Genetic risk for alcohol use disorder in relation to individual symptom criteria: Do polygenic indices provide unique information for understanding severity and heterogeneity?

Abstract

Background and aims: Previous studies have suggested that Alcohol Use Disorder (AUD) might result from separable genetic influences with symptoms reflecting differing degrees of genetic risk related to distinct risk mechanisms. The present study aimed to examine the genetic risk for individual AUD symptom criteria using a polygenic risk score (PRS) approach to assess the relative severity of each symptom and test for a multidimensional genetic structure of AUD.

Design: This retrospective study examined the correlation between genetic severity (i.e. PRS) and Item Response Theory (IRT) severity indices for each AUD symptom. Multiple Indicators Multiple Causes (MIMIC) models were employed to examine the effect of the PRS on individual AUD symptoms after accounting for overall AUD severity.

Setting and participants: The study made use of summary-level data produced in Finland, United Kingdom and the United States of America as well as individual-level data from 1639 participants of the University of California - San Francisco Family Alcoholism Study.

Measurements: Phenotypic measures included the 11 Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) AUD symptom criteria assessed using a modified version of the Semi-Structured Assessment for the Genetics of Alcoholism. AUD, Problematic Alcohol Use (PAU) and alcohol consumption [i.e. drinks per week (DPW)] PRSs were created using summary statistics obtained from published genome-wide association studies (GWAS).

Findings: Phenotypic and genotypic severities of AUD symptoms were statistically significantly correlated for the PAU PRS and AUD PRS (r range = 0.77-0.89), but not for the DPW PRS (r = 0.45). MIMIC models indicated that the PRSs statistically significantly predicted the AUD factor. Regression paths testing the direct effects of the PRSs on individual AUD symptoms, independent of the latent AUD factor, were statistically nonsignificant.

Conclusions: Polygenic risk scores derived from genome-wide association studies (GWAS) of alcohol use disorder (AUD) appear to influence AUD symptom expression through a single genetic factor that is highly correlated with the relative severity of individual symptoms measured at the phenotypic level. Item-level GWAS of AUD symptoms are needed to further parse heterogeneous symptom expression and allow for more nuanced tests of these conclusions.