Abstract
BACKGROUND: Caffeine, alcohol, nicotine and cannabis are commonly used psychoactive substances. While the use of these substances has been previously shown to be genetically correlated, causality between these substance use traits remains unclear. We aimed to revisit the genetic relationships among different measures of SU using genome-wide association study (GWAS) summary statistics from the UK Biobank, International Cannabis Consortium, and GWAS & Sequencing Consortium of Alcohol and Nicotine use.</p>
METHODS: We obtained GWAS summary statistics from the aforementioned consortia for ten substance use traits including various measures of alcohol consumption, caffeine consumption, cannabis initiation and smoking behaviours. We then conducted SNP-heritability (h2) estimation for individual SU traits, followed by genetic correlation analyses and two-sample Mendelian randomisation (MR) studies between substance use trait pairs.</p>
RESULTS: SNP h2 of the ten traits ranged from 0.03 to 0.11. After multiple testing correction, 29 of the 45 trait pairs showed evidence of being genetically correlated. MR analyses revealed that most SU traits were not causally associated with each other. However, we found evidence for an MR association between regular smoking initiation and caffeine consumption 40.17 mg; 95 % CI: [24.01, 56.33] increase in caffeine intake per doubling of odds in smoking initiation). Our findings were robust against horizontal pleiotropy, SNP-outliers, and the direction of causality was consistent in all MR analyses.</p>
CONCLUSIONS: Most of the substance traits were genetically correlated but there is little evidence to establish causality apart from the relationship between smoking initiation and caffeine consumption.</p>