Decoding the inflammatory signature of the major depressive episode: insights from peripheral immunophenotyping in active and remitted condition, a case-control study

Abstract

Depression is a prevalent and incapacitating condition with a significant impact on global morbidity and mortality. Although the immune system's role in its pathogenesis is increasingly recognized, there is a lack of comprehensive understanding regarding the involvement of innate and adaptive immune cells. To address this gap, we conducted a multicenter case-control study involving 121 participants matched for sex and age. These participants had either an active (or current) major depressive episode (MDE) (39 cases) or a remitted MDE (40 cases), including individuals with major depressive disorder or bipolar disorder. We compared these 79 patients to 42 healthy controls (HC), analyzing their immunological profiles. In blood samples, we determined the complete cell count and the monocyte subtypes and lymphocyte T-cell populations using flow cytometry. Additionally, we measured a panel of cytokines, chemokines, and neurotrophic factors in the plasma. Compared with HC, people endorsing a current MDE showed monocytosis (p = 0.001), increased high-sensitivity C-reactive protein (p = 0.002), and erythrocyte sedimentation rate (p = 0.003), and an altered proportion of specific monocyte subsets. CD4 lymphocytes presented increased median percentages of activation markers CD69⁺ (p = 0.007) and exhaustion markers PD1⁺ (p = 0.013) and LAG3⁺ (p = 0.014), as well as a higher frequency of CD4⁺CD25⁺FOXP3⁺ regulatory T cells (p = 0.003). Additionally, patients showed increased plasma levels of sTREM2 (p = 0.0089). These changes are more likely state markers, indicating the presence of an ongoing inflammatory response during an active MDE. The Random Forest model achieved remarkable classification accuracies of 83.8% for MDE vs. HC and 70% for differentiating active and remitted MDE. Interestingly, the cluster analysis identified three distinct immunological profiles among MDE patients. Cluster 1 has the highest number of leukocytes, mainly given by the increment in lymphocyte count and the lowest proinflammatory cytokine levels. Cluster 3 displayed the most robust inflammatory pattern, with high levels of TNFα, CX3CL1, IL-12p70, IL-17A, IL-23, and IL-33, associated with the highest level of IL-10, as well as β-NGF and the lowest level for BDNF. This profile is also associated with the highest absolute number and percentage of circulating monocytes and the lowest absolute number and percentage of circulating lymphocytes, denoting an active inflammatory process. Cluster 2 has some cardinal signs of more acute inflammation, such as elevated levels of CCL2 and increased levels of proinflammatory cytokines such as IL-1β, IFNγ, and CXCL8. Similarly, the absolute number of monocytes is closer to a HC value, as well as the percentage of lymphocytes, suggesting a possible initiation of the inflammatory process. The study provides new insights into the immune system's role in MDE, paving the ground for replication prospective studies targeting the development of diagnostic and prognostic tools and new therapeutic targets.

Fig. 1. Sensitive screening of circulating monocytes subsets by direct blood staining in patients with MDE showed an altered proportion compared to control.

A Representative dot plots, after fixation and RBCs lysis, show gating strategies based on forward scatter (FSC-A) versus side scatter (SCC-A) and then discrimination of singlets. After that, representative dot plots show the sequential gating based on the expression of HLA-DR versus CD86 first, and then, CD14 versus CD16. Finally, the 3 monocyte subsets are represented in blue for classical, in red for intermediate, and green for non-classical ones. B–D Independent data showing the percentage of classical (CD14⁺⁺CD16^-), intermediate (CD14⁺CD16⁺), and non-classical (CD14⁻CD16⁺⁺) monocytes determined by flow cytometry. The median and interquartile range of each group is shown. Major depression episode (MDE), active vs remitted MDE, Healthy Control (HC). Statistical differences among groups was calculated by Kruskal–Wallis rank sum test with Bonferroni adjustment for multiple comparisons. ****P-value < 0.0001; ***P-value < 0.001; ns: not significant.

Fig. 2. CD4 lymphocytes in patients with MDE showed increased activation and exhaustion markers with higher frequency of FOXP3 regulatory T cells.

A Representative dot plots, after fixation and RBCs lysis, showing gating strategies based on forward scatter (FSC) versus side scatter (SCC) and then discrimination of singlets. After that, representative dot plots show the sequential gating based on the expression of CD3 versus FSC first, and then, CD4 versus CD8. B–D Representative dot plots and the independent data showing the percentage of CD69⁺, PD1⁺ and, LAG3⁺ on CD4⁺ lymphocytes and, E the representative dot plot and percentage of CD4⁺CD25⁺FOXP3⁺ regulatory T cells, determined by flow cytometry. The median and interquartile range of each group is shown. Major Depression Episode (MDE), active vs remitted MDE, Healthy Control (HC). Statistical differences among groups were calculated by Kruskal–Wallis rank sum test with Bonferroni adjustment for multiple comparisons.**P-value < 0.01; *P-value < 0.05; ns: not significant.

Fig. 3. Inflammatory and Neuroinflammation panel measured in plasma of patients with MDE showed increased levels of sTREM2, IL-17A, and IL-6.

A, B Cytokines, chemokines, and neurotrophic factors, were assessed employing a customized LEGENDPlex system for a human inflammatory panel 1 (A), and a neuroinflammation panel (B) determined by flow cytometry. Each panel allow the simultaneous quantification of all assessed molecules in 50 µL of the sample. The concentrations of each cytokine are determined by a standard curve provided by the kit, see Supplemental Figs. S1A and S1B. C IL-6 was determined by high-sensitivity (HS) ELISA kit. Independent data for each molecule is graphed and the median of each group is shown. Human Inflammatory Panel 1 includes: IL-1β, IFNγ, IL-17A, IL-33, CXCL8, IL-10, IL-12p70 and IL-23, and the Human Neuroinflammation Panel includes: β-NGF, CX3CL1, BDNF, sTREM2, IL-18, TNFα and CCL2. Major Depression Episode (MDE), active vs remitted MDE, Healthy Control (HC). Statistical differences among groups was calculated by Kruskal–Wallis rank sum test with Bonferroni adjustment for multiple comparisons. **P-value < 0.01; *P-value < 0.05.

Fig. 4. Boruta and Random Forest validate the discriminatory power of biological markers in patients with MDE.

To select the most important variables for classifying individuals, Random Forest, in combination with the Boruta algorithm, was employed. A The first training set was performed with the 70% (N = 84) of the pool data discriminated by patients with MDE or HC. From this analysis, the most discriminatory variables were: lymphocyte percentage, monocytes absolute count, classical monocytes, non-classical monocytes, intermediate monocytes, hs-CRP, CCL2, CD4⁺PD1⁺, CD4⁺LAG3⁺ and CD4⁺CD25⁺ FOXP3⁺ Treg. B Secondly, the training model was applied to discriminating patients with active MDE, remitted, or HC. From this analysis, the most discriminatory variables were: classical monocytes, non-classical monocytes, intermediate monocytes, monocytes absolute count, ESR, hs-CRP, CD4⁺CD69⁺, CD4⁺LAG3⁺ and CD4⁺CD25⁺ FOXP3⁺ Treg.

Fig. 5. Clustering analysis shows different inflammatory groups within patients with MDE.

A clustering analysis was performed considering both active and remitted cases. A Initially, a principal component analysis was applied to reduce the dimensionality of the data and work with the most significant contributors. B The hierarchical clustering analysis revealed 3 main segregated groups based on the 9 selected principal components obtained. All clinical, biochemical, and immunological characteristics of each cluster are summarized in Tables 3 and 4 and Supplementary Tables S4 and S5.