Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 May 1.
Published in final edited form as: Drug Alcohol Depend. 2022 Mar 28;234:109430. doi: 10.1016/j.drugalcdep.2022.109430

Alcohol Use and Alcohol Use Disorder Differ in their Genetic Relationships with PTSD: A Genomic Structural Equation Modelling Approach

Kaitlin E Bountress 1, Leslie A Brick 2, Christina Sheerin 1, Andrew Grotzinger 3, Daniel Bustamante 1,4, Sage E Hawn 5,6, Nathan Gillespie 1, Robert M Kirkpatrick 1, Henry Kranzler 7,8, Rajendra Morey 9,10,11, Howard J Edenberg 12,13, Adam X Maihofer 14,15, Seth Disner 16,17, Allison Ashley-Koch 18, Roseann Peterson 1, Adriana Lori 19, Dan J Stein 20, Nathan Kimbrel 9,10, Caroline Nievergelt 14,15, Ole A Andreassen 21,22, Jurjen Luykx 23,24,25, Arash Javanbakht 26, Nagy A Youssef 27; The Psychiatric Genomics Consortium Posttraumatic Stress Disorder Working Group, Ananda B Amstadter 1
PMCID: PMC9018560  NIHMSID: NIHMS1796093  PMID: 35367939

Abstract

Purpose:

Posttraumatic Stress Disorder (PTSD) is associated with increased alcohol use and alcohol use disorder (AUD), which are all moderately heritable. Studies suggest the genetic association between PTSD and alcohol use differs from that of PTSD and AUD, but further analysis is needed.

Basic Procedures:

We used genomic Structural Equation Modeling (genomicSEM) to analyze summary statistics from large-scale genome-wide association studies (GWAS) of European Ancestry participants to investigate the genetic relationships between PTSD (both diagnosis and re-experiencing symptom severity) and a range of alcohol use and AUD phenotypes.

Main Findings:

When we differentiated genetic factors for alcohol use and AUD we observed improved model fit relative to models with all alcohol-related indicators loading onto a single factor. The genetic correlations (rG) of PTSD were quite discrepant for the alcohol use and AUD factors. This was true when modeled as a three-correlated-factor model (PTSD-AUD rG: .36, p<.001; PTSD-alcohol use rG: −.17, p<.001) and as a Bifactor model, in which the common and unique portions of alcohol phenotypes were pulled out into an AUD-specific factor (rG with PTSD: .40, p<.001), AU-specific factor (rG with PTSD: −.57, p<.001), and a common alcohol factor (rG with PTSD: .16, NS).

Principal Conclusions:

These results indicate the genetic architecture of alcohol use and AUD are differentially associated with PTSD. When the portions of variance unique to alcohol use and AUD are extracted, their genetic associations with PTSD vary substantially, suggesting different genetic architectures of alcohol phenotypes in people with PTSD.

Keywords: genomic Structural Equation Modeling (genomicSEM), Posttraumatic Stress Disorder (PTSD), Alcohol Use, Alcohol Use Disorder, Genetic Correlation, Genetic Architecture

1. Introduction

Traumatic experiences are common, with 50–70% of individuals experiencing at least one trauma during their lives (Benjet et al., 2016). Posttraumatic stress disorder (PTSD), the signature trauma-related disorder (Breslau, 2009), tends to co-occur with heavy alcohol use (AU, i.e., number of drinks consumed) (Vlahov et al., 2002) and alcohol use disorder (AUD) (Jakupcak et al., 2010). In twin studies, PTSD, alcohol use, and AUD are moderately heritable, with about 40–60% of the variance explained by genetic influences (Heath, Jardine, & Martin, 1989; Kaprio et al., 1987; Knopik et al., 2004; Stein, Jang, Taylor, Vernon, & Livesley, 2002; Verhulst, Neale, & Kendler, 2015). Genome-wide association studies (GWAS) have identified individual loci that are associated with each of these phenotypes.

Recent analyses using large-scale GWAS have examined genetic variation associated with PTSD (Nievergelt et al., 2019) and with re-experiencing symptoms of PTSD (Gelernter et al., 2019a). Re-experiencing symptoms are unique to PTSD and therefore help to differentiate PTSD from other disorders. The single nucleotide polymorphism (SNP)-based heritability of PTSD, derived from GWAS summary statistics, shows modest to moderate heritability (~5–20%), with estimates larger in women (10%) than men (1%) (Nievergelt et al., 2019).

Large scale GWAS have also identified loci that are associated with AUD (Henry R Kranzler et al., 2019), sometimes referred to as problematic alcohol use (Zhou et al., 2020) or alcohol dependence (Lai et al., 2019; Sun et al., 2019; Walters et al., 2018). The designations of AUD and alcohol dependence have been used to indicate a diagnostic threshold, whereas problematic alcohol use represents drinking associated with problematic consequences. Other researchers have conducted GWAS on a broader behavioral phenotype of alcohol use (Clarke et al., 2017; Johnson et al., 2019), specifically frequency and/or quantity of alcohol consumption. The SNP-based heritability of alcohol dependence and AUD ranged from 5.6% to 11.0% (Kranzler et al., 2019; Walters et al., 2018) and for PAU it was 6.8% (Zhou et al., 2020).

Twin studies have demonstrated 30% latent genetic overlap between PTSD and AUD (McLeod et al., 2001; Xian et al., 2000), but this question has not been applied to the PTSD-alcohol use association using this design. Molecular genetic analytic methods, such as linkage disequilibrium score regression (LDSR) (Bulik-Sullivan et al., 2015) allow for the estimation of pairwise, zero-order genetic correlations. Work by our group using LDSR has found significant genetic correlations between PTSD and AUD (rG range = 0.28–0.35; Bountress, et al., 2021; Sheerin et al., 2020). However, this association may be driven by women for whom the genetic correlation was moderate and significant, while non-significant for men (Sheerin et al., 2020). We also found significant, positive genetic associations between PTSD and AUD, but negative (significant and non-significant) associations between PTSD and alcohol use phenotypes (e.g., drinks per week, Alcohol Use Disorders Identification Test-Consumption [AUDIT-C] score) (Bountress et al., 2022). This pattern of findings is consistent with work on major depression, which showed positive genetic correlations with alcohol dependence and alcohol quantity, but negative genetic correlations with the frequency of alcohol consumption (Polimanti et al., 2019). Therefore, genetic associations between mental health disorders and more normative drinking phenotypes may be negative. Additionally, others have found that the strength of genetic correlations between measures of alcohol use and AUD vary considerably, with some moderate (AUDIT-C and alcohol dependence: rG=.33; Sanchez-Roige et al., 2019) and others large (AUDIT-problematic consequences of drinking [AUDIT-P] and AU: rG=.76; Sanchez-Roige et al., 2019) (AUDIT-C and AUDIT-P: rG=.80; Mallard et al., 2021). Thus, the genetic risk for normative versus problematic use is correlated but distinct. Taken together, it appears that the genetic architecture of AUD and alcohol use are different, and their genetic associations with other phenotypes (e.g., PTSD), are quite disparate.

Novel genetic approaches such as genomic Structural Equation Modeling (genomicSEM; Grotzinger et al., 2019) can enhance our understanding of the genetic pleiotropy among these phenotypes. Specifically, genomicSEM uses LDSR to fit multivariate models of genetic associations. It allows users to investigate alternative models from multiple phenotypes to identify the latent genetic factor structure, making it possible to index genetic overlap among phenotypes, as well as variance that is unique to each trait (e.g., alcohol- or PTSD-specific factors). Because genomicSEM can be estimated using summary statistics with varying degrees of sample overlap, it allows for the inclusion of relatively rare or difficult to ascertain clinical samples in the same model. A recent genomic SEM model found that PTSD and AUD load on a factor with autism and ADHD, but also cross-load on other factors (Grotzinger et al., 2020). Given what we know about the genetic correlations among alcohol phenotypes and PTSD, we aimed to test the fit of a number of models: 1) a unidimensional model that allows all PTSD and alcohol indicators to load on one factor, 2) a two-factor model that extracts a PTSD and a common alcohol factor, or 3) a three-factor model that extracts PTSD and two alcohol (alcohol use and AUD) factors.

To estimate these models, we used European Ancestry GWAS summary statistics for PTSD diagnosis, re-experiencing symptoms of PTSD, and a range of alcohol phenotypes, including drinks per week, the consumption score from the AUDIT-C, total score and problems score from the AUDIT, maximum alcohol intake, AUD, and alcohol dependence. We add to previous work by using a novel genetic approach (genomicSEM) to estimate the partial genetic associations between PTSD and several alcohol phenotypes. Our first hypothesis was that a model separating out PTSD, AUD, and alcohol use would fit better than those modeling them together, and that the genetic associations between PTSD and AUD phenotypes would be significant and positive, and between PTSD and alcohol use phenotypes would be non-significant or significant and negative. Our second hypothesis was that a model pulling out what is common to all the alcohol items (loading onto an “alcohol” factor), while leaving what is unique to AUD and alcohol use would show even better fit, with genetic correlations between PTSD and the common alcohol factor being near zero, PTSD and unique AUD being even more strongly positive and significant and PTSD and alcohol use being/becoming negative, potentially significant.

2. Methods

2.1. Summary of Cohorts

We obtained summary statistics for PTSD and alcohol use phenotypes using existing large-scale datasets from single studies and consortia, described below and summarized in Table 1. We only conducted analyses on those of European Ancestry due to the scarcity of large-scale summary statistics in other ancestral populations.

Table 1.

Descriptive Information about Included Phenotypes and Accompanying Cohorts.

Dataset Phenotype Accompanying GWAS N N effective Primary Sample make-up Sex Distribution Age Distribution
PGC-SUD Alcohol Dependence (AD) Case/Control Walters et al., 2019 45,568 34,780 Civilian 34–70% Male 64% of studies only included adults; others included those as young as 8+
MVP Alcohol Use Disorder (AUD) Case/Control Kranzler et al., 2019 267,391 152,332 Military ~93% Male 22–90 years old
MVP Max Alcohol Consumption Gelernter et al., 2019b 126,936 126,936 Military ~93% Male ~63 years old
UKB AUDIT Problems Sanchez-Roige et al., 2019a 121,604 121,604 Civilian 43.8% Male ~56 years old
23andMe AUDIT Total Sanchez-Roige et al., 2019b 20,328 20,328 Civilian 44.7% Male ~54 years old
UKB AUDIT Consumption Sanchez-Roige et al., 2019a 121,604 121,604 Civilian 43.8% Male ~56 years old
MVP AUDIT Consumption Kranzler et al., 2019 206,254 206,254 Military ~93% Male 22–90 years old
GSCAN and UKB Drinks per Week Liu et al., 2019 941,280 941,280 Civilian 47.8% Male Either ~40 or ~81 years old
MVP Re-Experiencing Symptoms Gelernter et al., 2019a 146,660 146,660 Military ~93% Male ~66 years old
PGC-PTSD PTSD Case/Control Nievergelt et al., 2019 48,471 37,723 Civilian ~50% Male ~52 years old
Open in a new tab

Notes: PGC-SUD: Psychiatrics Genomics Consortium Substance Use Disorder workgroup; MVP: Million Veteran Program, UKB: United Kingdom Biobank; GSCAN: GWAS & Sequencing Consortium of Alcohol and Nicotine use; PGC-PTSD: Psychiatrics Genomics Consortium Posttraumatic Stress Disorder workgroup.

2.2. PTSD Cohorts and Phenotypes

As shown in Table 1, PTSD case/control status came from the PGC-PTSD Freeze 1.5 (N = 48,471; Nievergelt et al., 2019). PTSD case status reflects primarily lifetime PTSD diagnosis, although it includes current PTSD diagnosis when lifetime diagnosis was unavailable.

The PTSD re-experiencing symptom cluster severity score (continuous measure) came from the PTSD Checklist for DSM-IV (PCL IV; (Wilkins, Lang, & Norman, 2011) in the MVP (Gelernter et al., 2019a), as it is the symptom cluster that is most distinctive for PTSD (N = 146,660).

2.3. Alcohol-Related Cohorts and Phenotypes

AUD case/control status came from the MVP dataset and was defined according to ICD-9 or ICD-10 codes for dependence or abuse diagnoses obtained from the Veteran’s Affairs electronic health records (EHR). Participants with at least one inpatient or two outpatient alcohol-related ICD-9/10 codes (from 2000–2018) were considered AUD cases (Kranzler et al., 2019). AUD case/control status was available for 267,391 participants in MVP. Alcohol dependence case/control data came from the PGC-SUD meta-analysis (Walters et al., 2018), which included over 20 datasets. Cases were defined as meeting criteria for a DSM-IV (and DSM-III-R for one study) diagnosis of alcohol dependence and all controls were alcohol exposed (N = 46,568).

Drinks per week (DPW), defined as the average number of drinks a participant reported drinking each week, aggregated across all types of alcohol, was examined in a combined approach with GSCAN consortium and UK Biobank (UKB) (Liu et al., 2019) (N = 941,280); see Table 1. In studies that reported binned response ranges (e.g., 1–4 drinks), the midpoint of the range was used (Liu et al., 2019). The AUDIT (Saunders, Aasland, Babor, De la Fuente, & Grant, 1993) was available in multiple forms and studies. First, the AUDIT total score (AUDIT-T), was available in the 23andMe dataset (Sanchez-Roige et al., 2018) for 20,328 participants. Second, data from the AUDIT-C subscale, which consists of three items that measure past-year typical quantity and frequency of drinking and frequency of heavy/binge drinking (Bush, Kivlahan, McDonell, Fihn, & Bradley, 1998), were available in two datasets: EHR data from the annual AUDIT-C assessment in MVP collected on individuals between 2007–2017 (N = 206,254; H. R. Kranzler et al., 2019) and as part of the full 10-item AUDIT in an online follow-up of the UKB (N = 121,604; Sanchez-Roige et al., 2019). Third, the AUDIT-P subscale, which consists of 7 items that focus on the problematic consequences of drinking, was used from the UKB (N = 121,604). Finally, in MVP, a quantitative measure of maximum habitual alcohol use in a typical month (MaxAlc; Gelernter, et al., 2019b) was used to reflect typical/habitual maximum use as opposed to maximum on a single occasion (N = 126, 936).

2.4. Genotyping, Quality Control, and Imputation

The existing summary statistics used in the present analyses have gone through quality control pipelines applied by the specific consortia (e.g., PGC quality control pipeline including filtering to remove SNPs with imputation information value < 0.90 and minor allele frequency/MAF < 0;01; Sullivan, 2010). The analytic pipeline for the current analyses incorporates additional filtering (keeping approximately 1,200,000 SNPs for each phenotype with the exception of the MVP analyses which keep approximately 625,000; see Grotzinger et al., 2019) including removing variants that are either not SNPs or are strand ambiguous, and removing SNPs based on a minimum N.

2.5. Genomic Structural Equation Modeling

We conducted analyses using the GenomicSEM package in R (version 0.0.3; https://github.com/GenomicSEM/GenomicSEM/wiki). GenomicSEM uses a two-stage SEM approach (Grotzinger et al., 2019). In the first stage, the genetic covariance matrix and sampling covariance matrix are estimated for each dataset (see Supplementary Table 1). In the second stage a SEM is specified and parameters are estimated by attempting to minimize the discrepancy between the model-implied genetic covariance matrix and the empirical covariance matrix. The fit of the model can then be evaluated using standard metrics, including the standardized root mean square residual (SRMR), model χ2, Akaike Information Criterion (AIC), and the Comparative Fit Index (CFI) (Browne, 1984; Savalei & Bentler, 2009).

Precomputed linkage disequilibrium (LD) scores were obtained from the 1000 Genomes Project Europeans subsample (https://data.broadinstitute.org/alkesgroup/LDSCORE/eur_w_ld_chr.tar.bz2). For case/control samples, liability scale estimates assumed a population prevalence of 10% for PTSD (Nievergelt et al., 2019), 15.9% for AD, and 16% for AUD as in prior work (Duncan et al., 2017; Walters et al., 2018).

2.6. Data Analytic Plan

We estimated four models and compared the fit of each. Specifically, we first asked whether a common factor model in which all alcohol-related and PTSD indicators loading on the same factor (Model A; see Figure 1A) would fit the data well, and whether either a two-factor model with all alcohol items loading on one factor and PTSD items loading on a second factor (Model B; Figure 1B) or a correlated three-factor model allowing for separate PTSD, AU, and AUD-related factors (Model C; Figure 1C) would provide a better fit. We hypothesized that a model with unique PTSD, AUD, and alcohol use factors would fit the data best, and the inter-factor genetic correlation between PTSD and AUD would be stronger and more positive than that for PTSD and AU, which would be non-significant or significant and negative.

Figure 1.

Figure 1.

Open in a new tab

Depiction of Tested Models.

Notes: Standardized loadings and correlations are depicted; *p<.05, **p<.01, ***p<.001.

Line textures and thicknesses vary as an indictor of which items load on which factor.

Secondly, we asked whether a more complicated, bifactor model that allows for factors common and specific to PTSD and alcohol phenotypes (Model D; Figure 1D) would fit the data well, and whether the inter-factor correlations between PTSD and AUD and PTSD and alcohol use would differ from one another when estimating a factor common to all alcohol phenotypes. In bifactor models, all items are allowed to load on one common factor and their specific group factors. The group factors are allowed to correlate with one another, but their correlations with the general factor are usually set to zero. Typically, within bifactor models, each item loads on the common factor and one specific factor. However, here the 23andMe AUDIT Total score item was allowed to load on both alcohol use and AUD, as it is comprised of items related to both consumption (AUDIT-C) and problems (AUDIT-P). We hypothesized that, within this framework, the genetic correlation between PTSD and the common alcohol factor would be non-significant, but that the genetic correlation with the factor capturing variance unique to AUD would be significant and positive, with the factor capturing variance unique to alcohol use significant and negative.

To determine which model best fit the data, we examined the substantive interpretability of each model and its loadings, including the genetic associations between PTSD and factors common and specific to alcohol use and AUD. We also examined goodness-of-fit indices with the standard cut-offs for good fit, including a CFI: ≥ .9 and SRMR ≤ .08 and lower AIC values indicating better fit and parsimony (Hu, 1999; Kenny, 2015). As the PTSD factor consisted of only two items, we constrained the loadings of these items on the factor to be equal so that this portion of the model was locally identified. We used the zero-order genetic correlations between PTSD and alcohol phenotypes generated from this same author group (Bountress et al., 2022) to inform which alcohol items would load onto the AU-related factor (i.e., drinks per week, AUDIT-C, and AUDIT-T) or the AUD-related factor (i.e., AUDIT-T, Max Alc, AUDIT-P, AUD, and AD).

3. Results

Regarding our first question, a common factor model (Model A; Figure 1A) did not fit the data well (χ2=637.67, AIC=677.67, CFI=.84, SRMR=.23; Table 2). The loadings indicated that this factor driven by the alcohol-related factors, while the loadings of the PTSD items were not statistically different from zero (Supplemental Table 2). The two-factor model (Model B; Figure 1B) fit somewhat better, but not well (χ2=506.97, AIC=546.97, CFI=.88, SRMR=.19). Once again, the loadings of the alcohol items onto the alcohol factor were all significant, but the PTSD item loadings on the PTSD factor were all non-significant (Supplemental Table 3). The genetic correlation between PTSD and the common alcohol factor was not significantly different from zero (rG:.03, NS). The three-factor model (Model C; Figure 1C; Supplemental Table 4) fit adequately (χ2=306.97, AIC=352.97, CFI=.93, SRMR=.11) and demonstrated improved fit compared to Models A and B. All of the loadings on these factors were significant except for the 23andMe AUDIT-T, which had a near-zero loading on AUD but loaded significantly on the alcohol use factor. There was a small-to-moderate positive, significant genetic correlation between PTSD and AUD (rG: .36, p<.001). The association between PTSD and alcohol use was negative, significant and small (rG: −.17, p<.001), and the association between AUD and alcohol use was positive, significant, and large1 (rG: .72, p<.001).

Table 2.

Fit indices for Models

Model X2 value (df) X2 p-value AIC CFI SRMR
A. Common Factor 637.67(35) p<.001 677.67 .84 .23
B. Two Correlated Factors 506.97(35) p<.001 546.97 .88 .19
C. Three Correlated Factors 306.97(32) p<.001 352.97 .93 .11
D. Bifactor 82.99(24) p<.001 144.99 .98 .06
Open in a new tab

To test our second question, specifically whether the associations between PTSD, AU, and AUD would shift when accounting for the common variation shared across all alcohol-related items, we fit a bifactor model (Model D; Figure 1D; Supplemental Table 5) in which the alcohol items loaded onto a common factor, residual alcohol use and residual AUD factors were estimated. The correlations across these factors were fixed to zero. This model also estimated the correlations between the PTSD factor and each of the three alcohol factors (the common factor and the residual alcohol use and AUD factors). Model D fit the data very well (χ2=82.99, AIC=144.99, CFI=.98, SRMR=.06). The genetic correlation between PTSD and the common alcohol factor was positive and non-significant (rG: .16, NS). The correlation between PTSD and AUD was positive, significant, and slightly larger than it had been in Model C (i.e., the correlated three factor model; rG:.40, p<.001). The correlation between PTSD and alcohol use was negative, significant, and much larger in magnitude than it had been in model C (rG: −.57, p<.001). Thus, extracting what is common to AUD and alcohol use served to increase each phenotype’s unique association with PTSD.

4. Discussion

We analyzed summary statistics from the largest available GWAS of PTSD and alcohol phenotypes to examine the genetic architecture of PTSD and two major alcohol phenotypes in EAs using genomic SEM, a novel, multivariate genetic approach. We extended our previous findings examining genetic associations between PTSD and alcohol phenotypes (Bountress et al., 2022), as these analyses pulled out what is common to all included alcohol outcomes, versus specific to alcohol use and AUD.

Our first hypothesis, specifically whether the multivariate genetic correlations between PTSD and alcohol factors would be significant and positive for PTSD-AUD but non-significant or significant and negative for PTSD-alcohol use, was supported. This is consistent with previous work demonstrating differential patterns of genetic associations between AUD and alcohol use phenotypes with PTSD (Bountress et al., 2022) and major depression (Polimanti et al., 2019). Neither a common factor model, in which all items across PTSD/AUD/ alcohol use loaded on a single factor, nor a two-factor model, in which all alcohol items loaded on one factor and PTSD items loaded on a second factor, fit the data well. Rather, a correlated three-factor model, in which AUD and alcohol use were separate factors, fit better. Consistent with our hypothesis, this model revealed a small-to-moderate positive genetic correlation between PTSD and AUD and a small negative genetic correlation between PTSD and AU. We also found a large and positive genetic correlation between the two alcohol factors, suggesting that AUD and alcohol use are related but not wholly overlapping genetically, consistent with existing research (Mallard et al., 2021; Sanchez-Roige et al., 2019).

We found support for our second hypothesis, specifically that a Bifactor model showed improved fit compared to the one-, two-, and three-factor models. After accounting for the common genetic variance between PTSD and the shared variance of the two alcohol factors (AUD, AU), we found that the unique associations between PTSD and the alcohol factors (AUD, AU) increased in magnitude, while the genetic correlation between PTSD and the common alcohol factor was non-significant. These findings suggest that similar genetic factors contribute to both PTSD and AUD, but that such factors are independent of genetic factors contributing to AU. Our findings are consistent with a wealth of previous research that has shown a consistent genetic link between PTSD and AUD phenotypes (Sartor et al., 2011; Sheerin et al., 2020; Hong Xian et al., 2000). However, they introduce support for a novel finding: that the association between PTSD and alcohol use among individuals of European Ancestry is negative and distinct. Thus, different genetic factors contribute to AUD, which may be more aligned with psychopathology or propensity towards problematic alcohol use. In contrast, the genetic factors contributing to AU may be more related to other facets of drinking behavior (e.g., social connectedness, releasing endorphins) that are plausibly adaptive and/or associated with resilience (Dunbar et al., 2017). This is consistent with recent work demonstrating unique components of genetic etiology for alcohol consumption versus alcohol-related problems and that alcohol-related problems is more strongly genetically correlated with psychopathology compared to consumption (Mallard et al., 2021).

Though the design of our study does not permit inferences of causality, multiple plausible mechanisms have been proposed to understand the nature of genetic correlations, especially in the context of substance use (Agrawal & Lynskey, 2014; María-Ríos & Morrow, 2020). A correlated liabilities explanation suggests that the two traits share common etiological and/or risk factors. For example, dysregulation of biological and neural mechanisms common to both disorders (e.g., those associated with stress, negative affective processes, neural reward processes, and cognitive processes) could lead to greater risk of comorbid PTSD-AUD (Norman et al., 2012). Likewise, individual differences in maladaptive behavioral and personality factors (e.g., impulsivity, cue reactivity) could underlie the propensity for risk in both PTSD and AUD, leading to comorbidity (María-Ríos & Morrow, 2020). Further, given that the PTSD-alcohol use link was negative and independent of the PTSD-AUD correlation, PTSD and alcohol use may be inversely correlated due to a different array of personality factors (e.g., those related to enjoyment, extroversion, broad socialization), rather than common dysregulation, as is seen with AUD.

Though not directly tested in this paper, PTSD and alcohol phenotypes may exert directional (and possibly bidirectional) associations on one another. Our previous work identified a putative causal effect of PTSD on AUD, supporting a model of self-medication in which individuals with PTSD are more likely to develop AUD (though PTSD was not causally related to drinks per week) (Bountress, et al., 2021). The pattern of positive and negative associations between PTSD and AUD and AU, respectively, aligns with the self-medication hypothesis, such that the consistent use of alcohol to cope with PTSD symptoms may lead to the development of AUD over time via negative reinforcement, whereas motivations for more normative alcohol use may be more aligned with enhancement and social drinking motives, which is incompatible with many of the symptom criteria of PTSD, including anhedonia and negative alternations in cognition and mood. Prior researchers have found a negative association between depression and alcohol use (Polimanti et al., 2019), consistent with this idea.

Given prior work and the present findings that underscore the genetic distinction between alcohol use and AUD, different mechanisms could account for the differences in the links between PTSD and AUD versus PTSD and AU. For example, genes related to PTSD may causally drive the association with AUD, while genes that link PTSD with alcohol use may be more consistent with a correlated liabilities model (i.e., non-causal). Novel research, with additional samples, is needed to understand the underlying mechanisms driving the negative genetic association observed between alcohol use and PTSD, as well as related phenotypes (e.g., major depression).

This study should be considered in light of several limitations. First, our analyses were performed in a sample of European Ancestry individuals and our results may not generalize to other populations. While some evidence suggests the presence of additive genetic similarity for alcohol dependence across individuals of European and African Ancestry (Brick, Keller, Knopik, McGeary, & Palmer, 2019), our prior work showed that the negative genetic correlation between PTSD-alcohol use observed in European Ancestry samples may be a positive correlation in African Ancestry samples (Bountress et al., 2022). More research in diverse samples is essential to understand the link between PTSD and alcohol phenotypes and advancing precision medicine efforts (Peterson et al., 2019). Relatedly, there are known sex differences both in PTSD (Kessler et al., 1995) and alcohol phenotypes (Grant et al., 2015), in the genetic influences on PTSD (Sartor et al., 2012), and in the genetic association between PTSD and alcohol dependence (Sheerin et al., 2020). Being unable to run these models split by sex is a limitation, and it is possible that the structure observed may be stronger or only hold for one sex (e.g., females). As sex-specific summary statistics become more ubiquitous and well-powered, we hope that future work will be able to test this question. Second, our PTSD factor was saturated and contained only two items, resulting in the need to constrain the loadings to be equal across phenotypes. Thus, our PTSD factor represented only the common variance between PTSD and re-experiencing symptoms. As more robust data on PTSD become available, work can evaluate more nuanced factor models of PTSD symptom domains. Finally, results from factor analyses can be sample specific and therefore may be influenced by whether individuals are sampled from the general population versus a clinical population, and may even be influenced by the social norms for alcohol use in the different countries included in our analyses. Replication of our findings across other large studies would strengthen the conclusions herein, especially given that the some of the phenotypes and data we included came from the MVP, which may not be reflective of the non-military, female population. Finally, another group recently found that individuals may over- or under-report their alcohol use (“misreports and longitudinal changes”; MLC) and this bias has serious implications for models of genetic architecture (Xue et al., 2021). Thus, a MLC correction at the GWAS level is likely to provide a more accurate depiction of such associations. Future researchers in this area may benefit from employing this correction when possible in future analyses.

In conclusion, our results support a growing body of work demonstrating that the genetic architecture of alcohol use and AUD are distinct. Further, among individuals of European Ancestry, the genetic associations of PTSD with alcohol use are negative while those between PTSD and AUD are positive. Our findings extend prior work by showing that, when the variance unique to alcohol use and AUD are extracted, the genetic associations with PTSD increases substantially for alcohol use and slightly for AUD.

Supplementary Material

1

Highlights.

  • Using genomic Structural Equation Modeling (gSEM), a Bifactor model fit best.

  • The genetic correlation between PTSD and AUD was significant and positive.

  • The genetic correlation between PTSD and alcohol use was significant and negative.

  • The genetic correlation between PTSD and the general alcohol factor was NS.

Funding Statement

The effort of co-authors was supported by NIAAA (1K01 AA028058 [KB], 1K01 AA025692[CS]), NIMH (MH020030-21A1 [DB], R01MH120219[ADG], R01 MH111671 [RAM], K01MH113848 [REP], R01MH124847-01[CN], T32MH019836[SEH), NIA (RF1AG073593[ADG]), U.S. Department of Veterans Affairs Rehabilitation Research and Development Service (IK2RX002922[SGD]), and The Brain & Behavior Research Foundation NARSAD grant 28632 P&S Fund [REP]. Financial support for the PTSD PGC was provided by the Cohen Veterans Bioscience, Stanley Center for Psychiatric Research at the Broad Institute, One Mind, and the National Institute of Mental Health (R01MH106595). The PGC-SUD Working Group receives support from the National Institute on Drug Abuse and the National Institute of Mental Health via MH109532. Statistical analyses for the PGC were carried out on the Genetic Cluster Computer (http://www.geneticcluster.org) hosted by SURFsara and financially supported by the Netherlands Scientific Organization (NWO 480-05-003), along with a supplement from the Dutch Brain Foundation and the VU University Amsterdam. Dr. Youssef received research support from Department of Veterans Affairs (VA), Department of Defense (DOD), and research support but not salary support from PCORI, MECTA Corporation, Vistagen, and Merck. The opinions in this paper are the authors and do not represent the views of the VA, DOD, the National Institutes of Health, or the United States Government.

Conflict of Interest:

Dr. Kranzler is a member of an advisory board for Dicerna Pharmaceuticals; a consultant to Sophrosyne Pharmaceuticals and Sobrera Pharmaceuticals; a member of the American Society of Clinical Psychopharmacology’s Alcohol Clinical Trials Initiative, which was supported in the last three years by AbbVie, Alkermes, Dicerna, Ethypharm, Indivior, Lilly, Lundbeck, Otsuka, Pfizer, Arbor, and Amygdala Neurosciences; and named as an inventor on PCT patent application #15/878,640 entitled: “Genotype-guided dosing of opioid agonists,” filed January 24, 2018.

No other authors have any conflicts of interest.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

Dropping the 23andMe AUDIT Total Score as an indicator of AUD did not impact model fit or the correlations between factors.

References

  1. Agrawal A, & Lynskey MT (2014). Cannabis controversies: how genetics can inform the study of comorbidity. Addiction, 109(3), 360–370. doi: 10.1111/add.12436 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Benjet C, Bromet E, Karam EG, Kessler RC, McLaughlin KA, Ruscio AM, … Koenen KC (2016). The epidemiology of traumatic event exposure worldwide: results from the World Mental Health Survey Consortium. Psychol Med, 46(2), 327–343. doi: 10.1017/s0033291715001981 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bountress K, Bustamante D, Subbie-Saenz de Viteri S, Chatzinakos C, Sheerin C, Peterson R, Webb BT, Daskalakis N, Edenberg H, Group PGCPW, Meyers J, & Amstadter A (2022). Differences in Genetic Correlations between Posttraumatic Stress Disorder and Alcohol Use Disorder-Related Phenotypes Compared to Alcohol Consumption-Related Phenotypes. medRxiv, 2022.2003.2002.22271415. 10.1101/2022.03.02.22271415 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bountress KE*, Wendt F*, Bustamante D, Agrawal A, Webb B, Gillespie N, … Amstadter A (2021). Potential causal effect of posttraumatic stress disorder on alcohol use disorder and alcohol consumption in individuals of European descent: A Mendelian Randomization Study. Alcohol Clin Exp Res. doi: 10.1111/acer.14649 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Breslau N (2009). Trauma and mental health in US inner-city populations. Gen Hosp Psychiatry, 31(6), 501–502. doi: 10.1016/j.genhosppsych.2009.07.001 [DOI] [PubMed] [Google Scholar]
  6. Brick LA, Keller MC, Knopik VS, McGeary JE, & Palmer RHC (2019). Shared additive genetic variation for alcohol dependence among subjects of African and European ancestry. Addict Biol, 24(1), 132–144. doi: 10.1111/adb.12578 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Browne MW (1984). Asymptotically distribution-free methods for the analysis of covariance structures. British journal of mathematical and statistical psychology, 37(1), 62–83. [DOI] [PubMed] [Google Scholar]
  8. Bush K, Kivlahan DR, McDonell MB, Fihn SD, & Bradley KA (1998). The AUDIT alcohol consumption questions (AUDIT-C): an effective brief screening test for problem drinking. Ambulatory Care Quality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test. Arch Intern Med, 158(16), 1789–1795. [DOI] [PubMed] [Google Scholar]
  9. Clarke T, Adams MJ, Davies G,Howard DM,Hall LS,Padmanabhan S,Murray AD,Smith BH,Campbell A Hayward C,Porteous DJ,D eary IJ,&McIntosh AM (2017). Genome-wide association study of alcohol consumption and genetic overlap with other health-related traits in UK Biobank (N=112 117). Mol Psychiatry, 00, 1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dunbar R, Launay J, Wlodarski R, Robertson C, Pearce E, Carney J, & MacCarron P (2017). Functional benefits of (modest) alcohol consumption. Adaptive Human Behavior and Physiology, 3(2), 118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Duncan LE, Ratanatharathorn A, Aiello AE, Almli LM, Amstadter AB, Ashley-Koch AE, … Koenen KC (2017). Largest GWAS of PTSD (N=20 070) yields genetic overlap with schizophrenia and sex differences in heritability. Mol Psychiatry. doi: 10.1038/mp.2017.77 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gelernter J, Sun N, Polimanti R, Pietrzak R, Levey DF, Bryois J, … Stein MB (2019a). Genome-wide association study of post-traumatic stress disorder reexperiencing symptoms in >165,000 US veterans. Nat Neurosci, 22(9), 1394–1401. doi: 10.1038/s41593-019-0447-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gelernter J, Sun N, Polimanti R, Pietrzak RH, Levey D, Lu Q, Hu Y, Li B, Radhakrishnan K, & Aslan M (2019b). Genome-wide association study of maximum habitual alcohol intake in> 140,000 US European and African American veterans yields novel risk loci. Biol Psychiatry, 86(5), 365–376. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Grant BF, Goldstein RB, Saha TD, Chou SP, Jung J, Zhang H, Pickering RP, Ruan WJ, Smith SM, & Huang B (2015). Epidemiology of DSM-5 alcohol use disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions III. JAMA psychiatry, 72(8), 757–766. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Grotzinger AD, Mallard TT, Akingbuwa WA, Ip HF, Adams MJ, Lewis CM, … Nivard MG (2020). Genetic Architecture of 11 Major Psychiatric Disorders at Biobehavioral, Functional Genomic, and Molecular Genetic Levels of Analysis. medRxiv, 2020.2009.2022.20196089. doi: 10.1101/2020.09.22.20196089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Grotzinger AD, Rhemtulla M, de Vlaming R, Ritchie SJ, Mallard TT, Hill WD, … Tucker-Drob EM (2019). Genomic structural equation modelling provides insights into the multivariate genetic architecture of complex traits. Nat Hum Behav, 3(5), 513–525. doi: 10.1038/s41562-019-0566-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Heath AC, Jardine R, & Martin NG (1989). Interactive effects of genotype and social environment on alcohol consumption in female twins. J Stud Alcohol, 50(1), 38–48. [DOI] [PubMed] [Google Scholar]
  18. Hu L B. PM (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. [Google Scholar]
  19. Jakupcak M, Tull MT, McDermott MJ, Kaysen D, Hunt S, & Simpson T (2010). PTSD symptom clusters in relationship to alcohol misuse among Iraq and Afghanistan war veterans seeking post-deployment VA health care. Addict Behav, 35(9), 840–843. doi: 10.1016/j.addbeh.2010.03.023 [DOI] [PubMed] [Google Scholar]
  20. Johnson EC, St Pierre CL, Meyers JL, Aliev F, McCutcheon VV, Lai D, … Agrawal A (2019). The Genetic Relationship Between Alcohol Consumption and Aspects of Problem Drinking in an Ascertained Sample. Alcohol Clin Exp Res, 43(6), 1113–1125. doi: 10.1111/acer.14064 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kaprio J, Koskenvuo M, Langinvainio H, Romanov K, Sarna S, & Rose RJ (1987). Genetic influences on use and abuse of alcohol: a study of 5638 adult Finnish twin brothers. Alcohol Clin Exp Res, 11(4), 349–356. [DOI] [PubMed] [Google Scholar]
  22. Kenny DA (2015). Measuring model fit. In.; Knopik VS, Heath AC, Madden PA, Bucholz KK, Slutske WS, Nelson EC, … Martin NG (2004). Genetic effects on alcohol dependence risk: re-evaluating the importance of psychiatric and other heritable risk factors. Psychol Med, 34(8), 1519–1530. [DOI] [PubMed] [Google Scholar]
  23. Kessler RC, Sonnega A, Bromet E, Hughes M, & Nelson CB (1995). Posttraumatic stress disorder in the National Comorbidity Survey. Archives of general psychiatry, 52(12), 1048–1060. [DOI] [PubMed] [Google Scholar]
  24. Kranzler HR, Zhou H, Kember RL, Vickers Smith R, Justice AC, Damrauer S, … Gelernter J (2019). Genome-wide association study of alcohol consumption and use disorder in 274,424 individuals from multiple populations. Nat Commun, 10(1), 1499. doi: 10.1038/s41467-019-09480-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lai D, Wetherill L, Bertelsen S, Carey CE, Kamarajan C, Kapoor M, … Foroud T (2019). Genome-wide association studies of alcohol dependence, DSM-IV criterion count and individual criteria. Genes Brain Behav, 18(6), e12579. doi: 10.1111/gbb.12579 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Liu M, Jiang Y, Wedow R, Li Y, Brazel DM, Chen F, … Vrieze S (2019). Association studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use. Nat Genet, 51(2), 237–244. doi: 10.1038/s41588-018-0307-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Mallard TT, Savage JE, Johnson EC, Huang Y, Edwards AC, Hottenga JJ, … Anokhin A (2021). Item-Level Genome-Wide Association Study of the Alcohol Use Disorders Identification Test in Three Population-Based Cohorts. American Journal of Psychiatry, appi. ajp. 2020.20091390. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. María-Ríos CE, & Morrow JD (2020). Mechanisms of shared vulnerability to post-traumatic stress disorder and substance use disorders. Frontiers in behavioral neuroscience, 14, 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. McLeod S, Koenen KC, Meyer J, Lyons MJ, Eisen S, True W, & Goldberg D (2001). Genetic and environmental influences on the relationship among combat exposure, posttraumatic stress disorder symptoms, and alcohol use. Journal of traumatic stress, 14(4), 259–275. [DOI] [PubMed] [Google Scholar]
  30. Nievergelt CM, Maihofer AX, Klengel T, Atkinson EG, Chen CY, Choi KW, … Koenen KC (2019). International meta-analysis of PTSD genome-wide association studies identifies sex- and ancestry-specific genetic risk loci. Nat Commun, 10(1), 4558. doi: 10.1038/s41467-019-12576-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Norman SB, Myers US, Wilkins KC, Goldsmith AA, Hristova V, Huang Z, … Robinson SK (2012). Review of biological mechanisms and pharmacological treatments of comorbid PTSD and substance use disorder. Neuropharmacology, 62(2), 542–551. doi: 10.1016/j.neuropharm.2011.04.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Peterson RE, Kuchenbaecker K, Walters RK, Chen C-Y, Popejoy AB, Periyasamy S, … Brick L (2019). Genome-wide association studies in ancestrally diverse populations: opportunities, methods, pitfalls, and recommendations. Cell, 179(3), 589–603. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Polimanti R, Peterson RE, Ong J-S, MacGregor S, Edwards AC, Clarke T-K, … Lind PA (2019). Evidence of causal effect of major depression on alcohol dependence: findings from the psychiatric genomics consortium. Psychological medicine, 49(7), 1218–1226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Sanchez-Roige S, Palmer AA, Fontanillas P, Elson SL, Adams MJ, Howard DM, … Clarke TK (2019a). Genome-Wide Association Study Meta-Analysis of the Alcohol Use Disorders Identification Test (AUDIT) in Two Population-Based Cohorts. Am J Psychiatry, 176(2), 107–118. doi: 10.1176/appi.ajp.2018.18040369 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Sanchez-Roige S, Fontanillas P, Elson SL, Team, a. R., Gray JC, de Wit H, … Palmer AA (2019b). Genome-wide association study of alcohol use disorder identification test (AUDIT) scores in 20 328 research participants of European ancestry. Addiction biology, 24(1), 121–131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Sartor CE, Grant JD, Lynskey MT, McCutcheon VV, Waldron M, Statham DJ, Bucholz KK, Madden PA, Heath AC, Martin NG, & Nelson EC (2012). Common heritable contributions to low-risk trauma, high-risk trauma, posttraumatic stress disorder, and major depression. Arch Gen Psychiatry, 69(3), 293–299. 10.1001/archgenpsychiatry.2011.1385 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Sartor CE, McCutcheon VV, Pommer NE, Nelson EC, Grant JD, Duncan AE, … Heath AC (2011). Common genetic and environmental contributions to post-traumatic stress disorder and alcohol dependence in young women. Psychol Med, 41(7), 1497–1505. doi: 10.1017/s0033291710002072 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Saunders JB, Aasland OG, Babor TF, De la Fuente JR, & Grant M (1993). Development of the alcohol use disorders identification test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption-II. Addiction, 88(6), 791–804. [DOI] [PubMed] [Google Scholar]
  39. Savalei V, & Bentler PM (2009). A two-stage approach to missing data: Theory and application to auxiliary variables. Structural Equation Modeling: A Multidisciplinary Journal, 16(3), 477–497. [Google Scholar]
  40. Sheerin CM, Bountress KE, Meyers JL, Saenz de Viteri SS, Shen H, Maihofer AX, Duncan LE, & Amstadter AB (2020). Shared molecular genetic risk of alcohol dependence and posttraumatic stress disorder (PTSD). Psychol Addict Behav. 10.1037/adb0000568 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Stein MB, Jang KL, Taylor S, Vernon PA, & Livesley WJ (2002). Genetic and environmental influences on trauma exposure and posttraumatic stress disorder symptoms: a twin study. Am J Psychiatry, 159(10), 1675–1681. doi: 10.1176/appi.ajp.159.10.1675 [DOI] [PubMed] [Google Scholar]
  42. Sullivan PF (2010). The psychiatric GWAS consortium: big science comes to psychiatry. Neuron, 68(2), 182–186. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Sun Y, Chang S, Wang F, Sun H, Ni Z, Yue W, … Shi J (2019). Genome-wide association study of alcohol dependence in male Han Chinese and cross-ethnic polygenic risk score comparison. Transl Psychiatry, 9(1), 249. doi: 10.1038/s41398-019-0586-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Verhulst B, Neale M, & Kendler K (2015). The heritability of alcohol use disorders: a meta-analysis of twin and adoption studies. Psychol Med, 45(05), 1061–1072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Vlahov D, Galea S, Resnick H, Ahern J, Boscarino JA, Bucuvales M, … Kilpatrick D (2002). Increased use of cigarettes, alcohol, and marijuana amond Manhattan, New York residents after the September 11th terrorist attacks. Am J Epidemiol, 155, 988–996. [DOI] [PubMed] [Google Scholar]
  46. Walters RK, Polimanti R, Johnson EC, McClintick JN, Adams MJ, Adkins AE, … andMe Research T (2018). Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders. Nature Neuroscience, 21(12), 1656–1669. doi: 10.1038/s41593-018-0275-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Wilkins KC, Lang AJ, & Norman SB (2011). Synthesis of the psychometric properties of the PTSD checklist (PCL) military, civilian, and specific versions. Depression and anxiety, 28(7), 596–606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Xian H, Chantarujikapong SI, Shrerrer JF, Eisen SA, Lyons MJ, Goldberg J, … True W (2000). Genetic and environmental influences on posttraumatic stress disorder, alcohol, and drug dependence in twin pairs. Drug and alcohol dependence, 61(1), 95–102. [DOI] [PubMed] [Google Scholar]
  49. Xue A, Jiang L, Zhu Z, Wray NR, Visscher PM, Zeng J, & Yang J (2021). Genome-wide analyses of behavioural traits are subject to bias by misreports and longitudinal changes. Nat Commun, 12(1), 20211. 10.1038/s41467-020-20237-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Zhou H, Sealock JM, Sanchez-Roige S, Clarke TK, Levey DF, Cheng Z, … Gelernter J (2020). Genome-wide meta-analysis of problematic alcohol use in 435,563 individuals yields insights into biology and relationships with other traits. Nat Neurosci, 23(7), 809–818. doi: 10.1038/s41593-020-0643-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1796093-supplement-1.docx (35.5KB, docx)

RESOURCES