Vulnerability theories propose that suboptimal levels of lipid markers and proinflammatory proteins predict future heightened depression. Scar models posit the reverse association. However, most studies that tested relationships between non-specific immune/endocrine markers and depression did not separate temporal inferences between people and within-person and how different immunometabolism markers related to unique depression symptoms. We thus used cross-lagged prospective network analyses (CLPN) to investigate this topic.

Community midlife women (n = 2224) completed the Center for Epidemiologic Studies-Depression scale and provided biomarker samples across five time-points spanning 9 years. CLPN identified significant relations (edges) among components (nodes) of depression (depressed mood, somatic symptoms, interpersonal issues), lipid markers [insulin, fasting glucose, triglycerides, low-density lipoprotein-cholesterol (LDL), high-density lipoprotein-cholesterol (HDL)], and proinflammatory proteins [C-reactive protein (CRP), fibrinogen], within and across time-points. All models adjusted for age, estradiol, follicle-stimulating hormone, and menopausal status.

In within-person temporal networks, higher CRP and HDL predicted all three depression components (d = 0.131–2.112). Increased LDL preceded higher depressed mood and interpersonal issues (v. somatic symptoms) (d = 0.251–0.327). Elevated triglycerides predicted more somatic symptoms (v. depressed mood and interpersonal problems) (d = 0.131). More interpersonal problems forecasted elevated fibrinogen and LDL levels (d = 0.129–0.331), and stronger somatic symptoms preceded higher fibrinogen levels (d = 0.188).

Results supported both vulnerability and scar models. Long-term dysregulated immunometabolism systems, social disengagement, and related patterns are possible mechanistic accounts. Cognitive-behavioral therapies that optimize nutrition and physical activity may effectively target depression.

Previous cross-lagged studies on depression and memory impairment among the elderly have revealed conflicting findings relating to the direction of influence between depression and memory impairment. The current study aims to clarify this direction of influence by examining the cross-lagged relationships between memory impairment and depression in an Asian sample of elderly community dwellers, as well as synthesizing previous relevant cross-lagged findings via a meta-analysis.

A total of 160 participants (Mage = 68.14, s.d. = 5.34) were assessed across two time points (average of 1.9 years apart) on measures of memory and depressive symptoms. The data were then fitted to a structural equation model to examine two cross-lagged effects (i.e. depressive symptoms→memory; memory→depressive symptoms). A total of 14 effect-sizes for each of the two cross-lagged directions were extracted from six studies (including the present; total N = 8324). These effects were then meta-analyzed using a three-level mixed effects model.

In the current sample, lower memory ability at baseline was associated with worse depressive symptoms levels at follow-up, after controlling for baseline depressive symptoms. However, the reverse effect was not significant; baseline depressive symptoms did not predict subsequent memory ability after controlling for baseline memory. The results of the meta-analysis revealed the same pattern of relationship between memory and depressive symptoms.

These results provide robust evidence that the relationship between memory impairment and depressive symptoms is unidirectional; memory impairment predicts subsequent depressive symptoms but not vice-versa. The implications of these findings are discussed