In initial efforts to understand who has healthy outcomes despite high genetic risk, we found that higher father–child relationship quality in adolescence promoted delays in alcohol initiation.141 With large, complex, multi‐generational pedigrees enriched for AUD, COGA provides an opportunity to look more closely not only at risk factors, but also factors that protect against the development of AUD. COGA ascertained probands in treatment for alcohol dependence, and a smaller number of comparison individuals from the same communities, and then recruited their families. Initial recruitment prioritized families with at least three first degree relatives meeting criteria for alcohol dependence (i.e., densely affected) although many families include more than three individuals with AUD, hence the higher than population prevalence of alcohol dependence and AUD (Table 1).
RECRUITMENT: A FOCUS ON FAMILIES
Individual reviews in this issue provide detailed illustrations of the ways in which COGA data have contributed towards advancing our understanding of the etiology, course and consequences of AUD, and pathways from onset to remission and relapse. COGA’s intergenerational design has, in addition to identifying genetic risk factors, contributed to our understanding of the role of social genetic mechanisms50, 52, 64, 65, 66 in the interplay between genetic liability and the socio‐environmental milieu (e.g., References 40, 48, 67, 68). Diversity in the data have driven gene discoveries within our dataset (e.g., Reference 44) and in collaboration with others (e.g., References 5, 55, 69). Our ability to develop iPSCs from individuals with different genetic loading is producing insights into properties of cells derived from persons with archival electrophysiological and behavioral phenotyping, and how the cells differentially respond to ethanol exposure. A notable contribution of COGA’s family design has been to disentangle antecedents of, and predisposition to AUD from its sequelae. By characterizing brain and behavior in offspring from families enriched for AUD liability—both genetic and environmental—prior to the onset of maladaptive drinking behaviors, COGA data have shown the importance of precursors of AUD in a neurobehavioral framework (e.g., References 23, 34, 70, 71, 72).
Many approaches to creating polygenic scores, from linkage disequilibrium (LD) clumping or pruning and thresholding approaches, to modern Bayesian methods, and even functional polygenic signatures, are available. While a high‐risk sample such as COGA can clearly contribute to characterizing genetic and environmental liability for AUD, it also presents a unique opportunity to study resilience and protective factors. The possibility of identifying such genetic “resilience” variants that may help protect against the development of an alcohol use disorder could provide insight into novel treatments or prevention efforts. For example, Hess et al.140 created a “polygenic resilience score” for schizophrenia by matching unaffected individuals at high genetic risk with risk‐matched cases, and then identifying genetic variants that contribute to resilience to schizophrenia and do not overlap with risk loci. Individuals with high genetic loading for AUD risk may also be resilient for entirely environmental reasons, such as strong familial and community support, or choosing not to drink after witnessing the effects of addiction on family members with alcohol use problems.
The genes with the clearest contribution to the risk for alcoholism andalcohol consumption are alcohol dehydrogenase 1B (ADH1B) andaldehyde dehydrogenase 2 (ALDH2; mitochondrial aldehydedehydrogenase), two genes central to the metabolism of alcohol (Figure 1)20. Alcohol is metabolized primarily in the liver, although thereis some metabolism in the upper GI tract and stomach. The first step in ethanolmetabolism is oxidation to acetaldehyde, catalyzed primarily by ADHs; there are 7closely related ADHs clustered on chromosome 4 (reviewed in20). The second step is metabolism of theacetaldehyde to acetate by ALDHs; again, there are many aldehyde dehydrogenases,among which ALDH2 has the largest impact on alcohol consumption20. Group meetings are available in most communities at low or no cost, and at convenient times and locations—including an increasing presence online. Combined with medications and behavioral treatment provided by health care professionals, mutual-support groups can offer a valuable added layer of support.
Taken together, these waves of longitudinal follow‐up provide a perspective of AUD risk and resilience across the lifespan. Parallel to the emphasis on increasing sample sizes to drive gene discovery is the growing recognition of the value of a sample like COGA with its family‐based design, deep phenotyping, longitudinal framework, multi‐modal data, wide age range, and ancestral diversity (see, Figure 2 for summary of key contributions enabled by COGA). Simply put, the family‐based COGA data are well‐suited to answer scientific questions that are not possible even in very large samples of unrelated individuals.
- Drug use and addiction represent a public health crisis, characterized by high social, emotional, and financial costs to families, communities, and society.
- In the study of complex disorders, it has become apparent that quitelarge sample sizes are critical if robust association results are to beidentified which replicate across studies.
- Most robust associations that have been reported in common disease haveemployed tens of thousands of samples and are now beginning to combine severalstudies of these magnitude into even larger meta analyses.
GWAS of AUD and related traits
The accompanying review (3. Brain Function) covers the available brain function data and genetics of alcohol use disorder national institute on alcohol abuse and alcoholism niaaa resulting findings in detail. COGA’s brain function data (see, 3. Brain Function) have also been paired with the project’s functional genomics pipeline (see, 5. Functional Genomics) to provide mechanistic insights. In an example of this, several variants within KCNJ6 (encoding the GIRK2 G‐protein coupled inwardly rectifying potassium channel) were identified as genome‐wide significant in our family‐based GWAS of a frontal theta EEG phenotype75 (an endophenotype for AUD14). COGA’s asset is its family‐based longitudinal design that supports an intensive clinical, behavioral, genetic, genomic and brain function data collection. As the project enters its late third decade of scientific exploration, we approach our contributions to the study of AUD with optimism.
EARLY RESULTS: CANDIDATE GENE STUDIES
- Compared to other genetic predictors, the genomic pattern identified here was also a more sensitive predictor of having two or more substance use disorders at once.
- COGA is a family based, diverse (~25% self‐identified African American, ~52% female) sample, including data on 17,878 individuals, ages 7–97 years, in 2246 families of which a proportion are densely affected for AUD.
- Individual reviews in this issue provide detailed illustrations of the ways in which COGA data have contributed towards advancing our understanding of the etiology, course and consequences of AUD, and pathways from onset to remission and relapse.
Meta-analyses, whichcombine results across a number of studies in order to attain the criticalsample sizes needed, are being developed. There were three major goals in the establishment of the NIAAA/COGA Sharing Repository in 1996, over 36 years ago. First, there was the perceived need to have quality‐assured biosamples from each COGA participant and to minimize differences between individual COGA samples due to potential collection, extraction or storage variables. Impetus to establish an ongoing biorepository also came from the belief that developing technologies for genome analysis would improve in the future and likely become more cost efficient (e.g., single nucleotide polymorphism and DNA sequencing) and that ample amounts of biosample of consistent quality would be required for such analyses. Third, there was the desire to collaborate with other groups by sharing COGA samples, thereby introducing more uniformity into research on the genetics of alcohol use disorder.
1. Resilience and protective factors
To provide a community‐facing forum for sharing our own research findings and also provide summaries of the state of scientific knowledge in the field of alcohol research, COGA has developed a series of resources for the public to understand how genetic and environmental factors contribute to the development of alcohol use problems. These were developed in collaboration with digital communication specialists and include short videos, text descriptions, interactive graphical elements, and key take‐aways, and can be found at cogastudy.org. In addition to generating functional genomic data, COGA has collaborated with other research groups and used curated gene expression, chromatin architecture and methylation data, from both humans and non‐human animals, to tease apart causal variants from the increasing number of genome‐wide significant loci emerging from large‐scale GWAS meta‐analyses of AUD and related traits.
PECRis located within broad linkage peaks for several alcohol-related traits,including alcoholism66,comorbid alcoholism and depression67, level of response to alcohol68, and amplitude of the P3(00)response69, 70. COGA was among the first studies to pursue GWAS genotyping, first for diagnostic and then, increasingly, for quantitative traits. This shift reflected, in part, the growing recognition that genes of large effect were likely to be the exception rather than the norm for complex traits like AUD, and that GWAS approaches would be necessary to elucidate the many genes and variants of individually small effect sizes contributing to disorders. The inclusion of data from different ancestral groups in this study cannot and should not be used to assign or categorize variable genetic risk for substance use disorder to specific populations. As genetic information is used to better understand human health and health inequities, expansive and inclusive data collection is essential.
The initial family‐based GWAS of COGA,74, 75 conducted in a second subset of the data, was analyzed using Genome‐Wide Association analyses with Family data (GWAF76). Subsequent GWAS combining the case–control and family‐based resources77 allowed for analyses using the Generalized Disequilibrium Test78 and linear mixed effects modeling with kinship matrices (lmekin) from the coxme package79 in R.80 As genotyping of the COGA sample proceeded, integrated analyses of imputed GWAS data across COGA’s dense and multigenerational families posed computational challenges. For instance, commonly used software such as PLINK62 or analytic modules that accounted for fewer degrees of relatedness (e.g., siblings or trios that could be analyzed using generalized estimating equations81, 82) did not adequately model these complex family structures, and Genome‐wide Complex Traits Analysis83 proved far too computationally burdensome for the high dimensional relatedness matrices arising in COGA. It is likely that, as for most complex diseases, alcohol dependence and AUDsare due to variations in hundreds of genes, interacting with different socialenvironments. An additional challenge in the search for genetic variants that affectthe risk for AUDs is that there is extensive clinical heterogeneity among thosemeeting criteria.
One study used a staged meta-analysis to explore comorbid alcoholand nicotine dependence and detected genome-wide evidence of association withSNPs spanning a region on chromosome 5 that includes both IPO11(importin 11) and HTR1A (5-hydroxytryptamine (serotonin)receptor 1A, G protein-coupled)78. Throughout this manuscript, we use the terminology of “alcohol use disorder” to discuss individuals meeting diagnostic criteria for case status, but we note that this has been variously defined in the COGA sample depending on the diagnostic system at the time of sample recruitment. Some alleles that reduce heavy drinking can,nevertheless, increase risk for disease in the subset of individuals who drinkheavily despite having them. Graphical summary of three of the most important contributions to understanding the etiology of alcohol use disorders that have come, in part, from genetic analyses in COGA.
Some of these genes have been identified, including twogenes of alcohol metabolism, ADH1B and ALDH2,that have the strongest known affects on risk for alcoholism. Studies arerevealing other genes in which variants impact risk for alcoholism or relatedtraits, including GABRA2, CHRM2,KCNJ6, and AUTS2. As larger samples areassembled and more variants analyzed, a much fuller picture of the many genesand pathways that impact risk will be discovered.
Clinical Trials on Alcohol and Alcohol Use Disorder
Binge drinkingis generally defined as a man consuming 5 standard drinks within 2 hours; women are typically smaller and have a lower percentage of body water, so 4 standarddrinks can reach similar alcohol levels. A standard drink is defined in the US as 12ounces of beer, 5 ounces of wine or 1.5 ounces of spirits, all of which approximate14 g of pure ethanol). The strong effects of binge drinking suggest that merelycalculating an average number of drinks per week is likely to obscure many effectsof alcohol, since it treats 2 standard drinks per day (14 per week) the same as 7drinks on each of two days per week. The GI tract is exposed to very high levels of alcohol as it passes throughthe mouth, esophagus, stomach and intestinal tract, and most ethanol passes throughthe liver before entering the circulation.
Linkage studies are relatively robust to populationdifferences in allele frequencies (because they test within-family inheritance), andcan find a signal even if different variants in the same gene or region areresponsible for the risk in different families. The drawback to this approach isthat linkage studies find broad regions of the genome, often containing manyhundreds of genes. In many cases, the initial linkage studies were followed by moredetailed genetic analyses employing single nucleotide polymorphisms (SNPs) that weregenotyped at high density across the linked regions. Some of the genes identifiedthrough this approach have been replicated across a number of studies and appear tobe robust genetic findings.