Depression trajectory modifies the association between estimated glucose disposal rate and incident stroke: a prospective cohort study

BACKGROUND: Insulin resistance is an established risk factor for stroke. The estimated glucose disposal rate (eGDR) serves as a surrogate marker for insulin sensitivity. Depression is linked to both insulin resistance and cerebrovascular disease. However, whether depression trajectories modify the association between eGDR and stroke risk remains unknown. METHODS: This prospective cohort study used data from the China Health and Retirement Longitudinal Study (CHARLS). Participants aged ≥ 45 years with baseline eGDR measurements (2011), complete depression assessments (2011, 2013, 2015), no prevalent stroke (2011-2015), and follow-up data (2018 or 2020) were included. Depression trajectories were identified using latent class growth analysis based on the 10-item Center for Epidemiologic Studies Depression Scale. Covariates were selected a priori with the aid of a directed acyclic graph. Multivariable logistic regression estimated odds ratios (ORs) for incident stroke, stratified by depression trajectory; the primary model adjusted for sociodemographic and lifestyle factors. Interaction was assessed using likelihood ratio tests. RESULTS: Among 4,592 participants (mean age 58.3 years; 54.9% female), 430 stroke cases (9.4%) occurred. Three trajectories were identified: Low (65.1%), Moderate (28.1%), and High (6.8%). In the primary model (Model 2 adjusted for age, sex, marital status, education, residence, smoking status, and alcohol consumption), each standard deviation increase in eGDR was associated with lower stroke odds overall (OR = 0.66, 95% CI: 0.60-0.73). This inverse association was strongest in the Low trajectory (OR = 0.58, 95% CI: 0.50-0.67) and was attenuated in the Moderate (OR = 0.74, 95% CI: 0.63-0.88) and High (OR = 0.74, 95% CI: 0.55-0.99) trajectories. The attenuation with greater cumulative depressive burden was supported by significant interactions using binary trajectory classification (Low vs. Moderate/High; P = 0.021) and continuous mean CESD-10 score (P = 0.035); the three-class interaction test was borderline (P = 0.067). CONCLUSIONS: Higher eGDR was associated with reduced stroke risk; this protective association was progressively attenuated with greater cumulative depressive symptom burden, being strongest among individuals with persistently low depressive symptoms. These findings suggest that cumulative depressive burden may attenuate the inverse association between insulin sensitivity and stroke risk.