Sibling Comparisons of Polygenic Scores: Evidence for Quasi-Causal Effects of Educational Attainment Polygenes on Substance Use Disorders

Low educational attainment is associated with higher rates of substance use disorder (SUD) diagnoses. We took a genetically informed perspective to understand the nature of these associations and examined whether (1)... [ view full abstract ]

Low educational attainment is associated with higher rates of substance use disorder (SUD) diagnoses. We took a genetically informed perspective to understand the nature of these associations and examined whether (1) genome-wide polygenic scores for educational attainment (EDU-GPS) also predict SUDs (i.e., pleiotropic effects), and (2) whether the associations between EDU-GPS and SUDs are likely to be “causal”. Our sample included 864 families of European-American siblings who were 25 years of age or older (2-12 siblings per family; 2317 total individuals; 53% female) collected as part of the Collaborative Study on the Genetics of Alcoholism (COGA). Our SUD outcomes included lifetime criterion counts for DSM-5 Alcohol Use Disorder (AUDSX); Cannabis Use Disorder (CUDSX); and Fagerstrom Nicotine Dependence (NDSX). Educational attainment polygenic scores (EDU-GPS) were calculated in the COGA sample using genome-wide association weights from the Social Science Genetic Association Consortium. We used mixed models to examine whether EDU-GPS was associated with the three SUD criterion counts at the population level. Covariates included age; cohort; sex; the number of variants available for polygenic scoring; and two genetic ancestry principal components. We found that higher EDU-GPS predicted lower AUDSX, CUDSX, and NDSX (ps < 0.01). We then used sibling comparisons to examine whether the population-level associations between EDU-GPS and SUDs were likely to be “causal”. Sibling comparisons control for the other genetic and environmental background factors that siblings share, and thus permit a quasi-causal test of the effects of EDU-GPS on SUDs. Those with higher EDU-GPS compared to their siblings met fewer SUD criteria, although the effect for AUDSX was only at the trend level (AUDSX: b = -5.84, p = 0.06; CUDSX: b = -119.81, p < 0.01; and NDSX: b = -6.65, p = 0.03). This suggests that EDU-GPS may have a causal, protective effect against the development of SUDs.