Using network analysis to examine links between individual depression symptoms, inflammatory markers, and covariates

Background: Studies investigating the link between Major Depressive Disorder (MDD) and inflammation have yielded inconsistent results. This may be due to two factors. First, studies differed regarding the specific inflammatory markers studied and covariates accounted for. Second, specific depression symptoms may be differentially related to inflammation, and traditional analyses of symptom total scores or diagnostic status may obfuscate such associations. Here we test these two possibilities.
Methods: In increasing levels of complexity, we estimated six regularized partial correlation networks in 2321 participants of the Netherlands Study of Depression and Anxiety study to explore the shared variance among depression modeled via sum-score, 9 DSM-5 symptoms, or 28 individual symptoms; inflammatory markers C-reactive protein (CRP), interleukin 6 (IL-6) and tumor necrosis factor α (TNF-α); excluding and including the covariates sex, age, body mass index (BMI), exercise, smoking, alcohol, and chronic diseases
Results: After covariate adjustment, the depression sum-score was associated with IL-6; all markers were related to sex; and CRP and IL-6 were associated with BMI. When modeling the 9 DSM-5 symptoms and CRP as replication of Jokela et al. (2016), CRP was associated with trouble sleeping, energy level, BMI, and sex. In a final adjusted model with all 28 individual symptoms, depression and inflammation were unrelated; CRP was associated with BMI and sex; IL-6 with BMI and age; and TNF-α with chronic diseases.
Conclusions: Differential relations between markers, symptoms, and covariates emerged. Links between depression and inflammation were largely to completely attenuated after covariate adjustment.

背景：关于重度抑郁症（MDD）与炎症之间联系的研究结果并不一致。这可能是由于两个因素造成的。首先，不同研究在所研究的特定炎症标志物和协变量考虑方面存在差异。其次，特定的抑郁症状可能与炎症存在不同的关联，而传统对症状总分或诊断状态的分析可能会模糊这种关联。本研究旨在检验这两种可能性。
方法：我们以递增的复杂度，在荷兰抑郁焦虑研究（Netherlands Study of Depression and Anxiety）的2321名参与者中估计了六个正则化部分相关网络，以探索通过总分、9个DSM-5症状或28个单独症状建模的抑郁之间的共有方差；炎症标志物C反应蛋白（CRP）、白细胞介素6（IL-6）和肿瘤坏死因子α（TNF-α）；在排除和包含性别、年龄、体重指数（BMI）、运动、吸烟、饮酒和慢性疾病等协变量。
结果：在调整协变量后，抑郁总分与IL-6相关；所有标志物都与性别相关；CRP和IL-6与BMI相关。当将9个DSM-5症状和CRP作为Jokela等（2016年）的复制研究时，CRP与睡眠困难、能量水平、BMI和性别相关。在最终的调整模型中，包含所有28个单独症状，抑郁和炎症无关联；CRP与BMI和性别相关；IL-6与BMI和年龄相关；TNF-α与慢性疾病相关。
结论：标志物、症状和协变量之间的差异关系显现。在调整协变量后，抑郁与炎症之间的联系在很大程度上甚至完全减弱。