Analysis of complete blood count and derived inflammatory indicators for bipolar disorder patients with different states

Purposes This study aimed to analyze the predictive ability of the complete blood count and derived inflammatory indicators for BD patients with different states to identify potential biomarkers. Methods We collected the data of BD in-patients from January 2021 to March 2023. The complete blood count and derived inflammatory indicators were computed by univariate analysis, logistic regression analysis, and receiver operating characteristic (ROC) curve analysis. Results In terms of BD patients, the levels of MON (p<0.0001), hs-CRP (p = 0.018), and NLR (p = 0.002) were independent risk factors in logistic regression analysis, as well as the cut-off values were 0.405 109/L, 2.600 mg/L, and 2.321, respectively. Regarding BD-M patients, the levels of MON (p<0.0001), hs-CRP (p = 0.012), and NLR (p = 0.002) were predictors in logistic regression analysis, and the cut-off values were, respectively, 0.505 109/L, 2.600 mg/L, and 2.620. Additionally, the levels of NLR (p = 0.006) and MHR (p<0.0001) were important indicators for BD-D and the cut-off values were 1.735 and 0.487, respectively. Furthermore, our findings showed that the level of MON (p = 0.001) was related to BD-mixed and the cut-off value was 0.340 109/L. Notably, MON+hs-CRP + NLR, MON+hs-CRP + NLR, and NLR + MHR had the highest diagnostic accuracy to predict BD, BD-M, and BD-D patients, respectively. Conclusion Our findings showed that distinct inflammatory indicators were closely associated with BD and its different states. Additionally, we also identified their cut-off values and optimal combined predictive indicators in different states of BD, helping us improve diagnostic accuracy and better assess them to manage early targeted interventions.


Introduction
Bipolar disorder (BD), which affects 2-3% of the general population (1), is a serious mental disease characterized by the presence of recurring (hypo)mania or depression phases (2), which brings a great burden on families and societies (3). BD is divided into three types: type I patients have at least one manic episode with or without major depressive episodes, type II patients have at least one hypomanic episode and one major depressive episode, and cyclothymic disorder (4). BD patients often have functional disabilities, reduced life quality, and even decreased life expectancy (5). Increasing evidence has been demonstrated that BD patients have almost 30 times suicide risk than healthy people, and high rates of comorbidity (6,7), such as cardiovascular disease, diabetes mellitus, and chronic obstructive pulmonary disease (8,9).
Although the etiopathogenesis of BD remains poorly understood, numerous studies have widely reported that the immune and inflammatory systems may be crucial contributors to the onset and progression of this life-threatening illness (10,11). Wei (12) found that the system inflammation response index (SIRI), neutrophil/high-density lipoprotein(HDL) ratio (NHR), and monocyte/HDL ratio (MHR) were predictors for the BD-M group and MHR was a predictor for the BD-D group. Misiak (13) discovered that the levels of interleukin-8 (IL-8), monocytechemoattractant protein-1 (MCP-1), eotaxin-1 and interferon-γinduced protein 10 (IP-10) were higher in BD patients than in healthy controls. Özdin (14) compared the neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), and monocyte to lymphocyte ratio (MLR) values in the manic and euthymic phases of the same patients and identified NLR and PLR as state markers and MLR as a trait marker of BD. Kirlioglu (15) investigated the relationship between inflammatory and metabolic markers and newly diagnosed BD but failed to find evidence supporting high-sensitive C-reactive protein and homocysteine as markers in newly diagnosed BD.
However, the relationship between inflammation factors and BD is still debated because some articles lack healthy people as the control group. As we know, few studies pay attention to the differences in admission laboratory indicators and optimal combined predictive indicators in patients with distinct types of BD. The study aims to investigate the inflammation predictors of BD subtypes compared with healthy people. At the same time, in this study, we not only used a single factor to predict the development of diseases but also used the combined method of multiple risk factors to improve the sensitivity of disease prediction, trying to improve the prediction efficiency through the combined method of multiple factors.

Patients and methods
Ethics approval and consent to participate This retrospective study was approved by the Institutional Review Board of our hospital (20220908) before collecting data. There was no need to obtain informed consent forms from patients because this was a retrospective study.

Patients
A retrospective analysis of all patients with BD between January 2021 and March 2023 in our hospital. Patients with BD, including BD manic episodes (BD-M), patients with BD depressive episodes (BD-D), and patients with BD mixed episodes (BD-mixed), as well as healthy subjects were considered as the BD group, the BD-M group, the BD-D group, the BD-mixed group, and the healthy group (HG), respectively.
Blood samples were taken within the same time frame (6.00-8.00 a.m.) among these groups. Regarding the BD patients, we collected the first blood samples that were taken on the second day after admission. The inclusion criteria were as follows: (1) patients 18-65 years old, (2) patients with BD in the acute phase based on the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) diagnostic criteria, and (3) patients with a first episode or who have not been treated for at least 2 weeks. The exclusion criteria were: (1) patients with other mental diseases, such as schizophrenia, (2) patients with diseases affecting inflammatory markers, such as acute inflammatory diseases, metabolic syndromes (e.g., infections, blood disorders, autoimmune diseases, diabetes, and heart failure), and (3) patients with incomplete data. To be better compared with the BD-M group, BD-D group, and BD-mixed group, data from 102 healthy subjects with age, gender, and BMI matched to the BD group were included as controls. Healthy subjects were recruited if they met the following inclusion criteria: (1) adult patients (between 18 and 65 years old), (2) patients without mental diseases, such as schizophrenia, BD, or depressive disorder, (3) patients without diseases affecting inflammatory markers such as acute inflammatory diseases, metabolic syndromes (e.g., infections, blood disorders, autoimmune diseases, diabetes, heart failure), and (4) completed data.

Statistics
We utilized SPSS (version 25.0 SPSS Inc., Chicago, IL) and regarded p < 0.05 as statistical significance. Regarding continuous variables, if data met normality criteria, all measurement data were presented as the mean ± SD (standard deviation) using the t-test, but if not, the Mann-Whitney U test was used to perform statistical analysis between groups. For count data, the Chi-square test was used. Furthermore, to identify the best predictors of BD, we used binary logistic regression analysis to detect its independent predictors. Additionally, receiver operator characteristic (ROC) curve analysis was used to identify the cut-off values for continuous variables and the area under the ROC curve (AUC) was used to determine the diagnostic ability, ranging from 0 to 100%, with more area meaning better ability. We choose the cut-off values for continuous variables by the maximum Youden index (sensitivity+specificity-1) in the ROC curve analysis.

Results
We collected a total of 785 BD patients who presented and were assessed for eligibility in this study. Finally, we collected a total of 470 patients due to the exclusion criteria, including 322 patients with BD manic episodes (BD-M), 105 patients with BD depressive episodes (BD-D), 43 patients with BD mixed episodes (BD-mixed), as well as 102 healthy subjects as the healthy group (HG), as shown in Figure 1.

Discussion
Recently, growing attention has been paid to the hot topic of the relationship between inflammation and BD. Wei (12) considered the SIRI, NHR, and MHR to be predictors to differentiate BD-M patients from healthy people and regarded the MHR as a predictor for differentiating BD-D patients from healthy people, which was inconsistent with the findings of Dadouli (16) that MLR was the only risk factor for BD-M patients and no indicator was found to be related to BD-D compared with the control group. Ongoing evidence has investigated the role of inflammation in BD, yet their true and concrete relationship is not completely understood (10-13). Additionally, most related studies focused on the comparison between BD patients and BD-M patients or between BD-D patients due to a lack of data from   healthy individuals. As we know, few studies have assessed the inflammatory indicators between BD patients and healthy individuals based on subgroups of BD. Our findings showed that numerous immunological factors were associated with BD and its different states by univariate analysis. In terms of BD patients, the levels of MON, hs-CRP, and NLR were independent risk factors in logistic regression analysis, as well as the cut-off values were 0.405 10 9 /L, 2.600 mg/L, and 2.321, respectively. Regarding BD-M patients, the levels of MON, hs-CRP, and NLR were predictors in logistic regression analysis, and the cut-off values were, respectively, 0.505 10 9 /L, 2.600 mg/L, and 2.620. Additionally, the level of NLR and MHR were close indicators for BD-D and the cut-off values were 1.735 and 0.487, respectively. Furthermore, our findings showed that the level of MON was related to BD-mixed and the cut-off value was 0.340 10 9 /L. Notably, MON+hs-CRP + NLR, MON+hs-CRP + NLR, and NLR + MHR had the highest diagnostic accuracy to predict the BD, BD-M, and BD-D patients, respectively.
Monocytes are the most crucial cells for the secretion of pro-inflammatory and pro-oxidant cytokines, which are essential elements of the innate immune reaction against pathogens (17). Increased monocyte counts and alterations have been demonstrated to be associated with affective disorders (18,19). In the present study, monocyte counts were found to be independent predictors of BD and BD-M patients compared with healthy subjects. Furthermore, we also identified the cut-off value of MON to predict BD subtypes. Wei (12) found no significantly elevated monocyte counts, which was inconsistent with our results. The potential reason that can explain the discrepancy with Wei (12) is that the BMI was not matched between control cases and BD patients in his study, which can influence the inflammation indexes. While increased MON was only used to differentiate between patients with BD-M and healthy controls in the study of Dadouli (16), which was partially similar to our findings. These discrepancies may be attributed to the different time frames in which cases and control blood samples were collected, or race differences in the research of Dadouli, which had great effects on the difference in our results. Notably, another important finding was that the cut-off value of MON to predict BD-M patients was higher than that of BD patients, implying that when monocyte counts were higher than 0.405 10 9 /L or 0.505 10 9 /L, patients were more likely to suffer from BD or BD-M, respectively. Frontiers in Psychiatry 07 frontiersin.org C-reactive protein (CRP) is a pentamer acute-phase protein produced by the liver and secreted into the blood. CRP plays an important role in the innate immune system, rising rapidly in response to infection and inflammation, declining sharply after the acute phase. Fernandes (20) found that the level of CRP was higher in BD patients than in the health group regardless of their mood states, while the CRP levels of BD-M patients were much higher than in other states. In addition, it has been reported that the hs-CRP level of BD-M patients was significantly higher than in the health group before treatment, and the hs-CRP level was significantly reduced after treatment (21). These results were partially similar to our findings. In our study, hs-CRP was found to be an independent predictor of BD and BD-M patients compared with healthy cases. Notably, the same cut-off values of hs-CRP were identified to predict BD and BD-M patients.
NLR, PLR, MLR, NHR, LHR, MHR, PHR, SII, and SIRI index are cost-effective and simple laboratory parameters, implying a balance of systemic inflammation in many diseases (22)(23)(24)(25). Our results showed that compared to HG, NLR was higher in BD patients, BD-M patients, and BD-D patients, it was considered a key predictor for BD patients, BD-M patients, and BD-D patients, which was consistent with Dadouli's findings. However, regarding patients with BD-D, MHR was found to be a risk factor compared with healthy individuals, and its cut-off value was demonstrated. Wei (12) found that the MHR and NHR were independent predictors for identifying BD-M patients from the healthy group in ROC curve analysis, while the MHR was an independent predictor for differentiating BD-D patients from the healthy group, which is consistent with our results.
Other blood indexes, such as lymphocyte number, PLR, or SII, have been proven to play crucial roles in many diseases (26,27), which are related to BD in recent studies (14,15,28). In our study, these indicators were significantly higher in BD patients or BD subtype groups than in healthy people, whereas they were demonstrated to have a negative relationship with BD in ROC curve analysis. A large sample and multi-center study are needed to assess their value in the prediction of BD.
Furthermore, we tried to find the combined predictors with the highest diagnostic accuracy in BD patients with different states compared with healthy subjects using inflammatory indicators. Regarding BD patients, MON+hs-CRP + NLR was found to have the highest diagnostic accuracy (AUC area = 0.746) to predict these patients, while MON+hs-CRP + NLR (AUC area = 0.775) was the best combined predictor of BD-M patients. In terms of BD-D patients, NLR + MHR (AUC area = 0.652) was the optimal indicator. Pointing out the bestcombined predictors and their cut-off values can effectively and maximally help us distinguish different stages of BD patients as early as possible. We still need a larger sample to further verify these results.
However, some limitations should not be neglected. First, we divided BD patients into groups based on their mood episodes Frontiers in Psychiatry 08 frontiersin.org regardless of the severity of their symptoms, so the predictors of the severity of symptoms were not analyzed in this retrospective study. Second, some confounders, such as smoking and diet, may influence the inflammation indicators. Third, we did not collect other potential biomarkers regarding inflammation in our study due to the retrospective study.
In conclusion, we found various inflammatory indicators associated with BD and its subtypes. Regarding BD patients and BD-M patients, the levels of MON, hs-CRP, and NLR were independent risk factors in logistic regression analysis, while the levels of NLR and MHR were independent risk factors for BD-D. However, MON was an independent risk factor for BD-mixed. Furthermore, our findings identified the cut-off values and optimally combined indicators to predict BD, BD-M, BD-D, and BD-mixed, which help us better assess them to manage early targeted interventions.

Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Ethics statement
This study involving humans was reviewed and approved by the Institutional Review Board of the First hospital of Hebei Medical University (20220908). Write informed consent was not required in accordance with local and national regulations.