographic columns were from the Agilent Company (California, USA), and the standard sample was from Sigma Company (California, USA). In total PD patients, 89 patients agreed lumbar puncture and after excluding patients used drugs with long T1/2, 76 patients' neurotransmitters in CSF were analyzed.

#### Statistical Analyses

Statistical analyses were performed using SPSS Statistics 20.0 (IBM Corporation, 220 New York, USA). P-value of less than an alpha level of 0.05 was defined as statistically significant.

Demographic information, motor and non-motor symptoms and ADL scores between PD-D and PD-ND group were compared. Continuous variables, if normally distributed, were presented as means ± SDs and the 2 groups were compared using 2-tailed t-test; if not normally distributed, the data were presented as median (quartile) and compared using nonparametric test. Discrete variables were compared using Chisquare test.

Binary logistic regression analysis was used to investigate the independent influencing factors of PD-D. The covariates with statistical differences in single-factor analysis were divided into a multivariate model; Backward elimination was applied to remove non-significant variables. Influencing factors were presented as odd ratios (OR) with 95% confidence intervals (CI).

Three kinds of neurotransmitters levels of CSF between PD-D and PD-ND groups were compared. Bonferroni correction was made and P was reduced to 0.017 (0.05/n = 0.05/3 ≈ 0.017). Pearson correlation analyses were conducted between HAMD scores and neurotransmitters levels in CSF in the PD group. Binary logistic regression analysis was used. The covariates contained the neurotransmitters which had statistical differences in single-factor analysis and the independent influencing factors of PD-D in the above mentioned binary logistic regression analysis; Backward elimination was applied to remove nonsignificant variables.

# RESULTS

# Frequency of Depression of PD

Among 478 PD patients, 282 (59.00%) were diagnosed with PD-D: of which, 76.95% (217/282 cases) had mild depression, 20.92% (59/282 cases) had moderate depression, and 2.13% (6/282 cases) had severe depression. The remaining 196 cases (41.00%) were in the PD-ND group.

# Assessment of Each Sub-factor of HAMD for PD-D and PD-ND Groups

The assessment of each sub-factor of HAMD was shown in **Table 1**. Anxiety/somatization was the most common subfactor reported in PD-D group at 99.30%, followed by retardation symptoms (97.51%) and hopelessness symptoms (89.71%). Anxiety/somatization also occurred in the largest frequency of PD-ND patients, but compared with those of PD-D, the frequency of each sub-factor of HAMD was much lower.

# Demographic Variables, Motor Function, Non-motor Symptoms, and ADL of PD-D and PD-ND Groups

Demographic variables, motor function, non-motor symptoms and ADL scores were compared between PD-D and PD-ND groups in **Table 2**.

Compared with the PD-ND group, the PD-D group showed a significantly earlier age of onset, remarkably longer disease duration, and significantly lower education level. There was no significant difference in the other demographic variables,


including sex, age, side of onset and the condition usage of the anti-Parkinson drugs between the two groups.

The PD-D group had a significantly advanced H-Y stage, significantly increased UPDRS III score and higher proportion of the PIGD subtype.

The PD-D group scored significantly lower MoCA score and higher scores of HAMA, FS, RBDSQ, SCOPA-AUT, and RLSRS scales when compared with the PD-ND group, demonstrating that the PD-D group had more severe cognitive impairment, anxiety, fatigue, RBD, autonomic dysfunction, and RLS.

Compared with the PD-ND group, the score of ADL scale in the PD-D group was significantly decreased, suggesting that the PD-D group had significantly poor ADL.

#### Factors Associated With PD-D

Binary logistic regression analysis was performed using the above variables that found statistical differences between PD-D and PD-ND groups. The results showed that the scores of UPDRS III, PIGD type, and the scores of HAMA and FS were independently associated with PD-D (**Table 3**).

### Levels of Neurotransmitters in CSF From PD-D and PD-ND Groups

The levels of DA, 5-HT, and NE in CSF from the PD-ND and PD-D groups were compared (**Table 4**). In the PD-D group, the levels of DA and 5-HT were all reduced compared with those in the PD-ND group. However, only DA level showed a significant difference after Bonferroni correction.

The correlational analyses between HAMD scores and neurotransmitters levels in CSF from PD patients were further conducted (**Table 5**). It was found that the HAMD scores had a negative correlation with DA level in CSF (r = −0.278, P < 0.05). However, no significant relationship between HAMD scores and the levels of 5-HT and NE in CSF was detected.

Binary logistic regression analysis was performed to investigate the relationship of PD-D and DA levels in CSF. The covariates contained the levels of DA in CSF and the scores of UPDRS III, PIGD type and the scores of HAMA and FS (**Table 6**). The results showed that the DA levels significantly negatively correlated with PD-D.

# DISCUSSION

Though PD was previously characterized by the classical motor symptoms, recent studies have suggested that non-motor symptoms play significant roles in the deterioration of the quality of life for PD patients (25, 26). As one of the mood disorders, depression is often covered by motor symptoms and fails to present as the chief complaint. It is important to diagnose PD-D, because it is one of the main determinants of quality of life for PD patients and lack of PD-D diagnosis results in heavy burden to the families of patients (27, 28).

The prevalence of PD-D differs greatly across studies, ranging from 2.7 to 90%, and around 35% of PD patients presented significant depressive symptoms clinically (29), which was much higher than that in general population (30). The variation might be resulted from different diagnostic criteria, rating scales, sample sizes, and study population (31). In the current study, a large sample containing 478 Chinese PD patients was established, of which, 59% of the population suffered from depression. Analysis of the severity of depression in PD patients indicated that mild, moderate, and severe depression accounted for 76.95, 20.92, and 2.13%, respectively, implying that PD-D was featured by the mild and moderate depression (9). Clinical trials have shown that mild depressive symptoms are a variable process and may result in remission or take a turn for more severe and persistent symptoms over time (32).

Previous studies rarely investigated the 7 sub-factors of HAMD-24 items in detail. In this investigation, the results showed that PD-D patients had the highest frequency of anxiety/somatization, which was followed by the symptoms of retardation and hopelessness. The cognitive disturbances, which contains suicide, had the fifth frequency. This suggested that the profile of depression in PD population was subtly different from that of the general population which showed a high rate of suicide (33).

In this study, the binary logistic regression analysis showed that the UPDRS III score, PIGD type, the scores of HAMA and FS were the significant influencing factors of PD-D. Motor symptoms were significantly evidently different between PD-D and PD-ND groups. Some earlier studies failed to find a significant relationship between motor symptoms and PD-D (9, 34); the inconsistency may be caused by patients' stronger perception of depression than actual disability in "on" phase.

TABLE 2 | Demographic variables, motor function, non-motor symptoms, and ADL of PD-D and PD-ND groups.


\**P* < *0.05;* \*\**P* < *0.01. LEDD, levodopa equivalent daily dose; Sinemet CR, controlled release Sinemet; Piribedil C, controlled release Piribedil; UPDRS III, Unified Parkinson's Disease Rating Scale III; TD, tremor-dominant; PIGD, postural instability/gait difficulty; MoCA, Montreal Cognitive Assessment Scale; HAMA, the 14-item Hamilton Anxiety Scale; FS, the 14-item Chalder Fatigue Scale; RBDSQ, the Rapid Eye Movement Sleep Behavior Disorder Screening Questionnaire; SCOPA-AUT, the Scale for Outcomes in PD for Autonomic Symptoms; RLSRS, Restless Legs Syndrome Severity Rating Scale.*

Worse motor symptoms may aggravate patients' psychological and physical burden, which would increase the sense of guilt and despair. Besides, our study found that PIGD type was independently associated with PD-D. Accordingly, good management of motor symptoms is important for PD-D.

In our previous study (35), PIGD patients had more severe or faster neurodegeneration than TD group. In addition, PIGD severity might be related to the depletion of homovanillic acid, one of the metabolites of DA, in CSF (35). Moreover, in the current study, the decrease of DA might correlate with PD-D.

TABLE 3 | Logistic regression analysis of factors associated with PD-D.


\**P* < *0.05,* \*\**P* < *0.01. UPDRS III, Unified Parkinson's Disease Rating Scale III; PIGD, postural instability/gait difficulty; HAMA, the 14-item Hamilton Anxiety Scale; FS, the 14 item Chalder Fatigue Scale; RBDSQ, the Rapid Eye Movement Sleep Behavior Disorder Screening Questionnaire; B, regression coefficient or intercept; OR, odds ratio; CI, confidence interval.*

TABLE 4 | Levels of neurotransmitters in CSF from PD-D and PD-ND groups.


\**P* < *0.017. DA, dopamine; 5-HT,5-hydroxytryptamine; NE, norepinephrine.*

TABLE 5 | Correlation of HAMD scores with the levels of neurotransmitters in CSF from PD patients.


\**P* < *0.017. DA, dopamine; 5-HT, 5-hydroxytryptamine; NE, norepinephrine.*

Therefore, the PIGD group might be vulnerable to PD-D. Much effort should be paid to PD patients with PIGD type to mitigate depression.

The comorbidity of depression and anxiety in PD patients was as high as 14–50% (36, 37). In this study, the logistic regression analysis demonstrated that anxiety was one of the risk factors of PD-D. Anxiety and depression in PD patients often exist together, suggesting that they may have a common biochemical basis. They may both be related to the extensive serotonergic alteration and a more limited dopaminergic breakdown (38). Moreover, increased levels of inflammatory markers in CSF were strongly associated with depression, anxiety and other nonmotor symptoms of PD (39). Nonetheless, research suggests that the comorbidity of depression and anxiety might be resulted from the superposition of different pathophysiological mechanisms of anxiety and depression rather than the common mechanism (37).

Fatigue was previously considered as a manifestation of PD-D, but in recent years, it has been recognized as a nonmotor symptom independent of depression. A previous study showed that fatigue was correlated with depression (40). We

TABLE 6 | Logistic regression analysis of the relationship of DA in CSF and PD-D.


\**P* < *0.05. UPDRS III, Unified Parkinson's Disease Rating Scale III; PIGD, postural instability/gait difficulty; HAMA, the 14-item Hamilton Anxiety Scale; DA, dopamine; B, Regression coefficient or intercept; OR, odds ratio; CI, confidence interval.*

further demonstrated that fatigue was one of the significant influencing factors of depression in PD patients. Our previous studies found that the decrease of serotonin dysfunction might be correlated with fatigue in PD patients (24). There was much interest in whether PD-D and fatigue shared the similar pathophysiologic mechanisms, for example, serotonergic dysfunction (41, 42). The underlying mechanism needs further investigation.

The research on the mechanisms of PD-D has not reached consensus. Studies showed that other chronic diseases, which also resulted in movement disorders, had lower incidences of depression than PD (28). What's more, depressive symptoms could appear before motor symptoms in PD patients (43). Therefore, as endogenous depression, PD-D is likely related to the characteristic pathology of PD. Theoretically, according to Braak stage of PD pathology (44), Lewy bodies deposit in the raphe nuclei, locus caeruleus, substantia nigra, and ventral tegmental area, and cause progressive loss of neurons and subsequent depletion of several neurotransmitters, including 5- HT, NE, and DA, etc.

At Braak stage II, the serotonergic neurons are affected. A study reported that PD-D patients had reduced levels of 5-hydroxyindole acetic acid (5-HIAA, a metabolite of 5-HT) in CSF than PD-ND patients (45, 46). However, Olivola and investigators failed to find the changes of 5-HT in the CSF of PD-D patients (47). These studies, including our study, showed the lack of correlation between depression and 5-HT or 5- HIAA in CSF. However, a positron emission tomography (PET) imaging study indicated a close correlation between PD-D and 5-HT transporter (48), although other similar studies failed to display consistent results (49, 50). Although there are many plausible theories explaining that deficits of the monoaminergic neurotransmitters are related to depression (51), the roles of serotonergic system on PD-D are less certain in human studies.

The level of DA is increasingly recognized as an important indicator for serious problems in PD-D; this is supported by pathological, experimental, and neuroimaging studies (52). The DA transporter availability was also proven to be reduced in PD-D in several PET studies (12, 53, 54). As a DA receptor agonist, pramipexole (0.125–1.0 mg three times per day) could improve the depressive symptoms in PD patients, and 80% of the improvements was caused by a direct effect of treatment on depressive symptoms while 20% worked through the alleviation of motor symptoms (55). Pramipexole could upregulate the expression of dopamine receptor D3 (56). The same effect was found in Ropinirole (57) and a prospective multicenter study showed that Ropinirole (a median dose of 10 mg) could improve both anxious and depressive symptoms in PD patients (58). The anti-depressant effect also was obtained in Rasagiline with higher doses than those used for the control of motor symptoms (1 mg/day) (59) and it might be related with its role for dopamine-enhancing (60). Unfortunately, none of these studies reported an association between the levels of neurotransmitters in CSF and PD-D. The results of our study supported the model that PD-D arised as a result of the dysfunction in dopaminergic pathways (61). On the other hand, DA helped to improve the motor symptoms, which was an independent influencing factor for PD-D. In this study, it showed that in PD-D patients, both the DA and 5-HT levels in CSF were decreased, however, only the level of DA in CSF was correlated with HAMD score. These results suggested that DA played a more significant role on PD-D. Therefore, it is plausible to speculate that dopaminergic-centered therapy may be much helpful for PD-D.

Most of NE neurons are distributed in the locus caeruleus. Previous investigations on the association between NE and PD-D were rather limited with relatively small sample sizes. Additionally, these studies chiefly focused on the damages of NE neurons and related innervations, and their correlation with PD-D. For example, a study observed that neuronal loss in locus caeruleus was different between PD-D and PD-ND groups (62). Another PET study using <sup>11</sup>C-RTI-32 as an in vivo marker for both DA and NE transporter binding, revealed that PD-D might be associated with the loss of NE innervations in the limbic system (12). Although the above studies showed clues implying the potential involvement of NE in PD-D, there was no direct evidence linking NE reduction in CSF to PD-D. In this study, the NE levels in CSF did not differ significantly between PD-D and PD-ND groups, and PD-D did not correlate with NE level in CSF. Therefore, NE might not be a critical neurotransmitter for PD-D.

This study has the following limitations. CSF samples were not obtained from all PD patients enrolled in this study due to difficulties including old age, hyperosteogeny, and intolerance of holding position for lumbar puncture, etc. CSF samples from PD patients with moderate and severe depression were also relatively limited. Thus, further investigations with more CSF samples, especially from moderate and severe PD-D patients and prolonged follow-up time are much needed to support the results from the current study. What's more, as a retrospective and observational study, our study provided limited grounds for drawing definite conclusions. Longitudinal studies are required to elucidate how the depression in PD patients affects the progression and prognosis of these patients.

In summary, PD patients have a high frequency of mild and moderate depression. Motor symptoms, PIGD type, anxiety and fatigue are the significant influencing factors of PD-D. DA plays a more important role on PD-D. Results from this study provide new insights for the management of the above risk factors of PD-D and possible route for reduction of PD-D by targeting dopaminergic system.

# AUTHOR CONTRIBUTIONS

T-HL drafting the manuscript, study design, analysis of data, accepts responsibility for conduct of research and will give final approval, acquisition of data, statistical analysis. PG study design, accepts responsibility for the conduct of research and will give final approval, acquisition of data. L-JZ, YH, S-YY, LL, ZJ, Q-JY, R-DW, L-XL, and Y-SP accepts responsibility for the conduct of research and will give final approval, acquisition of data. WZ study design, analysis of data, accepts responsibility for the conduct of research and will give final approval, acquisition of data, statistical analysis, study supervision.

# FUNDING

This work was supported by The National Key Research and Development Program of China (2016YFC1306000, 2016YFC1306300), National Key R&D Program of China— European Commission Horizon 2020 (2017YFE0118800— 779238), The National Natural Science Foundation of China (81571229, 81071015, 30770745), The Key Project of National Natural Science Foundation of China (81030062), The Key Project of Natural Science Foundation of Beijing, China (B) (kz201610025030), The Key Project of Natural Science Foundation of Beijing, China (4161004, kz200910025001), The Natural Science Foundation of Beijing, China (7082032), National Key Basic Research Program of China (2011CB504100), Important National Science & Technology Specific Projects (2011ZX09102-003-01), National Key Technology Research and Development Program of the Ministry of Science and Technology of China (2013BAI09B03), Project of Scientific and Technological Development of Traditional Chinese Medicine in Beijing (JJ2018-48), Project of Beijing Institute for Brain Disorders (BIBD-PXM2013\_014226\_07\_000084), High Level Technical Personnel Training Project of Beijing Health System, China (2009-3-26), Project of Construction of Innovative Teams and Teacher Career Development for Universities and Colleges Under Beijing Municipality (IDHT20140514), Capital Clinical Characteristic Application Research (Z12110700100000, Z121107001012161), Beijing Healthcare Research Project, China (JING-15-2, JING-15-3), Excellent Personnel Training Project of Beijing, China (20071D0300400076), Natural Science Foundation of Capital Medical University (PYZ2018077), Basic-Clinical Research Cooperation Funding of Capital Medical University, China (2015-JL-PT-X04, 10JL49, 14JL15), Youth Research Funding, Beijing Tiantan Hospital, Capital Medical University, China (2014-YQN-YS-18, 2015-YQN-15, 2015-YQN-05, 2015-YQN-14, 2015-YQN-17).

# REFERENCES


Parkinson's disease: a randomised, double-blind, placebo-controlled trial. Lancet Neurol. (2010) 9:573–80. doi: 10.1016/s1474-4422(10)70106-x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer FD and handling editor declared their shared affiliation at time of review.

Copyright © 2019 Lian, Guo, Zuo, Hu, Yu, Liu, Jin, Yu, Wang, Li, Piao and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Diversiform Etiologies for Post-stroke Depression

Zan Wang† , Yanmin Shi † , Fangfang Liu, Nan Jia, Junya Gao, Xiaomin Pang and Fang Deng\*

Department of Neurology and Neuroscience Center, The First Bethune Hospital of Jilin University, Changchun, China

After the onset of stroke, many patients suffer from emotional behavior changes. Approximately, one-third of stroke survivors are affected by post-stroke depression (PSD), making it a serious social and public health problem. Post-stroke depression (PSD) has an important impact on the course, recovery, and prognosis of stroke. The pathogenesis of PSD is very complex, involving many factors such as biological mechanism and social psychological mechanisms. This article provides a brief review of the hot issues related to etiologies of PSD.

#### Edited by:

Chunxue Wang, Beijing Tiantan Hospital, China

#### Reviewed by:

Chunyan Zhu, Anhui Medical University, China Yuan Yang, Huazhong University of Science and Technology, China

> \*Correspondence: Fang Deng defrances2000@163.com

†These authors have contributed equally to this work and are co-first authors

#### Specialty section:

This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry

Received: 14 October 2018 Accepted: 20 December 2018 Published: 23 January 2019

#### Citation:

Wang Z, Shi Y, Liu F, Jia N, Gao J, Pang X and Deng F (2019) Diversiform Etiologies for Post-stroke Depression. Front. Psychiatry 9:761. doi: 10.3389/fpsyt.2018.00761 Keywords: post-stroke depression, depression, stroke, biological mechanism, social psychological mechanisms, default mood network

# INTRODUCTION

Post-stroke depression (PSD) refers to persistent depression after a stroke. Expressed as loss of interest, decreased energy, decreased appetite, sleep disorders, low self-evaluation, self-blame, and even repeated self-injury, suicidal thoughts or behaviors. It is the most common emotional disorder after stroke. As early as 1977, Folstein et al. reported PSD for the first time, and its incidence rate was as high as 45% (1). Patients with major depression after stroke account for 10–25% of stroke patients, and those with mild depression account for 10–40% of stroke patients. Symptoms were most common in the third month after stroke, and the prevalence did not decrease in the following year (2). The clinical manifestations of patients with post-stroke depression (PSD) are more complicated. Clinically, the patient's performance is often divided into core symptoms and noncore symptoms. The main symptoms are: (1) most of the time patients feel unhappy, even painful; (2) lose interest and pleasure, and cannot get happiness from the things they usually love; (3) energy decline, easy to feel tired, even lose the belief of living, suicidal tendencies. Non-core symptoms are mainly: (1) weight loss, difficulty sleeping, insomnia and dreams, unexplained loss of appetite, pain, general malaise; (2) nervousness, anxiety; (3) self-evaluation decline, self-blame, worthless, hesitant, attention decreased, etc. The diagnosis of typical cases of PSD is not difficult. Patients with a history of stroke, low mood, lack of interest or loss of fun, plus some psychological, or physical symptoms can make a diagnosis. A considerable number of patients do not show obvious sadness and despair, but mainly a variety of physical symptoms, such as fatigue, anxiety, tension headache, loss of appetite, sleep disorders. Post-stroke depression affects the patient's cognitive function and quality of life, increases the patient's mortality and self-killing rate, and imposes a heavy burden on society and the family. However, there are still many ambiguities about the risk factors, etiologies of PSD. Therefore, early accurate etiologies of PSD is very important and should be taken seriously by clinicians. The research progress in the incidence, etiologies is summarized as follows.

# INCIDENCE AND PREVALENCE OF PSD

There are significant differences in the incidence of PSD, due to differences in study selection, time to assessment after stroke, assessment methods and diagnostic criteria (3). A systematic review of 14 studies involving the prevalence of PSD found that the peak of depression was 3–6 months after stroke, and the prevalence was 9–34%. The prevalence of depression remained at a high level until 1–3 years after stroke; the prevalence of mild depression after stroke was ∼8–22% (4). Hackett analyzed 51 studies: Using the Hamilton Depression Scale (HDRS), the lowest PSD rate was 26%. The highest incidence of PSD was 41% with the Montgomery-Asberg Depression Rating Scale and the Zung Depression Scale (5). Schöttke believes that the incidence of PSD was 31.1%, post-stroke anxiety prevalence was 20.4% (6). Chemerinski et al. analyzed 24 studies and classified patients with strokes from different sources. The results showed that the incidence of major depression in acute hospitalized stroke patients was 22%, mild depression was 17%; Out-patient stroke with severe depression was 23%, mild depression was 35%; community patients had severe depression of 13% and mild depression was 10% (7).

# ETIOLOGIES OF PSD

The pathogenesis of PSD is complex, involving many factors such as biological mechanism and social psychological mechanisms.

# Biological Mechanism

#### Monoamine Neurotransmitter Change

Numerous studies on depression have confirmed that noradrenergic and serotonergic neurons involved in emotional regulation in the brain are located in the brainstem. Its axons pass through the hypothalamus, basal ganglia, corpus callosum, and radial crown, and finally reach the frontal cortex. 5-HT and NE are both monoamine neurotransmitters, mainly involved in depression, anxiety, self-injury suicidal behavior, and sleep disorders. When stroke destroys the above related structures, it can cause a decrease in NE and 5-HT levels, and patients are more prone to depression (4). Some scholars have found that the concentration of serotonin metabolites in cerebrospinal fluid of patients with PSD is reduced (8). Combined with clinical application of antidepressants such as selective serotonin reuptake inhibitors (SSRIs), it is effective in the treatment of PSD. It was further confirmed that the occurrence of PSD is associated with a decrease in monoamine neurotransmitters (9).

Studies have found that the occurrence of PSD is related to neurotransmitters such as glutamate (Glu) and gammaaminobutyric acid (GABA). Hypoxia-induced hypoxia causes a decrease in ATP and changes in membrane permeability leading to K+ efflux and Ca2+ influx, leading to an increase in excitatory amino acids such as glutamate, while re-uptake is blocked, excitatory amino acids accumulate outside the cell, leading to post-synaptic Excessive excitation, degeneration, and necrosis of neurons. Increased Glu leads to post-synaptic neurons, excessive excitability, ulceration, and necrosis. Patients with PSD are often accompanied by changes in the level of glutamate in the frontal lobe. PSD patients have a significantly elevated glutamate/creatinine ratio in the frontal lobes and anterior cingulate gyrus. This change is associated with the Hamilton Depression Scale (HDRS) is associated with patients with high scores (10). Wang et al. used MRI spectroscopy studies to show that patients with PSD have higher glutamate levels than stroke patients without PSD (11). The mechanism of PSD is related to the imbalance of GABA expression, and a decrease in GABA can lead to a low level of NE (12).

#### Inflammation Mechanism

Inflammation refers to the defense response of living tissue to the stimulation of biological, physical, chemical, and other damage factors. Various inflammatory factors refer to cytokines involved in the inflammatory response, and are hydrophilic specific polypeptides or small molecular proteins secreted by activated immune cells. When inflammation occurs in the body, inflammatory factors inactivate the phosphorylation of the inhibitor of nuclear factor kappa B (IκB) resulting in a decrease in the inhibition of nuclear factor kappa B (NF-κB) by IκB, which causes NF-κB to enter the nucleus and bind to specific NF-κB. Inducing the transcription of related genes and promoting the expression of genes, causing an increase in anxiety and other depressive behaviors. IL-1, IL-2, IL-6, and TNFα are mainly produced by mononuclear macrophages, which are involved in the inflammatory response and promote the immune response. Both IL-10 and IL-13 are important inhibitory cytokines, mainly produced by Th2 cells, which inhibit the production of pro-inflammatory cytokines, suppress immune responses, and protect nerves. Inflammation plays a protective role in maintaining the homeostasis of the body. However, if it is overreacted, it can damage normal tissues and organs. Studies have found that inflammation under certain conditions can damage the internal balance of the body, causing metabolic disorders, leading to abnormal secretion of neurotransmitters, resulting in depression (13). Spalletta et al. believe that the occurrence of PSD may be related to immune activation, leading to increased secretion of cytokines, and proposed a "cytokine hypothesis." After stroke, astrocytes and microglia in the central nervous system produce cytokines and their receptors, including IL-1, IL-6, TNF-α, and IFN-γ. The phenomenon of immune activation and increased cytokines (14).

Studies have suggested that inflammatory factors may cause depression through neurodegeneration, decreased regeneration, decreased ω-3 fatty acids, decreased levels of tryptophan, and elevated levels of metabolites. Inflammatory factors interact with each other to form a network system that regulates immune responses (15, 16). Inflammatory factors can increase the activity of indoleamine-2,3-dioxygenase (IDO), increase the metabolism of tryptophan, and increase the concentration of quinolinic acid and kynurenine. The level of serotonin precursors synthesized by tryptophan is reduced, causing a decrease in serotonin concentration and accelerating depressive symptoms in patients (17, 18).

Yang et al. studies found that IL-18, IL-1, IL-6 play an important role in the occurrence, development and prognosis of PSD (19). Kim et al. performed polymorphisms of proinflammatory cytokine genes such as IL-1β, IL-6, IL-8, TNFα, and polymorphisms of anti-inflammatory cytokines such as IL-4 and IL-10 in patients with PSD. The results showed that the IL-10-1082A/A genotype was closely related to PSD, and the IL-4 + 33C/C genotype was only associated with heavy PSD (20). After a year of follow-up of PSD patients, Su et al. found that IL-10 levels in the depression group were significantly lower than those in the non-depressed group, and IL-10 levels were negatively correlated with depression. IL-10 may have antidepressant effects (21). Spalletta et al. believe that stroke promotes the release of inflammatory factors such as C-reactive protein (CRP), interleukin-1(IL-1), tumor necrosis factor-α (TNF-α), IL-6. These factors stimulate and produce a toxic effect on the monoamine neurotransmitter system. As a result, its function declines, causing depression to occur (22). Increased expression of TNF-α, IL-1β, and cortisol releasing factor was found in the hippocampus of PSD rats, and interaction between cortisol releasing factor and TNF-α signaling pathway was found in PSD (23). Studies by Reichenberg et al. have shown that induction of TNF-α production by experimental stimulation can induce depression and cognitive function changes in humans (24).

#### Hypothalamus-Pituitary-Adrenal Axis and Hypothalamus-Pituitary-Thyroid Axis

Studies have shown that inflammation can affect the function of the hypothalamic-pituitary-adrenal axis (HPA), promote the metabolism of monoamine transmitters, and induce depression (14). Pro-inflammatory factors activate HPA and promote excessive secretion of cortisol, and excess cortisol damages nerve cells through cytotoxicity. Abnormal activity of the HPA and hypothalamus-pituitary-thyroid axis (PHT), leading to elevated plasma cortisol. Elevated plasma cortisol induces the production of tryptophan pyrrolase and aminotransferase in the liver, which both degrade blood tryptophan (5-HT precursor) and tyrosine (NE precursor), resulting in 5-HT and NE synthesis are reduced, promoting, or aggravating the occurrence and development of PSD (25–27). Persistent and excessive secretion of cortisol can inhibit hippocampal neuronal regeneration and reduce neural plasticity in the pre-frontal cortex, leading to PSD.

#### Glial Cells

#### **Astrocytes**

Glial cells are widely distributed in the central and peripheral nervous systems and are mainly divided into three types: astrocytes, oligodendrocytes, and microglia. Astrocytes are the most important glial cells, which secrete a variety of neurotrophic factors, and play an important role in energy metabolism regulation, neurotrophic factor release, neuronal synaptic remodeling and nerve formation. A large body of evidence indicates that astrocyte dysfunction is an important factor in the onset of depression. Studies have reported that astrocyte hyperplasia is a characteristic response to central nervous system inflammation or injury. Astrocytes and microglia work together to regulate the release of pro-inflammatory cytokines and anti-inflammatory cytokines, maintaining the normal physiological functions of the brain. Astrocytes can release a variety of neurotrophic factors, such as nerve growth factor, brain-derived neurotrophic factor (BDNF), glial-derived neurotrophic factor (GDNF), fibroblast growth factor 2 (FGF 2) etc. Such neurotrophic factors regulate nerve function, promote nerve growth, and increase synaptic plasticity and delivery efficiency (28). Neurotrophic factor is a molecule that promotes the development and survival of neurons. It can prevent the pathological changes of ischemic brain injury, and can reduce the apoptosis of neurons and effectively improve the neurological function of patients after stroke. Previous experimental studies have found that neurotrophic factors are closely related to the condition and prognosis of PSD (29). A number of studies have found that levels of neurotrophic factors are significantly reduced in patients with PSD (30). A meta-analysis study showed a decrease in the expression of GDNF in the brain of patients with depression. Studies have shown that the reduction of GDNF and BDNF levels in the brain of patients with depression is associated with decreased hippocampal nerve regeneration. Antidepressants can increase the production of BDNF and GDNF in rat hippocampus (31–33). Selective serotonin reuptake inhibitors (SSRIs) significantly increased the expression of BDNF mRNA in astrocytes, suggesting that antidepressants can exert antidepressant effects by increasing astrocyte BDNF synthesis. Fluoxetine can increase the synthesis of GDNF and BDNF in astrocytes; amitriptyline can promote the synthesis and release of astrocytes FGF-2, BDNF, and GDNF (34, 35). Other studies have shown that astrocytes can synthesize antibodies and anti-inflammatory factors and inhibit the synthesis of proinflammatory factors. Astrocyte dysfunction can aggravate the inflammatory response and aggravate central nervous system damage (36).

#### **Microglia**

Microglia are immune cells of the central nervous system. When tissue damage or brain infection occurs, microglia are first activated to perform functions such as antigen recognition, phagocytosis, and antigen presentation (37). When activated, microglia can produce a large number of pro-inflammatory factors, causing degeneration and necrosis of neurons. Under normal circumstances, microglia are in a resting state, receive synaptic signals by sensing changes in the extracellular environment, thereby participating in intersynaptic interactions; and can also express neurotransmitters such as dopamine and serotonin. When the extracellular environment changes, microglia can be activated and undergo morphological changes, releasing inflammatory factors (38). Activated microglia are divided into two polarization states, M1 and M2. Polarized M1 microglia produce pro-inflammatory cytokines and neurotoxicity participate in the development of neural network dysfunction and promote inflammation. Polarized M2 microglia secrete anti-inflammatory mediators and neurotrophic factors involved in restoring homeostasis (39). Previous studies have shown that autopsy after suicide in patients with major depression found activation of microglia in the pre-frontal cortex and lateral anterior cingulate gyrus. Moreover, activation of microglia in the anterior cingulate cortex, hippocampus, and thalamus is associated with suicide caused by depression (40). The cause of depression is related to the secretion of pro-inflammatory factors by M1 microglia. A variety of antidepressants have anti-inflammatory effects and can reverse the M1 type polarization of microglia. Fluoxetine and citalopram, widely used clinically, regulate the immune system by inhibiting M1 polarization and improving M2 polarization of microglia, mediating the therapeutic effects of drugs (41).

#### Vitamin D

Vitamin D is a neurosteroid hormone, 25-hydroxyvitamin D is its main form in the blood circulation (42). Vitamin D is derived from food, especially from fish oil, which is synthesized on the skin and is affected by light. Vitamin D receptor (VDR) is located in an important area of the brain associated with depression and emotional behavior, such as cingulate gyrus, hippocampus, thalamus, hypothalamus, and substantia nigra (43). VDR is also present in immune cells and has an immunomodulatory effect (44). Vitamin D can regulate neurotransmitters such as serotonin in the brain through tryptophan-hydroxylase 2, vitamin D deficiency may lead to central morphological changes and decreased synthesis of neurotransmitters such as norepinephrine and dopamine (45). Studies by Puchacz et al. showed that vitamin D is involved in the regulation of the expression of the tyrosine hydroxylase gene, which catalyzes the production of levodopa by tyrosine in dopamine biosynthesis (46). Studies have found that vitamin D levels are negatively correlated with inflammatory markers, and the relationship between depression and inflammatory response can be regulated by the immune system (47). In the central nervous system, vitamin D acts as a neuroprotective factor through its antioxidant activity to increase the efficiency of neuronal projection and regulate the synthesis of neurotransmitters. A meta-analysis found that low vitamin D levels are associated with depression levels and are the biological basis for depression susceptibility (48). Studies have shown that supplemental reduced vitamin D levels contribute to the improvement of depressive symptoms in patients with depression, but different studies have also been reported (49). Han et al. also believe that vitamin D levels are positively correlated with PSD within 24 h after stroke onset (50). A prospective randomized controlled clinical trial by Shaffer et al. found that vitamin D supplementation helps prevent PSD. High serum vitamin D levels protect patients from PSD, and recent randomized controlled trials have shown that vitamin D supplementation can improve depressive symptoms in patients (51, 52).

#### Homocysteine

Some studies have concluded that high homocysteine levels are significantly associated with PSD. Stroke patients with high levels of homocysteine are relatively more prone to PSD (53). High homocysteine (Hhcy) has a direct toxic effect on blood vessels, causing further damage to the cerebral blood vessels, leading to the occurrence of depression (54). Hhcy affects the production and metabolism of monoamine neurotransmitters such as DA, 5-HT, NE, etc. These neurotransmitters play an important role in the pathogenesis of depression (55). Liu et al. studied 18 patients with ischemic stroke and selected three core regions of DMN (left parietal cortex, pre-frontal cortex, posterior cingulate ganglion/wedge anterior cortex). Then, the difference between the patient and the normal person ReHo was compared, and the ReHo of the posterior cingulate cortex of all stroke patients was found to be reduced. The functional connectivity (FC) analysis was performed using the posterior cingulate cortex for the region of interest, and the FC values of posterior cingulate cortex and anterior cingulate were found to be reduced (56).

#### Neural Network Dysfunction

Resting-state functional magnetic resonance imaging (rs-fMRI) indirectly reflects the functions of brain local and neural networks through signal changes, which have the advantages of non-invasiveness and reproducibility. Since the early 1990s, it has had an important impact on the development of neuroscience and psychology. In recent years, it has also begun to be used in the research of diseases such as PSD, and has gradually become one of the important means for studying the physiological and pathological activities of brain function. At present, the commonly used analytical methods for rs-fMRI are as follows: Regional homogeneity (ReHo), Amplitude of low-frequency fluctuation (ALFF), Functional connectivity (FC) (57). The most widely studied is the default mood network (DMN), which mainly involves the medial pre-frontal cortex, posterior cingulate gyrus/pre-wedge lobes, bilateral apical lobes (including angular gyrus), bilateral lateral temporal lobe, hippocampus, etc. Studies have shown that the default network is closely related to the monitoring of internal and external environments, the processing of emotions, introspection, the maintenance of thinking cognition, and the extraction of thought memories. A large number of studies have shown that there is an abnormality in the brain network under the resting state of depression, and DMN is an important neuropathological mechanism of depression (58, 59). Zhang et al. performed FC analysis on patients with cerebral infarction, and found that compared with non-PSD and normal people, PSD patients had changes in the right frontal gyrus and the left gyrus and the anterior cingulate gyrus. In addition, the FC values of the left temporal and anterior cingulate gyrus were significantly associated with the severity of depression (60). Liu et al. (56) studied 18 patients with ischemic stroke, and selected 3 core regions of DMN (left parietal cortex, pre-frontal cortex, posterior cingulate gyrus/anterior cranial cortex), and then compared patients and normal Human differences. The ReHo of the posterior cingulate cortex was found to decrease in all stroke patients, and the posterior cingulate gyrus was used as a functional area for FC analysis, and the FC value of the posterior cingulate gyrus and the anterior cingulate gyrus was found to be reduced (56). Zhang et al. studied 26 patients with PSD, and the results showed that the default emotional network, cognitive control network, and emotional network FC of PSD patients changed. The left lower parietal lobe, the left eyelid portion of the inferior temporal gyrus and the left anterior gyrus were significantly associated with the Hamilton Depression Rating Scale for PSD patients. Changes in the three neural networks may be associated with the development of PSD in the subacute phase of stroke (61).

At present, there are many studies on the relationship between PSD and stroke sites, but the conclusions are not the same. Current brain imaging studies of depression have shown that subcortical white matter damage leads to a susceptibility to depression by destroying certain neural circuits associated with emotions. Depression brain function changes mainly in the pre-frontal cortex, anterior cingulate gyrus, amygdala, ventral striatum, hippocampus, insula, thalamus, and basal ganglia. The pre-frontal cortex is thought to play a key role in cognitive and emotional activities, and functional abnormalities in these brain regions may present with affective and cognitive impairments. Current studies have shown that left hemisphere stroke is more prone to depression than right hemisphere stroke. The lesions near the frontal pole have a specific correlation with the degree of PSD. The lesions associated with PSD were: frontal lobe, left basal ganglia, and temporal lobe, and the incidence of lesions near the extreme was high (62). Carson et al. found that the occurrence of PSD was not related to stroke lesions after evaluation of related studies (63). PSD has nothing to do with the stroke site. The reasons for the difference between the two may be different sample sizes, differences in diagnostic criteria, ethnic differences, geographical differences, and so on (64). Vataja et al. found that the lesions of acute stroke were located in the left hemisphere, especially in the left anterior hemisphere, and their chance of developing depression was higher than in the right hemisphere. It is believed that the damage of the globus pallidus and the volume of injury are related to the diagnosis of depression within 3 months after stroke (65). Angeleri et al. observed the observation of the 3 years after the stroke, 3 years or more after stroke, the incidence of depression is not related to the lesion in the left or right hemisphere (66). Shimoda et al. observed the relationship between lesions and PSD at different time points after stroke. In the acute phase of stroke, PSD was associated with left anterior hemisphere lesions; at 3–6 months, PSD was related to the distance from the lesion to the frontal pole and lesion volume; and 1–2 years later, PSD was related to the distance from the right hemisphere lesion to the occipital pole and lesion volume (67).

Cognitive dysfunction is one of the common complications in stroke patients. Most studies have concluded that depression has a significant relationship with cognitive dysfunction. However, the specific relationship between the two is still controversial. Many studies have identified the most relevant factors for PSD and cognitive impairment: low education, speech impairment, stroke severity, and previous diabetes history (68). Murata et al. found that cognitive function improved with the improvement of PSD symptoms (69).

#### Genetic Background

Regarding the gene hypothesis, there is clear evidence that the shortening of the promoter region associated with the serotonin gene is associated with severe PSD. Studies have shown that individuals and families with a history of depression may be one of the risk factors for major depression after stroke (70, 71). The expression of some genes is considered to be a risk factor for PSD. Brain-derived neurotrophic factor (BDNF) plays an important role in the pathophysiology of PSD. It has been suggested that the single nucleotide polymorphisms rs1778929 and rs1187323 in the tyrosine receptor kinase B (TrkB) gene of BDNF are significantly associated with PSD (72). Kim et al. evaluated 222 stroke patients and followed up for 1 year. The increase in methylation status of 5-HTTLPR (The serotonin transporter-linked polymorphic region, 5-HTTLPR) SS genotype was associated with PSD. Higher levels of BDNF gene methylation were associated with PSD occurring at followup (73). The serotonin transporter gene (SLC6A4) has also been shown to play an important role in the pathophysiology of PSD. Studies have found that 2 weeks after stroke, higher SLC6A4 promoter methylation status is independently associated with PSD and is more pronounced 1 year after stroke, and is significantly associated with worsening depressive symptoms within 1 year (74). Studies have shown that apolipoprotein E (APOE) polymorphism is associated with PSD, APOE rs429358 polymorphism increases the probability of PSD, APOE rs429358- C allele may be post-stroke nerve Functional recovery is harmful (75).

#### Social Psychological Mechanisms

After cerebrovascular disease, most patients have different degrees of physical dysfunction, resulting in loss of work and life. The combined effects of family, society, and physiology lead to physiological and psychological imbalance in stroke patients. Psychosocial factors such as poor living ability, negative life events, family burden, social family support may all contribute to PSD. Acute stroke is a stressful event that increases the secretion of glucocorticoids, causing elevated blood glucose and abnormal neurotransmitters, leading to depression (76). The onset of PSD is not a single mechanism. Whyte et al. proposed that PSD, like other psychiatric diseases, is under the bio-psycho-social medical model, and that biological factors and psychological factors may contribute to the onset of PSD (4).

Studies have shown that the degree of education is negatively correlated with the occurrence of PSD, probably because patients with low levels of education have limited cognitive levels (77). Studies by Backhouse et al. showed that lower education levels were associated with an increased risk of PSD symptoms, but confidence intervals and heterogeneity were greater (78).

The relationship between age and PSD has been controversial. Previous studies have found that the younger patients with acute stroke, the higher the risk of PSD. This may be due to young people taking on greater family and social responsibility. After the stroke, the social roles and economic status of young patients are more prominent and the psychological acceptance is poor (79). Some studies have shown that age is positively correlated with the occurrence of depression. With the increase of age, the body's various functions are declining, frustration and attention to the body become more and more prominent. There are also many studies that do not have a clear correlation between age and PSD (80).

Studies have shown varying incidence rates for patients of different genders. Most studies have found that women are more likely to develop PSD earlier than men. This may be related to women's poor psychological quality, sensitivity, psychological and physiological imbalance. Another possible explanation is that women live longer than men, so women have an average age greater than men when they encounter a stroke (81). However, another study found that the prevalence of male PSD is higher than that of females. It may be that men have more family and social responsibility for men. Therefore, changes in work ability and social status caused by stroke will produce greater psychological stress in men (82). However, another part of the study concluded that there is no difference between the two (83).

In the study of pre-existing personality characteristics of patients with PSD, it is found that patients with neuroticism, introversion, emotional instability, and strong dependence are more likely to develop PSD (84). Pessimism, negative coping, introversion, acceptability, etc. are independent risk factors for PSD.

Regarding the marital status of patients, studies have shown that widowhood, divorce, and solitary living are closely related to depression. It may be related to the patient's loneliness, social isolation, long-term physical illness, and decreased stress ability.

The severity of stroke directly affects the quality of life of patients and has a great impact on the occurrence of PSD. Many studies have shown that the ability of daily living is related to the occurrence of PSD in the early stage of stroke. The more severe the neurological damage, the lower the ability of daily living, the greater the risk of PSD. There was a significant correlation between the activities of daily living (ADL) and the incidence of depression after stroke. Low ADL is an important factor leading to PSD. The severe physical dysfunction, low self-care ability and loss of working ability make patients have great psychological pressure (85).

Studies on chronic diseases and depression have shown that chronic diseases such as hypertension, diabetes, dyslipidemia, and respiratory diseases are also in the category of psychosomatic diseases. Due to the long course of chronic diseases and the difficulty of treatment, patients are often in a state of anxiety and depression. Most patients have low ability to recognize these mental illnesses. Current research shows that among vascular risk factors, only hypertension can predict PSD. Diabetes,

### REFERENCES


hyperlipidemia, obesity, and smoking were not independent predictors of PSD (86).

Social support can be divided into two categories, one is objective, visible or practical support; the other is subjective and empirical emotional support. Good social support will force the patient's psychological endurance, indirectly promote the recovery of stroke patients and improve their quality of life, while stroke patients who lack social support are more likely to develop PSD (87). Kotila et al. found that patients who lived in community-active areas after stroke had fewer PSD than those without community activity, suggesting that appropriate rehabilitation activities can reduce PSD (88).

## CONCLUSION

PSD seriously affects the quality of life of patients, which is a burden on individuals, families, and society. It is necessary to explore the pathogenesis and related influencing factors, establish appropriate diagnostic criteria and scales, and determine the best preventive and therapeutic measures. The pathogenesis of PSD is extremely complex and may be the result of multiple factors and multiple pathways. There are still many uncertainties in the neurobiological mechanisms of PSD. It is believed that with the improvement of science and technology, people will gradually uncover the mystery of the biological mechanism of PSD and provide a more accurate theoretical basis for the diagnosis and treatment of PSD.

### AUTHOR CONTRIBUTIONS

YS and ZW wrote the main manuscript text, contributed equally to this work and should be regarded as co-first author. FD contributed substantially to the conception and design of this work, drafting of the work, and revised it critically for important intellectual content. FL, NJ, JG, and XP collected the data etc.


hospitalised stroke patients. Aust N Z J Psychiatry (1992) 26:208–17. doi: 10.1177/000486749202600204


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wang, Shi, Liu, Jia, Gao, Pang and Deng. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Association Between Cortical Superficial Siderosis and Dementia in Patients With Cognitive Impairment: A Meta-Analysis

Chenheng Zhou1†, Keqin Liu2†, Shenqiang Yan<sup>3</sup> and Ying Jin<sup>4</sup> \*

<sup>1</sup> Department of Neurology, First People's Hospital of Wenling, Wenling, China, <sup>2</sup> Department of Neurology, Hangzhou First People's Hospital, Hangzhou, China, <sup>3</sup> Department of Neurology, The Second Affiliated Hospital of Zhejiang University, School of Medicine, Hangzhou, China, <sup>4</sup> Department of Integrated Traditional Chinese and Western Medicine, First People's Hospital of Wenling, Wenling, China

Background: It remains unclear whether cortical superficial siderosis (cSS) is associated with dementia and its subtypes. We thus performed a meta-analysis to evaluate the relationship between dementia and cSS.

Edited by: Yi Yang, Jilin University, China

#### Reviewed by:

Zhiyi Xie, Sichuan University, China Zhao Ye, Huashan Hospital Affiliated to Fudan University, China

> \*Correspondence: Ying Jin jinyinglkq@163.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Dementia, a section of the journal Frontiers in Neurology

Received: 15 September 2018 Accepted: 04 January 2019 Published: 29 January 2019

#### Citation:

Zhou C, Liu K, Yan S and Jin Y (2019) Association Between Cortical Superficial Siderosis and Dementia in Patients With Cognitive Impairment: A Meta-Analysis. Front. Neurol. 10:8. doi: 10.3389/fneur.2019.00008 Methods: We searched EMBASE, PubMed, and Web of Science for relevant studies assessing risk of dementia and prevalence of cSS in patients with cognitive impairment. Fixed-effects and random-effects models were performed.

Results: Seven eligible studies including 3,218 patients with definite cognitive impairment were pooled in meta-analysis. The prevalence of cSS was 3.4%. The pooled analysis demonstrates odds ratio for cSS and dementia to be 1.60 (95% CI 1.04–2.44; p = 0.031). Subgroup analysis further indicated a significant association between cSS and Alzheimer's disease (AD) (OR = 2.01, 95% CI 1.34–3.02; p < 0.001), but not non-AD dementia (OR = 0.700, 95% CI 0.435–1.128; p = 0.143).

Conclusions: Our meta-analysis of available published data demonstrates an increased prevalence of dementia in the subjects with pre-existing cSS, especially for AD. These findings suggest cSS to be a candidate imaging indicator for AD. Further longitudinal research is needed to investigate the clinical relevance.

Keywords: dementia, Alzheimer's disease, superficial siderosis, cognitive impairment, meta-analysis

# INTRODUCTION

Dementia is a major public health concern associated with the aging population, and currently affects millions of individuals worldwide, while Alzheimer's disease (AD) is the most common cause of dementia in the elderly (1). The pathogenesis of AD consists of two parts, of which one is the amyloid cascade hypothesis, linked to cerebral amyloid angiopathy (CAA), and the other is the vascular hypothesis, linked to cerebral small vessel disease (CSVD) (2).

Cortical superficial siderosis (cSS) is characterized by linear hypointensities over the cortical surface of the supratentorial cerebral convexities on gradient recalled echo (GRE) or susceptibility weighted imaging (SWI) (3). The underlying pathological mechanism of cSS remains elusive, and generally assumed to reflect recurrent blood leaking episodes in the subarachnoid space (4).

**83**

Patients with probable CAA manifested a much higher prevalence of cSS (34%) (5), compared with those from the general population (0.7%) (6). Moreover, Shams et al.'s study showed a link between cSS and the neuroimaging markers of CSVD (7). We therefore hypothesized that cSS itself might be a significant predictor for dementia, especially AD.

However, few studies investigated the relationship between cSS and dementia, and the results are controversial. Zonneveld et al. found a higher prevalence of cSS in patients with AD than those with mild cognitive impairment (MCI) (8), whereas a recent study failed to demonstrate the diagnostic significance of cSS for AD (9). We thus performed a meta-analysis to determine whether associations between cSS and dementia or AD exist in patients with cognitive impairment.

# MATERIALS AND METHODS

We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and Meta-analysis of Observational Studies in Epidemiology (MOOSE) statement (10, 11).

# Search Strategy and Eligibility Criteria

We searched appropriate articles by systematic queries of NCBI (PubMed), ISI Web of Science, and EMBASE databases on the 10th of September 2018, using the following search terms: "superficial siderosis" in association with "dementia" or "Alzheimer" or "cognition" or "cognitive." Articles not published in English were translated and case reports were excluded. The references of all identified publications were reviewed for any additional studies not indexed. Two authors identified potentially relevant studies, resolving any uncertainties with a third author.

Both retrospective and prospective studies were eligible for inclusion if they (1) assessed the cognitive status for each subject in the cohort, and (2) provided the detailed data of cSS in each group according to the cognitive status.

# Study Selection and Data Extraction

Two authors considered all titles and abstracts for eligibility in a systematic manner, went through all articles selected as relevant and extracted data independently. We extracted information on study design, MRI parameters for cSS detection, definition of cSS, criteria of neuropsychological assessment, number, and demographics of participants (including age and sex), mini mental state examination (MMSE) score of participants, number of participants with cSS, number of participants of different cognitive status, and the severity of cSS (focal or disseminated) by using a unified data form. Discrepancies were resolved by consensus.

### Data Analysis

We used a fixed effects model (Mantel and Haenszel method) to calculate the pooled ORs and corresponding 95% confidence intervals (CIs), with weights calculated using the inverse variance method, because of the relatively small number of the outcome events. Subgroup analysis was performed to isolate patients with AD only. Statistical heterogeneity was assessed using I-squared statistics with inspection of the forest plot. Publication bias was evaluated with Egger's test, Begg's test, and the funnel plot. We repeated all analyses using random-effects models. All statistical analysis was performed with Stata 11.2 (StataCorp LP, Texas, USA).

# RESULTS

We identified 79 articles from PubMed, 159 from EMBASE, and 76 from Web of Science in our initial search. Fourteen studies (all published) met our predetermined criteria, however, five of these were from a same cohort, and other three were from cohorts of intracerebral hemorrhage (ICH) population. Finally, seven studies were pooled in a meta-analysis (**Figure 1**) (7– 9, 12–15). Characteristics of the included studies are summarized in **Table 1**. The definition of cSS was almost the same across all included studies: hypointense linear structures within the subarachnoid space or in the superficial layers of the cerebral cortex on GRE or SWI.

Study demographics are summarized in **Table 2**. These studies were composed of 4,005 patients with cSS evaluation (study sample size range: 212–1,504), 110 (2.7%) of which had cSS on initial GRE or SWI, while 4.9% of AD patients had cSS. The severity of cSS was classified as focal (restricted to ≤3 sulci) or disseminated ( ≥4 sulci) in six studies (7–9, 13– 15), and with detailed data in three (7, 8, 14). The cognitive statuses were classified as AD, vascular dementia, other dementia or undetermined, MCI, subjective cognitive complains, or cognitively normal (**Table 2**).

#### TABLE 1 | Characteristics of included studies.


cSS, cortical superficial siderosis; the ADNI study, the Alzheimer's Disease Neuroimaging Initiative study; MCI, mild cognitive impairment; AD, Alzheimer's disease; GRE, gradientrecalled echo; ICD, international classification of diseases; SWI, susceptibility-weighted imaging; NINCDS-ADRDA, National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association; NINDS-AIREN, National Institute of Neurological Disorders and Stroke-Association Internationale pour la Recherché et l'Enseignement en Neurosciences; PiB-PET, pittsburg compound B-positron emission tomography; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition; IWG-MCI, International Working Group on Mild Cognitive Impairment.

\*ANDI is a longitudinal multicenter natural history study for AD, while the data of cSS and cognition was cross-sectional.


MMSE, mini mental state examination; AD, Alzheimer's disease; VaD, vascular dementia; MCI, mild cognitive impairment; SCC, subjective cognitive complains; CN, cognitively normal; cSS, cortical superficial siderosis; CMB, cerebral microbleed.

\*749 of 809 subjects had available data of CMB.

&Only the number of lobar CMB was given.

The cognitively normal patients (n = 171) and those with subjective cognitive complains (n = 616) were excluded, thus only patients with definite cognitive impairment were included in the meta-analysis. Among patients with dementia, 76 of 1,869 (4.1%) had cSS compared with 33 of 1,349 patients (2.4%) without dementia. Pooled analysis demonstrated OR for the presence of cSS and dementia to be 1.60 (95% CI 1.04–2.44; p = 0.031) with no evidence of statistical heterogeneity (I <sup>2</sup> = 0.0%, p = 0.621) (**Figure 2**). There was no evidence of a publication bias either from the result of Egger's test (p = 0.604) or Begg's test (p = 0.881), and the shape of the funnel plot seemed symmetrical (**Figure 3**). After excluding the AD Neuroimaging Initiative study (12), which only enrolled AD patients, the association of cSS with dementia remained significant (OR = 1.156, 95% CI 1.028–1.301; p = 0.016).

After excluding one study without data of dementia subtype (13), pooled analysis of the remaining six studies (7–9, 12, 14, 15), including 3,006 patients (96 with cSS), demonstrated OR for the presence of cSS and AD to be 2.01 (95% CI 1.34– 3.02; p < 0.001) with no evidence of statistical heterogeneity (I <sup>2</sup> = 0.0%, p = 0.592) (**Figure 4**), while no significant association was found between cSS and non-AD dementia (OR = 0.700, 95% CI 0.435–1.128; p = 0.143) (7–9, 14, 15). However, in the three studies with detailed data of cSS severity (7, 8, 14), the presence of disseminated cSS was not associated with dementia (OR = 0.873, 95% CI 0.337–2.260; p = 0.523), or AD (OR = 1.379, 95% CI 0.554–3.431; p = 0.976). All analyses were consistent when using a random-effects model.

#### DISCUSSION

Our meta-analysis in more than 3,000 subjects with cognitive impairment reported the existence of a positive relationship between cSS and AD, but not for non-AD dementia. These findings suggest cSS to be a candidate imaging indicator for AD.

Patients with cSS usually had lower cognitive scores (13). The Rotterdam Scan Study revealed a very low prevalence of cSS (0.7%) in a general population (6), while cSS was found in ∼3% of patients in a memory clinic population, and with a prevalence of 5% in patients with AD (7–9, 13–15). Only one (0.001%, 1/787) subject with normal cognitive function was reported to have cSS in the studies considered (7–9, 12, 15). The abnormally low prevalence would reduce the efficacy of detecting significant effects, therefore, we only enrolled cognitive impairment patients in the current meta-analysis. Lummel et al. included in their study 113 subjects with non-traumatic and non-aneurysmal cSS, the most common etiologies was CAA, followed by reversible cerebral vasoconstriction syndrome,

central nervous system vasculitis, and hyperperfusion syndrome (3). The clinical manifestations for cSS were: acute ICH: 49%; transient focal neurological episodes: 34%; cognitive impairment: 12%; generalized seizure: 4%; and headache: 2% (3).

The underlying mechanism of the pathological association between cSS and AD is not clear. The close relation between cSS and CAA might support the amyloid pathology. All of the individuals who presented with cSS in the Rotterdam Scan Study had cerebral microbleeds in lobar locations (6). The presence of cSS was also associated with lobar microbleeds in the memory clinic populations (7, 9, 15). In addition, the APOE genotype was more common in cases with cSS compared to those without (7, 14, 15). Immunohistochemistry staining showed severe CAA with A-β in the leptomeningeal and cortical vessels of a patient with both AD and cSS (16). Renard et al. evaluated cerebrospinal fluid amyloid-β 1–40 (Aβ40), amyloid-β 1–42 (Aβ42), total and phosphorylatedtau (t-tau and p-tau) in patients with symptomatic isolated cSS, and found that the patients with cSS showed higher ttau and lower Aβ42 compared to the controls, and lower t-tau, p-tau, and Aβ40 compared to the AD patients (17). Moreover, in vivo amyloid imaging using [11C] Pittsburg compound B (PiB)-PET was performed in a cognitively impaired population, and cSS was found to be associated with higher global PiB retention ratio, and not present in any of the patients with a negative PiB scan (14), further supporting the hypothesis that cSS reflects an amyloid rather than the ischemic etiology.

Cognitive impairment was more frequent in patients with disseminated cSS, while transient focal neurological episodes were more often found in those with focal cSS (3). In the patients with spontaneous ICH, disseminated cSS was a key risk factor of new-onset dementia and recurrent symptomatic ICH (18, 19). However, the presence of disseminated cSS was not associated with dementia incidence in a recent longitudinal study of patients with probable CAA (OR = 1.268, 95% CI 0.702–2.292; p = 0.431) (20). Similarly, the severity of cSS could not predict dementia or AD in our meta-analysis. To sum up, the presence of disseminated cSS could predict dementia in patients with ICH, while it was not associated with dementia incidence in memory clinic populations. Disseminated cSS seems more important in the subjects with ICH, thus the characteristics of cohorts might be the key point. Considering the small sample size of patients with disseminated cSS, future studies are needed to investigate the clinical relevance of cSS severity.

Our study had several limitations. First, our analysis had inherent biases associated with the use of observational studies, and most of them were cross-sectional studies. All studies were subject to selection bias because not every individual underwent GRE or SWI. Moreover, the causality between cSS and dementia is still unclear, and future longitudinal studies are needed to clarify this association. Second, the use of unadjusted data rendered our analysis vulnerable to confounding variables, such as the neuroimaging markers of CSVD. Third, the clinical diagnostic criteria for dementia and its subtype might be not quite similar in each cohort.

In conclusion, our analysis shows that the presence of cSS is associated with AD. Future large multicenter studies and individual patient data meta-analyses are needed to investigate the importance of cSS in the pathogenesis and longitudinal progression of AD in the subjects with cognitive impairment.

### AUTHOR CONTRIBUTIONS

YJ: design of the study, interpretation of data for the study, revision of the study for important intellectual content, and final approval of this version of the manuscript; CZ and KL: acquisition of data for the study, drafting of the study, revising the study for important intellectual content, and interpretation of data for the study; SY: acquisition of data for the study, drafting of the study, and revising the study for important intellectual content.

#### REFERENCES


# FUNDING

This study was supported by grant from the National Natural Science Foundation of China (81701150), and the Young Elite Scientists Sponsorship Program by CAST to SY (2017QNRC001).

in the ADNI cohort. Alzheimers Dement. (2013) 9(Suppl. 5):S116–23. doi: 10.1016/j.jalz.2012.10.011


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zhou, Liu, Yan and Jin. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Transient Global Amnesia Linked to Impairment of Brain Venous Drainage: An Ultrasound Investigation

Ke Han<sup>1</sup> , Han-Hwa Hu<sup>2</sup> \* † , A-Ching Chao<sup>3</sup> \* † , Feng-Chi Chang<sup>4</sup> , Chih-Ping Chung<sup>5</sup> , Hung-Yi Hsu<sup>6</sup> , Wen-Yung Sheng<sup>5</sup> and Jiang Wu<sup>7</sup>

#### Edited by:

Thomas Penzel, Charité Medical University of Berlin, Germany

#### Reviewed by:

Zixiao Li, Capital Medical University, China Ning Zhang, Beijing Tiantan Hospital, China

#### \*Correspondence:

Han-Hwa Hu hanhwa@hotmail.com A-Ching Chao achch@cc.kmu.edu.tw

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Applied Neuroimaging, a section of the journal Frontiers in Neurology

Received: 10 June 2018 Accepted: 17 January 2019 Published: 05 February 2019

#### Citation:

Han K, Hu H-H, Chao A-C, Chang F-C, Chung C-P, Hsu H-Y, Sheng W-Y and Wu J (2019) Transient Global Amnesia Linked to Impairment of Brain Venous Drainage: An Ultrasound Investigation. Front. Neurol. 10:67. doi: 10.3389/fneur.2019.00067 <sup>1</sup> Department of Neurology, The Seventh Affiliated Hospital, Sun Yat-sen University, Shenzhen, China, <sup>2</sup> Graduate Institute of Clinical Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan; Cerebrovascular Treatment and Research Center, College of Medicine, Taipei Medical University, Taipei, Taiwan; Department of Neurology,Taipei Medical University-Shaung Ho Hospital, Taipei, Taiwan, <sup>3</sup> Department of Neurology, College of Medicine, Kaohsiung Medical University and Department of Neurology, Kaohsiung Medical University Hospital, Kaohsiung, Taiwan, <sup>4</sup> Department of Radiology, Taipei Veterans General Hospital and National Yang Ming University, Taipei, Taiwan, <sup>5</sup> Department of Neurology, Taipei Veterans General Hospital and National Yang-Ming University, Taipei, Taiwan, <sup>6</sup> Department of Neurology, Tungs' Taichung Metro Harbor Hospital, Taichung, Taiwan, <sup>7</sup> Department of Neurology, First Hospital of Jilin University, Changchun, China

Background: Previous neuroimaging and ultrasound studies suggested that compression and stenosis of the internal jugular vein (IJV) in patients with transient global amnesia (TGA) may impair IJV drainage, while a patent IJV releases intracranial pressure caused by the Valsalva maneuver (VM).

Methods: Seventy-nine TGA patients with complete ultrasound examination data during admission were recruited prospectively to evaluate IJV drainage, which included the time-averaged mean velocity, and the cross-sectional lumen area of the IJV at the vein's middle (J2) and distal (J3) segments and the cross-sectional area during a 10-s VM to test for any retrograde or anti-grade flow. Forty-five TGA patients and 45 age- and sex-matched control subjects underwent complete contrastenhanced magnetic resonance (MR) venous studies, which included time-resolved imaging of contrast kinetics, contrast-enhanced axial T1-weighted MR imaging, and phase-contrast-based non-contrast enhanced magnetic resonance venography (MRV).

Results: In those subjects with complete MRV studies, the flow volumes exhibited at both the J2 and J3 segments of the left IJV and left vertebral vein (VV) were significantly lower in the TGA patients than in the control subjects. Although there was no significant difference in the flow volume of right IJV, the total of bilateral IJV, and VV flow volumes was still significantly lower in the TGA patients. As compared with the control subjects, the TGA patients exhibited significantly higher prevalence of completely blocked right IJV drainage at the J3 segment during the VM, but non-significantly higher for the left IJV at the J3 segment and for the right IJV at the J2 segment.

**89**

Conclusion: Our results confirmed that the total venous flow decreases in the IJVs and VVs of the patients with TGA. This is consistent with the findings of previous MR imaging studies that have reported about compression and stenosis of the draining veins. We also found that IJV drainage is relatively compromised during the VM in the patients with TGA.

Keywords: internal jugular vein (IJV), magnetic resonance venography (MRV), transient global amnesia (TGA), ultrasound, Valsalva maneuver (VM), vertebral vein (VV), hemodynamics

#### INTRODUCTION

Transient global amnesia (TGA) is defined as a sudden and transient inability to acquire new information (1). It can be triggered by certain events including Valsalva maneuver (VM)– like activities (1–3). Cerebral venous congestion/hypertension is one of the conditions that has been linked to TGA, which results from venous reflux while performing a VM in the subjects with internal jugular vein valve incompetence (IJVVI) (4–6). However, previous ultrasound and non-contrast venous magnetic resonance (MR) angiography studies have not supported a causal relationship between IJVVI and TGA (7–9). Previous studies have also reported that VM-induced pressure in the chest and abdomen is mainly transmitted to the intracranium via the epidural venous plexus (10, 11). Theoretically, bilateral IJV patency is needed for the brain venous drainage, which is regarded as a protective mechanism against intracranial venous hypertension. Stenosis or obstruction of the IJV hinders the brain venous drainage. This can directly cause intracranial hypertension (12, 13) and impair the protective function of IJV during the VM, which worsens the increased intracranial pressure further (11, 14). Consistent with this hypothesis of the venous outflow obstruction, we have previously showed that many patients with TGA exhibit stenosis or obstruction of the left brachiocephalic vein (BCV) (15). Using MR imaging, we have also demonstrated that the patients with TGA manifest a higher prevalence of compression/stenosis of the bilateral IJVs and left BCV, and transverse sinus (TS) hypoplasia, which supports the hypothesized role of abnormal brain venous drainage in the pathogenesis of TGA (16). Hence, we hypothesized that the compression/stenosis of the bilateral IJVs and left BCV would impede the brain venous drainage, which would result in reduced IJV flow volumes exhibited in the patients with TGA. We further hypothesized that in the patients with TGA, the IJV drainage would be particularly blocked during the VM, resulting in a phenomenon known as "IJV non-patency." This would impair the role of IJV drainage in releasing the intracranial pressure. We tested these hypotheses with ultrasound evaluations of the morphology and hemodynamics of the extracranial IJVs and vertebral veins (VVs) at rest and during the VM.

### METHODS

#### Study Design and Participants

The main study design and the participants have been described elsewhere (16). In brief, from January 2008 to December 2012, 79 patients with TGA were admitted to the Taipei Veterans General Hospital Neurology Department. All were examined by a neurologist, and TGA was diagnosed according to the criteria as modified and validated by Hodge and Warlow (1). All the TGA patients underwent complete ultrasound examination including analyses of the IJV responses during the VM; but of those, 34 patients had already underwent an emergency MRI at the emergency department to exclude the possibility of acute ischemic stroke, and hence did not undergo the complete MR venous study. The remaining 45 patients underwent complete ultrasound examinations and complete contrast-enhanced magnetic resonance (MR) venous studies, which included magnetic resonance imaging (MRI), magnetic resonance angiography (MRA), and magnetic resonance venography (MRV) assessments of the IJV drainage. We prospectively recruited 45 age- and sex-matched control subjects from the individuals who underwent physical checkups and presented no history of neurologic signs or symptoms. However, we did not recruit any controls for the subgroup of those 34 patients who did not undergo complete venous MR imaging studies.

Those 45 control subjects also underwent complete MRI, MRA, and MRV studies and ultrasound assessments. In accordance with the regulations of our government and the regulations of the Ethics Committee of Taipei Veterans General Hospital and in compliance with the Declaration of Helsinki, all the participants provided informed written consent with their signatures. The study protocol was approved by the Taipei Veterans General Hospital's institutional review board, and the study was conducted according to the institutional guidelines. All the participants gave the written informed consent.

# Ultrasound Acquisition

All the participants underwent color-coded duplex sonography with a 7-MHz iU22 linear transducer (Philips Medical Systems, Andover, MA), performed within 7 days after their TGA attacks, by a technician who had more than 10 years of experience in venous ultrasound studies and was blinded to the subjects' characteristics. The method for ultrasound examinations of the extracranial venous system has been reported elsewhere (17–20). In brief, the IJV's time-averaged mean velocity (TAMV, cm/s) and the cross-sectional lumen area (CSA, cm<sup>2</sup> ) were recorded at its middle (J2) and distal (J3) segments (19, 20). The location of J2 and J3 is the point where the common facial vein drains into the IJV (19, 20). The IJV segment above this point is J3, and below is J2. We acquired the TAMVs by directing the Doppler cursor parallel to the vein with the gate adjusted to comprise the entire lumen. We measured the TAMVs with the iU22's built-in software, and included at least three cardiac cycles on the Doppler spectrum. The probe was then turned by 90◦ at the same IJV segments to measure the CSAs, which were measured thrice as B-mode images. These images were averaged for later analysis. Both the CSAs and TAMVs were recorded at a brief apnea after the three respiratory states (20). The recordings made during the brief apnea after the three respiratory states are shown in the figures as follows: (1) normal respiration (resting or baseline) (**Figures 1A,D**), (2) deep inspiration (**Figures 1B,E**), and (3) expiration (**Figures 1C,F**). During the latter two respiratory states, the subjects were instructed to avoid strain-inducing breath-holding, because it could increase intra-thoracic pressure. We recorded the CSAs of the IJVs during a 10-s VM to test the incidence of reflow or no flow (**Figure 2**). The flow volume (FV) equals the TAMV multiplied by the CSA. Since it was difficult to obtain the CSAs of the VVs, we estimated the diameters of the VVs while modeling the VVs as having perfectly circular cross-sections. We measured the diameters of the VVs adjacent to the V2 segment of the vertebral artery. We also determined

the jugular venous reflux (JVR) at the baseline and during the VM. Our methods for performing the VM and detecting the JVR have been described elsewhere (17, 21).

All the ultrasound data and color imaging records were collected prospectively. Two trained neurologists, who were also blinded to the subjects' clinical characteristics, reviewed the collected data for the CSAs, TAMVs, and FVs for the IJVs, and the diameters and TAMVs for the VVs. A consensus meeting was conducted to discuss any problems or disagreements.

# MR Imaging Study

Our methods for MRI and grading of TS hypoplasia and stenosis/compression of the IJV and BCV have been described elsewhere (16, 20). In brief, we performed contrast-enhanced MRI of all the participants using a 1.5-T Excite II MRI device (GE Medical Systems, Waukesha, WI), which included time-resolved imaging of contrast kinetics (TRICKS), contrast-enhanced axial T1-weighted MRI (Contrast T1), and phase-contrast based noncontrast enhanced MRV. All the patients were examined within 10 days after their TGA attacks.

FIGURE 1 | Quantitative evaluation of the CSAs and TAMVs of the IJVs during the three respiratory states. The cross-sectional lumen areas (CSAs) and time-average-mean velocities (TAMVs) of the upper segment of the internal jugular vein (IJV, J3) were recorded during a brief apnea after the three respiratory statuses, namely: (A,D) at rest (normal respiratory status, (B,E); deep inspiration; and (C,F) expiration. As compared to that at rest, the CSA decreased (B) while the TAMV increased (E) during the deep inspiration, and the CSA increased (C) while the TAMV decreased (F) during the expiration. The figure was reproduced with the permission of Chao et al. (20).

pressure cannot be relieved effectively by the reopening the IJV in a 53-year-old TGA patient with significant IJV stenosis/obstruction. The arrows indicate the

beginning of the VM, and the arrowheads indicate the end of the VM. The figure was reproduced with the permission of Han et al. (16).

The IJV morphologies were assessed at the level of upper IJV (C1–2 level) and middle IJV (C3–5 level) using Contrast T1. The IJV compression/stenosis was graded according to the following criteria (16, 22): grade 0 = normal round or ovoid appearance; grade 1 = mild flattening; grade 2 = moderate flattening; grade 3 = severe flattening or non-visualized appearance (**Figure 1A**).

Based on the filling defect shown on TRICKS, the left BCV obstruction was graded as follows: grade 0 = normal or compression ≤20%; grade 1 = compression >20% and ≤80%; grade 2 = compression >80%; grade 3 = grade 2 + presence of different types of venous collaterals (23).

Based on MRV and Contrast T1 studies, TS morphology was graded based on the criteria modified from that given by Han et al. (16) and Cheng et al. (24). All the asymmetrical percentages were calculated relative to the contralateral TS, as follows: grade 0 = symmetry or asymmetry ≤10%; grade 1 = asymmetry >10% and ≤50%; grade 2 = asymmetry >50%; grade 3 = aplasia or TS signal absent. Hypoplasia was defined as an asymmetry <50%. We have illustrated the locations for performing MRV and Contrast T1 examinations in our previous paper (16). However, in this study, only the findings of Contrast T1 were used for the analysis of hypoplasia.

All the MR imaging scans were examined by a neuroradiologist and a neurologist. The intra-class correlation coefficient for grading indicated an inter-rater reliability of 0.91. If inconsistencies resulted between the reports given by the two MRI readers, they discussed and reached a "consensus."

# Statistical Analysis

Since this study was the first ultrasound study to evaluate the mechanism of jugular venous outflow impairment in the TGA patients, the sample size could not be pre-determined. The ultrasound data, such as the time-averaged mean velocity (TAMV), flow velocity (FV), and cross-sectional lumen (CSA) were skewed and were observed to be not normally distributed for the jugular venous flow. Thus, the expression of medians and inter-quartile ranges was used, and Wilcoxon rank sum analysis was performed. The chi-square test was performed for the categorical data. The same applies to the CSAs data for the TS, IJV, and VV. We compared the ultrasound data from the patients and control subjects using the Wilcoxon rank sum test. We used the chi-square tests to compare the prevalence of IJVVI or IJV patency during the VM. We defined the statistical significance as a 2-sided p < 0.05. All the analyses were performed with a SAS, 9.2 (SAS Institute, Cary, NC).

#### RESULTS

The demographic and clinical characteristics of all the 79 TGA patients, and the subset of 45 patients and 45 age- and sexmatched controls who underwent complete MRI, MRA, and MRV examinations are shown in **Table 1**. The patients and controls did not differ significantly in terms of their vascular risk factors. The TAMVs, FVs, and CSAs of the bilateral IJVs in the age- and sex-matched 45 patients and 45 controls are


TGA, Transient Global Amnesia.

shown in **Table 2**. As compared to the controls, the TGA patients exhibited significantly lower TAMVS at the J2 and J3 segments of the bilateral IJVs. They also exhibited significantly lower FVs in the left IJV at the J2 and J3 segments and in the left VV. The FVs were not significantly lower in the right IJV; however, the patients still exhibited significantly lower total flow volumes in the bilateral IJVs and VVs. In both the patients and controls, the CSAs of the bilateral IJVs were not significantly different either at the J2 or J3 segment. The TGA patients exhibited a significantly higher prevalence of IJVVI (patients vs. controls: 82 vs. 44%); however, the side-specific prevalence was significantly greater on the left side (patients vs. controls: 53 vs. 20%), while the right-sided prevalence being comparable between the patients and controls (patients vs. controls: 29 vs. 24%). Furthermore, we detected a significant difference in the prevalence of left-sided IJVVI between the subjects with and without the presence of left BCV compression/stenosis [19 (61%) vs. 14 (24%), p = 0.0044].

We have often observed that the IJV drainage flow usually appears ∼4–8 s after initiating the VM, so we simply defined the complete absence of IJV drainage flow at the J2 or J3 segment within 10 s of initiating the VM as IJV "non-patency." **Table 3** displays the relationship between the ultrasound findings of IJV non-patency during the VM, and the MRI findings of venous compression/obstruction or TS hypoplasia in those 90 study subjects (45 patients and 45 controls). For the left IJV, the prevalence of ultrasound-detected IJV non-patency during the VM was significantly greater at the J2 segment in the study subjects with an upstream TS hypoplasia than that in the patients without such hypoplasia (56.9 vs. 44.3%, respectively; p = 0.0425). For the right IJV, the prevalence of IJV non-patency at the J2 segment was significantly higher in the patients with IJV compression at C1 or C4 than in the patients without such compression (62.07 vs. 28.57%, respectively; p = 0.0111). We found no significant difference in the prevalence of bilateral IJV non-patency during the VM between those age- and sexmatched patients and controls who underwent complete MRI examinations. However, since there are no statistical differences in all the flow profiles between two groups of TGA patients with and without venous MR imaging as shown in **Table S1**, thus we included the 34 patients who underwent complete ultrasound examinations but incomplete MRI examinations for analysis, we found a significant difference in the prevalence of IJV non-patency during the VM between the patients and controls (**Table 4**). Specifically, we found that the patients exhibited significantly higher IJV non-patency at the right J3 segment (patients: 32.1%; controls: 11.6%; p = 0.0128), but not significantly higher in the left J3 segment (patients: 49.35%; controls: 37.21%), and the right J2 segment (patients: 44.00%; controls: 32.56%).

### DISCUSSION

Our results confirmed our first hypothesis that the patients with TGA having IJV stenosis/compression at various segments would exhibit significantly lower total FVs in the bilateral IJVs and VVs resulting in important consequences than that exhibited by the control subjects. More importantly, our findings are consistent with our second hypothesis that the prevalence of right IJV non-patency at various segments during the VM would be significantly higher in the study subjects with IJV stenosis/compression at various segments; and therefore, would be higher in the TGA patients than that in the controls, which supports our novel hypothesis of venous pathogenesis involved in the TGA attacks (16). Specifically, an insufficient IJV patency prevents the release of increased intracranial pressure and venous congestion/hypertension in the basilar plexus and cavernous sinus caused by VM-like maneuvers. Moreover, venous stasis and occlusion may cause constriction of cerebral arterioles (16), which further compromises cerebral hemodynamics.

### Venous Flow Velocity and FV in IJVs

This study, our previous study (20), and other studies (22, 23) all revealed that the IJV or BCV compression and stenosis significantly reduce the IJV FVs. These result in the venous drainage being routed through less efficient alternative routes, such as the tortuous path through the spinovertebral venous plexus. This obstruction of the venous drainage may also induce changes in the arterial blood flow, such as arterial constriction, through the venoarterial reflex (25). Such changes may explain the observation in our previous studies (21, 24)


TABLE 2 | Comparisons of the TAMVs, FVs, and CSAs of the IJVs between the age- and sex-matched TGA patients and control subjects.

CSA, Cross-Section Lumen Area; FV, Flow Volume; IJV, Internal Jugular Vein; TAMV, Time-Averaged Mean Velocity; TGA, Transient Global Amnesia; VV, Vertebral Vein; \*p < 0.05: when comparing the TGA patients and controls (ipsilateral IJV); \*\*p < 0.01: when comparing the TGA patients and controls (ipsilateral IJV).

TABLE 3 | The prevalence of no-reflow in the IJVs with and without venous compression/stenosis during the VM in the study subjects with complete MR venous examination.


MR, Magnetic Resonance; BCV, Brachiocephalic Vein; IJV, Internal Jugular Vein; TS, Transverse Sinus; VM, Valsalva Maneuver; \*p = 0.0425; \*\*p = 0.021; #p = 0.0111.

that, the patients with transient monocular blindness without carotid stenosis exhibited increased downstream resistance of the retrobulbar arteries (i.e., ophthalmic artery, posterior ciliary artery, and central artery) in association with significantly increased prevalence of compression/stenosis in the bilateral IJVs (26). In this study, we did not measure the venous FV in the spinovertebral venous plexus, because it is undetectable by the ultrasound; therefore, we do not know the changes in the total venous drainage from the bilateral TSs, though the patients exhibited significantly lower total FV in the bilateral IJVs and VVs.

### Prevalence of IJV Non-patency During the VM

Our results supported our second hypothesis that the prevalence of IJV non-patency during the VM would be higher in the TGA patients than that in the controls. We observed a significant difference in the prevalence of right IJV non-patency (patients: 32.1%; controls: 11.6%; p = 0.0128) as compared to that for TABLE 4 | The prevalence of no-reflow in the IJVs during the VM in all the 79 TGA patients and 45 controls among the study subjects with complete MR venous examination.


IJV, Internal Jugular Vein; TGA, Transient Global Amnesia; VM, Valsalva Maneuver; MR, Magnetic Resonance; \*p = 0.0128.

the left IJV non- patency (patients: 49.4%; controls: 37.2%). An abundance of evidence indicates that VM-induced pressure from the chest and abdomen is mainly transmitted to the intracranial compartment via the epidural venous plexus or vertebral venous plexus. Orthograde IJV outflows emerging shortly after the beginning of the VM, thus, serve as a mechanism for regulating the intracranial pressure and equalizing the pressure within the venous system (10, 11, 14, 16). Our findings indicate that the patients with TGA may experience defective intracranial pressure regulation during the VM-like movements. As described earlier, we restricted our definition of IJV non-patency to complete absence of the IJV drainage at the J3 or J2 segment during the first 10 s of the VM; but, this consideration regarding the IJV drainage might be unexhaustive. Several patients with partial or limited IJV drainage were excluded from this definition, which may explain why we observed a lower prevalence of IJV nonpatency during the VM. Further research is needed to develop a more sensitive and specific definition of IJV non-patency during the VM.

#### Incompetence of Jugular Venous Valves

Cerebral venous congestion/hypertension, which results from the venous reflux during the VM consequent to the IJVVI, is linked to TGA (4–6). However, previous ultrasound studies using either retrograde flow (27) or air bubbles (28) during the VM explained the involvement of only the proximal region of the IJV in the IJVVI, ignoring the rest of the IJV and the entire BCV, and possibly missing other important IJV/BCV abnormalities. Unsurprisingly, previous ultrasound and non-contrast MRA results have not supported a causal relationship between the IJVVI and TGA (7–9). Other than the IJVVI, we have previously described three ultrasound patterns of IJV abnormalities in the patients with TGA: (i) an isolated reverse flow in the left jugular vein branch (JB), (ii) a segmental reverse flow in the left distal IJV, and (iii) a continuous reverse flow in the left IJV and JB (29). All the three of these IJV patterns are suggestive of the venous outflow obstruction or compression/stenosis of the left BCV (29). Similar to other studies (4–6), even in this study, the overall prevalence of IJVVI was higher in the TGA patients than that in the controls (**Table 2**); but, the prevalence was higher in the left IJV only, and not in the right IJV. Furthermore, we found that the prevalence of IJVVI was significantly higher in the individuals with left BCV compression/stenosis. This suggests that the IJVVI might occur secondary to the BCV compression/stenosis on the left side. The fact that the IJVVI was more frequently observed on the right side, as reported in other studies (28), raises the question of whether the jugular venous valves are vulnerable in cases of IJV compression/stenosis due to pressure imbalances across the valves or whether the previous air bubble methodology overestimated the IJVVI prevalence (30). Further research is needed to address this question.

# Assessments of IJV Compression With Different Imaging Modalities

It is worth mentioning that despite the patients exhibiting reduced flow velocities and FVs in each of the segments of the bilateral IJVs, the diameters of their IJV segments were comparable to that of the controls (**Table 2**). Moreover, the patients' right IJVs were slightly wider than that of the controls. This may be partially explained by the Bernoulli's equation, which states that the pressure in the venous lumen is inversely proportional to the flow rate, so that the lumen diameter may be enlarged with flow stasis resulting from the venous stenosis/compression. As described earlier, the FV of the IJV decreases in cases of IJV or BCV stenosis/compression (20, 22, 23). However, discrepancies in the diagnosis of IJV compression/stenosis may be observed due to different types of venous examination (31). Catheter venography has been traditionally regarded as the gold standard for diagnosing venous disorders involving compression/stenosis, but it does not measure the flow velocity or FV; hence, it cannot prove or disprove the MRI or ultrasound findings that indicate decreased IJV FV resulting from compression/stenosis.

#### Study Limitations

This study has several limitations, particularly regarding the use of ultrasound to study the IJV. First, there are no ideal, fixed locations for measuring the CSA and TAMV in the J2 and J3 segments; and these measurements may vary if the IJVs are non-uniform in diameter, which may occur in the cases where the IJV is affected by segmental dilatation, narrowing, or compression. However, the FV was calculated by multiplying the TAMV by the CSA, and it was theoretically correct according to the Bernoulli's equation. We measured the CSA and TAMV at the widest available lumen of the J2 and J3 segments to minimize the bias. This allowed us to detect the differences in both the FVs and TAMVs between the patients and controls. Second, it is unexhaustive to consider complete absence of IJV drainage during the VM while explaining IJV non-patency. Several of our subjects exhibited limited or intermittent flow, which suggests partial IJV blockage. However, we excluded these subjects from our definition of IJV non-patency, which may have caused us to underestimate the prevalence of IJV non-patency during the VM. Third, left BCV blockage may disappear during deep inspiration (15), which usually precedes the VM; hence, paradoxical reopening of the left BCV and left IJV occurs during the VM. This causes further underestimation of the prevalence of non-patency.

# CONCLUSION

Our results further confirmed that decrease in the total flow of the IJVs and VVs reflect impaired venous drainage in the patients with TGA, which is consistent with the findings of previous MRI studies that reported about the compression/stenosis of the bilateral IJVs and the left BCV. Second, our findings support the hypothesis that the compression/stenosis of the bilateral IJVs and left BCV may block the IJV drainage during the VM, which may prevent the release of intracranial hypertension caused by the VM. Further study is needed to obtain a more sensitive and specific definition of IJV non-patency during the VM.

# AUTHOR CONTRIBUTIONS

H-HH and JW conceived and designed the experiments. KH, A-CC, F-CC, C-PC, and H-YH performed the experiments. KH, W-YS, and H-HH analyzed the data. A-CC, F-CC, C-PC, and H-YH contributed materials/analysis tools. KH and H-HH wrote the paper.

# FUNDING

This work was supported by the Taiwan National Science Council Research Grant (NSC101-2314-B-037-069-MY2) and KMUH 104-4R53 to A-CC. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

# ACKNOWLEDGMENTS

We are thankful to all the individuals who participated in the study.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fneur. 2019.00067/full#supplementary-material

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Han, Hu, Chao, Chang, Chung, Hsu, Sheng and Wu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Neuroimaging Marker Based on Diffusion Tensor Imaging and Cognitive Impairment Due to Cerebral White Matter Lesions

Na Wei 1,2,3,4, Yiming Deng2,3,5, Li Yao<sup>6</sup> , Weili Jia<sup>1</sup> , Jinfang Wang1,7, Qingli Shi <sup>1</sup> , Hongyan Chen<sup>8</sup> , Yuesong Pan<sup>2</sup> , Hongyi Yan<sup>2</sup> , Yumei Zhang1,2,3,4 \* and Yongjun Wang1,2,3,4 \*

*<sup>1</sup> Department of Neurology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>2</sup> China National Clinical Research Center for Neurological Diseases, Beijing, China, <sup>3</sup> Center of Stroke, Beijing Institute for Brain Disorders, Beijing, China, <sup>4</sup> Beijing Key Laboratory of Translational Medicine for Cerebrovascular Disease, Beijing, China, <sup>5</sup> Department of Interventional Neuroradiology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, <sup>6</sup> College of Information Science and Technology, Beijing Normal University, Beijing, China, <sup>7</sup> Department of Neurology, General Hospital of The Yang Tze River Shipping, Wuhan Brain Hospital, Wuhan, China, <sup>8</sup> Department of Neuroimaging, Beijing Neurosurgery Institute, Capital Medical University, Beijing, China*

#### Edited by:

*Yi Yang, First Affiliated Hospital of Jilin University, China*

#### Reviewed by:

*Liqun Zhang, St George's Hospital, United Kingdom Xianzeng Liu, Peking University People's Hospital, China*

#### \*Correspondence:

*Yumei Zhang zhangyumei95@163.com Yongjun Wang yongjunwang@ncrcnd.org.cn*

#### Specialty section:

*This article was submitted to Applied Neuroimaging, a section of the journal Frontiers in Neurology*

Received: *06 September 2018* Accepted: *21 January 2019* Published: *13 February 2019*

#### Citation:

*Wei N, Deng Y, Yao L, Jia W, Wang J, Shi Q, Chen H, Pan Y, Yan H, Zhang Y and Wang Y (2019) A Neuroimaging Marker Based on Diffusion Tensor Imaging and Cognitive Impairment Due to Cerebral White Matter Lesions. Front. Neurol. 10:81. doi: 10.3389/fneur.2019.00081* Background: The peak width of skeletonized mean diffusivity (PSMD) is a new, fully automated, robust imaging marker for cerebral small vessel disease (SVD), strongly associated with processing speed. However, it has never been applied to cerebral white matter lesions (WMLs). Our study aimed to investigate the correlation between PSMD and cognition, particularly in the executive function of patients with WMLs.

Methods: A total of 111 WML patients and 50 healthy controls (HCs) were enrolled, and their demographic information and cardiovascular disease risk factors were recorded. Subjects were divided into three groups: WMLs with normal cognition (WMLs-NC), WMLs with vascular cognitive impairment (WMLs-VCI), and HCs. They underwent conventional head magnetic resonance imaging and diffusion tensor imaging (DTI), followed by neuropsychological and psychological examinations, including the Montreal Cognitive Assessment (MoCA), and the executive function tests. We compared executive function and PSMD among the three groups and analyzed the correlation between PSMD and cognitive function in all subjects.

Results: There were no significant differences in demographic characteristics (age, sex, education level, and cardiovascular disease risk factors) among the three groups (*P* > 0.05), but there were significant differences in global cognition (*P* < 0.0001), executive function (*P* < 0.0001), and PSMD (*P* < 0.0001). The average PSMD value (×10−<sup>4</sup> mm<sup>2</sup> /s) was 2.40 ± 0.23, 2.68 ± 0.30, and 4.51 ± 0.39 in the HC, WMLs-NC, and WMLs-VCI groups, respectively. There was no correlation between PSMD and cognition in the HC group, but PSMD was significantly correlated with MoCA scores (*r* = −0.3785, *P* < 0.0001) and executive function (*r* = −0.4744, *P* < 0.0001) in the WMLs-NC group and in the WMLs-VCI group (*r* = −0.4448, *P* < 0.0001 and *r* = −0.6279, *P* < 0.0001, respectively).

**97**

Conclusions: WML patients have higher PSMD and worse cognitive performance than HCs, and PSMD is strongly associated with global cognition and executive functions in WML patients. This result provides new insights into the pathophysiology of cognitive impairment in WML patients. PSMD could be a surrogate marker for disease progression and could thus be used in therapeutic trials involving WML patients.

Keywords: white matter lesions, vascular cognitive impairment, magnetic resonance imaging, diffusion tensor imaging, white matter structural integrity

## INTRODUCTION

On computed tomography (CT), cerebral white matter lesions (WMLs) appear as hypodense bilateral and symmetrical areas in the WM of the periventricular region and centrum semiovale (1) and are indicators of cerebral small vessel disease (SVD). The Leukoaraiosis And Disability (LADIS) study confirmed a significant impairment of cognitive function in WML patients (2–4). WMLs are closely correlated to cognitive impairments in attention, executive function, and information processing speed (5). However, some patients may have severe cognitive dysfunction in the absence of widespread WMLs on magnetic resonance imaging (MRI). One of the factors may be the loss of microstructure integrity in the largest part of white matter, which can be visualized with conventional MRI, but can be investigated with DTI.

DTI is a sensitive technique for evaluating disease progression, allowing the quantification of microstructural tissue changes (6). The typical diffusion change pattern in WMLs consists of a decrease in fractional anisotropy (FA) and an increase in mean diffusivity (MD). However, the reliability in multicenter and long-term studies seems questionable. In addition, there are some limitations to the wide application of DTI measures due to the large amount of data postprocessing and the subjective operation errors in methods based on brain regions of interest.

Recently, Baykara et al. (7) proposed the assessment of SVD through DTI parameters which use a WM skeleton to measure the peak width of mean diffusivity in the brain. The new imaging marker was called peak width of skeletonized mean diffusivity (PSMD). Calculations of this marker appear to be robust and promising for studies of large populations. The derived measures were strongly correlated with processing speed and performed better than other neuroimaging markers for SVD, such as WM hyperintensities (WMHs) and the numbers and volumes of lacunes. The method eliminated cerebrospinal fluid contamination and increased the sensitivity in capturing SVDrelated changes. A longitudinal analysis revealed the smallest sample size estimate for PSMD when compared with whole brain mean diffusivity peak height, normalized WMH volume, brain parenchymal fraction, processing speed score, or normalized lacune volume. PSMD may thus have a great practical value for clinical research and applications. However, no study has assessed the relationship between PSMD and cognitive function in patients with WMLs, yet.

In the present study, we aimed to examine the relationship between whole WM microstructural integrity, as assessed by PSMD, and cognitive function, in patients with WMLs. We hypothesized that WML patients would show poorer cognitive performance and a higher PSMD than HCs. Further, we aimed to assess the correlation between this new DTI marker and cognition, particularly executive function, in patients with WMLs.

# MATERIALS AND METHODS

#### Subjects

WML patients were recruited from the neurology clinic of the Beijing Tiantan Hospital, Capital Medical University, China, between January 2014 to March 2017. WML was diagnosed independently and unanimously by two radiologists, who visually evaluated the fluid-attenuated inversion recovery (FLAIR) MR images without knowledge of the participants' clinical profiles. The inclusion criteria for WML patients were as follows: (a) age 50–85 years and (b) presence of WMHs on MRI scans, according to a revised version of the Fazekas scale (8). The exclusion criteria were as follows: (a) cardiac or renal failure, cancer, or other severe systemic diseases; (b) unrelated neurological diseases such as epilepsy, traumatic brain injury, or multiple sclerosis; (c) chronic cerebral infarction or other lesions; (d) leukoencephalopathy of non-vascular origin; (e) dementia of non-vascular origin; (f) psychiatric diseases or drug addiction; (g) consciousness disruption or aphasia; or (h) inability or refusal to undergo brain MRI. Initially, 113 WML patients were enrolled. In addition, 48 age-, sex-, and education level-matched normal volunteers were recruited as control subjects. Their age ranged between 50 and 85 years, and their MRI results were normal. The exclusion criteria for the healthy controls were the same as those for the WML patients.

All subjects were administered the Beijing version of the MoCA (9) and the Clinical Dementia Rating (CDR) scale under the supervision of a physician. Based on the results of these cognitive tests, the subjects were divided into three groups: (a) WML patients with normal cognition (WMLs-NC), defined as MoCA ≥ 26 and CDR = 0; (b) WML patients with vascular cognitive impairment (WMLs-VCI), defined as MoCA < 26 and CDR > 0; and (c) healthy controls (HC), defined as MoCA ≥ 26 and CDR = 0.

This study was approved by the Ethics Committee of the Beijing Tian Tan Hospital. All patients or their legal representatives provided written informed consent.

#### MRI Scanning Protocol

MRI scans of all participants were performed using a 3.0T Signa scanner (Magnetom Trio Tim, Siemens, Germany). The general MRI protocol included the following sequences: T1 weighted 3-dimensional magnetization prepared rapid gradient echo (MPRAGE) sequence (TR/TE/TI 2300/3.28/1200 ms; flip angle 9◦ ; voxel size 1.0 × 1.0 × 1.0 mm), a FLAIR sequence (TR/TE/TI 8000/94/2200 ms; voxel size 1.0 × 1.2 × 5.0 mm, interslice gap 1 mm), and DTI sequences (TR/TE 4900/93 ms; voxel size 2.5 × 2.5 × 2.5 mm; 4 unweighted scans, 30 directions with b-value 1,000 s/mm<sup>2</sup> ). Two radiologists blinded to the clinical information assessed the MRI data.

#### Processing of PSMD

DTI data were corrected for quick visual inspection to exclude large artifacts. The fully automated calculation of the new marker consists of two steps: Skeletonization of the DTI data and histogram analysis. All study samples were processed with the same pipeline. First, DTI data were skeletonized using the Tract-Based Spatial Statistics procedure included in the Functional Magnetic Resonance Imaging of the Brain (FMRIB) software library (FSL) (10). The fractional anisotropy (FA) data of each subject were projected onto a skeleton derived from a standard space template. The mean diffusivity image was then projected onto the skeleton using FA-derived projection parameters. This process avoids contamination of the skeleton through partial volume effects of cerebrospinal fluid and fornix. The fully automated PSMD calculation pipeline is available at http://www. psmd-marker.com/ as a shell script including all processing steps (including pre-processing). No further human intervention is required during the processing pipeline.

#### Neuropsychological Testing

Neuropsychological assessment followed the LADIS protocol (11). In the test battery, the MoCA was considered to be a measure of global cognitive function. Executive function was assessed by computing compound measures from the Stroop color and word test (SCWT), trail-making test (TMT), symboldigital replacement task (DST), and verbal fluency test (VFT).

#### Statistical Analyses

SPSS 23.0 was used for data processing, and SAS 9.4 for statistical analysis. To allow direct comparisons between the imaging marker and neuropsychological tests results, we generated z scores, representing the position of a score value within the score distribution. Executive functions = z scores of ((Stroop3-2) + (TMB-TMA) + symbol digit + verbal fluency)/4.The sign of each z-score was changed if necessary, to make positive scores correspond to better performance. The results of the MoCA were also transformed into z scores.

The numerical variables were reported as mean ± standard deviation (SD), and as median and interquartile ranges for parameters with skewed distributions. Normally distributed continuous variables were compared by one-way analysis of variance(ANOVA), and the Kruskal-Wallis test was used to compare non-normally distributed variables. A chi squared test was used to compare categorical variables. Multivariate regression analysis was used to assess the relative contribution of PSMD measures to performance in different cognitive domains. Regression analysis was performed with two levels of adjustment for covariates: Model 1 adjusted for age, sex, and level of education; model 2 extended model 1 by the addition of hypertension, diabetes, hyperlipidemia, coronary heart disease, smoking status, drinking, and BMI. Furthermore, using Pearson correlation analysis, we examined the association between PSMD value and cognitive functions. P-values <0.05 were considered statistically significant.

# RESULTS

A total of 163 subjects were enrolled in this study. Among them, there were 35 and 78 patients in the WMLs-NC and WMLs-VCI groups, respectively, and 48 healthy controls. The baseline characteristics of all subjects are listed in **Table 1**. There were no statistically significant differences in age, sex, years of education, or incidence of cerebral vessel risk factors in terms of hypertension, hyperlipidemia, coronary heart disease, smoking, drinking, and BMI among the three groups (P < 0.05).

As shown in **Table 2**, there were significant differences in cognitive status among the three groups, as measured by the MoCA z score (P < 0.0001), and in executive functions (P < 0.0001). There were significant differences in PSMD among the three groups (P < 0.0001). The mean PSMD values were 2.40 ± 0.23 × 10−<sup>4</sup> mm<sup>2</sup> /s, 2.68 ± 0.30 × 10−<sup>4</sup> mm<sup>2</sup> /s, and 4.51 ± 0.39 × 10−<sup>4</sup> mm<sup>2</sup> /s in the healthy controls, WMLs-NC, and WMLs-VCI, respectively.

As seen in **Table 3**, when assessing the relationship between PSMD and global cognitive function, no significant associations were seen with test results in the HC or WMLs-NC groups after correction for age, sex, and education level (model 1) or after adjustment for hypertension, diabetes, hyperlipidemia, coronary heart disease, smoking status, drinking, and BMI (model 2). However, the association with global cognitive function in the WMLs-VCI group was significant in the fully adjusted model (ß=-0.513; standard error [SE] = 0.091; P < 0.001). When assessing the relationship between PSMD and executive function, the association remained significant in both model 1 and model 2. In the latter model the coefficients were: HCs: ß = −2.155; SE = 0.714; P = 0.005; WMLs-NC: ß = −1.629; SE = 0.741; P = 0.039; and WMLs-VCI: ß = −0.372; SE = 0.107; P < 0.001.

As seen in **Table 4**, there were no significant associations between PSMD and cognitive performance in the healthy control samples (P = 0.56–0.88). Negative correlation was found between PSMD and global and executive function in the WMLs-CN samples, with correlation coefficients −0.3785 and −0.4744, respectively, and also in the WMLs-VCI samples, with correlation coefficients −0.4448 and −0.6279, respectively.

The correlation analysis between cognitive performance and PSMD in healthy controls and WML patients is shown in **Figure 1**. Linear regression showed no significant association between PSMD and MoCA scores (**Figure 1A**) or executive function (**Figure 1B**) in the healthy controls. The WMLs-CN group showed associations between PSMD and MoCA

#### TABLE 1 | Characteristics of the study population.


*Categorical variables are expressed as number (percentage), and continuous ones as mean* ± *SD. HC, healthy controls; WMLs-CN, cognitively normal white matter lesion patients; WMLs-VCI, white matter lesion patients with vascular cognitive impairment; BMI, body mass index. <sup>a</sup>The P-value was obtained by chi squared (*χ *2 ) test. <sup>b</sup>The P-value was obtained by ANOVA.*

TABLE 2 | Cognitive function measures and the peak width of skeletonized mean diffusivity in the study population.


*Variables and z scores are shown as mean* ± *SD. HC, healthy controls; WMLs-CN, cognitively normal white matter lesion patients; WMLs-VCI, white matter lesion patients with vascular cognitive impairment; MoCA, Montreal Cognitive Assessment; SCWT, Stroop color and word test; TMT, trail-making test; DST, symbol-digital replacement task, VFT, verbal fluency test; PSMD, peak width of skeletonized mean diffusivity.*

scores (**Figure 1C**) as well as executive function (**Figure 1D**). The WMLs-VCI group showed strong associations between PSMD and MoCA scores (**Figure 1E**) as well as executive function (**Figure 1F**).

#### DISCUSSION

In this study, we analyzed PSMD and cognitive function in HC subjects and WMLs patients, with or without cognitive impairment. We also investigated the relationship between PSMD and cognitive functions.

There were no significant differences in age, sex, years of education, and vascular risk factors between the WML patients and the healthy controls, largely eliminating the influence of possible confounders on cognitive assessment. Extensive neuropsychological assessment was performed by two investigators, and multivariate regression analysis included adjustment for potential confounders. Global cognitive function was measured using MoCA scores, which have been proposed as a screening tool for vascular cognitive impairment (12). Pasi et al. demonstrated that DTI-measured WM microstructural damage is more related to MoCA results than to mini mental state examination performances in SVD patients (13), indicating that MoCA is suitable for the cognitive screening of patients with small vessel disease. Yuan et al. demonstrated that the cognitive domains affected in patients with WMLs were attention, executive function, and information processing speed (2012).


TABLE 3 | Associations between the peak width of skeletonized mean diffusivity and cognitive performance.

*Model 1: adjusted for age, sex, and level of education. Model 2: same as model 1, additionally adjusted for hypertension, diabetes, hyperlipidemia, coronary heart disease, smoking status, drinking and BMI. SE, standard error.*

TABLE 4 | Correlation between cognitive function and the peak width of skeletonized mean diffusivity.


*HC, health controls; WMLs-CN, cognitively normal white matter lesion patients; WMLs-VCI, white matter lesion patients with vascular cognitive impairment. r, Pearson's correlation coefficient.*

The loss of memory is not common in patients with cognitive impairment due to WMLs, which is one of the differences between vascular dementia and Alzheimer's disease (14).

In our study, executive function was assessed by computing compound measures from the SCWT, TMT, DST, and VFT, which examine the cognitive domain of psychomotor speed, fluency, concept shifting, and attention. We found that the global cognitive and executive functions of WML patients were significantly worse than those of healthy subjects. Further, WMLs-NC patients had better executive function performance than WMLs-VCI patients. These findings are consistent with those of other studies (15).

The characteristics of cognitive impairment in individuals with WMLs depend on the location, degree, and size of the lesions. Diffusion tensor tractography study showed that the location of WMLs was related to the damaged cognitive domain (2). As cognitive disturbances in subjects with cerebral small vessel disease are related to microstructural integrity of multiple WM fibers (within WMH and normal-appearing WM) connecting different cortical and subcortical regions, we examined the relation between the microstructural integrity of the whole WM and cognitive performance in subjects with WMLs. In addition, whole brain histogram analysis is particularly appropriate when quantifying total disease burden (16). In this study, we measured PSMD as an imaging marker to measure the microstructural integrity of the whole WM in HC subjects and WML patients. Our results showed that WML patients had higher PSMD values than control subjects. Moreover, we found that in WML patients the severity of cognitive impairment increased with PSMD.

DTI measures are more sensitive than conventional MRI markers in capturing changes associated with SVD. Histograms of MR parameter values measured in the whole brain are increasingly being used to characterize subtle disease that affects large parts of the brain. Studies have shown that histogram peak height measures were associated with cognitive function and can capture disease burden in SVD (17, 18). PSMD is a combination of DTI, skeletonization and histogram analysis of WM tracks, and is therefore superior in assessing the burden of disease (19).

In this study, we examined the relationship between PSMD and cognitive performance in the three groups. We found that there were no significant associations between PSMD and cognition performance in the healthy controls. However, in the WML group, PSMD was associated with executive dysfunction, a pattern that has been associated with SVD (20). The associations were also reflected in the MoCA scores, which give a global measure of cognitive function. Our study found a clear inverse relationship between PSMD and cognition: High PSMD values were associated with lower scores for cognitive functions, and especially for executive performance. As PSMD is a wholebrain measure computed from DTI scans, but not a local estimate of possible changes in the microstructure within the brain, subjects with the same WMH loads may show different cognitive performances. We demonstrated that PSMD was highly correlated with executive function in WML patients. Moreover, the correlation was more significant in WML patients with cognitive impairment. This imaging marker is therefore highly sensitive to vascular cognitive impairment and could therefore be used in addition to conventional MRI to investigate cognitive dysfunction.

A major strength of this study is that all subjects were assessed by multiple MRI sequences, including 3D T1, FLAIR, and 30-direction DTI acquisition. Therefore, accurate and comprehensive original image data were acquired. Another strength is the use of novel imaging techniques: Image data are processed by the online scanning software. All the processing steps are simple and fully automated, without any manual intervention. Furthermore, our study is a single-center study, with all subjects examined by only two investigators, using manual segmentation of the WML, without prior knowledge of the clinical data.

Our study also had limitations. First, the study was hospitalbased, and patients who did not meet the inclusion criteria were excluded. This may result in selection bias and influence the measurement of cognitive function in WML patients. Second, PSMD mostly reflects SVD-related alterations or primary neurodegenerative pathology, and the subjects should be grouped based on the degree of WM damage, but this could not be done in the present study. In future studies, it is desirable to combine the measurements of cognitive impairment and WMH load with PSMD, in order to elucidate the correlations between brain microstructural integrity and cognitive function associated with WMLs. Third, our study is cross-sectional, and the causal interpretation of the results is limited. Future studies with larger sample sizes may further clarify the interactions between cognitive impairment and brain microstructural characteristics.

To summarize, the present study comprehensively investigated the characteristics of PSMD in WML patients and compared them with those of HC subjects. We demonstrated that PSMD is significantly correlated with cognitive impairment in WML patients. Our findings suggest that microstructural integrity of the whole WM should be considered when investigating the relationship between WMLs and cognitive function. PSMD could therefore serve as an addition to a conventional MRI in order to investigate cognitive dysfunction. This result provides new insights into the pathophysiology of cognitive impairment in WML patients. PSMD could be a surrogate marker for disease progression and could thus be used in therapeutic trials involving WML patients.

## AUTHOR CONTRIBUTIONS

NW and YZ designed the study and drafted the manuscript. YD, WJ, QS, and JW performed the measurements and collected the data. HC performed the brain MRI data acquisition. LY participated in the brain imaging data analysis. YP and HY performed the statistical analysis. YZ and YW participated in the critical discussion of the manuscript.

# REFERENCES


### FUNDING

This study was funded by the Beijing Municipal Administration of Hospitals Clinical Medicine Development of Special Funding Support (ZYLX201836), the National Natural Science Foundation of China (NSFC: 81371201), the National Key Research and Development Plan (2018YFC2002302), the National Key Technology Research and Development Program of the Ministry of Science and Technology of The People's Republic of China (2015BAI12B02), Beijing Institute For Brain Disorders (1152130306), grants (D151100002015003/D131100002313002) from the Beijing Municipal Science & Technology Commission, and Beijing Municipal Administration of Hospitals' Youth Programme (QML20180506).

# ACKNOWLEDGMENTS

We acknowledge all the participants, colleagues, nurses, and imaging technicians.

attack and stroke a population-based study. Stroke (2010) 41"1290–3. doi: 10.1161/STROKEAHA.110.579888


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wei, Deng, Yao, Jia, Wang, Shi, Chen, Pan, Yan, Zhang and Wang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effect of Inflammation on the Process of Stroke Rehabilitation and Poststroke Depression

Meidan Fang<sup>1</sup> , Lili Zhong<sup>2</sup> , Xin Jin<sup>3</sup> , Ranji Cui <sup>2</sup> , Wei Yang<sup>2</sup> , Shuohui Gao<sup>4</sup> \*, Jing Lv <sup>5</sup> \*, Bingjin Li <sup>2</sup> \* and Tongjun Liu<sup>1</sup> \*

<sup>1</sup> Department of General Surgery, Second Hospital of Jilin University, Changchun, China, <sup>2</sup> Jilin Provincial Key Laboratory on Molecular and Chemical Genetic, Second Hospital of Jilin University, Changchun, China, <sup>3</sup> Department of Oncology and Hematology, Second Hospital of Jilin University, Changchun, China, <sup>4</sup> Department of Gastrointestinal Colorectal Surgery, China-Japan Union Hospital of Jilin University, Changchun, China, <sup>5</sup> Chang Chun University of Chinese Medicine, Changchun, China

A considerable body of evidence has shown that inflammation plays an important role in the process of stroke rehabilitation and development of poststroke depression (PSD). However, the specific molecular and cellular mechanisms involved remain unclear. In this review, we summarize how neuroinflammation affects stroke rehabilitation and PSD. We mainly focus on the immune/inflammatory response, involving astrocytes, microglia, monocyte-derived macrophages, cytokines (tumor necrosis factor alpha, interleukin 1), and microRNAs (microRNA-124, microRNA 133b). This review provides new insights into the effect of inflammation on the process of stroke rehabilitation and PSD and potentially offer new therapeutic targets of stroke and PSD.

Keywords: stroke, immune/inflammation, rehabilitation, poststroke depression, pharmacotherapy

# INTRODUCTION

Stroke is defined as permanent tissue damage caused by a sudden loss of brain blood supply as a result of occlusion or a hemorrhage. Stroke includes two main types, ischemic stroke and intracerebral hemorrhage (ICH). Approximately 85% of strokes belong to the ischemic type and 12% are ICHs (3% are subarachnoid hemorrhage). Neural plasticity can be affected by different risk factors of stroke, medical management, and anti-inflammatory interventions during the process of stroke rehabilitation.

The immune/inflammatory response can be triggered by several factors, such as ischemia or hemorrhage. Microglia, astrocytes, and endotheliocytes are involved in immune/inflammatory activation induced by ischemic stroke. These cells can communicate with each other by proinflammatory and anti-inflammatory factors, such as cytokines and adhesion molecules. Inflammatory cells, such as neutrophils and macrophages, are activated, reach the ischemic area, and contribute to the inflammatory response (1). These immune responses following the initial ischemic insult can be long lasting and subsequently modulate synaptic plasticity alterations during the process of stroke rehabilitation (2). In ICH, ambient microglia and astrocytes can also exert modulatory effects during ICH rehabilitation (3).

Poststroke depression (PSD), a critical psychiatric complication of stroke, involves several major symptoms including sleep and appetite disturbance, psychomotor agitation, and fatigue (4). As the inflammatory response may modulate neuroplasticity during stroke and altered neuroplasticity may be associated with the onset of PSD, the stroke-induced immune response in the brain can also affect the PSD process. It was found that several inflammatory markers, pro-inflammatory cytokines, and the pro-inflammatory/anti-inflammatory ratios were increased and the complementary expression was reduced in the PSD process (5).

#### Edited by:

Yuan-Pang Wang, University of São Paulo, Brazil

#### Reviewed by:

Chengqi Xu, Huazhong University of Science and Technology, China Weili Zhang, Fuwai Hospital (CAMS), China

#### \*Correspondence:

Shuohui Gao gaoshuohui@foxmail.com Jing Lv lvjing1267@163.com Bingjin Li libingjin@jlu.edu.cn Tongjun Liu tongjunliu@163.com

#### Specialty section:

This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry

Received: 30 March 2018 Accepted: 13 March 2019 Published: 11 April 2019

#### Citation:

Fang M, Zhong L, Jin X, Cui R, Yang W, Gao S, Lv J, Li B and Liu T (2019) Effect of Inflammation on the Process of Stroke Rehabilitation and Poststroke Depression. Front. Psychiatry 10:184. doi: 10.3389/fpsyt.2019.00184

In this review, we describe the two different types of stroke (ischemic stroke and ICH) and respectively summarize the effects of inflammation on the process of stroke rehabilitation. We then describe PSD and summarize the effects of inflammation on PSD. Finally, we discuss the potential efficacy of using anti-inflammatory medication for stroke rehabilitation and PSD prevention.

# STROKE

Stroke refers to several conditions caused by occlusion or a hemorrhage of the brain blood vessels. Stroke is a worldwide neurological disease with few effective treatments and preventative measures (6). Neuroinflammation involves damage-associated molecular patterns, instead of microbial pathogens (7). Importantly, neuroinflammation plays a key role in several neurological diseases such as a hemorrhage and ischemia (8). There are certain complicated connections between immune/inflammatory processes and stroke rehabilitation.

# Ischemic Stroke

Ischemic stroke, the most common type of brain ischemic injury in humans, is the leading cause of mortality and long-term disability (9). Ischemic stroke is mainly caused by an ischemic core induced by brain artery occlusion, surrounded neuronal loss, and glial scarring (10). In ischemic stroke, the most relevant inflammatory-cellular component is the microglial and astrocytic responses, chemokines and cytokines, and infiltrating peripheral blood cells (9).

There are two types of models in ischemic stroke: the middle cerebral artery (MCA) occlusion model and the photothrombotic MCA stroke model. The photothrombotic MCA stroke model is created by a laser beam irradiating a photosensitizing dye in the MCA. The latter can slowly substitute the former because of the ease of application and reproducibility of the model. However, the latter delays microglial and astrocytic invasion of the ischemic core but elevates the levels of inflammatory cytokines or chemokines and their infiltration from the circulatory system (11). In an experimental striatal stroke model induced by endothelin-1, focal ischemic neuronal loss appeared, with intense microglia activation in 3–14 postlesion days (maximum at 7 postlesion days). Astrocytosis was also maximal at 7 postlesion days (12). In ischemic brains, local inflammation involves astrocytes, activated resident microglia, and infiltrating monocytes or monocyte-derived macrophages (MDMs), with upregulated expression of proinflammatory factors [interleukin (IL)-6, nitric oxide synthase-2, IL-1β, tumor necrosis factor alpha (TNF-α)] and anti-inflammatory factors (CCL22,Ym1, CXCL13,TGFβ,CD163) (13).

#### Astrocytes

Astrocytes are the largest specialized cells in the central nervous system (CNS). Astrocytes play a significant role in neural development and neuroprotection via supporting synaptic connections, ionic homeostasis, and glutamate clearance. It is considered that astrocytes are involved in the local inflammatory response via modulating proinflammatory and anti-inflammatory cytokines (14–16). It was found that astrocytes can enhance neuronal viability through the uptake of glutamate and the release of neurotrophins; astrocytes also compromise neuronal viability by producing inflammatory cytokines or releasing glutamate, and contribute to angiogenesis and neuronal plasticity several days after stroke (17).

The ring- or crescent- shaped "peri-infarct" form is mainly localized around the infarct region and significantly grows after stroke. Microglia and macrophages are mainly localized in the lesion infarct core, rather than in the infarct region (18). One of the pathological alterations of the infarct region is reactive astrogliosis and the formation of glial scarring. Astrocytes in the "peri-infarct" region respond adaptively to stroke, which is known as reactive astrogliosis. Astrocytes can proliferate and be centrally involved in glial scar formation in the "peri-infarct" region, which separates the damaged infarct tissue from the normal tissue. The intertwined connection of astrocytes in and around the infarct region forms the mature glial scar and impedes neuronal rehabilitation after stroke. Early dysfunction and subsequent function recovery after stroke, through the destruction and remodeling of intertwined connection around the infarct region, is associated with neuroinflammation (18, 19).

Reactive astrogliosis and glial scar formation after stroke is considerable during the rehabilitation process, with a change in gene expression, morphology, and proliferation of reactive astrocytes (14, 16). In addition, the main characteristic of astrogliosis is hypertrophic astrocytes with a high expression of proinflammatory cytokines, neurotrophic factors, and neuronal and proliferation markers (8). As major components of the neuroinflammatory process after ischemic stroke, reactive astrocytes have both positive and negative effects on pathological progression (18). Reactive astrogliosis actively protects the neurons in the CNS and regulates their homeostasis to limit the size of the infarct region in the early stage of ischemic stroke. However, if not resolved in time, reactive astrogliosis can also inhibit plasticity and regeneration in the CNS (20). At the early stage of ischemia, perivascular astrocytes can release excess cytokines, which subsequently activate metalloproteases and disrupt the blood-brain barrier (BBB) and vasogenic edema. At the later stage of ischemia, perivascular astrocytes can uptake excess extracellular glutamate, contributing to the regeneration of the BBB (21).

After ischemic stroke, maladapted morphological and functional plasticity of astrocytes occurs in the neurovascular unit, which may result in disorders of the neurovascular unit and disrupt the BBB and astrocyte membrane homeostasis in the CNS during stroke rehabilitation (22). In response to oxidative stress, a typical feature of reactive astrocytes is the high expression of intermediate filament proteins (nanofilament proteins) and remodeling of the intermediate filament system in astrocytes,

**Abbreviations:** BBB, blood-brain barrier; CNS, central nervous system; ICH, intracerebral hemorrhage; IL, interleukin; MCA, middle cerebral artery; MDM, monocyte-derived macrophage; MiRNA, microRNA; MSC, multipotent mesenchymal stromal cell; PSD, poststroke depression; TNF-a, tumor necrosis factor alpha.

with a high expression of many characteristic morphological hallmarks. A characteristic morphological hallmark of reactive astrocytes is the presence of hypertrophic astrocytes with increased production of nanofilament proteins, glial fibrillary acidic protein, vimentin, nestin, and synemin (20, 23). Another typical feature of reactive astrocytes is the expression and remodeling of ion channels, which modulate the function of astrocytes by altering the transporters and neurotransmitter receptors. Consequently, alterations in neuronal excitability might lead to secondary neurological disease, such as ischemia and epilepsy during stroke rehabilitation (10). Although astrocytes are not electrically-excitable cells, they can mediate neuron-glia bidirectional interactions through modulating the Ca2<sup>+</sup> signaling of synapses. It was reported that astrocytes can enhance Ca2<sup>+</sup> excitability and modulate synaptic function and plasticity during stroke rehabilitation (16). G protein-coupled calcium-sensing receptor expression is also a feature of reactive astrocytes with astrocyte hypertrophy and high expression of glial fibrillary acid protein in ischemic stroke (24).

#### Microglia

Microglia are resident immune cells involved in physiological and pathological processes in the CNS. Physiologically, microglia are long-living resident immune cells that support a stable chemical and physical microenvironment in the CNS. Pathologically, microglia are dynamic immune cells that respond to nervous damage, repair, and regeneration in the CNS. Microglia can be activated and recruited by the injury signals or stimulation and can elicit a quick response to infection or injury by releasing proinflammatory or anti-inflammatory cytokines. The BBB in the CNS consists of microglia, astrocytes, endothelial cells, and pericytes and selectively separates the sensitive brain parenchyma from the circulatory system. Microglia bidirectionally survey the influx of blood-borne components into the CNS and may stimulate the BBB to open, to extravasate leukocytes resulting in angiogenesis (15, 25, 26).

Another pathological change in ischemic stroke is the activation of resident microglia and infiltrating monocytes/macrophages (27). Activated microglia have both positive and negative effects on the pathological progression of ischemia. Early activated microglia contribute to ischemic injury by releasing TNF and IL-1 and can engulf the cellular debris and invading pathogens. Activated microglia also contribute to resolving the inflammatory response by producing IL-10 and TGFβ and inhibiting the ischemia-induced astrocytic response as a neuroprotective effect during stroke rehabilitation (21, 28, 29). Activated microglia participate in attenuating neuronal apoptosis and enhancing neurogenesis after ischemic stroke (30) and they can contribute to nervous reconstruction and repair during stroke rehabilitation together with reactive astrocytes (4). Nevertheless, chronically activated microglia may cause neuronal death by releasing excessive inflammatory mediators (28). Activated microglia appear after ischemia, and microglial survival depends on signaling through the colony-stimulating factor I receptor during stroke rehabilitation. Therefore, depletion of microglia via colony-stimulating factor I receptor inhibitor PLX3397 exacerbates ischemic infarction and

augments the production of inflammatory mediators, leukocyte infiltration, and cell death after ischemic stroke (29).

For instance, P2X4 receptors (P2X4Rs) on microglia modulate the inflammatory response to ischemia. In acute ischemia, P2X4R activation leads to microglial activation and proliferation to exacerbate the inflammatory response of ischemia. In chronic ischemia, stimulation of P2X4Rs on microglia leads to release of brain-derived neurotrophic factor to support synaptic plasticity and strengthen behavioral rehabilitation. Therefore, knockout of P2X4R on microglia protects against stroke at the early stage of ischemia but exacerbates behavioral recovery at the late stage of ischemia (27).

Modulating microglial overreaction and microglia-mediated neuroinflammation is considered a therapeutic strategy against ischemic damage. For instance, triggering receptor expressed on myeloid cells 2 (TREM2) was mostly expressed in microglia, but not in neurons, astrocytes, or oligodendrocytes in ischemic stroke. TREM2 responds to inflammation after ischemia to protect against cerebral ischemia/reperfusion. Targeting TREM2 to inhibit the inflammatory response in ischemic stroke may be a new therapeutic option (31). Electroacupuncture is also reported as a safe and effective therapy to attenuate the overactivation of Iba-1 and ED1 positive microglia and the expression of TNF-α, IL-1β, and IL-6 and leads to reduced neurological and sensorimotor impairment in ischemia (32).

#### MDMs

MDMs recruited to the injured area at the early stage of ischemia contribute to behavioral rehabilitation by resolving the inflammatory response. The infiltrating monocytes compromise the neurogenesis from endogenous new striatal neurons from neural stem/progenitor cells. The depletion of circulating monocytes early after ischemic stroke most likely increases the short-term survival of the newly formed neoblasts to enhance neurogenesis, using the anti-CCR2 antibody MC21 (33). Incubation of exogenous peroxiredoxin with murine RAW264.7 macrophages leads to nuclear translocation of transcription factor κB p65 and production of proinflammatory mediators (NO, TNF-α, IL-6) (34). Transcription factor κB is also essential to the upregulation of pro-inflammatory genes, which participate in microglial activation and proliferation during stroke rehabilitation (35).

#### Two Phenotypes of Microglia and MDMs

Microglia and MDMs differentiate toward two phenotypes: the M1 phenotype is the classical one, pro-inflammatory, and detrimental, whereas the M2 phenotype is the alternative one, anti-inflammatory, and protective. The two phenotypes of microglia and MDMs suggest their dual roles. The M1 phenotype, which is activated by toll-like receptors or IF-r, promotes injury, whereas the M2 macrophage or N2 neutrophil phenotype, which is activated by regulatory mediators, such as ILs 4, 10, 13; or TGFβ, prompts tissue remodeling and repair (dualistic role) (21).

These mononuclear phagocytes including microglia and macrophages respond to ischemic stroke dynamically, from the M1 phenotype to the M2 phenotype. After stroke onset, monocytes and microglia infiltrate into the infarct core, peaking 3 days after stroke. Before day 7, MDMs with the pro-inflammatory phenotype dominate, and at day 7, half of the infiltrating MDMs are found to be of the proinflammatory phenotype and the other half of the anti-inflammatory phenotype, but the anti-inflammatory phenotype dominates during the subsequent 2 weeks. Similarly, microglia are predominantly of the proinflammatory phenotype at days 3 and 7 after stroke (12, 36). Therefore, instead of broad suppression, there is a need of shifting the polarization of microglia/macrophages into the protective, anti-inflammatory M2 phenotype during stroke rehabilitation (36, 37). For example, ST2, a member of the IL family, and its ligand IL-33 play critical roles in neuroinflammatory responses after ischemic stroke. There is increased expression of ST2 in microglia during stroke rehabilitation, which enhances the expression of M2 polarization markers on microglia/macrophages and impairs the expression of M1 polarization markers. The absence of ST2 shifted the polarization of microglia/macrophages into a proinflammatory M1-like phenotype. There is also increased expression of IL-33 in astrocytes during stroke rehabilitation, and IL-33 and ST2 serve as immune regulatory brakes on the process of stroke rehabilitation (7).

#### Cytokines

Some cytokines and chemokines have been found to affect the inflammatory response to stroke in the process of stroke rehabilitation. Two important inflammatory mediators of the neuroinflammatory response during stroke rehabilitation are TNF-α and IL-1. We next describe how these two cytokines affect stroke rehabilitation.

#### **TNF-**α

A common proinflammatory cytokine is TNF-α, which is involved in every phase of the stroke rehabilitation process. When there are certain stimuli, such as ischemia or hemorrhage, TNF-α is synthesized and released by astrocytes, microglia, or neurons in response to the stimuli and is involved in many pathophysiological processes of ischemic stroke or ICH. TNF-α can activate microglia and astrocytes and have a modulatory effect on BBB permeability, and may also have several positive and negative effects on synaptic transmission and synaptic plasticity during stroke rehabilitation (1, 38).

Inhibition of TNF-α R1 signaling can reportedly preserve brain plasticity during stroke rehabilitation. Etanercept, which is a biologic TNF antagonist, can decrease microglial activation in experimental stroke models and has been used therapeutically in animal stroke models. It has been shown that intravenous administration of etanercept is not therapeutic during stroke rehabilitation because biologic TNF inhibitors can be reengineered for BBB penetration. However, intravenous IgG-TNFR fusion protein is reported to have a therapeutic effect on stroke rehabilitation by significantly reducing stroke volume and neural damage (1, 39, 40).

#### **Interleukin-1**

Another common inflammatory cytokine is IL-1, which affects both systemic and local inflammation and is also an important driver of central and peripheral immune responses to infection or injury. There is considerable experimental and clinical evidence that it is valuable to inhibit IL-1 by IL-1 receptor antagonism as an effective treatment in ischemic stroke. The IL-1 receptor antagonist appears to be a promising treatment target in stroke and is being studied for its therapeutic potential (41).

#### MicroRNAs

MicroRNAs (miRNAs) are small non-coding RNA molecules that regulate gene expression post-transcriptionally by inhibiting the translation of select target genes. MiRNAs are involved in chronic microglial inflammation and lead to progression of neurological diseases such as Alzheimer's disease, amyotrophic lateral sclerosis and stroke (15). In astrocytes and microglia, miRNAs are critical regulators in the mitochondrial response to ischemic stroke. Thereby, MiRNA-targeted therapies have become a viable intervention to optimize mitochondrial function in astrocytes and microglia and improve clinical outcome after ischemic stroke (42).

MiRNA-124 and MiRNA 133b are reportedly involved in the inflammatory response in stroke rehabilitation. Their effects are discussed in detail below.

#### **MiRNA-124**

MiRNA-124 is the most common brain-specific MiRNA in the CNS and has recently been reported to shift the polarization of activated microglia and infiltration of macrophages into the anti-inflammatory M2 phenotype and while also maintaining microglial activation in the acquiescent state. Early injection of MiR-124 significantly increases the number of microglia/macrophages of the M2-like phenotype and neuronal survival and reduces ischemic core formation by inhibiting reactive astrocytes (36, 37). Moreover, liposomated miR-124 administration before the peak of the proinflammatory process in ischemic stroke can shift the predominantly proinflammatory microglia/macrophage phenotype into the anti-inflammatory M2 phenotype most effectively and enhance stroke rehabilitation in the subacute phase (36).

#### **MicroRNA 133b**

Compared with naïve multipotent mesenchymal stromal cells (MSCs), MSCs with overexpressed MiRNA 133b significantly contribute to stroke rehabilitation in animal models of MCA occlusion. Exosomes releasenaïvenaive MSCs are beneficial mediators in the MSC treatment of ischemic stroke. Ex-miR-133b+ significantly increases the release of exosomes from astrocytes by promoting neurite branching and elongation of cortical embryonic neurons, whereas Ex-miR-133b- significantly decreases the release (43).

#### ICH

ICH is the most critical subtype of stroke and lacks effective treatment (44). ICH also leads to neuronal loss, cerebral edema, and neuropathological alterations, including activation of astrocytes and microglia/macrophages and the invasion of neutrophils and T lymphocytes from the blood circulation after ICH (34). There are two types of ICH models: the collagenase-induced model and the autologous arterial whole blood-induced model (45). ICH also leads to neuronal loss, cerebral edema, and neuropathological alterations, including microglial/macrophage and astrocytic activation, and neutrophil and T lymphocyte invasion after ICH (34). After ICH, microglia and astrocytes in brain tissue adjacent to the hematoma may modulate brain cellular plasticity (3). Microglia are among the first non-neuronal cells in the innate immune response to ICH. Microglia become activated by the classical pro-inflammatory M1 phenotype or alternative anti-inflammatory M2 phenotype (44). Astrocytes have differential roles in the recovery pattern of ischemic and hemorrhagic stroke. However, there is similar long-term GFAP-positive astrocytic plasticity after both ischemic stroke and ICH (46). Astrocyte HO-1 overexpression shows distinct neuroprotection after ICH (47). Moreover, ICH stimulates expression and release of Prx 1, activation of tolllike receptor4/nuclear factor κB, and production of cytokines (TNF-α, IL-6, and IL-17) (34). Prostaglandins such as PGE2 also mediate secondary brain injury in the inflammatory response to ICH. The EP2 receptor, which can be activated by PGE2, is expressed in neurons but not in astrocytes or microglia after ICH. The neuronal EP2 receptor shows neuroprotection after ICH by suppressing inflammatory responses, oxidative stress, and matrix metalloproteinase-219 activity, which is involved in brain injury after ICH (45).

# PSD

PSD is a critical psychiatric complication after stroke that frequently occurs at ∼3–6 months and remains for 2–3 years after ischemic stroke or ICH. It is reported that the prevalence rate of PSD is ∼33% in ischemic stroke and it is 15% at 1 year after ICH. It is independently associated with increased morbidity and mortality in stroke because PSD may hinder rehabilitation. To wit, alleviating PSD can improve the outcomes and quality of life in patients after stroke. PSD is reportedly associated with late worsening of disability, but not with initial damage severity after stroke (4, 48, 49).

The mechanisms between cerebrovascular diseases and depressive disorders are intertwined. As the inflammatory response in stroke affects stroke rehabilitation, some studies have confirmed that an immunological hypothesis is one of the pathophysiological mechanisms of PSD and the inflammatory response in PSD affects its outcome. However, the specific mechanisms of the inflammatory effects on stroke and PSD reportedly differ. It has been shown that patients with PSD have early increased inflammatory markers (such as highsensitivity C-reactive protein, ferritin, neopterin, and glutamate), increased proinflammatory cytokines (TNF-α, IL-6, IFN-γ), increased pro-inflammatory/anti-inflammatory ratios (TNFα/IL-10, IL-6/IL-10), and lowered complement expression (5). Recent investigations have revealed imbalances in inflammatory cytokine levels and increased oxidative DNA damage in association with PSD, as well as the involvement of inflammatory and immune mechanisms and elevated oxidative stress level in PSD (49–52). It has also been reported that cytokines can drive intrinsic apoptotic factors to increase the risk of PSD through intracellular calcium and glutamate excitotoxicity after ischemic damage; pro-inflammatory cytokines may amplify the proinflammatory processes by activating indoleamine 2, 3-dioxygenase and reducing serotonin production, which sequentially results in PSD. Cytokines can provoke the dysregulation of several growth factors, such as BDNF, fibroblast growth factor-2, and contribute to PSD and other comorbidities (53–55).

# PHARMACOTHERAPY

# Pharmacotherapy in Stroke Rehabilitation

Using anti-inflammatory drugs along with neurorehabilitation therapy is useful for neuroprotection and functional recovery (56). Anti-inflammatory drugs may enhance brain plasticity after stroke but need to be used in conjunction with neurorehabilitation therapy (57). Therefore, anti-inflammatory treatment has the most potential as a therapy to enhance neurorehabilitation after stroke (58).

The findings of recent studies on anti-inflammatory treatments for stroke are listed below. Multimodal intervention of minocycline (pharmacotherapy, anti-inflammatory drug used to modulate the dynamics of the immune system) together with cerebral stimulation using transcranial direct-current stimulation and repetitive transcranial magnetic stimulation (neurorehabilitation therapy used to enhance functional recovery after stroke) may augment plasticity, rehabilitation, and neurorestoration (48). Simvastatin (statin class of cholesterollowering drugs), alters the release of cytokines and trophic factors from microglia, including IL-β, TNF-α, and brain derived neurotrophic factor in a cholesterol-dependent manner, but inhibits phagocytosis in a cholesterol-independent manner (59). Vinpocetine (a potent anti-inflammatory agent) improves neuronal plasticity and reduces the release of inflammatory cytokines and chemokines from microglia, macrophages, endothelial cells, and vascular smooth muscle cells (60). Omega-3 polyunsaturated fatty acids provide anti-inflammatory neuroprotective function in ischemic stroke by targeting astroglia and microglia and preventing the release of cyclooxygenase 2, hypoxia-inducible factor 1α, nitric oxide synthase, and IL-1β and have clinical potential as a therapeutic treatment in stroke (61). Trypsin inhibitor ulinastatin (anti-inflammatory drug) provides neuroprotective function in synaptic plasticity and spatial memory in cerebral ischemia-reperfusion injury (62). The melanocortin MC4 receptor agonist RO27-3225 (used to reduce expression of TNF-α, BAX, ERK1/2, JNK1/2, and cascapse-3 and counteract prolonged/recurrent inflammatory and apoptotic responses) provides neuroprotective function and promotes functional recovery in ischemic stroke (63). Scutellarin (a potential Chinese herbal extract, a putative therapeutic agent, used to improve neurological function), ameliorates neuroinflammation by suppressing microglial activation and enhances astrocytic reaction by upregulating the expression of neurotrophic factors (8, 28).

Sex differences are involved in the frequency of intracellular astrocyte Ca2+ elevation and microglial volume immediately in ischemic stroke and are foundational for future sex-specific stroke therapeutic treatments (64). Female sexual hormones (estradiol and progesterone, anti-inflammatory), modulate the cellular and immune response to ischemic stroke (9). The corticotropin-releasing hormone type 1 receptor actively alters neuronal injury and inflammation, neuronal plasticity, and functional recovery in ischemic stroke (65).

The vascular endothelial growth factor mediates reactive astrocyte transdifferentiation into new mature neurons and enhances neurogenesis in ischemic stroke (66). The vascular endothelial growth factor also suppresses the inflammatory response in ischemic stroke to promote neuronal plasticity and neuronal remodeling (67). High-mobility group box 1 (amphoterin or HMG1) promotes neuronal necrosis and influx of damaging inflammatory cells in the acute stage of ischemic stroke but promotes beneficial plasticity and neuronal recovery in the delayed stage after stroke (68). The vagus nerve system regulates the immune system through the cholinergic anti-inflammatory pathway (69). Acetylcholine-alpha 7 nicotinic acetylcholine receptor on macrophages or microglia also provides neuroprotection through the cholinergic anti-inflammatory pathway. Nicotine (an acetylcholine receptor agonist, anti-inflammatory), inhibits microglial activation and production of proinflammatory cytokines and cholinesterase by activated astrocytes, which is partly medicated by COX-2 (70). Therefore, treatments inhibiting cyclooxygenases enhance neuronal plasticity after ischemic stroke (71). Therapeutic hypothermia is another potential treatment for ischemic stroke. Novel neurotensin receptor1 (NTR1) agonists induce hypothermia to inhibit microglial activation and decrease the expression of proinflammatory (M1) chemokines and cytokines and protect against neuronal damage in ischemic stroke and ICH (6).

# Pharmacotherapy in PSD

Anti-cytokine modulators are new therapeutic targets for the treatment of PSD, especially in subjects affected by inflammatory

# REFERENCES


processes. For instance, an investigation revealed that antiinflammatory treatment, such as acetylsalicylic acid, nonsteroidal anti-inflammatory drugs, and statins decrease the risk of PSD, and inflammation contributes to PSD depending on the onset of PSD (72).

# CONCLUSIONS

Stroke comprises ischemic stroke and ICH. The immunoinflammatory process is involved in neural plasticity following events such as a hemorrhage or ischemic stroke. After ischemia, astrocytes, microglia, and MDMs play important roles during rehabilitation with the modulation of cytokines or chemokines, such as TNF-α and IL-1. Moreover, MiRNAs are also important posttranscriptional regulators in these glial mitochondrial responses to cerebral ischemia. ICH involves processes similar and different to those seen in ischemia, including neuronal injury, astrocytic and microglial/macrophage activation, and neutrophil and T lymphocyte invasion after ICH. Immunological hypothesis is also one of the pathophysiological mechanisms of PSD. To date, many pharmacotherapies have been suggested as having an anti-inflammatory function in stroke rehabilitation, including those involving minocycline, melanocortin, omega-3 polyunsaturated fatty acids, UTI, statin, vinpocetine, RO27- 3225, scutellarin, and sexual hormones. Other potential therapies involve the vascular endothelial growth factor, high-mobility group box 1, corticotropin-releasing hormone type 1 receptor, the vagus nerve system, nicotine and cyclooxygenase 2, and therapeutic hypothermia. In PSD, very few anti-inflammatory treatments have been studied, including the use of acetylsalicylic acid, non-steroidal anti-inflammatory drugs, and statins.

# AUTHOR CONTRIBUTIONS

MF, LZ, and XJ wrote the first draft. RC and WY provided the organization and framework of the article. SG, JL, BL, and TL provided the critical revisions. All authors approved the final version of the manuscript for submission.

# FUNDING

The work was supported by the Natural Science Foundation of China (NSFC). Grant Nos. 81772291.


a depressive phenotype during recovery from ischemic stroke. Brain Behav Immunity. (2017) 66:302–12. doi: 10.1016/j.bbi.2017.07.155


has no influence on the different recovery patterns. Behav Brain Res. (2015) 278:257–61. doi: 10.1016/j.bbr.2014.10.005


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Fang, Zhong, Jin, Cui, Yang, Gao, Lv, Li and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Anxiety in Patients With Acute Ischemic Stroke: Risk Factors and Effects on Functional Status

#### *Edited by:*

*Yi Yang, First Affiliated Hospital of Jilin University, China*

#### *Reviewed by:*

*Alexander Nikolaevich Savostyanov, State Scientific-Research Institute of Physiology and Basic Medicine, Russia Qing Hao, Icahn School of Medicine at Mount Sinai, United States*

> *\*Correspondence: Yang-Kun Chen cykun78@163.com*

*†These authors have contributed equally to this work.*

#### *Specialty section:*

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

*Received: 14 September 2018 Accepted: 04 April 2019 Published: 17 April 2019*

#### *Citation:*

*Li W, Xiao W-M, Chen Y-K, Qu J-F, Liu Y-L, Fang X-W, Weng H-Y and Luo G-P (2019) Anxiety in Patients With Acute Ischemic Stroke: Risk Factors and Effects on Functional Status. Front. Psychiatry 10:257. doi: 10.3389/fpsyt.2019.00257*

*Wei Li1†, Wei-Min Xiao1†, Yang-Kun Chen1\*, Jian-Feng Qu1, Yong-Lin Liu1, Xue-Wen Fang2, Han-Yu Weng1 and Gen-Pei Luo1*

*1 Department of Neurology, Dongguan People's Hospital, Dongguan, China, 2 Department of Radiology, Dongguan People's Hospital, Dongguan, China*

Background: Anxiety is prevalent after a stroke. The pathophysiological mechanisms underlying the development of poststroke anxiety (PSA) remain unclear. The aim of this study was to investigate the clinical and neuroimaging risk factors for development of PSA and examine the effects of PSA on activities of daily living (ADL) and quality of life (QOL) in Chinese patients with ischemic stroke.

Methods: Two hundred nineteen patients with acute ischemic stroke were recruited to the study. A series of comprehensive assessments, including Hamilton Anxiety Rating Scale (HARS), Hamilton Depression Rating Scale (HDRS), Lawton ADL Scale, and the Stroke-Specific Quality of Life (SSQOL) Scale, were conducted in the acute stage and 3 months after stroke. Magnetic resonance imaging assessment focused on evaluation of infarctions, white matter lesions, and brain atrophy.

Results: In the acute stage and 3 months after stroke, 34 (16%) and 33 (15%) patients had PSA, respectively. Multiple logistic regression analysis indicated that HDRS (OR = 1.269, 95% CI = 1.182–1.364, *P* < 0.001) and acute infarcts in cerebral hemispheric white matter (CHWM; OR = 2.902, 95% CI = 1.052–8.007, *P* = 0.040) were significant correlates of PSA in the acute stage of stroke. Three months after stroke, these correlates remained significant predictors, along with male sex. Multiple linear regressions showed that age, NIHSS, HARS, and HDRS in the acute stage were significant predictors for both ADL and SSQOL at 3 months after stroke.

Conclusion: Depressive symptoms are the major correlates of PSA while more severe PSA is associated with poorer ADL and health-related QOL. Acute lesions involving CHWM may correlate with PSA in ischemic stroke patients with mild-to-moderate neurologic deficits, supporting a lesion-location hypothesis in PSA.

Keywords: anxiety, depression, functional status, quality of life, stroke, cerebral hemispheric white matter

# INTRODUCTION

Anxiety is prevalent after stroke and occurs in about one-quarter of stroke survivors (1, 2). Poststroke anxiety (PSA) may have a negative impact on quality of life (QOL) of stroke survivors, affecting their rehabilitation (3). Furthermore, one prospective study found that severe anxiety symptoms were associated with increased risk for incident stroke, independent of other risk factors (4). Despite high prevalence of anxiety after stroke, understanding of PSA is limited.

Risk factors related to PSA include depression (5–7), cognitive impairment (5, 8), fatigue (9), age (10–12), female sex (10, 12), lesion location (13, 14), and sleep disturbance (9, 15), indicating PSA might be multifactorial. Apart from stress due to acute ischemic stroke, the biological mechanisms of PSA should also be considered. Anxiety-related neural circuits span a wide range of brain structures, including subcortical white matter and the limbic system (16, 17). Neuroimaging techniques, e.g., magnetic resonance imaging (MRI), can locate the infarction precisely. Thus, studying the neuroimaging correlates of PSA may be helpful in understanding the pathophysiology of PSA. However, few studies have evaluated the association between PSA and neuroimaging variables. The lesion-location hypothesis of PSA might be presumed as the infarction may damage brain structures involved in anxiety. Recently, a large-scale MRI study involving 239 stroke patients was performed, but no association was found between brain lesion location and PSA (18). Thus, the underlying pathophysiological mechanisms of development of PSA remain unclear.

Functional status, including the ability to perform activities of daily living (ADL) and QOL, is an important outcome of stroke in many studies (3, 19, 20). However, few studies have explored the subsequent effect of PSA in the acute phase on patient QOL and functional outcomes in the chronic stage.

The purpose of this prospective study was twofold. The first purpose was to investigate the associated clinical and MRI risk factors for PSA, testing the lesion-location hypothesis of PSA. The second purpose was to examine the effects of PSA on patient ADL and QOL following ischemic stroke. We assumed that lesions in specific locations (e.g., structures related to emotional modulation) might be more likely to result in PSA. Severity of PSA in the acute stage is a significant factor independently contributing to poor ADL and QOL in the chronic stage.

### METHODS

#### Participants and Setting

Patients with first-ever or recurrent acute ischemic stroke admitted to the Department of Neurology, Dongguan People Hospital, between July 2013 and June 2014 were screened for this study. Patients were enrolled in the study if they met the following criteria: 1) age 40 to 80 years; 2) had an acute, first, or recurrent ischemic stroke that occurred within 7 days prior to admission; if they had a previous stroke, the modified Rankin Scale score before the index stroke was <2. Patients were excluded if they 1) had significant neurological illness other than stroke, e.g., Parkinson's disease, brain tumor, or multiple sclerosis; 2) had no MRI scans or poor-quality MRI scans on admission; 3) had a severe stroke, which received a National Institutes of Health Stroke Scale (NIHSS) total score of ≥15; 4) had severe aphasia (defined as NIHSS best language subscore ≥2) or dysarthria; 5) had severe cognitive impairment, defined by a Mini-Mental State Examination (MMSE) total score of <17; 6) had a history of anxiety disorders, depression, substance abuse/dependence, or other psychiatric disorders before the index stroke; and 7) had comorbid severe diseases of the heart, lung, kidney, liver, or malignant tumors. This study was carried out in accordance with the recommendations of the World Medical Association's Declaration of Helsinki. The study protocol was approved by the Ethics Committee of Dongguan People's Hospital. Consent forms were obtained from the patients or their legally authorized representative.

### Collection of Demographic and Clinical Data

Patient demographics (age, sex, and education level) and clinical characteristics including vascular risk factors (e.g., hypertension, diabetes mellitus, hyperlipidemia, and smoking history) and previous stroke history were collected from medical records. The severity of stroke was assessed by the NIHSS from medical records.

#### Assessment of PSA

The Chinese version of the 14-item Hamilton Anxiety Rating Scale (HARS) (21, 22) was used to evaluate anxiety symptoms in all participants in the acute stage when they were medically stable (5–14 days after the index stroke) and at the 3-month follow-up. Assessments of clinical anxiety were performed by two trained neurologists (WL and HW) who were blinded to the MRI results of the stroke survivors. The Chinese version of the 14-item HARS has been widely used in the Chinese population, as well as in Chinese stroke patients (22), indicating good reliability and validity. PSA in this study was defined by a HARS score ≥14 (22). HARS was repeatedly conducted at 3 months after stroke by the same raters. PSA was defined as a HARS score ≥14 at 3 months after stroke. If the patients were diagnosed with anxiety at baseline and received anti-anxiety treatment, they were also judged to have PSA even if they had a HARS score lower than 14.

#### Assessment of Other Psychological Status in the Acute Stage of Stroke

The Chinese version of the MMSE (scores range from 0 to 30, with lower scores indicating greater deficits) (23) was used to measure basic cognitive function by the two trained neurologists (WL and HW). They also administered the Chinese version of the 24-item Hamilton Depression Rating Scale (HDRS) (22, 24), which was used to evaluate the severity of depressive symptoms, with an internal consistent Cronbach's α = 0.88–0.99 (22).

## Assessment of Functional Status at 3 Months After Stroke

The two raters administered the Lawton ADL Scale (25) and the Chinese version of the Stroke-Specific Quality of Life (SSQOL) Scale (26). The Lawton ADL Scale, which contains six items assessing self-maintenance and eight items evaluating instrumental activities, was used to measure functional level of patients with stroke. Each item was rated from 1 to 4, and the total ADL score was calculated by summing the scores of all items. Higher scores indicate poorer performance. The test–retest kappa of the Chinese version of the Lawton ADL Scale is 0.502 (22). The Chinese version of the SSQOL Scale, which was used to assess patient QOL and proved to have good reliability and validity, consists of 49 questions grouped into 12 domains, with scores rated from 1 (worst outcome) to 5 (best outcome). The internal consistent reliability is high (Cronbach's α = 0.76) (26, 27).

Before the first interview, the two neurologists selected 10 patients with ischemic stroke to test the interrater reliability of the rating instruments. The intraclass correlation coefficients (ICCs) of the above scales between the two raters ranged between 0.83 and 0.91.

#### Magnetic Resonance Imaging Assessment

MRI acquisition was performed using a 1.5-T scanner (Achieva Phillip Medical System, Best, the Netherlands) within 7 days of the index stroke. The sequences of MRI scanning included diffusionweighted imaging (DWI), gradient echo sequences, and T1- and T2-weighted, fluid-attenuated inversion recovery sequences. A trained neurologist (YL), who was blinded to patient clinical information, assessed the MRI variables as follows:


5. Medial temporal lobe atrophy (MTLA). MTLA was measured using Schelten's scale (31). This visual rating scale yields standard images with different severity of MTL atrophy on coronal MRI sections, ranging from 0 to 4, from "no atrophy" to "severe atrophy." The MTLA score was determined using the sum of left and right medial temporal lobes.

Intrarater reliability activities were performed on 10 patients by the same MRI rater at two time points (interval ≥2 months). The intrarater agreements of the MRI measurements were good to excellent, as reported in our previous study (32).

#### Statistical Analysis

All statistical tests were performed using SPSS for Windows (Release 16.0, SPSS Inc., Chicago, IL, USA). In the acute stage of stroke, all patients were divided into two groups, the PSA and non-PSA groups, according to the HARS cutoff. The demographic and clinical variables were compared between the PSA and non-PSA groups using χ2 test, two independent *t* tests, or Mann–Whitney *U* tests, as appropriate, in order to screen for potential predictors. Variables with *P* < 0.1 in univariate comparisons were entered as independent variables in multiple stepwise logistic regression analysis with PSA as the dependent variable. The same statistical procedures were performed at 3 months after stroke. Subsequently, multiple linear regressions were performed to explore the effects of HARS in the acute stage on ADL and SSQOL at 3 months after stroke (ADL and SSQOL were used as dependent variables) after adjusting for age, sex, NIHSS, and HDRS. The significance level was set at 0.05 (two-sided).

# RESULTS

A total of 435 patients aged 40 to 80 years with acute ischemic stroke were admitted and screened. Two hundred nineteen patients (50.3%) fulfilled the study criteria and were included in the study. Compared to those who were excluded, participating patients were younger (61.4 ± 11.2 *vs.* 64.8 ± 12.7 years; *P* < 0.001), had a lower NIHSS score at admission (median, 3.0 [range, 0–15] *vs*. 5.0 (0–35), *P* < 0.001), but had a comparable frequency of male sex (73.1% *vs*. 68.1%; *P =* 0.252).

### Demographic and Clinical Characteristics

The study cohort consisted of 219 patients who satisfied the study criteria (**Table 1**). One patient died and three patients were lost to follow-up before the 3-month assessment. In the acute stage and 3 months after the index stroke, there were 34 (15.5%) and 33 (15.1%) patients who were judged to have PSA, respectively. Compared to patients without PSA, patients with PSA were more likely to be female and to have more severe depressive symptoms (**Table 1**). No MRI variables were significantly different between the two groups, although patients with PSA trended toward more CHWM infarcts in both the acute stage and 3 months after stroke (*P =* 0.075 and *P =* 0.071, respectively; **Table 2**).

TABLE 1 | Comparisons of demographic and clinical variables between the PSA and non-PSA groups.


*PSA, poststroke anxiety; NIHSS, National Institutes of Health Stroke Scale; MMSE, Mini-Mental State Examination; HDRS, Hamilton Depression Rating Scale; HARS, Hamilton Anxiety Rating Scale.*

*\*Mean (SD), t test; †n (%), chi-square test; ‡ median (25%Q–75%Q), Mann–Whitney U test.*

TABLE 2 | Comparisons of MRI variables between the PSA and non-PSA groups.


*PSA, poststroke anxiety; CHWM, cerebral hemispheric white matter; PVH, periventricular hyperintensities; DWMH, deep white matter hyperintensities; VBR, ventricle-to-brain ratio; MTLA, medial temporal lobe atrophy.*

*\*n (%), chi-square test; †median (25%Q–75%Q), Mann–Whitney U test; ‡mean (SD), t test.*

#### Correlates of Poststroke Anxiety in the Acute Stage of Stroke

#### HDRS, sex, and acute infarcts in CHWM were evaluated by multiple logistic regressions. HDRS (odds ratio [OR] = 1.269, 95% CI = 1.182–1.364, *P* < 0.001) and acute infarcts in CHWM (OR = 2.902, 95% CI = 1.052–8.007, *P =* 0.040) were significant correlates of PSA in the acute stage of stroke (**Table 3**).

#### Correlates of Poststroke Anxiety at 3 Months After Stroke

HDRS, sex, NIHSS, MMSE, and acute infarcts in CHWM were evaluated by multiple logistic regression. HDRS (OR = 1.232, 95% CI = 1.150–1.320, *P* < 0.001), female sex (OR = 3.214, 95% CI = 1.124–9.189, *P =* 0.029), and acute infarcts in CHWM (OR = 2.904, 95% CI = 1.033–8.162, *P =* 0.043) significantly correlated with PSA (**Table 4**).

TABLE 3 | Correlates of PSA in the acute stage of ischemic stroke.


*Total R2 = 0.536; PSA, poststroke anxiety; HDRS, Hamilton Depression Rating Scale; CHWM, cerebral hemispheric white matter.*



*Total R2 = 0.491; PSA, poststroke anxiety; HDRS, Hamilton Depression Rating Scale; MMSE, Mini-Mental State Examination; CHWM, cerebral hemispheric white matter.*

#### Effects of Anxiety in the Acute Stage on Activities of Daily Living and Stroke-Specific Quality of Life at 3 Months After Stroke

Multiple linear regressions showed that age, NIHSS, HARS, and HDRS in the acute stage were significant predictors for both ADL and SSQOL at 3 months after stroke (**Table 5**). Patients with PSA in the acute stage were more likely to have a poorer performance in ADL and SSQOL at 3 months after stroke.

#### Sensitivity Analysis After Excluding Patients With Previous Stroke

Analyses including only patients with their first-ever stroke are summarized in the supplemental tables. Acute infarcts in CHWM remained a significant correlate of PSA in the acute stage, but not at 3 months after stroke. HARS score in the acute stage significantly contributed to poorer ADL and SSQOL 3 months after stroke after adjusting for age, sex, NIHSS, and HDRS.

TABLE 5 | The effects of PSA in the acute stage on ADL and SSQOL at 3 months after stroke.


*PSA, poststroke anxiety; ADL, activities of daily living; SSQOL, Stroke-Specific Quality of Life; HDRS, Hamilton Depression Rating Scale; HARS, Hamilton Anxiety Rating Scale; NIHSS, National Institutes of Health Stroke Scale.*

#### DISCUSSION

In this prospective and longitudinal study, we found that frequency of PSA in the acute stage and 3 months after a mildto-moderate ischemic stroke was 15.5% and 15.1%, respectively. HDRS and acute infarcts in CHWM correlated with PSA in both the acute stage and 3 months after stroke. Severity of PSA was a significant indicator for both ADL and SSQOL. To the best of our knowledge, studies investigating the effects of PSA on functional status are very limited. Our study represents a significant contribution to literature on the significance of PSA.

Anxiety symptoms were common after stroke. A meta-analysis study estimated that PSA affected 25% of stroke survivors (1). A summary of studies on PSA is shown in **Table 6**. In our study, we used HARS to assess anxiety symptoms with a cutoff of mean HARS ≥14 and found that frequency of PSA was about 15% in stroke survivors, which was lower than most previous studies (33–37). This may be due to the inclusion of a stroke sample with relatively mild neurological deficits (median NIHSS, 4) and exclusion of severe neurologic deficits or aphasia. Patients excluded from this study might be more likely to have PSA. Differences in assessment tools for PSA might also contribute to the differences between our results and previous studies.

Sudden occurrence of neurological deficits might cause stress or anxiety in stroke patients. We assessed PSA at two time points, including the acute stage (5–14 days) and months after stroke. The time point of 5–14 days was chosen for the first time point because this is when patients are typically medically stable. The 3-month time point was selected as the acute effects of stress related to an adverse life event might have diminished, and is a common time point used in previous studies (5, 13). As PSA might be multifactorial, we collected comprehensive data to the extent possible, including clinical, physical, psychological, and neuroimaging variables. The present study showed that PSD was significantly associated with PSA both in the acute stage and 3 months after stroke, indicating that PSD and PSA may share a common pathophysiological mechanism. This comorbidity has been confirmed by other studies (2, 5, 37).

Available clinical data on the relationship between PSA and lesion location are conflicting. Tang et al*.* (13) found that patients with acute frontal lobe infarction were more likely to have PSA. Similar to other studies (8, 18, 40), we could not locate a single lesion location that was directly related to PSA. However, we found that patients with acute infarction in the CHWM were more likely to have PSA in the acute stage of stroke, as well as 3 months after stroke. This finding has not been reported previously. CHWM includes a wide range of regions of connected neural fibers in the cerebral hemisphere, e.g., corpus callosum, corona radiata, centrums semiovale, and internal capsule. Neural circuits associated with emotion regulation are widely distributed in the cerebral hemispheres, such as the fronto-subcortical circuits or the limbic system (41, 42). Brain white matter abnormalities have also been implicated in development of anxiety (17). Thus, acute CHWM lesions are logical potential contributors to PSA. However, the wide range of OR in CHWM in prediction of PSA indicates that this preliminary finding should be carefully repeated in further studies. Furthermore, after excluding patients


TABLE 6 | Summary of studies assessing the correlates or prevalence of anxiety in patients with ischemic stroke.

*Diagnostic and Statistical Manual of Mental Disorders-IV-Text Revision; EQ-5D5L, EuroQoL-5D5L; HADS-A, Hospital Anxiety and Depression Scale–Anxiety; MMSE, Mini-Mental State Examination; BI, Barthel Activities of Daily Living* 

*Index; BAI, Beck Anxiety Inventory; DMN, Default Mode Network; MDA, malondialdehyde; GPX, glutathione peroxidase; SOD, superoxide dismutase; CAT, catalase; NR, not recorded.*

with previous strokes, CHWM only contributed to PSA in the acute stage, but not 3 months after stroke.

The role of CHWM in development of PSA remains unclear. Recently, studies have focused on lesions involving neuronal network or circuits rather than single locations. Fornito et al. (43) reported that functional neuronal network disruption may be more critical than lesion location to explain PSA. Vicentini et al. (12) reported that PSA was not associated with infarct location but correlated with disruption of the default mode network (DMN) in the brain. Accordingly, examining the effects of the integrity of brain networks or neural circuits rather than a single location on PSA might be another direction for further research.

The severity of PSA in the acute stage was inversely associated with performance of ADL in this study. Schultz et al. (10) reported that association of anxiety and impairment in ADL were present only at the initial evaluation (in the acute stage of ischemic stroke), with independent effects only for women. It can be postulated that PSA patients may have poor adherence to rehabilitative efforts because of a significant decrease in both physical and mental energy, which, in turn, impairs performance of ADL.

Stroke frequently reduces the level of health-related QOL (HRQOL) of survivors. Our study indicated that the severity of PSA in the acute stage was a significant contributor to poorer SSQOL 3 months after stroke. PSA may reduce physical and mental energy, motivation, and activity, which then inversely affects HRQOL. A cross-sectional study also found that poorer QOL was associated with greater levels of physical disability, anxiety, and depression, and reduced social interaction (44). Thus, assessment and intervention of PSA as well as PSD in the acute stage may be helpful to predict outcomes of functional status.

Our study has several strengths. First, we conducted a faceto-face interview to evaluate functional and psychological measures, which was rarely reported in other studies. Second, we obtained comprehensive MRI data from all participants. However, there were also several limitations to our study. First, only patients with mild-to-moderate ischemic stroke without severe cognitive impairment and aphasia were recruited, which limits the generalization of our findings. Second, we did not collect medication or rehabilitation after discharge, though most

#### REFERENCES


patients would have follow-up visits with neurologists or general physicians in community clinics. Third, we only used a screening tool (HDRS) rather than the standard psychiatric interview to define PSA, as there are no sufficient psychiatric professionals in our hospital. Lastly, the associations between PSA and QOL or ADL did not indicate causality due to the study design.

In general, anxiety is common in the acute and chronic stages of ischemic stroke with mild-to-moderate neurologic deficits. The lesion-location hypothesis of PSA might be relevant but remains uncertain. PSA in the acute stage may have a significant impact on ADL and HRQOL in stroke patients in the chronic stage. Early detection of anxiety symptoms may facilitate functional recovery and improve QOL in stroke patients. Careful evaluation of PSA should be integrated into clinical care of stroke patients.

#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of "Operational Guidelines for Ethics Committees That Review Biomedical Research, World Health Organization (2000), Ethics Committee of Dongguan People's Hospital" with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the "Ethics Committee of Dongguan People's Hospital."

#### AUTHOR CONTRIBUTIONS

YC and WX designed the study. WL, JQ, and GL screened and collected the patients. WL and HW performed the psychological assessments. XF designed and trained the MRI assessment. YL assessed the MRI variables. WL and YC wrote the manuscript.

### FUNDING

This study was funded by the Medical Scientific Research Foundation of Guangdong Province, China (Grant No: B2011349).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Li, Xiao, Chen, Qu, Liu, Fang, Weng and Luo. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Does the Use of Antidepressants Accelerate the Disease Progress in Creutzfeldt–Jakob Disease Patients With Depression? A Case Report and A Systematic Review

*Yifan Liang, Yan Li, Huibin Wang, Xi Cheng, Meiting Guan, Shanshan Zhong and Chuansheng Zhao\**

*Department of Neurology, The First Hospital of China Medical University, Shenyang, China*

#### *Edited by:*

*Yi Yang, First Affiliated Hospital of Jilin University, China*

#### *Reviewed by:*

*Michael X. Zhu, University of Texas Health Science Center at Houston, United States Yan-Jiang Wang, Third Military Medical University, China*

*\*Correspondence:*

*Chuansheng Zhao, cszhao@cmu.edu.cn*

#### *Specialty section:*

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

*Received: 04 October 2018 Accepted: 16 April 2019 Published: 03 May 2019*

#### *Citation:*

*Liang Y, Li Y, Wang H, Cheng X, Guan M, Zhong S and Zhao C (2019) Does the Use of Antidepressants Accelerate the Disease Progress in Creutzfeldt–Jakob Disease Patients With Depression? A Case Report and A Systematic Review. Front. Psychiatry 10:297. doi: 10.3389/fpsyt.2019.00297*

Background: Creutzfeldt–Jakob disease (CJD) is a fatal neurodegenerative disorder characterized by rapidly progressive dementia. Growing evidence suggests that antidepressant usage was associated with dementia. Given the commonality of depression in CJD, it is necessary to investigate the effect of antidepressants on CJD.

Methods: First, we report a case of sporadic CJD (sCJD) with depression where the condition worsened rapidly after using a serotonin and noradrenaline reuptake inhibitor (SNRI) antidepressant. Second, a systematic literature survey was conducted to investigate the effect of antidepressants on the survival time of sCJD patients with depression. Thirteen cases plus our case were included for qualitative analysis. Twelve subjects were included in the Kaplan–Meier survival and Cox regression analysis. Finally, we provide a postulation of pathophysiological mechanism in CJD.

Results: The median survival time of all patients was 6.0 months, of which patients with SNRIs were significantly shorter than those with first-generation antidepressants (2.0 vs. 6.0 months; log rank, *P* = .008) and relatively shorter than those with nonselective serotonin reuptake inhibitors (SSRIs; 4.0 vs. 6.0 months; log rank, *P* = .090). In comparison with first-generation antidepressants, the use of SNRIs [hazard ratio (HR), 23.028; 95% confidence interval (CI), 1.401 to 378.461; *P* = .028] remained independently associated with shorter survival time.

Conclusions: The use of antidepressants, especially SNRIs, was associated with a shorter survival time of sCJD patients. The possible changes in neurotransmitters should be emphasized. Scientifically, this study may provide insights into the mechanism of CJD. Clinically, it may contribute to the early diagnosis of CJD.

Keywords: depression, Creutzfeldt–Jakob disease, antidepressant, neurotransmitters, sleep wake disorders

# INTRODUCTION

Depression is common in the elderly. Its prevalence rate is as high as 11.19%, and this increases progressively with worsening cognitive impairment (1). The presence of depression is an acknowledged risk factor for dementia (2); it can even double the risk for Alzheimer's disease (AD) (3, 4). Many reasons lie behind the prescription of antidepressant drugs, which increased dramatically from 1999

**120**

to 2014 (5). However, some studies have questioned whether antidepressants confer any benefits in terms of cognitive decline (6–9). Recently, a meta-analysis indicated that antidepressant usage was associated with AD/dementia (10). Tricyclic antidepressants (TCA) may be associated with a reduced risk (11) or no risk of dementia (12) for depressed patients, whereas nonselective serotonin reuptake inhibitors (SSRIs) antidepressant drugs, including monoamine oxidase (MAO) inhibitors, and serotonin and noradrenaline reuptake inhibitors (SNRIs) have been reported to possess an intermediate risk (11–13).

Creutzfeldt–Jakob disease (CJD), a fatal neurodegenerative disorder characterized by rapidly progressive dementia, is divided into the sporadic (sCJD), familial (fCJD), variant (vCJD), and iatrogenic (iCJD) subtypes (14). sCJD accounts for the majority, i.e., 85% of all CJD cases, with an annual worldwide incidence of one to two cases/million population (15). Although rare, the overall mortality rate of sCJD has been increasing since 1993 (16). There is no effective treatment for CJD, so it is important to identify modifiable risk factors for CJD and to delay disease progression. Psychiatric manifestations are often the first symptoms to appear in vCJD. Recent studies have shown that they are also more prevalent in sCJD than previously thought. Most commonly, these are exhibited as depression in 16%–37% of cases (17–19). While treatment of the psychiatric symptoms with hypnotics, anxiolytics, or antipsychotic drugs may offer relief to CJD patients, it appears that antidepressant drugs are ineffective (19). As is the case in some other dementias, the implication is that antidepressants may also fail to benefit CJD patients.

Here, we report a case of sCJD in which depression was the first symptom, and the condition worsened rapidly after the administration of an SNRI antidepressant. Subsequently, a systematic review of the literature was undertaken to explore the characteristics of sCJD patients with depressive symptoms as well as the effect of treatment with antidepressants on sCJD. This report aims to provide novel insights into the underlying causes and treatment of CJD and dementia.

# CASE REPORT

Ms. S was a 63-year-old female with no previous medical or psychiatric history. In July 2017, she presented with dizziness, weakness, chronic shoulder pain, and high blood pressure. She informed her family that she felt helpless and sick. The preliminary examination revealed nothing but multiple lacunar infarcts in brain magnetic resonance imaging (MRI) scans. On September 17, 2017, she exhibited anhedonia, fear, anxiety, impatience, and a propensity to cry after being annoyed with others. She was examined in the psychiatric unit of the local hospital. Her value on the Self-rating Depression Scale (SDS) was 53.75, which pointed to mild depression, whereas on the Hastgawa Dementia Scale (HDS), she scored 13.0, which suggested probable dementia (education: primary school). The memory quotient (MQ) of Wechsler Memory Scale (WMS) was 59. Her sleep was normal. She was diagnosed with depression, and sertraline 50 mg/day was prescribed. Her symptoms nonetheless worsened with insomnia, garrulity, irritability, and gait imbalance. Her memory function deteriorated, and she became disoriented. The psychiatrist changed the antidepressant drug to venlafaxine 75 mg/day on October 8, 2017. However, instead of improving, the condition rapidly worsened. Her speech became hypophonic and monotonous with a paucity of content. She was sleepy during the day and sometimes burst into tears. Her arms curled up, indicating panic. She developed psychomotor retardation, responded poorly to questions, experienced visual hallucination, and suffered from a rigid posture with paroxysmal myoclonus and an inability to walk. The changes in her symptoms were initially considered to be side effects of venlafaxine. Two weeks later, she had deteriorated further and was unable to talk, exhibiting dysphagia and suffering from urinary incontinence. The symptoms did not improve after the withdrawal of the antidepressant. An assessment of her electroencephalogram (EEG) revealed generalized slow activity (**Figure 1A**). She was then transferred to the neurologic ward of our hospital where the following neurological findings were detected: akinetic mutism (AM), normal muscle strength, increased muscle tension, brisk tendon reflexes, and unresponsive pathologic reflexes. We performed a hematology screen for endocrine, metabolic, autoimmune, neoplastic, and infectious diseases, which were all negative. Cerebrospinal fluid (CSF) studies, including a paraneoplastic, an autoimmune antibody panel, and a tubercular, fungal antibody survey were also negative. Fluid-attenuated inversion recovery (FLAIR) of the brain MRI showed hyperintensities in the bilateral frontal lobe, corona radiate, and near the anterior and posterior horns of the lateral ventricle (**Figure 2A**). Diffusion-weighted imaging (DWI) showed hyperintensities in the bilateral caudate nucleus and putamen (**Figure 2B**). A diagnosis of CJD was considered. One week after admission, the second EEG was performed, revealing partially periodic sharp wave complexes (PSWC; **Figure 1B**). No gene mutations associated with genetic CJD were found, but methionine homozygotes were detected at codon 129 of the prion protein gene. The final diagnosis was probable sCJD according to the diagnostic criteria for sCJD (20, 21). Antibiotics, antiviral, and corticosteroid therapies had been tried since admission, but none of them worked. Ultimately, she was discharged from the hospital.

# METHODS

## Search Strategy and Study Selection

We searched PubMed, EMBASE, and PsycINFO up to May 2018 for previous cases using the key words "Creutzfeldt–Jakob disease AND depression." The reports were restricted to those published and unpublished in English and those including human subjects. The inclusion criteria were as follows: case reports, case series, previous literature reviews, or systematic reviews describing sCJD patients with depression as the first symptom and receiving the treatment for depression. The CJD patients had to meet the WHO or 2009 Consortium diagnostic criteria for definite or probable sCJD (20, 21). To minimize confounders, such as the effect of other medications on outcomes, the included cases were those in which depression was the only symptom diagnosed initially. Two authors independently decided on the selection.

#### Data Extraction

periodic sharp wave complexes (PSWC).

The data extracted included study name, study characteristics, patient characteristics, and the duration, institutional care, symptoms, examinations, treatments, and diagnosis of distinct phases. The duration of CJD patients was divided into three phases based on the main symptoms. The first phase was the prodromal phase, with mental manifestations; the second phase was the intermediate phase, with progressive dementia, myoclonus, psychiatric disorder, pyramidal signs, and extrapyramidal performance; the third phase was the late phase, with incontinence, AM, coma, or decorticate rigidity. If an article did not distinguish between the duration of the second and third phases, we utilized a value of half of the total duration of the two periods as their respective durations. Data were graded by two authors independently according to the Oxford Centre for Evidence-Based Medicine levels of evidence (22).

### Statistical Analysis

A systematic analysis was performed. Categorical variables were described using proportions and continuous variables using medians and interquartile range (IQR). A Kaplan–Meier

survival analysis was conducted in those patients for whom we had data on the three-phase duration and the use of antidepressants. Antidepressants were categorized into three classes, SSRIs, newer non-SSRI antidepressants (mostly SNRIs), and first-generation antidepressants (mostly TCA) according to the Anatomical Therapeutical Chemical (ATC) classification system (World Health Organization, 1999). The log rank test was used to compare the survival distributions of different groups. Finally, a multivariate Cox regression analysis with Enter was undertaken to determine the predictors of survival. Due to the small number of cases, we considered only three factors: gender, age, and antidepressant type. The model with a significant score test and a smaller deviance in likelihood ratio test will be preferred. Significance was set at *P* < .05 (two-sided test). Statistical analysis was completed using SPSS v17.0 (SPSS Inc., Chicago, IL, USA).

# RESULTS

### Study Identification and Characteristics

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed (23). The PRISMA flow diagram is depicted in **eFigure** in the **Supplementary Material**. In our literature search, we identified 13 cases from 12 articles that met our inclusion criteria for qualitative analysis (24–35). Subsequently, 11 cases from 10 articles were included for qualitative analysis (24–26, 28, 30–35). With the addition of our case, a total of 12 subjects could be included in the Kaplan– Meier survival and multivariate Cox regression analysis.

The characteristics and evidence levels of the 14 cases published from 1993 to 2017 are shown in **eTable** in the **Supplementary Material**. All included articles were case reports. The age of all subjects was 58.8 (55.5–61.5) years with 11 (79%) being female. After administration of antidepressants, only 1 case out of 13 (8%) showed improved depressive symptoms.

### Survival Time of Sporadic Creutzfeldt– Jakob Disease Patients with Different Antidepressants

A Kaplan–Meier survival curve for all of the sCJD patients who had used antidepressants is shown in **Figure 3**. The median survival time of all of the cases was 6.0 months. The cumulative incidences with survival times less than 3, 6, and 12 months were 30.0%, 90.0%, and 100%, respectively. All of the patients died within 1 year after onset.

The use of antidepressants in 12 cases is as follows: 3 (25%) were given SNRIs (1 censored), 4 (33%) were administered SSRIs (2 censored), and 5 (42%) were treated with first-generation antidepressants. The median survival times for cases with SNRIs, SSRIs, and first-generation antidepressants were 2.0, 4.0, and 6.0 months, respectively. The median survival time of patients with SNRIs was significantly shorter than those treated with firstgeneration antidepressants (log rank, *P* = .008) and relatively shorter than those with SSRIs (log rank, *P* = .090). Furthermore, the median survival time of patients receiving SSRIs was nonsignificantly shorter than those with the first-generation antidepressants (log rank, *P* = .615).

### Predictors of Survival Time in Sporadic Creutzfeldt–Jakob Disease Patients With Depression

The Cox regression model including age and antidepressant types (**Table 1**) was preferred (likelihood ratio test, deviance = 25.469; score test, *P* = .043). Compared to first-generation antidepressants, the use of SNRIs [hazard ratio (HR), 23.028; 95% confidence interval (CI), 1.401 to 378.461; *P* = .028] remained


*sCJD, sporadic Creutzfeldt–Jakob disease; HR, hazard ratio; C, confidence interval; NA, not applicable (outcome not investigated); SNRIs, serotonin and noradrenaline reuptake inhibitors; SSRIs, selective serotonin reuptake inhibitors.* 

*aAdjusted for age and antidepressant types.* 

*\*P < .05.*

independently associated with significantly shorter survival time in sCJD patients with depression.

## DISCUSSION

Since depression is one of the most common global mental health conditions, the use of antidepressant drugs has increased dramatically with almost half of the prescriptions being for some off-label indication (36). However, our investigation revealed that almost none of the sCJD patients experienced any relief of their depressive symptoms after the antidepressant treatment. Furthermore, the median survival time of sCJD patients receiving SNRI therapy was shorter than the average survival of sCJD patients (2.3 vs. 4.6–17.4 months) (16). Thus, antidepressants do not seem to have any beneficial effect on sCJD patients with depression, a finding consistent with previous clinical studies not only on sCJD patients but also those with dementia (10, 19). Likewise, the efficacy characteristics of antidepressants indicate that antidepressants appear to display relatively poor efficacy in people older than 65 years (37). Based on the neurotransmitter receptor hypothesis of antidepressant drugs, the amount of neurotransmitter changes rapidly after an antidepressant is administered. But the clinical effects appear only weeks later (usually 6–12 weeks) (37). Due to the rapid progress of the disease, sCJD patients often use antidepressants for only a brief period. Therefore, the drugs usually cannot achieve clinical efficacy but are instead likely to exert unwanted side effects.

The question arises as to why patients with sCJD receiving antidepressants seem to deteriorate faster. Since antidepressants mainly alter neurotransmitter levels, we postulate that this deterioration must be related to these changes. Several independent lines of evidence support this postulation. sCJD resembles the degenerative dementias. Studies of AD, the best-known of the degenerative dementias, have proved that the accumulations of β-amyloid (Aβ) and tau proteins damage neurons and synapses, whereas the change in neurotransmitters such as acetylcholine (ACh) occurs at the initial stage (38). Similarly, the cause of sCJD neuropathological changes has also proved to be a reversible process, such as synaptic or neuronal dysfunction (39). Interestingly, patients with sCJD also have higher concentrations of Aβ and tau proteins in their serum and CSF (40, 41). Aβ may be propagated in a prion-like manner (42, 43). Similar observations have been made for tau (44). Because sCJD and AD share these common features (45), perhaps we can also attempt to delay the progress of sCJD by regulating the level of neurotransmitters. Furthermore, the typical lesions in MRI and histologic appearance in sCJD consist of cortical, basal ganglia, and cerebellum (46, 47). It was observed that the clinical target areas in the brainstem of prion-infected mice were the locus coeruleus, the nucleus of the solitary tract, and the pre-Bötzinger complex (48). These brain areas are exactly those in which the neuronal cell bodies generating neurotransmitters are mainly located or the areas innervated by their axonal projections.

How do these neurotransmitters modulate disease progression? According to our study, the survival period of sCJD patients is related to the type of antidepressants. By analyzing the pharmacological characteristics, we postulate that elevations in 5-hydroxytryptamine (5-HT) and norepinephrine (NE) may worsen the condition, although the sedative effects mediated by anti-histamine (HA), anti-ACh, and blockade of α-1 adrenoceptors may contribute to the relief of symptoms. Acute stimulation of the 5-HT can produce symptoms similar to sCJD (37). Neurotransmitters exist in many brain areas, but which area plays the key role? When SSRI treatment is initiated, the concentrations of 5-HT are elevated to a much greater extent at the somatodendritic area located in the midbrain raphe, rather than in the brain areas where the axons terminate (37). Therefore, SSRIs may exert more significant effects on the brainstem in patients with sCJD. However, the pathological changes in sCJD do not occur in the brainstem but rather in the projection pathways of neurotransmitters, such as cortical, basal ganglia, and cerebellum. This raises the question of how changes in the brainstem's neurotransmitter activities affect other brain areas. Taking into account the symptoms (such as myoclonus that occurs at night, AM) of patients with sCJD, we postulate that one pathway through which brainstem's neurotransmitter activities trigger cognitive impairment encountered in sCJD patients may be through its disruption of sleep centers in the brainstem. AM is a disorder caused by damage to the ascending reticular activating system (ARAS) centered on the brainstem. Arousal is regulated by ARAS, which is influenced in large part by five key neurotransmitters: HA, dopamine (DA), NE, 5-HT, and ACh. Changes in these neurotransmitters can cause sleep disorders, i.e., rapid eye movement (REM) sleep without atonia (RSWA), and nonrapid eye movement (NREM) sleep disruption. Clinical studies have shown that SSRIs and SNRIs are associated with a higher prevalence of RSWA (49), explained in part by the theories about REM sleep initiation that advocate for a double switch model, possibly triggered by neurons located in the brainstem (50). Sleep disorders can cause many symptoms similar to sCJD, such as psychiatric symptoms (fear, anger, aggressive behavior, etc.), increased muscle tone, and most notably, cognitive impairment. For example, REM sleep behavior disorder (RBD) is by far the strongest clinical predictor of onset of neurodegenerative diseases (51). The presence of RBD in Parkinson's disease (PD) is associated with higher risk of cognitive decline (52). The reduced NREM slowwave activity (SWA) generation was associated with impaired hippocampus-dependent memory consolidation (53). The Aβ burden in the medial prefrontal cortex correlates significantly with the severity of impairment in NREM SWA generation (53). Even one night of sleep deprivation could result in a significant increase in Aβ burden in the brain (54). Thus, the dual excitatory effects of 5-HT and NE may exacerbate the sleep deprivation encountered in sCJD patients, causing a cascading effect and then triggering cognitive impairment.

Why is the effect of neurotransmitters so rapid in sCJD patients? One reason may be the pathological overactivity of the brain's serotonergic system in this disease. This hypothesis is supported by the evidence that the mean tryptophan hydroxylase (TPH)-positive cell size was significantly greater and cells were more intensely stained in CJD compared to controls (55). This may result in an increase in release of 5-HT. Coupled with the cascade effect of neurotransmitters, the actual effects may be amplified. The increase in 5-HT also reduced the release of DA in the prefrontal zone by negative feedback regulation (37). The reduction of DA may cause some symptoms similar to sCJD, such as cognitive impairment and apathy. Another reason could be that synapses in sCJD may be more vulnerable. The pathological features of CJD indicate that the vacuole in the cytoplasm is the cystic dilation of neurons and necrosis of the necrotic membrane. The cell membrane damage of CJD seems to be more serious than AD, where amyloid plaques form outside the neurons and neuron fibers entangle within the neurons. Damage to the synaptic membrane leads to a decrease in neurotransmitter receptors. In response to this change, the remaining receptors may be in a hypersensitivity state, or the number of receptors may increase (37), which may further enhance the effects of neurotransmitters such as 5-HT.

Many of the families of patients with sCJD complain of the delay in diagnosis and the plethora of misdiagnoses (56). Expediating a sCJD diagnosis is of great significance. Almost half of patients were misdiagnosed first as "psychiatric patients" (57). Consequently, it is very important for psychiatrists to consider CJD among the possible differential diagnoses in elderly patients. Our investigation suggests that it may be helpful to use imaging such as functional MRI and positron emission computed tomography (PET) to detect earlier changes in patients with sCJD.

Of course, our study has some limitations. First, the number of available cases is too small, and in many cases, the description of psychiatric symptoms and details of the antidepressants were inadequate. Second, because case reports tend to report exceptional situations, there is inevitably some bias. However, we think it is a reasonable approach to study CJD by undertaking case analysis or studies of one single individual. Due to the rapid progression of CJD, studies on population samples often overlook certain unique changes.

Investigations into sCJD have mainly focused on autopsybased pathology, but little is known about neurophysiological changes. We hope this study will draw attention to the depressive symptoms of sCJD patients and the underlying neurophysiological mechanisms.

#### CONCLUSIONS

The use of antidepressants was associated with a shorter survival time of sCJD patients, especially the use of SNRIs. The possible neurotransmitter changes may be due to a pathophysiological mechanism in CJD. Functional imaging and use of the polysomnogram to detect earlier changes in sCJD patients may be worth trying.

### ETHICS STATEMENT

This case study was carried out in accordance with the recommendations of the Ethical Committee of China Medical University. The case study has been approved by the Ethics Committee of China Medical University. The subject gave written informed consent in accordance with the Declaration of Helsinki. Written informed consent was also obtained from each patient for the publication of this case report.

# AUTHOR CONTRIBUTIONS

CZ had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. CZ and YLia contributed to the study concept and design. YLi contributed to the case report.

# REFERENCES


YLia, YLi, MG, and SZ contributed to the acquisition, analysis, or interpretation of data. YLia, HW, and XC drafted the manuscript. CZ conducted the critical revision of the manuscript for important intellectual content. All authors performed the statistical analysis.

# FUNDING

This work was supported by The Liaoning Province Key Research and Development Project Critical Project (no. 2017225005, CZ), The Shenyang Municipal Bureau of Science and Technology International Exchange and Cooperation Project (no. 17-129- 6-00, CZ), and China Medical University High-level Innovation Team Training Plan (no. 2017CXTD02, CZ).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyt.2019.00297/ full#supplementary-material


**Conflict of Interest Statements:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Liang, Li, Wang, Cheng, Guan, Zhong and Zhao. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Clinical Study of Restless Leg Syndrome Accompanied by Psychological Symptoms Induced by High-Dose Treatment With Madopar

#### *Lei Zhu1, Jing Li1, Chongyang Ren1, Mei Zhang1\*, Min Xue1, Chuanqing Yu1 and Weili Zhang2,3\**

*1 Department of Neurology, The First People's Hospital of Huainan, The First Affiliated Hospital of Anhui University of Science and Technology, Huainan, China, 2 State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China, 3 Beijing Institute for Brain Disorders, Center for Brain Disorders Research, Capital Medical University, Beijing, China*

#### *Edited by:*

*Chunxue Wang, Beijing Tiantan Hospital, China*

#### *Reviewed by:*

*Li Cao, Shanghai Jiao Tong University, China Chun-Feng Liu, Second Affiliated Hospital of Soochow University, China*

#### *\*Correspondence:*

*Mei Zhang hnzhangmei2008@163.com Weili Zhang zhangweili1747@yahoo.com*

#### *Specialty section:*

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

*Received: 04 September 2018 Accepted: 08 May 2019 Published: 24 May 2019*

#### *Citation:*

*Zhu L, Li J, Ren C, Zhang M, Xue M, Yu C and Zhang W (2019) Clinic Study of Restless Leg Syndrome Accompanied by Psychological Symptoms Induced by High-Dose Treatment With Madopar. Front. Psychiatry 10:360. doi: 10.3389/fpsyt.2019.00360*

Objectives: Some neurological disorders demonstrate indistinguishable psychological symptoms at an early stage, especially when accompanied by jitters similar to those in Parkinson's disease. During dopamine replacement therapy, some patients display restless leg syndrome (RLS)-like symptoms. Therefore, we aimed to analyze treatment strategies and the prognosis of RLS caused by high-dose Madopar.

Methods: Nine patients who were misdiagnosed with Parkinson's disease, taking a high dose of Madopar, and showed symptoms of anxiety, depression, and somatization were recruited. Clinical data were collected, and strategies of treatment and prognosis were analyzed.

Results: Seven patients demonstrated varying degrees of anxiety and depression, and the other two cases were misdiagnosed as Parkinson's disease. During Madopar treatment, patients gradually showed aggravated symptoms, including swelling, numbness, pain, and other sensory abnormalities in both lower extremities, which spread to both upper extremities in a few patients. Among the seven patients, symptoms of anxiety, depression, insomnia, and somatization significantly worsened during the observation period. The average time from taking Madopar to the appearance of RLS was 2.6 ± 0.6 months, the average time to clinical diagnosis was 18.17 ± 9.40 months, and the average dosage of Madopar was 1.44 ± 0.21 g per day. Gradually reducing the Madopar dosage and administering a small dose of long-acting dopamine preparation greatly alleviated the symptoms after 3 months.

Conclusion: A high dose of Madopar can cause RLS-like symptoms accompanied by anxiety, depression, insomnia, and other mental health symptoms. These symptoms should be more closely monitored by clinicians.

Keywords: Madopar, restless leg syndrome, anxiety, depression, psychological symptoms

### INTRODUCTION

Psychiatric symptoms, such as anxiety, depression, insomnia, and somatization, are the clinical manifestations of common psychiatric and neurological diseases. Many neurological disorders show these indistinguishable psychological symptoms in the early stages, especially when symptoms similar to Parkinson's disease (e.g., difficulty walking, stiff limbs, and tremors) are present. These patients are likely to be misdiagnosed with Parkinson's disease and are treated with dopamine replacement therapy; however, in rare cases, increasing the dosage of dopamine can elicit restless leg syndrome (RLS). Extended durations of these psychiatric symptoms can be detrimental to the patient's physical and mental health.

The present study assessed a group of patients who were misdiagnosed with Parkinson's disease and were administered large doses of Madopar. All the patients exhibited rare RLS-like symptoms, such as difficulty in walking, stiff limbs, and tremors, which were accompanied by anxiety, depression, and other psychiatric symptoms. Clinical data of all patients were collected, and strategies of treatment and prognosis were analyzed.

#### MATERIALS AND METHODS

#### Patients

The present study was approved by the Ethics Committee of the First People's Hospital of Huainan, and written informed consent was provided. Twelve patients demonstrating symptoms of anxiety, depression, and somatization due to misdiagnosis of Parkinson's disease and taking a large dose of Madopar were identified and recruited from January 2010 to December 2017. Two Parkinson's disease patients did not meet the inclusion criteria, and one patient declined to follow up. Therefore, nine patients (47–78 years old) were enrolled. All nine patients were hospitalized, and after a detailed evaluation of medical history, rigorous neurological physical examination, and related auxiliary tests, they were determined to not meet the criteria for Parkinson's disease according to the British Parkinson's Disease Society (1). Anxiety, depression, insomnia, and somatization symptoms were diagnosed as depression and anxiety disorders according to the *Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition* (DSM-5) (2). All patients underwent general routine examinations as well as biochemical and imaging examinations. No abnormalities were noted except for the primary disease. The diagnosis for RLS was based on clinical criteria (3) and included an urge to move the legs, usually associated with unpleasant sensations; symptoms occurring during periods of rest, such as sitting or lying down; symptoms relieved by movement; and worsened symptoms in the evening or night.

The education level of all patients was above primary school, and they could independently complete the questionnaire without communication barriers. All patients agreed to follow up.

#### Laboratory and Imaging Examinations

Routine blood, urine, fecal, serum glucose level, liver and kidney function, thyroxine, and electrolyte laboratory and physical examinations were conducted. Electroencephalogram and brain magnetic resonance imaging were performed in all patients.

#### Clinical Evaluation and Follow-Up

Severity of RLS was evaluated on the basis of the International RLS Rating Scale (IRLS-RS) (4). The diagnosis and severity of insomnia, anxiety, and depression in all patients were assessed by two neurologists and a psychiatrist according to the Insomnia Severity Index (ISI) (5), Hamilton Anxiety Rating Scale (Hamilton) (6), Hamilton Depression Rating Scale (HDRS) (7), and DSM-5 diagnostic criteria (2) combined with clinical symptoms and signs. Follow-up data for all patients with RLS were obtained during face-to-face or telephone interviews.

#### Clinical Research Flow

The clinical study flow is shown in **Figure 1**.

#### Statistical Analysis

All statistical analyses were performed using Statistical Product and Service Solutions (SPSS) version 19.0 (SPSS Inc., Chicago, IL, USA). The normality of the distribution was assessed using the Kolmogorov–Smirnov test. Normally distributed quantitative data were presented as "mean ± standard deviation (SD)." The international RLS scores of patients before and after treatment were compared by *t*-test. The anxiety, depression, and insomnia

scores of patients before and after taking a large dose of Madopar were compared using the Student's *t*-test*. P* values <0.05 were considered significant.

#### RESULTS

Nine patients took Madopar orally due to being misdiagnosed with Parkinson's disease, and the starting dosage ranged from 1/2 to 1 tablet (0.25 g/tablet). All patients gradually increased the amount of medication administered. Some were under the guidance of a doctor, but then to achieve the "curative effect," patients increased the amount of medication themselves. Some patients increased their doses by themselves from the beginning (i.e., without the doctor's assistance). The amount of medication in most patients was 2–3 tablets per dose, 3–4 times per day, which was at maximum 5 tablets per dose, 3–5 times a day in one case. When the average dosage reached 6–8 tablets per day and the duration of administration lasted 2–4 weeks, the onset of bilateral lower limb discomfort appeared. Initially, the symptoms were minimal, which did not alert the attention of the patients. As the medication dosage and duration increased, so did the symptoms, which appeared as unexplained abnormal sensations in both lower extremities to varying degrees, such as numbness, swelling, crawling, burning, and traction pain at night. The symptoms could be temporarily reduced by activity, which forced patients to stay out of bed for exercises, which affected their sleep. As a result, patients typically increased the dose of Madopar, which could reduce the symptoms, especially when the symptoms were unbearable. The increasing dosage of Madopar could effectively improve the symptoms, and thus forced patients to increase the amount of medication.

During this cycle, when symptoms appeared during the daytime, the upper limbs and occasionally the entire body displayed varying degrees of involvement. As shown in **Table 1**, the average time from the use of the Madopar to the onset of RLS symptoms was 2.61 ± 0.60 months. For symptoms to appear, the minimum of the average daily dose of Madopar was 1.44 ± 0.21 g; moreover, the average duration for the nine patients with RLS from the time of high-dose Madopar administration to the time of hospital admissions was 18.17 ± 9.40 months. The original symptoms of anxiety, depression, insomnia, and general discomfort worsened in seven patients before onset of the disease. The other two cases displayed anxiety, depression, and insomnia, as well as whole-body burning-like and mobile pain, accompanied by the gradually aggravated discomfort of the bilateral lower limbs. As shown in **Table 2**, the symptoms of anxiety, depression, and insomnia were significantly worse in all nine patients after taking a large dose of Madopar (*P* < 0.0001, *P* < 0.05, *P* < 0.0001).

All nine cases were asked to gradually reduce their dose of Madopar. Low doses of long-acting dopamine agents, dopamine receptor agonists, α2δ calcium channel ligands, clonazepam, and other drug treatments were administered. All psychiatric symptoms were greatly alleviated but did not fully disappear and lasted for approximately 2 years. The severity of symptoms in seven patients with more than 6 months of disease course was significantly improved, but after 3 months of treatment, there was no obvious further improvement of symptoms and fluctuations were present.

As shown in **Table 3**, there was a significant difference in the IRLS scores 1 month before and after treatment. During the first month of follow-up, IRLS scores of all patients were significantly lower than the initial assessment [21.22 + 2.05 points (indicative of severe symptoms) compared to 35.33 ± 2.40 points; *P* < 0.0001]. At the 3-month follow-up, the IRLS scores of patients were significantly lower than the first month of follow-up (*P =* 0.001). Finally, at the 6-month follow-up, the IRLS scores were 13.89 ± 5.06 points, indicating moderate severity. Before treatment and at the 1-month follow-up, there was statistically significant difference between IRLS scores (*P* < 0.0001); when the 3-month follow-up was compared to this, although the symptoms were improved, it was not found to be statistically significant (*P* = 0.33). And, at the 12-, 18-, and 21-month follow-ups, when compared with the 6-month follow-up, two cases at the 21-month follow-up demonstrated that the IRLS score continued to decrease, reaching a mild level of severity; however, the subsequent treatment did not provide additional benefits to the remaining seven cases. The RLS symptoms showed no obvious improvements compared to the 6-month follow-up and were still classified as moderately severe symptoms. Another patient died of primary disease (cirrhosis and hemorrhage of upper digestive tract) during the 12 months of follow-up. This demonstrated that improvements were no longer obvious after 3 months, which suggested that early diagnosis and treatment might be the key factor to improving prognosis.

#### DISCUSSION

RLS is a common nervous system sensory dyskinesia disease, and the clinical manifestations are extreme discomfort during rest and nocturnal sleep. Symptoms can be remitted through movement of the lower extremities, which forces patients to continue to move their limbs, therefore disturbing sleep and rest. According to the etiology, RLS can be divided into two subtypes: primary and secondary. The former etiology is unclear and may be heredity, while the latter is often due to iron deficiency, pregnancy, chronic renal failure, and other causes. In the present study, we reported for the first time that nine patients who had no family history of primary or secondary RLS showed RLS-like symptoms accompanied by anxiety and depressive symptoms, as induced by high-dose Madopar.

Previous reports and case studies have suggested that certain medications may cause or exacerbate RLS. These medications include several classes of antidepressants, including tricyclic antidepressants (8) such as imipramine (9); selective serotonin or norepinephrine reuptake inhibitors (10), such as citalopram (11), escitalopram (12), fluoxetine (13), sertraline (14), paroxetine (15), trazodone (16), venlafaxine (17), dutoxetine (18), and mirtazapine (19); and neuroleptics that have significant dopaminergic blockade (20), such as olanzapine (21), risperidone (22), and quetiapine (23). In addition, antihistamines operating on the H1 receptor (24) and selected antiemetics with dopamine antagonism such as metoclopramide (25) and prochlorperazine (26) have also been associated with RLS. However, as none of the nine patients we observed took the aforementioned drugs, RLS-like symptoms caused by these drugs were excluded.



TABLE 2 | Comparison of anxiety, depression, and insomnia scores of the nine patients.

*B, before taking Madopar; A, after taking large-dose Madopar; HARS, Hamilton Anxiety Rating Scale; HDRS, 24-item Hamilton Depression Rating Scale; ISI, Insomnia Severity Index; p-value, Student's t-test.*

The clinical manifestations of anxiety, depression, and somatoform disorders are complex and diverse. Other than the emotional aspects, these illnesses can manifest as different forms of somatic symptoms, such as dizziness, headache, limb weakness (especially in the lower extremities), and difficulty walking, which is likely to be misdiagnosed as a primary disease. Among the nine patients in the present study, seven had varying degrees of anxiety and depression. For example, Case 5 was misdiagnosed with Parkinson's disease and treated with Madopar due to the spasm gait of the double lower limb after encephalitis. Since the effect of treatment was not obvious, the patient increased the dosage by themselves. RLS-like symptoms and severe anxiety, insomnia, and other psychiatric symptoms occurred with an increasing amount of medication. Case 2 was diagnosed with a sleep-related leg spasm according to the predisease clinical manifestation, and then the patient was misdiagnosed with Parkinson's disease and treated with a high dose of Madopar, which induced RLS-like symptoms and anxiety. As one of the effective drugs to treat RLS, Madopar has been widely used in clinical practice; however, the onset of RLS is very rare and its pathogenesis should be further discussed.

In 2016, single-photon emission computed tomography imaging was used to study the pathophysiological mechanisms of RLS at the Tri-Service General Hospital, National Defense Medical Center (27). The results showed a significantly reduced uptake in striatal dopamine transporter (DAT) density and activity in RLS patients (27). This study supported that symptoms of RLS resulted from the striatum due to dopaminergic system dysfunction (27).

To date, many studies have shown that using drugs such as levodopa and other dopamine agonists can significantly improve the symptoms of RLS; therefore, the central dopaminergic nervous system (particularly the nigra-striatum system or intermediate cortical system) has been considered to be associated with the onset of RLS. In the present group, patients displayed RLS symptoms after the use of long-term high doses of Madopar (from the use of Madopar to the onset of RLS-like symptoms), suggesting that the cause was the dysfunction of the central dopamine system. A number of studies have reported that the use of dopamine drugs in the treatment of RLS may promote symptom deterioration, reverse jump, and other adverse reactions. This is especially true after the long-term use of levodopa, as the proportion of deteriorated symptoms is 18–80% (28). Since these adverse reactions are more common in patients with long-term high doses of levodopa treatment, it suggests that the mechanism of symptom deterioration may be associated with dopamine overdose in the central nervous system (29). Therefore, we speculate that the mechanism of RLSlike symptoms in the nine patients of the present study may be caused by excessive dopamine in the central nervous system after the use of long-term high doses of dopaminergic agents.

High concentrations of dopamine can excite D1 receptors and cause D1 receptor-related pain, which results in periodic limb movements. Several studies have shown that certain concentrations of external toxic substances [such as levodopa, dopamine (DA)] may damage dopamine transporters (DAT), therefore significantly reducing their abundance. Additionally, compared to the mitochondria, DAT is more sensitive to injury stimulation from external toxic substances. Before the cells' mass death, the number of DAT on the cell membranes is significantly reduced. The remaining DAT functions display compensatory hyperfunction and are therefore eliminated due to the reciprocal inhibition, which allows them to ingest more dopamine and its metabolites into the cells. This results in a large number of free radicals and the inhibition of the mitochondrial respiratory chain, and eventually causes cell death. Therefore, it is speculated that the long-term use of dopamine in patients of the present study may lead to a reduction in the number of dopamine receptors in the brain and spinal cord or a decrease in DAT function, finally resulting in dopaminergic systemic dysfunction and the occurrence of RLS symptoms.

In the present study, nine patients with long-term high doses of dopaminergic drugs displayed RLS symptoms, while the original symptoms of anxiety, depression, insomnia, and somatization appeared or were aggravated. Current epidemiological studies report that RLS is a common cause of insomnia, and the rate of comorbidity with depression and anxiety is high (30). Winkelmann et al. (31) assessed 238 RLS patients with a standardized diagnostic interview [Munich-Composite International Diagnostic Interview for DSM-IV (*Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition*)]. Rates of anxiety and depressive disorders were compared between them and 2,265 community respondents from a nationally representative sample. RLS patients revealed an increased risk of having anxiety and depressive disorders with particularly strong associations with panic disorder, generalized anxiety disorder, and major depression (31). Moreover, the Baltimore Epidemiologic Catchment Area follow-up study suggested a strong association between RLS and major depressive disorder and/or panic disorder (32). An anonymous survey study in Appalachia suggested that those


with RLS were significantly more likely to indicate a history of depression and anxiety and report sleep impairments both 4 and 7 days/week, with a mean sleep duration <5 h/night (33). These associations increased in both strength and magnitude with increasing symptom frequency (33). More recently, a study on the clinical characteristics of RLS in adult patients from Peking Union Medical College Hospital demonstrated that primary RLS patients suffer from poor sleep and are more susceptible to anxiety and depression (34). The scores of Hospital Anxiety and Depression Scale for depression and anxiety were significantly correlated with those of the Pittsburgh Sleep Quality Index and IRLS (34).

The underlying cause of the high incidence of RLS and anxiety and depression is unclear. It is possible that these illnesses may share a basic pathophysiological mechanism leading to their development. Pan et al. (35) explored the regional gray matter (GM) density in depressed drugnaïve RLS patients using voxel-based morphometry, which showed that GM density of the bilateral anterior cingulate cortex (ACC) was significantly reduced in RLS patients with depressive symptoms (RLS-D) compared to RLS patients without depressive symptoms or healthy controls. Additionally, a significant negative correlation between right ACC density and HDRS scores and duration of depressive symptoms in patients with RLS-D was found (35). It was speculated that depressive symptoms are associated with GM abnormalities in the ACC of patients with RLS.

In the present study, anxiety was more obvious than the depressive symptoms in our patients; however, there is no evidence to suggest its mechanism to date, thereby requiring further discussion. By administering small doses of longacting dopamine agents, dopamine receptor agonists, α2δ calcium channel ligands, clonazepam, and other treatment to this group of patients, the IRLS score gradually declined and symptoms improved, but RLS symptoms did not completely disappear. Here, the prognosis of RLS was significantly different from secondary RLS patients, as they typically demonstrated complete disappearance of symptoms after cause elimination. It is suggested that long-term high doses of Madopar cause excessive accumulation of dopamine in the central nervous system, thereby irreversibly decreasing the number of dopamine receptors or DAT function, resulting in the persistence of clinical RLS-like symptoms.

The present study reports drug-induced (high-dose levodopa) RLS, which is different from the idiopathic RLS in treatment; however, the method treatment is identical. First, the dose of levodopa was gradually reduced, but as the clinical symptoms of the patients were severe, measures needed to be taken accordingly, depending on the patient's condition and their accompanying anxiety, depression, and insomnia. We referenced the European guidelines on management of restless legs syndrome:report of a joint task force by the European Federation of Neurological Societies, the European Neurological Society and the European Sleep Research Society (36). According to the recommendations and precautions for drug treatment, the principle of individualization was followed, simultaneously supplemented by physical therapy (hot water bath before sleep,

TABLE 3 | Treatment and outcome of treatment of the nine patients and follow-up.

limb massage, etc). Thus, the clinical symptoms of the nine patients were relieved to varying degrees. The specific method recommends the minimization and withdrawal of the use of Madopar. The therapy was changed to a small dose of a longacting dopamine agent (Xining 100–200 mg/day) to reduce the risk of dosage increase. Madopar was finally discontinued in five of the nine patients; however, Case 1 was found to have difficulty when the Madopar dose was reduced to 3 pills/day. A small dose of a dopamine receptor agonist and a small-tomoderate dose of a α2δ calcium channel ligand (pregabalin) were added depending on the severity of the patient's symptoms. To improve the symptoms of anxiety, depression, and insomnia, the therapy was supplemented by small-dose paroxetine, duloxetine, and olanzapine; an obvious curative effect was achieved by all of them.

In summary, clinicians should be mindful of differential diagnoses when patients present with walking difficulties and limb stiffness. In addition to the original diseases, anxiety, depression, and somatization disorders should be considered. In particular, clinicians should strengthen the management of patients who use dopaminergic agents to reduce the great physical and mental adverse events due to misdiagnosis and mistreatment.

# CONSENT FOR PUBLICATION

The nine patients gave written consent for both participation and publishing the data in a scientific journal. They understood

### REFERENCES


that the information will be published without their names attached, but that full anonymity cannot be guaranteed. They understood that the material may be published and placed on worldwide website and journals. Both the printed version and the website are seen and read by doctors, journalists, and members of the public. The material will not be used for advertising or packaging.

#### ETHICS STATEMENT

This study was approved by the Ethics Committee of the First People Hospital of Huainan, and written informed consent was obtained from the patient for publication of this case report.

### AUTHOR CONTRIBUTIONS

LZ, MZ, WZ: study design and critical revision of the manuscript. LZ, MZ: collection and interpretation of data. LZ: analysis data and drafting of the manuscript. JL, CR, MX, CY: collection data. All authors approved the final version for publication.

# FUNDING

This work was supported by the Chinese Academy of Medical Sciences Innovation Fund for Medical Sciences (2016-I2M-1-006).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Zhu, Li, Ren, Zhang, Xue, Yu and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Compromised Dynamic Cerebral Autoregulation in Patients With Depression

*Ming-Ya Luo1†, Zhen-Ni Guo2†, Yang Qu1, Peng Zhang1, Zan Wang2, Hang Jin1, Hong-Yin Ma1, Shan Lv1, Xin Sun1\* and Yi Yang1\**

*1 Department of Neurology, First Hospital of Jilin University, Chang Chun, China 2 Clinical Trial and Research Center for Stroke, Department of Neurology, First Hospital of Jilin University, Chang Chun, China*

Background: Patients with depression tend to have various comorbid neurological symptoms, but the mechanisms remain unclear. The purpose of this study was to analyze the characteristics of dynamic cerebral autoregulation in depressed patients.

#### *Edited by:*

*Julian Macoveanu, Copenhagen University Hospital, Denmark*

#### *Reviewed by:*

*Anirban Dutta, University at Buffalo, United States Janusz Skrzypecki, Medical University of Warsaw, Poland*

#### *\*Correspondence:*

*Yi Yang doctoryangyi@163.com Xin Sun sjnksunxin@163.com*

*†These authors have contributed equally to this work.*

#### *Specialty section:*

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry Received: 14 October 2018*

*Accepted: 13 May 2019 Published: 31 May 2019*

#### *Citation:*

*Luo M-Y, Guo Z-N, Qu Y, Zhang P, Wang Z, Jin H, Ma H-Y, Lv S, Sun X and Yang Y (2019) Compromised Dynamic Cerebral Autoregulation in Patients with Depression. Front. Psychiatry 10:373. doi: 10.3389/fpsyt.2019.00373*

Methods: Patients (aged ≥ 18 years) who were diagnosed with depression [17-item Hamilton Depression Rating Scale (HAMD) > 17] or suspected of depression (HAMD > 7) were enrolled in this study. Medically healthy volunteers were recruited as controls. The subjects also received the 7-item HAMD. We simultaneously recorded noninvasive continuous arterial blood pressure and bilateral middle cerebral artery blood flow velocity from each subject. Cerebral autoregulation was assessed by analyzing the phase difference using transfer function analysis.

Results: This study enrolled 54 patients with suspected depression, 45 patients with depression, and 48 healthy volunteers. The mean phase difference values were significantly lower in the patients with depression (F = 9.071, P < 0.001). In the multiple regression analysis, depression was negatively correlated with the phase difference values.

Conclusions: Dynamic cerebral autoregulation was compromised in patients with depression and negatively correlated with the depression score. Improving dynamic cerebral autoregulation may be a potential therapeutic method for treating the neurological symptoms of depression.

Keywords: depression, dynamic cerebral autoregulation, transcranial Doppler, transfer function, cerebral hemodynamics

# INTRODUCTION

Depression is the most common psychiatric disorder, a leading cause of disability, and affects nearly 15% of the population (1, 2). Core features of this disorder include depressed mood, loss of interest or pleasure, irritability, change in appetite and sleep, and neurocognitive dysfunctions (3, 4). In addition to suicide ideation and behavior, patients with depression also tend to have comorbid medical illnesses, such as cancer, cardiovascular diseases, and diabetes (5, 6). Depression is associated with an increased risk of stroke morbidity and mortality. These combined conditions generally worsen patient outcomes (7–12).

Despite the prevalence of depression and its considerable burden on global health, knowledge about its pathogenesis remains rudimentary. Previous studies have revealed global and regional changes in the cerebral blood flow of patients with depression compared to healthy individuals (13–15). Cerebral blood flow abnormalities in depression differ in patients, with a varying age of onset (16), disparate responses to antidepressant treatment (17), and diverse family histories (18). Longitudinal research also shows the apparent elevation of regional cerebral perfusion in remissive depression compared to current depression (19). The mechanism of the unusual cerebral blood flow in depressed patients is complex and incompletely understood, and cerebral autoregulation may play a role.

Cerebral autoregulation is the innate ability to maintain appropriate brain perfusion during blood pressure changes. It can be dynamically assessed with transfer function analysis (TFA) between spontaneous fluctuations of arterial blood pressure (ABP) and cerebral blood flow velocity (CBFV) (20, 21). To date, cerebral autoregulation has not been well analyzed in patients with depression. In the present study, we hypothesize that dynamic cerebral autoregulation is compromised in patients with depression, and we use TFA to assess dynamic cerebral autoregulation in depressed patients and explore its relationship with the degree of depression.

### METHODS

#### Participants and Clinical Assessment

Patients whose first complaint was poor sleep and with 17-item Hamilton Depression Rating Scale (HAMD) scores > 7 were included from the Department of Neurology, First Hospital of Jinlin University, from September 2017 to June 2018. Two blinded clinical psychiatrists evaluated the patients' mental health status. All patients had never been treated with antidepressants before. Patients with a history of cerebrovascular diseases (that is, cerebrovascular stenosis and stroke), frequent arrhythmias, anemia and unstable blood pressure, and hyperthyroidism were excluded from the study as controls. The patients with hypertension or diabetes took medications, and their blood pressure and blood glucose levels were well controlled. These patients were divided into two groups, those with depression (HAMD ≥ 17) and those suspected of depression (17 > HAMD ≥ 7). Physical health status was assessed using a questionnaire covering cardiovascular, nervous system, thyroid, and metabolic diseases, and information regarding age, smoking, and drinking habits. A total of 48 medically and psychiatrically healthy volunteers were recruited as controls. Liver and kidney function, blood glucose, blood lipid, blood pressure, electrocardiography, transcranial Doppler (EMS-9 PB, Delica, Shenzhen, China), and carotid ultrasound (IU22, Phillips, Andover, MA, USA) tests were used to exclude subjects who did not meet the study standards.

#### Cerebral Autoregulation Assessment Monitoring

Before the dynamic cerebral autoregulation examination, all of the patients were instructed to avoid caffeine, nicotine, alcohol, and all kinds of sleep medications for at least 24 h. The assessments

were performed in a quiet, dedicated monitoring room with minimal external stimuli. The subjects were instructed to breathe spontaneously and assumed a supine position with a head elevation of 30° when baseline ABP (automatic blood pressure monitor, Omron 711) was measured. Signals were recorded after a 10-min rest. Beat-to-beat ABP was noninvasively recorded through servo-controlled finger plethysmography (Finometer Model 1, FMS, Netherlands), and continuous bilateral middle cerebral artery blood flow velocity was recorded with 2-MHz probes with an insonation depth of 45 to 60 mm attached to a customized head frame (MultiDop X2, DWL, Sipplingen, Germany). Stable end-tidal carbon dioxide (CO2) levels were confirmed through a capnograph with a face mask attached to the nasal cannula. Each participant's blood pressure and blood flow velocity were recorded for 10 min. The data were then stored for further dynamic cerebral autoregulation examination analysis.

#### Analysis of Dynamic Cerebral Autoregulation

The recorded data were analyzed blindly using a laptop computer equipped with MATLAB (MathWorks, Natick, MA, USA). The beat-to-beat alignment of the data was acquired using a crosscorrelation function to eliminate possible time lags. By using a cross-correlation function between ABP and CBFV, we may calculate the correlation at each time lag (by sample). We can then find the time lag with the maximum correlation, suggesting that the two signals are synchronized at this time lag. This is considered as the time delay between ABP and CBFV, which is likely caused by the data acquisition devices. We used a thirdorder Butterworth low-pass filter (cutoff at 0.5 Hz) as an antialiasing filter before downsampling the data to 1 Hz. A TFA was applied for evaluating cerebral autoregulation (22). TFA is a frequency domain analysis that calculates the "phase shift" between the CBFV and blood pressure waveforms in the 0.06– 0.12 Hz frequency domain to evaluate cerebral autoregulation where the derived parameters were considered most relevant to autoregulation hemodynamics (23). In the current study, a phase shift was accepted for later statistical analysis only if the calculated coherence of one measurement was >0.49 within 0.06–0.12 Hz (24–26), in order to ensure that there was at least 49% linearity between ABP and CBFV. Otherwise, it is invalid to use TFA for the assessment, as it is a linear model.

### Statistical Analysis

The statistical data were analyzed using Statistical Program for Social Sciences version 21.0 (SPSS, IBM, West Grove, PA, USA). Continuous and discrete variables were respectively compared between the patients and healthy controls using analysis of variance. Liner multiple regression was used to explore the association between the phases and characteristics of the patients. The relationships between the phase difference values and the HAMD scores were analyzed using Spearman's rank-order correlation analysis. Multiple linear regression analysis was used to investigate the effects of the covariates on the phase difference. P values < 0.05 were considered statistically significant.

# RESULTS

#### Baseline Characteristics and Phase Difference

The characteristics of the participants are listed in **Table 1**. This study analyzed 45 patients (median age = 47.95 ± 13.30 years, 12 males) with depression, 154 (median age = 47.19 ± 4.48 years, 57 males) suspected of depression, and 48 healthy controls (median age = 47.29 ± 12.24 years, 19 males). The prevalence of hypertension and hyperlipidemia in the patients with depression and those suspected of depression was higher than in the control group.

There was no significant difference between left and right phase difference values in the patients and the controls. The phases in the patients suspected of depression and in those with depression were significantly lower than in the corresponding hemispheres of the healthy controls (**Table 1**, **Figure 1**).

#### Multiple Linear Regression Analysis

Age, sex, diabetes, hypertension, hyperlipidemia, tobacco smoking habits, drinking habits, left middle cerebral artery, right middle cerebral artery, mean ABP, and heart rate did not influence the phase averages. However, when the level of depression increased (as evaluated by the HAMD score), the phase difference values were negatively correlated to the HAMD scores (7–17: 95% CI −13.825 to −2.911, P = 0.003; 17: 95% CI −19.725 to −5.802, P < 0.001) (**Table 2**).

### DISCUSSION

This study found that dynamic cerebral autoregulation was impaired in patients with depression and the phase difference value was negatively correlated with the HAMD score. Higher levels of depressive symptoms were associated with increased risk of neurological diseases such as stroke or TIA (27). The patients with depression also suffered from dizziness (28). Neurological diseases in patients with depression can be connected to impaired dynamic cerebral autoregulation.

The potential mechanisms underlying this phenomenon are unknown, but there are theoretical possibilities. Recent studies showed that the modulation of neurotransmitters can malfunction in depression (29, 30), and neurotransmitters such as serotonins have a major impact on cerebral vessel tone and could affect cerebral autoregulation (31, 32). In addition, the clinical evidence suggests increased pro-inflammatory markers in patients with depression (increased TNF-α, CRP, and IL-6) (33, 34). These inflammatory factors (35, 36) also can affect the endothelial cells and impair cerebral autoregulation (37). Therefore, depression may influence dynamic cerebral autoregulation through neuroendocrine and immunological/inflammation pathways.

Although various studies (38, 39) indicated that diabetes and hypertension are related to dynamic cerebral autoregulation, we did not find that they have significant influence on cerebral autoregulation. One possible explanation for this disparate finding might be related to the incidence of diabetes and hypertension. This study has a low rate of diabetes and hypertension, and we need to have a larger sample size to further explore their relationship.

In this study, the phase values were correlated with the depression levels. As the HAMD scores increased, the phase difference values (dynamic cerebral autoregulation) tended to decrease. The potential mechanisms are unclear. We supposed that the level of serotonins might be a potential mechanism to explain the relationship between the HAMD scores and the phase values. It has been proven that dysfunction of serotonin is related to major depressive disorder (40). Serotonins also have an impact on cerebrovascular function (41). The study of Edvinsson et al. suggested that serotoninergic projections played a significant role in the regulation of cerebral microvascular tone (42). Nevertheless, a study using PET found that citalopram (a selective serotonin reuptake inhibitor) led to alteration of cerebral hemodynamics (32). In addition, the negative correlation between phase values and the HAMD scale suggests

TABLE 1 | Baseline characteristics and phase differences in the patients and controls.


*\*The difference was statistically significant compared to the control group (P < 0.025).*

*ABP, arterial blood pressure; LCMA, left cerebral middle artery; RCMA, right cerebral middle artery.*

FIGURE 1 | The phase differences of left (A) and right (B) derived from the transfer function within significant interval 0.06–0.12 Hz are plotted. Phase difference values (parameter of dynamic cerebral autoregulation) were significantly compromised in patients with depression and suspected of depression compared with the controls, indicating an impairment of dynamic cerebral autoregulation in patients with depression.

TABLE 2 | Multiple regression coefficients for the mean phases of the left and right hemispheres.


*ABP, arterial blood pressure; LCMA, left cerebral middle artery; RCMA, right cerebral middle artery.*

the potential impact of depressive symptoms on dynamic cerebral autoregulation.

The impairment of dynamic cerebral autoregulation in depression indicates that cerebral vascular function may be a therapeutic target of depression. Therefore, improving dynamic cerebral autoregulation may potentially alleviate the neurological symptoms in patients with depression.

This study has some limitations. This was an observational study without in-depth research mechanisms. In addition, the sample size is relatively small. In the future, larger sample sizes and animal studies are needed. In terms of the TFA method, a phase shift is considered valid for further statistical analysis only when the linearity between ABP and CBFV is greater than 49% (24), and TFA is currently the only method that has been studied by multiple centers and standardized by a white paper for the assessment of cerebral autoregulation (23).

#### CONCLUSIONS

Dynamic cerebral autoregulation was compromised in patients with depression and negatively correlated with depression scores. The mechanism of impaired cerebral autoregulation may play a role not only in the development of cerebrovascular diseases but also as a potential therapeutic method for treating the neurological symptoms of depression.

#### REFERENCES


#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Declaration of Helsinki and the Ethics Committee of the First Hospital of Jilin University with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Ethics Committee of the First Hospital of Jilin University.

#### AUTHOR CONTRIBUTIONS

YY, ML, and ZG drafted the manuscript. ZG and ML revised the manuscript. SL, XS, and HM drew the figures. ZW, PZ, HJ was in charge of acquisition of data. YQ and SL performed the data analysis. ML and PZ performed the statistical analysis. XS and YY conceived and designed the manuscript. All authors read and approved the final manuscript.

#### FUNDING

This article was supported by the National Natural Science Foundation of China (Grant No. 81571123), the National Key R&D Program of China (2016YFC1301600) and JLUSTIRT (2017TD-12) to YY.


and impaired blood–brain barrier integrity. *World J Biol Psychiatry* (2012) 13(7):482–92. doi: 10.3109/15622975.2011.583941


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Luo, Guo, Qu, Zhang, Wang, Jin, Ma, Lv, Sun and Yang. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Brain Iron Deposits in Thalamus Is an Independent Factor for Depressive Symptoms Based on Quantitative Susceptibility Mapping in an Older Adults Community Population

*Wenhua Zhang1†, Ying Zhou1†, Qingqing Li1, Jinjin Xu1, Shenqiang Yan1, Jinsong Cai2, Yeerfan Jiaerken2 and Min Lou1\**

#### *Edited by:*

*Yi Yang, First Affiliated Hospital of Jilin University, China*

#### *Reviewed by:*

*Thomas Tourdias, Université de Bordeaux, France Audrey Fan, Stanford University, United States*

#### *\*Correspondence:*

*Min Lou lm99@zju.edu.cn; loumingxc@vip.sina.com*

*†These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Mood and Anxiety Disorders, a section of the journal Frontiers in Psychiatry*

*Received: 14 October 2018 Accepted: 12 September 2019 Published: 15 October 2019*

#### *Citation:*

*Zhang W, Zhou Y, Li Q, Xu J, Yan S, Cai J, Jiaerken Y and Lou M (2019) Brain Iron Deposits in Thalamus Is an Independent Factor for Depressive Symptoms Based on Quantitative Susceptibility Mapping in an Older Adults Community Population. Front. Psychiatry 10:734. doi: 10.3389/fpsyt.2019.00734*

*1 Department of Neurology, the Second Affiliated Hospital of Zhejiang University, School of Medicine, Hangzhou, China, 2 Department of Radiology, the Second Affiliated Hospital of Zhejiang University, School of Medicine, Hangzhou, China*

Objectives: With the trend of an aging population, an increasing prevalence of late-life depression has been identified. Several studies demonstrated that iron deposition was significantly related to the severity of symptoms in patients with depression. However, whether brain iron deposits influence depressive symptoms is so far unclear in the community of older adults. We measured iron deposition in deep intracranial nucleus by quantitative susceptibility mapping (QSM) and aimed to explore the relationship between iron deposition and depressive symptoms.

Methods: We reviewed the data of a community population from CIRCLE study, which is a single-center prospective observational study that enrolled individuals above 40 years old with cerebral small vessel disease (SVD), while free of known dementia or stroke. We evaluated regional iron deposits on QSM, measured the volume of white matter hyperintensities (WMHs) on T2 fluid-attenuated inversion recovery, and assessed depressive symptoms by Hamilton depression scale (HDRS). We defined depressive symptom as HDRS > 7.

Results: A total of 185 participants were enrolled. Participants in depressive symptom group had higher QSM value in thalamus than control group (18.79 ± 14.94 vs 13.29 ± 7.64, *p =* 0.003). The QSM value in the thalamus was an independent factor for the presence of depressive symptoms (OR = 1.055; 95% CI: 1.011-1.100; p = 0.013). The regional QSM values in other areas were not associated with HDRS score (all p > 0.05). No significant correlations were observed between WMHs volume and HDRS score (p > 0.05), or regional QSM values and WMHs volume (all p > 0.05).

Conclusions: Our study demonstrated that iron deposits in the thalamus were related to the depressive symptoms in older adults.

Keywords: iron deposits, white matter hyperintensities, depressive symptoms, quantitative susceptibility mapping, thalamus

# INTRODUCTION

With the trend of an aging population, an increasing prevalence of late-life depression has been identified. The devastating effects of depression in older adults have been reported, including increases in suicide, hastened cognitive decline, worsening physical comorbidities, higher caregiver burden and all-cause mortality (1, 2). However, the mechanism of depression in older adults remains uncertain.

As age increases, intracranial iron deposits increase (3, 4). Studies have indicated that increased iron deposits in the deep gray matter of the brain are closely related to neurodegenerative diseases, such as Parkinson's disease (PD) and Alzheimer's disease (AD) (5). Recent studies also show that increased intracranial iron deposits are associated with emotional behaviors among PD patients, especially depression (6). Moreover, Yao et al. (7) also demonstrated that patients with major depression disorder had a significantly increased susceptibility value in the bilateral putamen than patients with mild-moderate depression or control subjects. Therefore, we presume that in older adults, brain iron deposits might be related to depressive symptoms.

White matter hyperintensities (WMHs), demonstrating related to ischemic, inflammatory and protein deposition, which are commonly seen as confluent or patchy hyperintense areas on T2 weighted or fluid-attenuated inversion recovery (FLAIR) scans, have been reported to be related to both iron deposition and depression in older adults (8). However, the relationship between iron deposits and WMHs is still controversial. Yan et al. (9) observed a significant association between iron deposits in globus pallidus and WMHs volume among patients admitted to hospital, while other study revealed that intracranial iron deposition was not associated with the volume of WMHs (10). Besides, numerous studies have demonstrated that there is a significant but weak association between WMHs and depression (11). However, the relationship between iron deposits, WMHs and depression in older adults remains uncertain.

In previous studies, R2\* was usually used to measure iron content, which was easily affected by different factors, such as calcification. The current study has suggested that quantitative susceptibility mapping (QSM) measures iron deposits more accurately (12). Moreover, postmortem study also found that QSM values were directly proportional to iron content (13). Therefore, in this study we evaluated iron deposition in deep intracranial nucleus on QSM, measured the volume of WMHs on T2-FLAIR, and assessed depressive symptoms using the Hamilton depression scale (HDRS), with the aim to explore the relationship between iron deposition and depressive symptoms, and the role of WMHs among them.

# MATERIALS AND METHODS

#### Subjects

The CIRCLE study (ClinicalTrials.gov ID: NCT03542734) was a single-center prospective observational study that enrolled community residents, which aimed to explore the predictors of small vessel disease (SVD) and cognitive deficits. We reviewed the data of consecutive individuals from CIRCLE cohort between 2017 October and 2018 July. Detailed inclusion criteria was: (1) age above 40; (2) SVD imaging markers (WMHs with Fazekas score 1-3 in periventricular or deep white matter, lacunes, microbleeds) visible on MRI; (3) free of known dementia or stroke (both cerebral infarction and hemorrhage); (4) without any MRI contraindications; (5) free of serious head injury (resulting in the loss of consciousness) or received intracranial surgery; (6) not suffering from cancer. Participants with poor image quality due to motion artifact or with the history of psychotropic drugs were excluded. All participants received neuropsychological testing, retinal digital images and multimodal MRI.

# MRI Protocol

All subjects underwent a multi-model MRI by a 3.0 T MR (HDXT, GE Healthcare, United States) scanner using an 8- channel brain phased array coil, including T1, T2 fluid-attenuated inversion recovery (FLAIR), and susceptibility weighted imaging (SWI) sequence. In order to minimize head motion, foam pads were inserted into the space between the subject's head and the MRI head coil. An axial T2 FLAIR sequence was used to measure the WMHs volume with the following parameters: repetition time = 8000 ms, echo time = 150 ms, FOV = 24 cm × 24 cm, matrix size = 256 × 256, inversion time = 2100 ms, slice thickness = 4.0 mm with no gap (continuous) between slices, and in-plane spatial resolution of 0.4688 mm/pixel × 0.4688 mm/pixel. The whole brain was imaged. The SWI sequence was in an axial orientation parallel to the anterior commissure to posterior commissure line and covered the whole lateral ventricles, using a three-dimension multi-echo gradient-echo sequence with 11 equally spaced echoes: echo time = 4.5 ms [first echo], inter-echo spacing = 4.5 ms, repetition time = 34 ms, FOV = 24 cm × 24 cm, matrix size = 416 × 384, slice thickness = 2.0 mm with no gap between slices, and in-plane spatial resolution of 0.93 mm/pixel × 0.93 mm/ pixel. Flow compensation was applied.

## Volume Assessment of WMHS

First, the axial T2 FLAIR images were segmented automatically through the lesion segmentation tool (LST) in MATLAB (R2014a) pipeline integrating SPM12 (Wellcome Department of Neurology, University College of London, UK). Then the automatically segmented lesions were manually checked and corrected on mricron (http://www.nitrc.org/projects/mricron) by two experienced neuro-radiologists (WZ and YJ) who were blinded to all other imaging and clinical data after it was coregistered to the T1 images through SPM12. The manual correction process included: (1) division of deep white matter hyperintensities (DWMHs) and periventricular white matter hyperintensities (PVHs); (2) correction of non-white matter area being labeled as WMHs; (3) WMHs area not adequately labeled as WMHs or normal-appearing white matter falsely labeled as WMHs. Afterwards, the volume of WMHs was measured automatically on mricron.

#### Measurement of the QSM Values

According to the published methods (14), the QSM reconstruction was achieved through the use of a C++ software developed and validated by Wang and his colleagues (15), and is based on nonlinear morphology-enabled dipole inversion (MEDI), which makes use of the consistency between the susceptibility maps and magnitude images obtained from the Spoiled Gradient Recalled Echo (SPGR) acquisitions. The susceptibility maps are obtained through estimating (nonlinearly) the phase maps, which have to be unwrapped and subsequently undergo dipole inversion. The regularization parameter has been fixed and set to 1000. The regions of interest (ROIs) were manually drawn on QSM maps by two experienced neuro-radiologists (YZ and QL) who were blinded to all other imaging and clinical data. ROIs were put on the slices where the boundaries of target nuclei could be seen most clearly. Susceptibility values were averaged within each ROI from three successive slices. Both left and right sides of the target nuclei were measured, and the average values were calculated based on the volume. Globus pallidus, head of caudate nucleus, putamen, red nucleus, substantia nigra, thalamus, and dentate nucleus were contained in the ROIs (**Figure 1**). The segmentation function of spm12 in MATLAB (R2014a) was used to get the ROIs of white matter and gray matter. Absolute QSM values of the ROIs were measured automatically on mricron software.

#### Clinical Assessment

We used the Hamilton Depression Rating Scale (HDRS) to assess depressive symptoms and Mini–Mental State Examination (MMSE) to assess cognitive ability at the same time as the MRI scan. According to the HDRS score, the participants were subdivided into two groups: depressive symptom group (HDRS > 7) and control group (HDRS ≤ 7) (16).

# Statistical Analysis

Since the WMHs volume was skewed towards the left of mean, we performed natural log transformations of WMHs volume before the correlation analysis. The log-transformed WMHs volume appeared to be acceptably normative. Independent samples' two-tailed t-test was used to compare the demographics, vascular factors, HDRS score and imaging data between depressive symptom group and control subjects. Fisher's Exact test was used for categorical data. We also conducted logistic regression analysis to provide an odds ratio statistic to facilitate comparison with other known risk factors. Partial Pearson's correlation analysis was conducted to determine the correlation among regional QSM values, logtransformed WMHs volume and HDRS scores, by adjusting for baseline sociodemographic and vascular risk factors. Statistical significance was set at a probability value of < 0.05. All statistical analysis was performed with SPSS 17.0 (SPSS Inc., Chicago, USA).

# RESULTS

# Subject Characteristics

185 consecutive participants were enrolled in this study, after 12 participants were excluded due to poor image quality and 2 participants were excluded due to the history of psychotropic drug use. **Table 1** shows the sociodemographic characteristics, vascular risk factors, volume of WMHs and regional QSM values.

TABLE 1 | Sociodemographic characteristics, vascular risk factors, white matter hyperintensities (WMHs) volume and regional quantitative susceptibility mapping (QSM) values in all included participants.


*MMSE, Mini–Mental State Examination; HDRS, Hamilton Depression Rating Scale DWMHs, deep white matter hyperintensities; PVHs, periventricular white matter hyperintensities.*

#### Reliability of the QSM Value Measurements

The intraclass correlation coefficients (ICCs) were 0.88 for red nucleus, 0.99 for substantia nigra, 0.87 for globus pallidus, 0.96 for putamen, 0.91 for head of caudate, 0.97 for thalamus and 0.97 for dentate nucleus. ICCs were described in detail elsewhere (17).

#### Proof of QSM Data

Pearson's correlation analysis was conducted to determine the correlation between the mean QSM value of deep intracranial nuclei from the control group in the present study and the mean iron distribution in postmortem samples as reported by Hallgren and Sourander in 1958 (18), which measured the content of iron in different brain regions and found the relationship between increased age and increased iron deposition. Globus pallidus, red nucleus, substantia nigra, putamen, dentate nucleus, and thalamus were selected as reference region for comparison. A significant correlation (r = 0.932, *p =* 0.007; **Figure 2**) was found and it supported that the QSM data provide a quantitative measure of iron.

#### Comparison of Iron Deposits and WMHS Volume Between Depressive Symptom Group and Control Group

**Table 2** shows the sociodemographic characteristics, vascular risk factors, volume of WMHs and regional QSM values of depressive symptom group and control group. The HDRS scores of the depressive symptom group were significantly higher than the control group (11.43 ± 5.32 vs 1.32 ± 2.06, *p* < 0.001). There were no differences between two groups in gender, age or vascular risk factors. Participants in depressive symptom group had lower years of education (6.29 ± 4.42 vs 8.64 ± 4.69, *p =* 0.022), lower MMSE score (22.83 ± 4.38 vs 25.53 ± 5.73, *p =* 0.028), higher volume of total WMHs (10.87 ± 18.65 vs 5.54 ± 6.63, *p =* 0.006), higher volume of PVHs (7.18 ± 11.70 vs 3.93 ± 4.65, p = 0.015), higher volume of DWMHs (4.03 ± 7.66 vs 1.63 ± 2.65, p = 0.003) and higher QSM value in thalamus (18.79 ± 14.94 vs 13.29 ± 7.64, *p =* 0.003; **Figure 1**) than the control group. No differences were observed in other deep nuclei that had been measured. No difference was observed in total grey matter (376.56 ± 21.96 vs 375.22 ± 28.48, *p* > 0.05) or white matter (176.42 ± 27.10 vs 170.66 ± 27.08, *p* > 0.05), either.

The binary logistic regression model revealed that the QSM value in thalamus was an independent factor for the presence of depressive symptoms (OR = 1.052; 95% CI: 1.010-1.096; *p =* 0.015) after adjusting for years of education and MMSE score. Furthermore, the QSM value in thalamus was still an independent factor for depressive symptoms (OR = 1.055; 95% CI: 1.011-1.100; *p* = 0.013), after adjusting for years of education, MMSE score and the volume of WMHs. In addition, volume of total WMHs, volume of PVHs, or volume of DWMHs were not influencing factors for depressive symptoms after adjusting for years of education and MMSE score (all *p* > 0.05).

#### Correlation Analysis Between Iron Deposits, WMHS Volume and HDRS Score in Depressive Symptom Group

As presented in **Table 3**, in the depressive symptom group, none of the QSM values were significantly correlated with HDRS score (all *p* > 0.05), after adjusting for age, gender, year of education, MMSE score and baseline vascular risk factors. Further adjusting for the volume of WMHs did not change the results (all *p* > 0.05).

In addition, in the depressive symptom group, no significant correlation was observed between log-transformed WMHs volume and HDRS score (all *p* > 0.05), and no significant correlations were observed between log-transformed WMHs volume and regional QSM value (all *p* > 0.05).

nuclei from the control group and the mean iron distribution in postmortem samples as reported by Hallgren and Sourande.

TABLE 2 | Comparison of sociodemographic characters, vascular risk factors, white matter hyperintensities (WMHs) volume, and regional quantitative susceptibility mapping (QSM) values between depressive symptom group and control group.


*MMSE, Mini–Mental State Examination; HDRS, Hamilton Depression Rating Scale*

*DWMHs, deep white matter hyperintensities; PVHs, periventricular white matter hyperintensities*

TABLE 3 | Associations between regional quantitative susceptibility mapping (QSM) values, white matter hyperintensities (WMHs) volume HDRS and Hamilton depression scale (HDRS) score in depressive symptom group.


*Model 1 adjusted for age, sex, years of education, MMSE scores and vascular risk factors (hypertension, diabetes, hyperlipidemia); Model 2 adjusted for age, sex, years of education, MMSE scores, vascular risk factors (hypertension, diabetes, hyperlipidemia) and log-transformed WMHs volume; age, sex, years of education, MMSE scores and vascular risk factors (hypertension, diabetes, hyperlipidemia, smoke) were also adjusted for analyzing the association between regional QSM value and log-transformed WMHs volume.*

### DISCUSSION

Our main findings include: (1) in older adults, iron deposits in the thalamus was an independent factor for depressive symptoms, even after adjusting for WMHs volume; (2) the severity of iron deposits is not related to severity of depression; (3) WMHs volume was not associated with presence of depressive symptoms or brain iron deposits.

In general, the regional QSM value in our study was relatively low, almost the same as the QSM value of healthy older adults

in the previous study (the QSM value of Thalamus: our study vs Bettes et al.'s study 0.014 ± 0.009 vs 0.021 ± 0.0008 ppm) (19).

Previous studies have demonstrated that iron deposits in the thalamus were related to the degree of depression among depression patients and post-stroke patients (7, 20). Our study further confirmed that even in older adults, iron deposits in the thalamus were related to depressive symptoms. Although the pathogenesis of depression is still not sufficiently clear, iron deposits might lead to depressive symptoms by multiple mechanisms.

It has been reported that there were abnormal connections in the thalamus-temporal lobes and thalamus-cortex areas in patients with depression (21). Meanwhile, studies have shown that excessive iron could affect functional connectivity (22). Therefore, we speculate that thalamic iron deposits might cause local neuron and neurotransmitter dysfunction, which affects functional connections, finally leading to depressive symptoms.

In addition, the monoamine hypothesis might mediate the relationship between iron deposits and depression. The lack of monoamines, including serotonin, dopamine, norepinephrine, and epinephrine, would lead to depression. Moreover, deficiency of a certain neurotransmitter could lead to a certain depressive symptom (23). Studies have reported that brain iron deposits could affect monoamine function (24). Therefore, brain iron deposits might reduce the activity of monoamines, leading to the occurrence of depressive symptoms.

Furthermore, some studies suggested that inflammation and oxidative stress also could influence depression (25). Cytokines and other pro-inflammatory mediators were involved in the pathophysiological process of mood regulation, such as neurotransmitter metabolism, neuroendocrine function, anterior cingulate cortical activity, and synaptic plasticity (26). Indeed, animal experiments have proven depression-like behaviors could be induced by brain iron overload through apoptosis pathways among adult rats (27). Moreover, Dixon et al. raised the notion of ferroptosis in 2012, which refers to a form of regulated cell death characterized by the iron-dependent accumulation. The accumulation of intracellular iron would induce accumulation of lipid reactive oxygen species (ROS) and the over-accumulation of lipid ROS results in oxidative stress which finally leads to lethal levels for cell deaths (28). Therefore, iron deposits may also cause depressive symptoms by causing an inflammatory reaction and apoptosis.

Yao et al. (7) identified susceptibility values were higher in putamen of patients with major depressive disorder. Whereas, we did not find a correlation between the iron deposits in putamen and depressive symptoms. In addition, our QSM values of putamen met the range of healthy older adults reported before (19, 29). Therefore, given the differences in the study population, we suspect that the putamen might affect different stages of depression. This hypothesis needs to be confirmed in future.

We did not find a relationship between iron deposits and the severity of depression in the depressive symptom group. The mild symptoms of our study population might explain it. We conducted research in old healthy community populations. Although some of their HDRS scores were greater than 7 points, it could only be considered as depressive symptoms rather than depression. In addition, a previous study which found that iron deposits in thalamus were associated with depression severity did not consider cognitive ability, and the cognitive decline in older adults was likely to affect depression scores.

Surprisingly, we did not find the relationship between volume of WMHs and depressive symptoms after adjusting for years of education and MMSE score. According to the vascular depression hypothesis, WMHs may result in mechanistic disconnection and hypoperfusion, which links cerebrovascular diseases with the depression (8). The heterogeneity of the research population might explain the contradiction between our findings and previous research results. Our included population was not diagnosed with depression but was considered to have depressive symptoms based on their HDRS scores. Even in the depressive symptoms group, the HDRS scores were not high (average HDRS score was 11.38). The damage of white matter in our participants might be too slight to affect the depressive symptoms.

We also found no correlation between WMHs and iron deposits. It might also be explained by the difference of research population. The previous positive finding about the relationship between iron deposits and volume of WMHs was based on a population of in-hospital patients, who had severe white matter damage (average volume of WMHs was 35.95 ml) (9), while our current study included a community population with an average WMHs volume of only 6.22 ml. Pathological studies have demonstrated that myelin has a strong ability to store iron without causing damage (30), which may indicate that the damage of iron deposits to myelin is a late manifestation. Moreover, iron deposits might be an early stage of neurodegeneration and might not produce WMHs (10). One possible but hypothetical scenario would be that the early impact of iron deposits on depressive symptoms might be caused by abnormal functional connection, monoamine dysfunction and inflammation reaction, while during the advanced period, iron deposition would aggravate white matter damage and accelerate emotional disorders.

Our study had strength in its methodology. QSM was regarded as a more accurate way to measure iron deposits because it could avoid the loss of signals and the influence of calcifications when measured by R2\* (31). We innovatively analyzed the relationship in older adults as there was only one study in the past that examined the relationship between iron deposits and depression by QSM among patients diagnosed as depression. In addition, we excluded the effects of WMHs and cognitive ability, ensuring that the effect of iron deposits on depressive symptoms was an independent process.

Our study had limitations. First, while the total size is close to 200 patients, the group with depressive symptoms is small which can impact the robustness of the results. Second, it was a crosssectional study that could only analyze the correlation among depression, WMHs and brain iron, and cannot analyze the cause and effect. Further longitudinal study is needed to clarify this. Third, we did not carry out laboratory tests. Therefore, the health status of the enrolled population has not been fully checked, and some factors that may influence emotional behaviors could be ignored. Fourth, voxel-based QSM analysis might further prompt the phenomenon and mechanism, and further study is needed to clarify this. Fifth, the QSM values were measured by manual delineation. Although the consistency was good, there were still measurement errors.

In summary, our finding indicated that in a community population, thalamic iron deposits were an independent factor for depressive symptoms, but WMHs volumes were irrelevant to either increased iron deposits or depressive symptoms. Our results may help to investigate the underlying pathophysiological mechanism of depression in the future studies. It's worthy to explore the relationship between depressive and specific parts of the thalamus in future.

### ETHICS STATEMENT

All subjects had given written informed consent prior to the study, and the protocol was approved by the local ethics committee. All clinical investigation has been conducted according to the principles expressed in the Declaration of Helsinki.

## AUTHOR CONTRIBUTIONS

WZ and YZ drafted and revised the manuscript, participated in study concept and design, conducted the statistical analyses, analyzed, and interpreted the data. ML participated in study concept and design, data interpretation and made a major contribution in revising the manuscript. QL, JX, and SY

#### REFERENCES


participated in the study design and made contribution in revising the manuscript. YJ and JC assisted in designing the MRI sequences and imaging analysis.

### FUNDING

This study was supported by National Key Research and Development Program of China (2016YFC1300504), National Natural Science Foundation of China (81622017, 81701150), Science Technology Department of Zhejiang Province (2018C04011), Chinese Cardiovascular Association-V.G Fund (2017-CCA-VG-004), and Basic Public Interests of Research Plan of Zhejiang Province (GF18H090006).


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Zhang, Zhou, Li, Xu, Yan, Cai, Jiaerken and Lou. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

digital media

of impactful research

article's readership