We investigated the relationship between serum levels of inflammatory markers and symptomatic severity of bipolar disorder among 228 drug-naïve patients with bipolar disorder. Our study has three major findings. First, in BDD, both pro-inflammatory markers and anti-inflammatory markers are lower than those in HC in spite of the symptomatic severity. Second, in BDM, the serum levels of pro-inflammatory markers first elevate then decrease with the worsening of the manic symptoms. Third, the imbalance of pro-inflammatory markers and anti-inflammatory markers is seen in both BDD and BDM, which is shown by an alteration (decrease in BDD and increase in BDM) in the IL-6/IL-10 ratio when the symptoms are mild and a decrease in the IL-17/IL-10 ratio when the symptoms are severe.
Contrary to the view from studies in major depressive episode (MDD)  that serum levels of both pro-inflammatory cytokines and ant-inflammatory cytokines slightly increase in depressive episode, our study finds that both pro-inflammatory cytokines and anti-inflammatory cytokines decrease in bipolar depression. In addition, the decrease of the ratios of IL-6/IL-10 and IL-17/IL-10 found in this study also suggests that immune-inhibition is prominent in bipolar depression, especially in severe BD. Different from studies in MDD, few studies have ever assessed the relationship between the serum levels of inflammatory markers and depressive symptoms from studies in BDD. Several studies with small sample size have found that bipolar depression is associated with a pro-inflammatory state [24, 25]. However, in two review studies [26, 27] which meta-analyzed the cytokines alteration in BD, no consistent conclusion about the relationship between cytokines and bipolar depression was reached since few studies on bipolar depression were included and the sample sizes of the included studies were all small. In addition, the heterogeneity between studies also played a part in this inconsistency. A recent comparison study base on a machine learning approach found that compared to healthy control, BDD had higher levels of C-C Motif Chemokine Ligand (CCL)3, CCL4, CCL5, CCL11, CCL25, CCL27, CXCL11, IL-9 and TNF-α. However, in this study, all the participants were medicated, potential confounders including age, gender, BMI, and tobacco or alcohol use were not controlled. In addition, the average age of the participants were bigger than those in our study (46.59±10.8 VS. 27.8±10.7 ), which might be another factor that makes their conclusion different from ours. In addition, the foregoing conclusion could not be duplicated by other studies either. For example, a study from German showed that both pro-inflammatory and anti-inflammatory markers, but not CRP were inversely correlated with the severity and symptoms of major depression. A recent study from China found that the serum levels of IL-13 and TNF-αwere significantly lower in BDD than in MDD, and the serum levels of IL-4 and TNF-αincreased in the treatment response subgroup of BDD. In recent years, anti-inflammatory agents were expected to be a promising treatment target of MDD and BDD. However, negative results from large clinical trials [33, 34]not only overturns the conclusion from small studies that the adjunctive use of anti-inflammatory drugs might help improve depressive symptoms in MDD or BDD, but also make us rethink the relationship between serum levels of inflammatory cytokines and depressive symptoms. Moreover, pro-inflammatory cytokines are not always neurotoxic. Instead, low-dose of pro-inflammatory cytokines such as IL-6, TNF-a, and IFN-γ have been proved to have a neuroprotective role.
The relationship between serum levels of inflammatory markers and BDM has been widely studied. Most of previous studies find that both pro-inflammatory cytokines (IL-1RA, IL-1, Il-2, sIL-2R[39, 40], IL-6[38, 41], sIL-6R, TNF-α, CXCL10, CXCL11, CRP, IL-17, IFN-γ and IL-18) and anti-inflammatory markers (IL-4 and IL-10) elevate in the manic episode. However, our study only partially supports this view. In our study, the elevation of pro-inflammatory markers was seen only in mild and moderate BDM, while in severe BDM the serum levels of pro-inflammatory were found to decrease. This finding is consistent with our previous hypothesis about the dynamic change of inflammatory markers over the course of a multi-system inflammatory disease. As far as we know, this has rarely been explored before. Although serum levels of some inflammatory cytokines like sIL-2R, IL-17, neural cell adhesion molecule 1(NCAM-1), carcinoembryonic antigen (CEA) and IFN-γ have been reported to be positively correlated with severity of manic symptoms, the relationship between serum levels of inflammatory markers, according to our study, seems to be more complicated than a positive correlation.
Inflammatory ratios except N/L, which are less affected by exercise, BMI, and other confounding factors than other commonly used markers of inflammation, have been rarely studied in BD before. Partially in line with previous reports that N/L ratio is higher in BDM than in BDD or MDD, our study finds that N/L ratio is significantly higher in severe BDM than in HC, suggesting an imbalance in favor of innate immunity . In addition, we also find that the ratios of IL-6/IL-10 and IL-17/IL-10 vary with the clinical phase of BD and severity of symptoms: in BDD, the IL-6/IL-10 ratio first decrease in mild BDD and then the IL-17/IL-10 ratio decrease in severe BDD; while in BDM, the IL-6/IL-10 ratio first elevate in mild BDM and then decrease with the IL-17/IL-10 ratio in severe BDM. That is to say, the IL-6/IL-10 imbalance is an early immune response to BD, while IL-17/IL-10 imbalance is an indicator of deterioration of BD. To our knowledge, these have not been reported before. Although the pathophysiological meaning of IL-6/IL-10 and IL-17/IL-10 balance is far from being interpreted clearly, it is possible to suppose that they play a role in homing of inflammatory cells and therefore in the outcome of inflammation.
This study has several strengths. First of all, it comprises of one of the largest sample of BD patients ever examined. Second, all the subjects were Han Chinese and drug-naïve, most of them were young, and active physical diseases and recent substance abuse were excluded, which provided us a good sample with great homogeneity. Third, generalized linear model was used to statistically analyze the data, whereas as many potential confounding factors as possible were adjusted for, so the power of test greatly improved. Fourth, as many inflammatory cytokines as possible were measured at the same time, which makes the assessment of the immune state of the subjects more multidimensional. Finally, all the blood samples were treated in the same way and in the same laboratory, thus minimized the effect of measurement deviation. However, several limitations should be addressed when interpreting the foregoing conclusion. First, its cross-sectional design limits causal conclusion about the relationship between serum levels of inflammatory markers and BD. Second, symptoms of BD were treated as a binary variable—depressive or manic symptoms, while other domains of symptoms like psychotic symptoms, or cognitive symptoms were not concerned, which had been proved to be associated with the levels of inflammation. Third, although we have the largest sample of BD ever examined, the sample might be still not big enough to detect some small but clinically meaningful effect of inflammatory marker on the symptomatic severity of BD, especially when the sample is divided into several subgroups. Fourth, some but fortunately not much inflammatory markers' data was too dispersed to build a good generalized linear model, therefore caution should be exercised when interpretating the corresponding results.