Reweighting UK Biobank corrects for pervasive selection bias due to volunteering.

BACKGROUND: Biobanks typically rely on volunteer-based sampling. This results in large samples (power) at the cost of representativeness (bias). The problem of volunteer bias is debated. Here, we (i) show that volunteering biases associations in UK Biobank (UKB) and (ii) estimate inverse probability (IP) weights that correct for volunteer bias in UKB. METHODS: Drawing on UK Census data, we constructed a subsample representative of UKB's target population, which consists of all individuals invited to participate. Based on demographic variables shared between the UK Census and UKB, we estimated IP weights (IPWs) for each UKB participant. We compared 21 weighted and unweighted bivariate associations between these demographic variables to assess volunteer bias. RESULTS: Volunteer bias in all associations, as naively estimated in UKB, was substantial-in some cases so severe that unweighted estimates had the opposite sign of the association in the target population. For example, older individuals in UKB reported being in better health, in contrast to evidence from the UK Census. Using IPWs in weighted regressions reduced 87% of volunteer bias on average. Volunteer-based sampling reduced the effective sample size of UKB substantially, to 32% of its original size. CONCLUSIONS: Estimates from large-scale biobanks may be misleading due to volunteer bias. We recommend IP weighting to correct for such bias. To aid in the construction of the next generation of biobanks, we provide suggestions on how to best ensure representativeness in a volunteer-based design. For UKB, IPWs have been made available.

Risk and Protective Factors of Lifetime Cocaine-Associated Chest Pain.

Introduction: Cocaine users often present with repetitive events of cocaine-associated chest pain (CACP), clinically resembling acute coronary syndromes. The aim of the study is to describe the specific risk factors for CACP. Method: Cocaine users (n = 316) were recruited for a multicenter cross-sectional study. Lifetime CACP history, sociodemographic factors, and lifetime use of cocaine and other substances were assessed. Thirty single nucleotide polymorphisms (SNPs) of NOS3, ROCK2, EDN1, GUCY1A3, and ALDH2 genes, suggested by the literature on coronary spasms, were selected. The associations with CACP history were tested using the chi-square test, Student's t-test and logistic regression. Results: Among the 316 subjects [78.5% men, mean age 37.5 years, (standard-deviation ±8.7)], 190 (60.1%) were daily cocaine users and 103 (32.6%) reported a lifetime CACP history. Among those with a lifetime CACP history, the median was 10 events per individual. In multivariate analysis, lifetime CACP history was associated with daily cocaine use [odds-ratio (OR) 3.24; 95% confidence intervals (1.29-9.33)], rapid route of cocaine use [OR 2.33 (1.20-4.64) vs. intranasal use], and lifetime amphetamine use [daily amphetamine use: OR 2.80 (1.25-6.32) and non-daily amphetamine use: OR 2.14 (1.15-4.04) vs. never used]. Patients with lifetime opioid maintenance treatment (OMT) reported significantly less lifetime CACP history [OR 0.35 (0.16-0.76)]. None of the selected SNPs was associated with CACP history after multiple testing corrections. Conclusions: Clinical variables describing the intensity of stimulant use were positively associated with lifetime CACP history, while OMT was negatively associated with it. Specific harm reduction strategies can target these risk factors.

Genetic correlates of socio-economic status influence the pattern of shared heritability across mental health traits.

Epidemiological studies show high comorbidity between different mental health problems, indicating that individuals with a diagnosis of one disorder are more likely to develop other mental health problems. Genetic studies reveal substantial sharing of genetic factors across mental health traits. However, mental health is also genetically correlated with socio-economic status (SES), and it is therefore important to investigate and disentangle the genetic relationship between mental health and SES. We used summary statistics from large genome-wide association studies (average N ~ 160,000) to estimate the genetic overlap across nine psychiatric disorders and seven substance use traits and explored the genetic influence of three different indicators of SES. Using genomic structural equation modelling, we show significant changes in patterns of genetic correlations after partialling out SES-associated genetic variation. Our approach allows the separation of disease-specific genetic variation and genetic variation shared with SES, thereby improving our understanding of the genetic architecture of mental health.

Genetic heterogeneity in self-reported depressive symptoms identified through genetic analyses of the PHQ-9.

BACKGROUND: Depression is a clinically heterogeneous disorder. Previous large-scale genetic studies of depression have explored genetic risk factors of depression case-control status or aggregated sums of depressive symptoms, ignoring possible clinical or genetic heterogeneity. METHODS: We analyse data from 148 752 subjects of white British ancestry in the UK Biobank who completed nine items of a self-rated measure of current depressive symptoms: the Patient Health Questionnaire (PHQ-9). Genome-Wide Association analyses were conducted for nine symptoms and two composite measures. LD Score Regression was used to calculate SNP-based heritability (h2SNP) and genetic correlations (rg) across symptoms and to investigate genetic correlations with 25 external phenotypes. Genomic structural equation modelling was used to test the genetic factor structure across the nine symptoms. RESULTS: We identified nine genome-wide significant genomic loci (8 novel), with no overlap in loci across symptoms. h2SNP ranged from 6% (concentration problems) to 9% (appetite changes). Genetic correlations ranged from 0.54 to 0.96 (all p < 1.39 × 10-3) with 30 of 36 correlations being significantly smaller than one. A two-factor model provided the best fit to the genetic covariance matrix, with factors representing 'psychological' and 'somatic' symptoms. The genetic correlations with external phenotypes showed large variation across the nine symptoms. CONCLUSIONS: Patterns of SNP associations and genetic correlations differ across the nine symptoms, suggesting that current depressive symptoms are genetically heterogeneous. Our study highlights the value of symptom-level analyses in understanding the genetic architecture of a psychiatric trait. Future studies should investigate whether genetic heterogeneity is recapitulated in clinical symptoms of major depression.

Potential influence of socioeconomic status on genetic correlations between alcohol consumption measures and mental health.

BACKGROUND: Frequency and quantity of alcohol consumption are metrics commonly used to measure alcohol consumption behaviors. Epidemiological studies indicate that these alcohol consumption measures are differentially associated with (mental) health outcomes and socioeconomic status (SES). The current study aims to elucidate to what extent genetic risk factors are shared between frequency and quantity of alcohol consumption, and how these alcohol consumption measures are genetically associated with four broad phenotypic categories: (i) SES; (ii) substance use disorders; (iii) other psychiatric disorders; and (iv) psychological/personality traits. METHODS: Genome-Wide Association analyses were conducted to test genetic associations with alcohol consumption frequency (N = 438 308) and alcohol consumption quantity (N = 307 098 regular alcohol drinkers) within UK Biobank. For the other phenotypes, we used genome-wide association studies summary statistics. Genetic correlations (rg) between the alcohol measures and other phenotypes were estimated using LD score regression. RESULTS: We found a substantial genetic correlation between the frequency and quantity of alcohol consumption (rg = 0.52). Nevertheless, both measures consistently showed opposite genetic correlations with SES traits, and many substance use, psychiatric, and psychological/personality traits. High alcohol consumption frequency was genetically associated with high SES and low risk of substance use disorders and other psychiatric disorders, whereas the opposite applies for high alcohol consumption quantity. CONCLUSIONS: Although the frequency and quantity of alcohol consumption show substantial genetic overlap, they consistently show opposite patterns of genetic associations with SES-related phenotypes. Future studies should carefully consider the potential influence of SES on the shared genetic etiology between alcohol and adverse (mental) health outcomes.

Post-GWAS analysis of six substance use traits improves the identification and functional interpretation of genetic risk loci.

BACKGROUND: Little is known about the functional mechanisms through which genetic loci associated with substance use traits ascertain their effect. This study aims to identify and functionally annotate loci associated with substance use traits based on their role in genetic regulation of gene expression. METHODS: We evaluated expression Quantitative Trait Loci (eQTLs) from 13 brain regions and whole blood of the Genotype-Tissue Expression (GTEx) database, and from whole blood of the Depression Genes and Networks (DGN) database. The role of single eQTLs was examined for six substance use traits: alcohol consumption (N = 537,349), cigarettes per day (CPD; N = 263,954), former vs. current smoker (N = 312,821), age of smoking initiation (N = 262,990), ever smoker (N = 632,802), and cocaine dependence (N = 4,769). Subsequently, we conducted a gene level analysis of gene expression on these substance use traits using S-PrediXcan. RESULTS: Using an FDR-adjusted p-value <0.05 we found 2,976 novel candidate genetic loci for substance use traits, and identified genes and tissues through which these loci potentially exert their effects. Using S-PrediXcan, we identified significantly associated genes for all substance traits. DISCUSSION: Annotating genes based on transcriptomic regulation improves the identification and functional characterization of candidate loci and genes for substance use traits.

QT length during methadone maintenance treatment: gene × dose interaction.

Methadone is known to be a risk factor for sudden death by enlarging ECG QT corrected (QTc) interval. For other medical conditions, QTc lengthening has been described as the result of interactions between pharmacological treatments and genetic factors. Former heroin-dependent subjects under methadone maintenance treatment in remission for at last 3 months were recruited. We studied the association between QTc length (Bazett formula) and 126 SNPs located on five genes (KCNE1, KCNQ1, KCNH2, NOS1AP and SCN5A) previously associated with drug-induced QT prolongation. Both SNP-based and gene-based approaches were used, and we tested also the interaction of the top SNP with methadone dosage to predict the QTc length. In our sample of 154 patients, current methadone daily dose was associated with QTc length (rPearson  = 0.26; P = 10-3 ). Only one SNP, rs11911509 on KCNE1, remained significantly associated with QT length after correction for multiple testing (P = 3.84 × 10-4 ; pcorrected  = 0.049). Using a gene-based approach, KCNE1 was also significantly associated with QTc length (pempirical  = 0.02). We found a significant interaction between methadone dosage and rs11911509 minor allele count (allele A vs. C; P = 0.01). Stratified analysis revealed that the correlation between QTc length and methadone dosage was restricted only to AA carriers of this top SNP. Patients' genetic background should be taken into account in the case of clinically relevant QT enlargement during methadone maintenance treatment.

Exploring the role of low-frequency and rare exonic variants in alcohol and tobacco use.

BACKGROUND: Alcohol and tobacco use are heritable phenotypes. However, only a small number of common genetic variants have been identified, and common variants account for a modest proportion of the heritability. Therefore, this study aims to investigate the role of low-frequency and rare variants in alcohol and tobacco use. METHODS: We meta-analyzed ExomeChip association results from eight discovery cohorts and included 12,466 subjects and 7432 smokers in the analysis of alcohol consumption and tobacco use, respectively. The ExomeChip interrogates low-frequency and rare exonic variants, and in addition a small pool of common variants. We investigated top variants in an independent sample in which ICD-9 diagnoses of "alcoholism" (N = 25,508) and "tobacco use disorder" (N = 27,068) had been assessed. In addition to the single variant analysis, we performed gene-based, polygenic risk score (PRS), and pathway analyses. RESULTS: The meta-analysis did not yield exome-wide significant results. When we jointly analyzed our top results with the independent sample, no low-frequency or rare variants reached significance for alcohol consumption or tobacco use. However, two common variants that were present on the ExomeChip, rs16969968 (p = 2.39 × 10-7) and rs8034191 (p = 6.31 × 10-7) located in CHRNA5 and AGPHD1 at 15q25.1, showed evidence for association with tobacco use. DISCUSSION: Low-frequency and rare exonic variants with large effects do not play a major role in alcohol and tobacco use, nor does the aggregate effect of ExomeChip variants. However, our results confirmed the role of the CHRNA5-CHRNA3-CHRNB4 cluster of nicotinic acetylcholine receptor subunit genes in tobacco use.

A tutorial on conducting genome-wide association studies: Quality control and statistical analysis.

OBJECTIVES: Genome-wide association studies (GWAS) have become increasingly popular to identify associations between single nucleotide polymorphisms (SNPs) and phenotypic traits. The GWAS method is commonly applied within the social sciences. However, statistical analyses will need to be carefully conducted and the use of dedicated genetics software will be required. This tutorial aims to provide a guideline for conducting genetic analyses. METHODS: We discuss and explain key concepts and illustrate how to conduct GWAS using example scripts provided through GitHub (https://github.com/MareesAT/GWA_tutorial/). In addition to the illustration of standard GWAS, we will also show how to apply polygenic risk score (PRS) analysis. PRS does not aim to identify individual SNPs but aggregates information from SNPs across the genome in order to provide individual-level scores of genetic risk. RESULTS: The simulated data and scripts that will be illustrated in the current tutorial provide hands-on practice with genetic analyses. The scripts are based on PLINK, PRSice, and R, which are commonly used, freely available software tools that are accessible for novice users. CONCLUSIONS: By providing theoretical background and hands-on experience, we aim to make GWAS more accessible to researchers without formal training in the field.