Genetic, environmental, and behavioral correlates of lifetime suicide attempt: Analysis of additive and interactive effects in two cohorts of US Army soldiers

Abstract: Recently developed measures of genetic liability to suicide attempt may convey unique information regarding an individual’s risk of suicidal behaviour. We calculated a polygenic risk score for suicide attempt (SA-PRS) for soldiers of European ancestry who participated in the Army STARRS New Soldier Study (NSS; n = 6573) or Pre/Post Deployment Study (PPDS; n = 4900). Multivariable logistic regression models were fit within each sample to estimate the association of SA-PRS with lifetime suicide attempt (LSA), and to examine whether SA-PRS displayed additive or interactive effects with environmental and behavioural risk/protective factors (lifetime trauma burden, childhood maltreatment, negative urgency impulsivity, social network size, perceived mattering, and dispositional optimism). Age, sex, and within-ancestry variation were included as covariates. Observed prevalence of LSA was 6.3% and 4.2% in the NSS and PPDS samples, respectively. In the NSS model, SA-PRS and environmental/behavioural factors displayed strictly additive effects on odds of LSA. Results indicated an estimated 21% increase in odds of LSA per 1 SD increase in SA-PRS [adjusted odds ratio (AOR; 95% CI) = 1.21 (1.09–1.35)]. In PPDS, the effect of SA-PRS varied by reports of optimism [AOR = 0.85 (0.74–0.98) for SA-PRS x optimism effect]. Individuals reporting low and average optimism had 37% and 16% increased odds of LSA per 1 SD increase in SA-PRS, respectively, whereas SA-PRS was not associated with LSA in those reporting high optimism. Overall, results suggested the SA-PRS had predictive value over and above several environmental and behavioural risk factors for LSA. Moreover, elevated SA-PRS may be more concerning in the presence of environmental and behavioural risk factors (e.g., high trauma burden; low optimism). Given the relatively small effect magnitudes, the cost and incremental benefits of utilizing SA-PRS for risk targeting must also be considered in future work.