The Contributions of Multiple Polygenic Scores in Predicting Liability for Major Depressive Disorder and Its Comorbidity with Alcohol Use Disorder.

Abstract

The inclusion of polygenic scores (PGS) from genetically correlated traits such as Major Depressive Disorder (MDD) and alcohol use disorder (AUD) may improve the prediction of these outcomes and their comorbidity. Despite the importance of this work, few studies have evaluated the efficacy of this possibility. The current study evaluates the use of MDD and AUD PGS individually and together to improve the prediction of MDD, AUD, and comorbid MDD-AUD using a sample of European, African, or Admixed American Ancestry participants from the National Longitudinal Study of Adolescent to Adult Health (N = 7,965). Cross-ancestry MDD and AUD PGS were created using PRS-CSx. The best fitting model of comorbid MDD-AUD in the whole sample included PGS for MDD and AUD (PGS_MDD OR: 1.26, 95% CI 1.16-1.35, p = 2.69 × 10^- 6; PGS_AUD OR: 1.77, 95% CI 1.66-1.87, p = 3.49 × 10^- 28), explaining an additional 4.88% of variance compared to a model only including sociodemographic covariates. For MDD, the best fitting model included the MDD PGS (OR: 1.25, 95% CI 1.17-1.33, p = 2.05 × 10^- 8), explaining an additional 0.65% of variance. For AUD, the best fitting model included the AUD PGS (OR: 1.37, 95% CI 1.32-1.43, p = 1.25 × 10^-28), which explained an additional 1.52% of variance. Inclusion of both PGS did not significantly improve the prediction of individual MDD or AUD. Inclusion of PGS for MDD and AUD significantly improved prediction for comorbid MDD-AUD, but not in MDD or AUD. These results help clarify the role of utilizing genetically correlated PGS in improving prediction of MDD, AUD, and comorbid MDD-AUD.