SA110 - SHARED POLYGENIC RISK ANALYSIS OF ALCOHOLISM AND 5 MAJOR PSYCHIATRIC DISORDERS

Abstract

Background Alcohol is the most commonly used and abused substance. Addictions to alcohol and other legal/illicit drugs are heritable disorders influenced by a large number of variants of small effect, frequently being a comorbidity of other psychiatric disorders. This comorbidity between a mental illness and a Substance Use Disorder (SUD) is consistently contributing to an increased disability in individuals. In spite of this evidence of important comorbidity, shared genetic factors remain partially understudied. Previous GWAS of SUD showed evidence of genetic overlap with schizophrenia, bipolar disorder and major depressive disorder. We tested individual associations using Polygenic Risk Scores (PRS) from the Psychiatric Genomics Consortium's Cross-Disorder meta-analysis as discovery samples, and each of the five psychiatric disorders included in the study (ADHD, AUT, BIP, MDD, and SCZ), including 2014 PGC schizophrenia mega-analysis results, and, as target sample, a 1144 individuals case/control sample of alcohol abuse/dependence from Galicia, northwest Spain. Methods A total of 572 substance dependent patients were included. Inclusion criteria were: age between 18 and 65, born in Galicia and follow DSM-IV criteria for dependence of at least one substance and abuse/dependence of other substance. The substances registered in the study were: alcohol, tobacco, cannabis, cocaine, opiates, hypnotics, stimulants, hallucinogens and solvents. Most patients (N=540) were diagnosed with alcohol abuse/dependence. A total number of 604 controls from Galicia were included (mean age at recruitment: 40.26, SD: 10.70). DNA was extracted from blood samples and genotyping was performed using Illumina Infinium PsychArray. Variant imputation was executed with SHAPEIT and IMPUTE2, and PRS models were derived from the summary statistics of the PGC cross-disorder meta-analysis (CROSS) and 2014 PGC schizophrenia mega-analysis (SCZ2). Extensive data cleaning and Quality Control (QC) was employed. SNPs within the extended MHC region were excluded from part of the analysis. Associations between every PRS model and our data were tested using logistics regression. Results After QC and imputation, a total number of 503 alcohol cases and 587 controls remained (728 males and 362 females) and genotypic data remained in a total number of 6294324 SNPs. Covariates were included in regression analysis for correction (missing genotyping rate, significant MDS dimensions, sex and age). General CROSS-disorder PRSs (5 disorders included) showed a strong association with alcohol dependence (P=7.39e-06, ΔR=2%). After analyzing each disease separately, significant results were found in bipolar disorder (P=0.0064, ΔR=0.7%) and schizophrenia (P=7.06e-05, ΔR=1.5%). A consistent result was achieved using the greater sample of SCZ2 were obtained a consistent (P=4.49e-10, ΔR=4.1%). SCZ2 and general CROSS-disorders results are robust after Bonferroni multiple test correction (7 thresholds and 7 pathologies employed). Discussion The use of a greater sample in SCZ2 improves results in schizophrenia, showing the consistency of these in this pathology. Moreover, the variability percent explained is high considering previous data of shared genetic susceptibility between schizophrenia and other psychiatric disorders. Despite the fact that the sample used in major depression disorder was bigger than the one in bipolar disorder, these results were not significant. It is necessary to consider that a fraction of PGC individuals show comorbidity with SUDs. As evidenced in this study, an increase in the number of PGC discovery individuals sample will improve the ability to detect this shared genetic susceptibility.