Body mass index and depressive symptoms: instrumental-variables regression with genetic risk score

The causal role of obesity in the development of depression remains uncertain. We applied instrumental-variables regression (Mendelian randomization) to examine the association of adolescent and adult body mass index (BMI) with adult depressive symptoms. Participants were from the Young Finns prospective cohort study (n = 1731 persons, 2844 person-observations), with repeated measurements of BMI and depressive symptoms (modified Beck's Depression Inventory). Genetic risk score of 31 single nucleotide polymorphisms previously identified as robust genetic markers of body weight was used as a proxy for variation in BMI. In standard linear regression analysis, higher adult depressive symptoms were predicted by higher adolescent BMI (B = 0.33, CI = 0.06-0.60, P = 0.017) and adult BMI (B = 0.47, CI = 0.32-0.63, P < 0.001). These associations were replicated in instrumental-variables analysis with genetic risk score as instrument (B = 1.96, CI = 0.03-3.90, P = 0.047 for adolescent BMI; B = 1.08, CI = 0.11-2.04, P = 0.030 for adult BMI). The association for adolescent BMI was significantly stronger in the instrumented analysis compared to standard regression (P = 0.04). These findings provide additional evidence to support a causal role for high BMI in increasing symptoms of depression. However, the present analysis also demonstrates potential limitations of applying Mendelian randomization when using complex phenotypes.