Evidence suggests that the impact of adult stressful life events (SLE) on health depends on an individual’s environmental and genetic predisposition to psychopathology. However, the influence of the genomic and exposomic (the sum of exposures) liabilities for schizophrenia on the association between SLE and health outcomes in the general population has not been studied. The current longitudinal study therefore aims to evaluate whether the association of recent SLE with mental as well as physical health is moderated by polygenic risk score for schizophrenia (PRS-S) and the exposome score for schizophrenia (ES-S, an aggregated environmental exposure score, akin to PRS-S).
Data from four waves of the Netherlands Mental Health Survey and Incidence Study-II (NEMESIS-II), a prospective survey in the Dutch general population aged 18 – 64, was used. The baseline data (T0) of NEMESIS-II, including 6646 participants, were collected from 2007 to 2009 and were followed up at year 3 (T1), year 6 (T2), and year 9 (T3). The Short-Form-36 Health Survey was used to repeatedly assess mental and physical health. Repeated assessment of SLE were based on the Brugha Life events section. Guided by our previous work, ES-S was constructed as the weighted sum of important baseline environmental factors, which were previously associated with schizophrenia. A subsample of 3099 participants (at T0) were genotyped and PRS-S was trained on the latest Psychiatric Genetics Consortium schizophrenia meta-analysis, adjusted for the first three population principal components. The analyses were conducted using random intercept multilevel regression models with age, sex, and education included as a priori covariates. Models were clustered for multiple assessments per individual, and the random intercept of mental or physical health at T-1 was added to the model to control for previous health status. We evaluated the main effects of SLE, ES-S, and PRS-S, and tested the moderating effects of ES-S and PRS-S on the impact of SLE on physical and mental health. As sensitivity analyses, we repeated the analyses using lagged SLE at T-1.
SLE and ES-S were associated with decreased mental (SLE: β = -2.53, P <0.001; ES-S: β = -2.93, P <0.001) and physical health (SLE: β = -2.62, P < 0.001; ES-S, β = -3.10, P < 0.001), while PRS-S was only associated with decreased mental health (β = -1.02, P <0.001) and not with physical health (β = -0.57, P = 0.061). ES-S moderated the association of SLE with mental health (β = -0.87, P = 0.002) but not with physical health (β = -0.60, P = 0.069). There was no evidence for gene–environment correlation between SLE and PRS-S (OR = 1.02, P = 0.527); and PRS-S did not moderate the association of SLE with mental (β = -0.38, P = 0.241) or physical health (β = -0.44, P = 0.252). The sensitivity analyses with lagged SLE at the previous time point converged with the results. SLE at T-1 was associated with mental (β = -0.90, P < 0.001) and physical health (β = -1.87, P < 0.001). ES-S moderated the association of SLE at T-1 with mental (β = -0.52 P = 0.029) but not physical health (β = -0.45, P = 0.116). PRS-S did not moderate the association of SLE at T-1 with mental (β = -0.03, P = 0.917) or physical health (β = 0.43, P = 0.233).
In support of the diathesis-stress model, we have provided novel evidence showing that exposure to adult SLE, particularly in individuals with high ES-S, is associated with decreased mental health. However, we did not find evidence for a gene-environment interaction in the context of SLExPRS-S. Our findings also add to the literature showing that genomic and exposomic liability for schizophrenia is pleiotropically associated with health outcomes in the general population.