Sex Differences in Depression: Insights from Multimodal Gray Matter Morphology and Peripheral Inflammatory Factors

Abstract

Major depressive disorder (MDD) exhibits notable sex differences in prevalence and clinical and neurobiological manifestations. Though the relationship between peripheral inflammation and MDD-related brain changes is well studied, the role of sex as a modifying factor is underexplored. This study aims to assess how sex influences brain and inflammatory markers in MDD. We utilized voxel-based and surface-based morphometry to analyze gray matter (GM) structure, along with GM-based spatial statistics (GBSS) to examine GM microstructure among treatment-naive patients with depression (n = 174) and age-matched healthy controls (n = 133). We uncovered sex-by-diagnosis interactions in several limbic system structures, the frontoparietal operculum and temporal regions. Post hoc analyses revealed that male patients exhibit pronounced brain abnormalities, while no significant differences were noted in females despite their higher depressive scores. Additionally, heightened inflammation levels in MDD were observed in both sexes, with sex-specific effects on sex-specific brain phenotypes, particularly including a general negative correlation in males. Intriguingly, mediation analyses highlight the specific role of the parahippocampal gyrus (PHG) in mediating interleukin (IL)-8 and depression in men. The findings suggest that in clinical practice, it would be beneficial to prioritize sex-specific assessments and interventions for MDD. This includes recognizing the possibility that male patients may experience significant brain alterations, especially when identifying male patients who may underreport symptoms. Possible limitations encompass a small sample size and the cross-sectional design. In future research, the incorporation of longitudinal studies or diverse populations, while considering illness duration, will enhance our understanding of how inflammation interacts with brain changes in depression.

Keywords: IL-8; brain morphology; gray-matter-based spatial statistics; inflammatory factors; major depressive disorder; microstructure; sex difference.

Figure 1

Brief pipeline for the multimodal metric of GM. (A) Structural-level metric of GM. (B) Microstructural-level metric of GM. Note: GM, gray matter; SBM, surface-based morphometry; VBM, voxel-based morphometry. GBSS, GM-based spatial statistic; MD, mean diffusivity; FA, fractional anisotropy.

Figure 2

Main effects of MDD diagnosis. The main effect of MDD on cortical thickness (A), GM FA (B), and MD (C). Areas in blue-green indicate regions where the cortical thickness (A) and GM FA (B) are significantly lower in the MDD group compared with HCs regardless of sex. The red-yellow areas are regions where the MD (C) is significantly higher in the MDD group compared with HCs. All statistical significances were determined at p < 0.05 after applying FWE correction for multiple comparisons following TFCE while controlling for age and BMI. Note: MDD, major depressive disorder; GM, gray matter; HC, healthy control; MD, mean diffusivity; FA, fractional anisotropy; TFCE, threshold-free cluster enhancement; FEW, family-wise error rate.

Figure 3

Sex-by-diagnosis interaction on GMV and cortical thickness. (A) The red-yellow areas indicate a significant sex-by-diagnosis interaction on GMV (p < 0.05 after FWE correction for multiple comparisons following TFCE). (B) Histograms exhibit the validation of region of interest (ROI) analysis in the original individual space for each cluster determined by the interaction of sex and diagnosis on GMV. (C) (Left): The red-yellow areas indicate a significant sex-by-diagnosis interaction on cortical thickness (p < 0.05 after FWE correction for multiple comparisons following TFCE). (Right): The histograms display ROI validation in raw space for each cluster determined by the interaction of sex and diagnosis on cortical thickness. Note: GMV, gray matter volume; Vol_(rel), relative volume; MDD, major depressive disorder; HC, healthy control; F, female; M, male.

Figure 4

The impact of MDD on males. Areas in blue-green indicate regions where the cortical thickness (A) and GMV (B) are significantly decreased in male patients with MDD. (C) The effect of MDD on GM MD in males. Areas in red-yellow indicate regions where GM MD is significantly increased in male patients with MDD. All statistical significances were determined at p < 0.05 after applying FWE correction for multiple comparisons following TFCE while controlling for age and BMI. Note: MDD, major depressive disorder; GM, gray matter; HC, healthy control; MD, mean diffusivity; FA, fractional anisotropy; TFCE, threshold-free cluster enhancement; FEW, family-wise error rate.

Figure 5

Sex differences in inflammatory factors associated with MDD. Altered plasma concentrations of MMP8 (A), active MMP8 (B), the pro-inflammatory cytokines TNF-α (C), IL-6 (D), and IL-8, and (E) the anti-inflammatory cytokine IL-10 (F) in both male and female individuals diagnosed with MDD while controlling for age and BMI. Significant sex-by-diagnosis interactions were found for IL-8 ((E), right). (G) Correlations between IL-8 and HAMD (left) or BDI (right) in female individuals with MDD, controlling for age and BMI. (H) Correlations between IL-8 and HAMD (left) or BDI (right) in male individuals with MDD, controlling for age and BMI. Note: MMP8, matrix metalloproteinase-8; TNF, tumor necrosis factor; IL, interleukin. * p < 0.05.

Figure 6

Correlations among inflammation, structural imaging phenotypes, and depression. (A,B) The correlation matrix demonstrates the associations between inflammatory factors and structural imaging phenotypes with sex-by-diagnosis interaction in females (A) and males (B) while controlling for age and BMI. The ellipses in the plot symbolize the correlation between variable pairs. Larger and more elongated ellipses denote stronger correlations, while smaller or circular ones suggest weaker or no correlations. Red, upward-pointing ellipses indicate positive correlations, whereas blue, downward-pointing ones represent negative correlations. Statistically significant correlations are marked with asterisks (p < 0.05). (C–E) Examining the mediating role of Vol_cluster1 in the association between IL-8 and depression across all participants (C), as well as separately for females (D) and males (E). (F–H) Investigating the mediating role of Vol_cluster4 in the relationship between IL-8 and depression among all participants (F), as well as individually for females (G) and males (H). Note: vol, volume; th, thickness. * p < 0.05, ** p < 0.01, *** p < 0.001.