A nomogram for predicting cerebral white matter lesions in elderly men

Abstract

Objective:

This study aimed to develop a nomogram tool to predict cerebral white matter lesions (WMLs) in elderly men.

Methods:

Based on a retrospective cohort from January 2017 to December 2019, a multivariate logistic analysis was performed to construct a nomogram for predicting WMLs. The nomogram was further validated using a follow-up cohort between January 2020 and December 2022. The calibration curve, receiver operating characteristics (ROC) curves, and the decision curves analysis (DCA) were used to evaluate discrimination and calibration of this nomogram.

Result:

A total of 436 male patients were enrolled in this study, and all 436 patients were used as the training cohort and 163 follow-up patients as the validation cohort. A multivariate logistic analysis showed that age, cystatin C, uric acid, total cholesterol, platelet, and the use of antiplatelet drugs were independently associated with WMLs. Based on these variables, a nomogram was developed. The nomogram displayed excellent predictive power with the area under the ROC curve of 0.951 [95% confidence interval (CI), 0.929–0.972] in the training cohort and 0.915 (95% CI, 0.864–0.966) in the validation cohort. The calibration of the nomogram was also good, as indicated by the Hosmer–Lemeshow test with p-value of 0.594 in the training cohort and 0.178 in the validation cohort. The DCA showed that the nomogram holds good clinical application value.

Conclusion:

We have developed and validated a novel nomogram tool for identifying elderly men at high risk of WMLs, which exhibits excellent predictive power, discrimination, and calibration.

Introduction

Cerebral white matter lesions (WMLs) are not only an important part of cerebral small vessel disease (SCVD), but also one of the most common causes of cognitive impairment in the elderly, and considered to be an early sign of brain damage (1, 2). The incidence of WMLs in the elderly has been reported to range from 11% to 21%, with rates reaching as high as 94% in advanced-aged individuals (3). Given the increasing global aging population, research on WMLs has gained significant attention (4). Growing evidence showed that heavier WMLs burden was closely associated with more severe cognitive impairment (5, 6). Some studies have shown that the location of WMLs was closely related to the prognosis of patients with acute ischemic stroke (7), and the severity of WMLs may affect the functional prognosis of ischemic stroke (8). Other studies suggest that WMLs can increase the risk of dementia and death in stroke patients (3). Furthermore, a prospective study revealed that severe WMLs can predict composite endpoints of death, pneumonia, and falls (9).

Some studies have identified the correlations of age, hypertension, female gender, high density lipoprotein (HDL) levels, and the use of antihypertensive medications with occurrence of WMLs (10, 11). However, accurately predicting the occurrence of WMLs remains a challenge. Only one recent study reported a prediction model focused on the elderly (10). However, constructing models that incorporate a wider range of clinically relevant variables could help enhance the accuracy of WMLs prediction, which is crucial for identifying the appropriate population for early intervention.

In this context, the current study aims to develop a predictive nomogram based on comprehensive baseline characteristics of elderly men.

Methods

Patients and study design

We retrospectively enrolled consecutive patients who underwent comprehensive medical examination including brain magnetic resonance imaging (MRI) between January 2017 and December 2019. Eligible patients were used as a training cohort to develop a nomogram for predicting WMLs. After this time, all these patients were retrospectively followed up for 3 years (between January 2020 and December 2022) as the validation cohort. Patients fulfilling the following eligibility criteria were included for analysis: (a) age ≥ 60 years; (b) complete brain MRI including T2-weighted images and fluid-attenuated inversion recovery (FLAIR); (c) complete medical examination. The exclusion criteria: (a) incomplete clinical data; (b) large area cerebral infarction, cerebral hemorrhage, subarachnoid hemorrhage and craniocerebral trauma; (c) severe liver dysfunction, severe cardiac dysfunction, hypothyroidism, gout, autoimmune system diseases, severe systemic infection; (d) taking immunosuppressant recently; (e) taking B vitamins and folic acid recently; (f) taking uric acid-lowering drugs recently; (g) other conditions known to cause WMLs, such as carbon monoxide poisoning, hypoxic encephalopathy, various types of hydrocephalus, immune white matter demyelination and so on; (h) diseases that may interfere with the analysis of WMLs, such as brain tumors, cysts and history of craniocerebral surgery, etc. This study was approved by the Ethics Committee of General Hospital of Northern Theater Command, Shenyang and waived the need for informed consent from all subjects.

Data collection

The following demographic and clinical characteristics were recorded: age, weight, smoking and drinking history, admission systolic blood pressure (SBP) and admission diastolic blood pressure (DBP), medical comorbidity [hypertension, diabetes, coronary heart disease (CHD), atrial fibrillation (AF), left heart failure (LHF) and tumor]; the use of statins, antiplatelet drugs and anticoagulants. Laboratory test values including serum total cholesterol (TC), triglyceride (TG), high density lipoprotein (HDL), low density lipoprotein (LDL), lipoprotein-A (LP-A), albumin (ALB), total protein (TP), high-sensitivity C-reactive protein (h-CRP), platelets (PLT), glycosylated hemoglobin (HbA1C), total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), alanine aminotransferase (ALT), gamma glutamyl transpeptidase (GGT), homocysteine (Hcy), cystatin C (Cys C), uric acid (UA), creatinine (CREA), urea nitrogen (UREA) and brain natriuretic peptide (BNP).

WMLs assessment

The WMLs were evaluated utilizing the Fazekas scale on MRI (12), and were defined as any occurrence of WMLs (namely a Fazekas score of at least 1 point). In the scale, periventricular and deep WMLs were rated separately, and periventricular WMLs were graded according to the following patterns: 0 = absent; 1 = caps or pencil-thin lining; 2 = smooth halo; and 3 = irregular periventricular WMLs extending into a deep WMLs. Deep WMLs were graded according to the following patterns: 0 = absent; 1 = punctate foci; 2 = beginning confluence of foci; and 3 = large fused areas. A total Fazekas score, ranging from 0 to 6, was acquired by summing the periventricular and deep WMLs scores. Based on the Fazekas scale, WMLs did not include old lacunar infarctions and/or white matter lesions at some other locations such as basal ganglia, deep gray matter and cortical/subcortical areas. The neuroimaging data were evaluated independently by two experienced neurologists who were blinded to the clinical data. Any disagreement between the two assessors was resolved by discussion until a consensus was achieved.

Statistical analysis

The patient data in the training cohort were used to develop the prediction model, and the patient data in the validation cohort were used to validate the model. Gaussian distributed data were expressed as means ± standard deviation (SD) and compared by student’s t-test, while non-Gaussian distributed data were expressed as medians (interquartile range, Q2–Q3) and compared by the Mann–Whitney U test. Categorical data are expressed using frequencies and ratios (%) and Fisher’s exact tests or the χ2 tests were used for categorical variables. SPSS 26.0 was used for statistical analyses. All variables with a probability value < 0.10 in the univariate analysis entered into a multivariate logistic regression analysis using a backwards stepwise method. Stata software (version 17.0) was used to build the nomogram prediction model. The nomogram converts each independent risk factor included in the model into an assessment point system. The total points obtained determine the final risk assessment value. The odds ratio (OR) and 95% confidence interval (CI) of each significant risk factor in the final logistic regression model were calculated. Performances of the nomogram were assessed in the validation cohort. Discrimination of the nomogram was assessed using the area under the ROC curve (AUC). An AUC of 0.51–0.7 indicates low accuracy, 0.71–0.9 indicates moderate accuracy, and 0.91–1.0 indicates high accuracy to discriminate between patients with WMLs and without WMLs. The calibration of the nomogram was evaluated by the Hosmer-Lemeshow test (p > 0.05 indicates good calibration). In addition, a decision curve analysis (DCA) of the model was developed to quantify the net benefit rate at different threshold probabilities to assess the clinical validity of the model. All statistical tests were 2-tailed. We deemed statistical significance at p = 0.05.

Results

Demographic and clinical features

As shown in Figure 1, 1,276 patients were screened for eligibility and 436 patients were included in the training cohort after excluding 840 individuals. The patients in the training cohort were followed for 3 years. However, due to incomplete follow-up information, 273 patients were subsequently excluded. As a result, a total of 163 follow-up patients were included in the validation cohort. The demographics and clinical characteristics of the patients in the two cohorts were shown in Table 1. There was a good balance in the baseline characteristics between the training and validation cohort except for tumor history, statins history, uric acid and serum total protein. The average age of patients in the training cohort and validation cohort were 73.93 ± 14.15 and 75.22 ± 11.84 years old, respectively. The prevalence rate of WMLs in the training cohort was 65.6%, whereas in the validation cohort was as high as 82.6%.

Figure 1

Predictors for WMLs and nomogram construction

In the training cohort, a univariate analysis was conducted to investigate the association of several variables with WMLs. These variables include age, medical comorbidities (hypertension, CHD, tumor), SBP, use of statins, use of antiplatelet drugs, laboratory tests (Hcy, Cys C, UA, CREA, UREA, h-CRP, TC, TG, HDL, LDL, TP, ALB, BNP, HbA1C, PLT; Table 2). Finally, six variables remained significant in the multivariate logistic model, including age, cystatin C, total cholesterol, platelet, uric acid and use of antiplatelet drugs (Table 2). The predictive nomogram for WMLs that integrated these six variables was constructed and depicted in Figure 2. Based on this nomogram, the probability of predicting WMLs in patients can be calculated as follows: P_WMLs = 1/(1 + e^x), x = −(−15.29 + 0.099 × age − 0.174 × antiplatelet drugs (yes) + 0 × antiplatelet drugs (no) + 6.455 × cystatin C − 0.611 × total cholesterol +0.023 × platelet +0.006 × uric acid).

Figure 2

Assessment, validation of predictive accuracy of the nomogram for WMLs

The ROC (Figures 3A,B) and calibration curve were plotted (Figures 3C,D) to verify the accuracy and discrimination of this model. The AUC in the training cohort was 0.951 (95% CI, 0.929–0.972), while 0.915 (95% CI, 0.864–0.966) in the validation cohort, which suggested that our model had good accuracy. The calibration curve of the training cohort and validation cohort showed that there was excellent consistency between the predicted survival probability and the actual survival probability. The Hosmer-Lemeshow test for the training cohort yielded a p-value of 0.594, suggesting that the nomogram had a perfect fit. The calibration of the nomogram was further confirmed with the validation cohort, for which the Hosmer-Lemeshow test yielded a p-value of 0.178, suggesting no departure from the good fit of the nomogram. DCA was also plotted to evaluate the usefulness of the nomogram in the clinical utility (Figures 4A,B) and the results showed that the predictive nomogram provided sound clinical guidance with a good net benefit.

Figure 3

Figure 4

Discussion

In the current study, we developed and validated a predictive nomogram to identify the increased risk of WMLs in the male elderly patients over 60 years old. The newly developed model incorporated several key factors, including age, TC, PLT, Cys C, UA, and use of antiplatelet drugs. The resulting nomogram showed excellent predictive power as well as good discrimination and calibration.

Among these six variables, age, Cys C, and platelet were found to have a greater impact on WMLs. It was expected that age was found as a main contributor of this model, because age has been well demonstrated to be associated with WMLs (13–15). Given the inhibitory effect of increased serum Cys C on the contractile function of arterioles (16), and the damaging effect on the blood–brain barrier (17), the close association of Cys C with WMLs was explainable in the current study. In agreement with prior studies, platelet was found to be associated with WMLs in this study. Kuriyama et al. (18) found that the risk of WMLs in the elderly with high platelet activity increased by about threefold. Higher platelet aggregation rate have been linked to a higher incidence of WMLs, with a risk significantly higher than classical vascular risk factors such as hypertension, diabetes, and age (19). In addition, aspirin administration in patients with WMLs not only prevent cerebral infarction, but also curb the progress of WMLs (20), which was in agreement with the current finding that antiplatelet drug was independently associated with WMLs. Collectively, these results suggest the potential association of platelet and use of antiplatelet drug with WMLs.

In the current study, UA was also found to be associated with WMLs. Prior studies have found that hyperuricemia was an independent risk factor for WMLs (21). The underlying mechanisms included its effect on vascular endothelial cell dysfunction and atherosclerosis (22). Remarkably, our study found that lower TC levels may lead to an increased risk of WMLs, which seems contradictory with previous studies reporting that high cholesterol levels increased the risk of WMLs (23, 24). We argue that there are two possible reasons. First, as the main components of cell membrane and nerve myelin sheath, cholesterol plays a fundamental role in the development of the central nervous system and the creation and maintenance of new synapses (25, 26). This could explain the effect of significantly lower cholesterol levels on chronic cerebral injury, such as the developments of WMLs. Second, many studies have confirmed that intensive lipid reduction will greatly increase the risk of cerebral micro bleeds (CMBs) (27, 28), while there was a close correlation between CMBs and the formation of WMLs (29), which may explain the association of significantly lower cholesterol levels with WMLs. Furthermore, the current finding was supported by some studies reporting that the increase of TC was considered to be a protective factor of WMLs (30, 31). In addition, some studies did not find a correlation between TC and WMLs (32). Although there was a higher prevalence of diabetes and hypertension in the current study, we did not find the association of diabetes, hypertension and blood pressure with WMLs, which were known to be associated with WMLs (33–35). One possible explanation is that the blood pressure and blood glucose were well controlled in this cohort (36, 37). In addition, high prevalence of WMLs in both cohorts due to elderly patients (older than 60) and high prevalence (>60%) of hypertension may mask their associations, so their associations with severity of WMLs deserve to be further investigated. However, some studies did not find a correlation between diabetes and WMLs (38, 39), which was in line with our findings.

Our study possesses two potential strengths. First, our nomogram was created based on analysis of comprehensive and routinely available baseline variables in a training cohort, which was further validated in a follow-up cohort. Second, Cys C, UA, TC, platelet, and the use of antiplatelet drugs were incorporated into the prediction model for the first time and had excellent predictive power, and good discrimination and calibration. Nevertheless, we acknowledge several limitations. First, this was a retrospective observational study, which had common limitations including bias, potential lack of baseline data and inevitable confounding effect, although baseline data were comprehensively collected. Second, this was a single-center study without an external validation cohort, but it was validated by a follow-up cohort. Third, as almost all the patients in our center were male, our study did not include female patients. So, this finding is only applicable to male patients. Fourth, we did not collect the data about other brain lesions or underlying disease, which might have effect on this finding given their associations with WMLs. Finally, further confirmation of these findings in non-Chinese populations would be needed, given the differences in body mass and comorbid factors compared with other populations.

Conclusion

A novel nomogram tool has been developed to predict WMLs with excellent predicting power, good discrimination and calibration. The model should be helpful to identify elderly men at high risk of cerebral WMLs.

References

• 1.

  Pantoni L Poggesi A Inzitari D . The relation between white-matter lesions and cognition. Curr Opin Neurol. (2007) 20:390–7. doi: 10.1097/WCO.0b013e328172d661

  • CrossRef
  • Google Scholar

• 2.

  Ble A Ranzini M Zurlo A Menozzi L Atti AR Munari MR et al . Leukoaraiosis is associated with functional impairment in older patients with Alzheimer’s disease but not vascular dementia. J Nutr Health Aging. (2006) 10:31–5. PMID:

  • Pubmed Abstract
  • Google Scholar

• 3.

  Debette S Markus HS . The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta-analysis. BMJ. (2010) 341:c3666. doi: 10.1136/bmj.c3666

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4.

  Altaf N Daniels L Morgan PS Lowe J Gladman J MacSweeney ST et al . Cerebral white matter hyperintense lesions are associated with unstable carotid plaques. Eur J Vasc Endovasc Surg. (2006) 31:8–13. doi: 10.1016/j.ejvs.2005.08.026

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5.

  Caunca MR De Leon-Benedetti A Latour L Leigh R Wright CB . Neuroimaging of cerebral small vessel disease and age-related cognitive changes. Front Aging Neurosci. (2019) 11:145. doi: 10.3389/fnagi.2019.00145

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6.

  Kloppenborg RP Nederkoorn PJ Geerlings MI van den Berg E . Presence and progression of white matter hyperintensities and cognition: a meta-analysis. Neurology. (2014) 82:2127–38. doi: 10.1212/WNL.0000000000000505

  • CrossRef
  • Google Scholar

• 7.

  Jagani M Kallmes DF Brinjikji W . Correlation between clot density and recanalization success or stroke etiology in acute ischemic stroke patients. Interv Neuroradiol. (2017) 23:274–8. doi: 10.1177/1591019917694478

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 8.

  Onteddu SR Goddeau RJ Minaeian A Henninger N . Clinical impact of leukoaraiosis burden and chronological age on neurological deficit recovery and 90-day outcome after minor ischemic stroke. J Neurol Sci. (2015) 359:418–23. doi: 10.1016/j.jns.2015.10.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9.

  Briley DP Haroon S Sergent SM Thomas S . Does leukoaraiosis predict morbidity and mortality?Neurology. (2000) 54:90–4. doi: 10.1212/WNL.54.1.90

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10.

  Li J Xie XM . Nomograms for predicting the incidence of white matter lesions: a real-world study based on elderly patients. Eur Rev Med Pharmacol Sci. (2023) 27:151–8. doi: 10.26355/eurrev_202301_30866

  • CrossRef
  • Google Scholar

• 11.

  Sawada H Udaka F Izumi Y Nishinaka K Kawakami H Nakamura S et al . Cerebral white matter lesions are not associated with apoE genotype but with age and female sex in Alzheimer's disease. J Neurol Neurosurg Psychiatry. (2000) 68:653–6. doi: 10.1136/jnnp.68.5.653

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12.

  Fazekas F Chawluk JB Alavi A Hurtig HI Zimmerman RA . MR signal abnormalities at 1.5 T in Alzheimer's dementia and normal aging. AJR Am J Roentgenol. (1987) 149:351–6. doi: 10.2214/ajr.149.2.351

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13.

  Franchetti MK Bharadwaj PK Nguyen LA van Etten EJ Klimentidis YC Hishaw GA et al . Interaction of age and self-reported physical sports activity on white matter Hyperintensity volume in healthy older adults. Front Aging Neurosci. (2020) 12:576025. doi: 10.3389/fnagi.2020.576025

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14.

  Shinkawa Y Yoshida T Onaka Y Ichinose M Ishii K . Mathematical modeling for the prediction of cerebral white matter lesions based on clinical examination data. PLoS One. (2019) 14:e215142. doi: 10.1371/journal.pone.0215142

  • CrossRef
  • Google Scholar

• 15.

  Vedala K Nagabandi AK Looney S Bruno A . Factors associated with Leukoaraiosis severity in acute stroke patients. J Stroke Cerebrovasc Dis. (2019) 28:1897–901. doi: 10.1016/j.jstrokecerebrovasdis.2019.04.003

  • CrossRef
  • Google Scholar

• 16.

  Lafarge JC Naour N Clement K Guerre-Millo M . Cathepsins and cystatin C in atherosclerosis and obesity. Biochimie. (2010) 92:1580–6. doi: 10.1016/j.biochi.2010.04.011

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17.

  Kanhai DA de Kleijn DP Kappelle LJ Uiterwaal CS van der Graaf Y Pasterkamp G et al . Extracellular vesicle protein levels are related to brain atrophy and cerebral white matter lesions in patients with manifest vascular disease: the SMART-MR study. BMJ Open. (2014) 4:e3824. doi: 10.1136/bmjopen-2013-003824

  • CrossRef
  • Google Scholar

• 18.

  Kuriyama N Mizuno T Yasuike H Matsuno H Kawashita E Tamura A et al . CD62-mediated activation of platelets in cerebral white matter lesions in patients with cognitive decline. Arch Gerontol Geriatr. (2016) 62:118–24. doi: 10.1016/j.archger.2015.09.001

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19.

  Fujita S Kawaguchi T . Association of platelet hyper-aggregability with leukoaraiosis. Acta Neurol Scand. (2002) 105:445–9. doi: 10.1034/j.1600-0404.2002.01208.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20.

  Fujita S Kawaguchi T Uehara T Fukushima K . Progress of leukoaraiosis is inhibited by correction of platelet hyper-aggregability. Int Psychogeriatr. (2005) 17:689–98. doi: 10.1017/S104161020500164X

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21.

  Sampat R Young S Rosen A Bernhard D Millington D Factor S et al . Potential mechanisms for low uric acid in Parkinson disease. J Neural Transm. (2016) 123:365–70. doi: 10.1007/s00702-015-1503-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22.

  Li JJ Huang YH Lin YY Li MM Chen YF Cai RW . The association of uric acid with leukoaraiosis. J Int Med Res. (2017) 45:75–81. doi: 10.1177/0300060516674353

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23.

  Park K Yasuda N Toyonaga S Yamada SM Nakabayashi H Nakasato M et al . Significant association between leukoaraiosis and metabolic syndrome in healthy subjects. Neurology. (2007) 69:974–8. doi: 10.1212/01.wnl.0000266562.54684.bf

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24.

  Dufouil C de Kersaint-Gilly A Besancon V Levy C Auffray E Brunnereau L et al . Longitudinal study of blood pressure and white matter hyperintensities: the EVA MRI cohort. Neurology. (2001) 56:921–6. doi: 10.1212/WNL.56.7.921

  • CrossRef
  • Google Scholar

• 25.

  Goritz C Mauch DH Pfrieger FW . Multiple mechanisms mediate cholesterol-induced synaptogenesis in a CNS neuron. Mol Cell Neurosci. (2005) 29:190–201. doi: 10.1016/j.mcn.2005.02.006

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26.

  Dietschy JM Turley SD . Cholesterol metabolism in the brain. Curr Opin Lipidol. (2001) 12:105–12. doi: 10.1097/00041433-200104000-00003

  • CrossRef
  • Google Scholar

• 27.

  Mitaki S Nagai A Oguro H Yamaguchi S . Serum lipid fractions and cerebral microbleeds in a healthy Japanese population. Cerebrovasc Dis. (2017) 43:186–91. doi: 10.1159/000456623

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 28.

  Yano K Reed DM MacLean CJ . Serum cholesterol and hemorrhagic stroke in the Honolulu heart program. Stroke. (1989) 20:1460–5. doi: 10.1161/01.STR.20.11.1460

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29.

  Yamada S Saiki M Satow T Fukuda A Ito M Minami S et al . Periventricular and deep white matter leukoaraiosis have a closer association with cerebral microbleeds than age. Eur J Neurol. (2012) 19:98–104. doi: 10.1111/j.1468-1331.2011.03451.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30.

  Ohwaki K Yano E Tamura A Inoue T Saito I . Hypercholesterolemia is associated with a lower risk of cerebral ischemic small vessel disease detected on brain checkups. Clin Neurol Neurosurg. (2013) 115:669–72. doi: 10.1016/j.clineuro.2012.07.025

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31.

  Jimenez-Conde J Biffi A Rahman R Kanakis A Butler C Sonni S et al . Hyperlipidemia and reduced white matter hyperintensity volume in patients with ischemic stroke. Stroke. (2010) 41:437–42. doi: 10.1161/STROKEAHA.109.563502

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32.

  Saji N Shimizu H Kawarai T Tadano M Kita Y Yokono K . Increased brachial-ankle pulse wave velocity is independently associated with white matter hyperintensities. Neuroepidemiology. (2011) 36:252–7. doi: 10.1159/000328260

  • CrossRef
  • Google Scholar

• 33.

  Marseglia A Fratiglioni L Kalpouzos G Wang R Bäckman L Xu W . Prediabetes and diabetes accelerate cognitive decline and predict microvascular lesions: a population-based cohort study. Alzheimers Dement. (2019) 15:25–33. doi: 10.1016/j.jalz.2018.06.3060

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34.

  Allan CL Zsoldos E Filippini N Sexton CE Topiwala A Valkanova V et al . Lifetime hypertension as a predictor of brain structure in older adults: cohort study with a 28-year follow-up. Br J Psychiatry. (2015) 206:308–15. doi: 10.1192/bjp.bp.114.153536

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35.

  Pantoni L Garcia JH . The significance of cerebral white matter abnormalities 100 years after Binswanger's report. A review. Stroke. (1995) 26:1293–301. doi: 10.1161/01.STR.26.7.1293

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36.

  Kern KC Wright CB Bergfield KL Fitzhugh MC Chen K Moeller JR et al . Blood pressure control in aging predicts cerebral atrophy related to small-vessel white matter lesions. Front Aging Neurosci. (2017) 9:132. doi: 10.3389/fnagi.2017.00132

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37.

  Del BA Ciolli L Borgheresi L Poggesi A Inzitari D Pantoni L . Is type 2 diabetes related to leukoaraiosis? An updated review. Acta Neurol Scand. (2015) 132:147–55. doi: 10.1111/ane.12398

  • CrossRef
  • Google Scholar

• 38.

  Yu X Wang G Zhan J Zhang Z Feng T Xu J . Risk factors of pure leukoaraiosis and the association with preclinical carotid atherosclerosis. Atherosclerosis. (2018) 275:328–32. doi: 10.1016/j.atherosclerosis.2018.06.869

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39.

  Guan J Yan C Gao Q Li J Wang L Hong M et al . Analysis of risk factors in patients with leukoaraiosis. Medicine. (2017) 96:e6153. doi: 10.1097/MD.0000000000006153

  • Pubmed Abstract
  • CrossRef
  • Google Scholar