Environmental influences on alcohol use: Informing research on the joint effects of genes and the environment in diverse U.S. populations

Abstract

Background and Objectives

This review aimed to inform the current state of alcohol research on the joint effects of genes and the environment conducted in U.S. racial/ethnic minority populations, focusing on African Americans, Latinos/Hispanics, Asians, and American Indians.

Methods

A key-word and author-based search was conducted and supplemented with direct contact to researchers in this area to ensure a comprehensive inclusion of published, peer-reviewed studies. These studies were considered in terms of the racial/ethnic population groups, phenotypes, genetic variants, and environmental influences covered. Research findings from alcohol epidemiologic studies were highlighted to introduce some potential environmental variables for future studies of gene and environment (G–E) relationships.

Results

Twenty-six (N=26) studies were reviewed. They predominantly involved African American and Asian samples and had a very limited focus on Latinos/Hispanics and American Indians. There was a wide range of alcohol-related phenotypes examined, and studies almost exclusively used a candidate gene approach. Environmental influences focused on the most proximate social network relationships with family and peers. There was far less examination of community- and societal-level environmental influences on drinking. Epidemiologic studies informing the selection of potential environmental factors at these higher-order levels suggest inclusion of indicators of drinking norms, alcohol availability, socioeconomic disadvantage, and unfair treatment.

Conclusions

The review of current literature identified a critical gap in the study of environments: There is the need to study exposures at community and societal levels.

Scientific Significance

These initial studies provide an important foundation for evolving the dialogue and generating other investigations of G–E relationships in diverse racial/ethnic groups.

1. Introduction

There is a recognizable shortage of genetic studies of alcohol consumption and related phenotypes in U.S. racial/ethnic minority populations (the focus of this special issue). This is occurring alongside a growing interest in understanding how environmental conditions enhance or reduce genetic effects in light of long-held evidence that sociocultural factors and the built environment are associated with levels of alcohol consumption and risk for related problems in U.S. racial/ethnic minority populations (e.g., Karriker-Jaffe et al.¹, Zapolski et al.², Zemore et al.³). The current review seeks to bring together two large areas of alcohol research in racial/ethnic minority populations. The first are studies of relationships between genes and the environment (G–E) and the second are alcohol epidemiologic studies, and both literatures include a range of features of drinking, from normative behaviors to binge or heavy drinking and alcohol use disorder (AUD).

We conceptualize G–E relationships from an integrated perspective, to mean that genetic and environmental effects come together (in different ways) to contribute to alcohol-related phenotypes⁴. Three fundamental relationships include: 1) additive effects of genes and the environment; 2) gene-environment correlations; and 3) gene × environment interactions (G×E). An additive relationship is characterized by independent genetic and environmental contributions to an alcohol outcome. Gene-environment correlations and G×E, respectively, describe mediation and moderation relationships in predicting an outcome. A gene-environment correlation identifies a selection effect where genetic vulnerability is associated with exposure to an environment. GxE relationships indicate that genetic vulnerability – or protection – can vary depending on the level of environmental exposure, or, equally, that the effect of the environmental exposure can be different depending on genotype.

The study of GxE relationships has garnered attention and excitement, but also caution and concern^5,6. Therefore, we seek to encourage careful and systematic analyses of G–E relationships as an important tool to further our understanding of the complex etiology of AUD and to identify environmental targets for prevention and intervention efforts. Previous papers have outlined recommendations for rigorous GxE research practices (e.g., Dick et al.⁵). We focus here on the selection of environmental factors that have both theoretical plausibility and empirical precedent for inclusion in the study of G–E relationships.

This report begins with a presentation of two theoretical mechanisms for conceptualizing environmental influences on alcohol consumption and related phenotypes, offered as a model to help synthesize and critique the extant literature in this area. We then review studies that have tested hypotheses about the interplay between measured genes and the environment in U.S. racial/ethnic minority populations, focusing on African Americans, Latinos/Hispanics, Asians, and American Indians. By measured gene studies, we refer to those that have used molecular methods and examined genetic variation in a given region or at a specific marker on a chromosome. (See Agrawal et al. in this issue for a discussion of latent genetic studies using twin designs.) However, because these genetic studies are few, we also identify some key environmental exposures shown to be associated with greater risks or protections for drinking and AUD in these populations.

2. Theoretical mechanisms

We use the term environment to broadly refer to community- or societal-level factors like per capita alcohol consumption, drinking norms, and alcohol outlet density, as well as social network factors like peer relationships and family circumstances (e.g., childhood abuse/neglect). This conceptualization of the environment is consistent with a multi-level perspective on health^7,8 and risk behavior including substance use⁹ in that we emphasize contexts in which individuals are embedded, including peer groups, organizations, physical and social spaces, and the broader culture. This multi-level perspective is reflected in the model presented here (See Figure 1) as a means of organizing environmental exposures across three spheres of influence: social network, community, and societal. Additionally, we categorize these environments under two central mechanisms, social control and trigger, that are thought to modify genetic predisposition to AUD and that have previously been the focus of GxE studies. Earlier literature (e.g. Dick & Kendler¹⁰, Shanahan & Hofer¹¹, and Sher et al.¹²) offers a solid place to start in identifying examples of environmental social controls and triggers important to AUD, although issues relevant to specific racial/ethnic groups have not been widely considered in prior research. Therefore, here, we sought to introduce these mechanisms in order to set the stage for a wider discussion of environmental conditions related to alcohol use for racial/ethnic minority groups.

Figure 1.

Theoretical mechanisms for genetic and environmental influences on AUD

Social control

Elements of social control include both protective and risk environments that primarily operate to constrain or enable opportunities to use alcohol, thereby influencing the contribution of genetic effects to AUD. Shanahan and Hofer¹¹ emphasized elements of social control such as social norms or structural constraints that help maintain social order by decreasing deviant behaviors such as excessive alcohol use. They highlighted involvement with community institutions including churches and social norms surrounding alcohol use. For example, the Higuchi et al.¹³ study is often highlighted in discussions of the relationship of societal-level effects on genetic influences for alcoholism, as the protective ALDH2*2 variant was reported to be weaker in younger Japanese cohorts, suggesting a link to more liberal drinking norms and an increase in per capita alcohol consumption. In the U.S., there is evidence for strong cohort effects on drinking, with some birth cohorts drinking much more heavily than others^14,15; however, corresponding studies in U.S. samples on the relationship of these changes to genetic influences for alcohol use have not been conducted. Dick and Kendler¹⁰ and Sher et al.¹² also underscored more proximate environmental social controls related to peer and family networks, including role transitions like marriage or marriage-like relationships, parental monitoring, and relationships with deviant or antisocial peers. Kendler et al.¹⁶ showed that low parental monitoring and high peer deviance were associated with increased genetic influences on alcohol use in youth. Similar social control processes also may be happening on the community level. Neighborhood environments characterized by urbanicity and increased access to alcohol (which have been associated with increased genetic influences on alcohol consumption) may be associated with lower monitoring and control of others’ behavior, particularly in concert with high residential mobility^16,17.

Social trigger

By contrast, adverse environmental conditions may act as social stressors that trigger or strengthen genetic effects for AUD; evidence points to an individual’s responsiveness or sensitivity to stressful conditions being influenced by genetic factors¹⁸. Nurturing environments conceptually may be placed at the other end of this spectrum, but have not been the focus of G–E studies of AUD. For example, Turkheimer et al.¹⁹ examined a continuum of socioeconomic conditions, showing different relationships for genetic and environment effects with intelligence when comparing impoverished and affluent family environments. Conversely, studies of GxE triggering mechanisms for AUD, as well as for major depression and aggression, have focused more on contextual variables like stressful life events and childhood maltreatment^10,11. Two recent measured-gene studies identified significant interactions between genetic effects and childhood adversities (e.g., abuse, parental divorce or death, witnessed violence) in predicting AUD: Meyers et al.²⁰ and Sartor et al.²¹ reported, respectively, an enhanced association of the ADH1B risk variant and a reduced association of the ADH1B protective variant with AUD under adverse childhood conditions. These studies are complemented by evidence suggesting strong main effects of these stressors on AUD, including robust evidence for adverse effects of early child abuse and parental loss on alcohol outcomes in adulthood^22–24. Additional indicators of environmental stress may be relevant moderators of genetic influences, including at community and societal levels. Boardman et al.⁸ suggest a reframing of the ‘environment’ in studies of G–E relationships to include macro-level factors that characterize where people live and work when assessing their exposure to risks and resources. One potential trigger for AUD is neighborhood disadvantage, which has consistently been associated with illicit drug use, and in some studies has also shown associations with heavy drinking and AUD²⁵. Later we return to a discussion of neighborhood disadvantage as an environmental variable that may be relevant to studies of AUD and G–E relationships for ethnic/racial minority populations.

3. Studies of joint G–E effects in U.S. ethnic minority populations

We conducted a focused survey of the peer-reviewed, published literature for measured gene and environment alcohol studies that focus on U.S. racial and ethnic minority populations to inform the current state of the science in this area. The results of our literature review are presented in Table 1. Key terms for this search included “gene-environment”, “gene × environment”, “alcohol”, “substance use”, “divers*”, “ethnic*”, “race”, “racial”, “Latin*”, “Hispanic”, “African American”, “Asian American”, “Native American”, and “American Indian”. The primary search engine used was Google Scholar. In addition, an author search based on key researchers in this area was conducted and a number of researchers were contacted directly to explore any current research efforts being conducted and additional relevant researchers that should be considered. Below, we have reviewed the identified studies in terms of the racial/ethnic population groups, phenotypes, genetic variants, and environmental influences studied. We also present some of the challenges for GxE research evidenced in these studies.

Table 1.

Studies of genetic and environmental influences with diverse population groups: Summary characteristics

Citation	Sample Size (subgroups)	Age Group	Gene	Drinking Phenotype*	Environmental Influences
					Controls	Triggers
African American
Beach et al., 2010	337	Youth	DRD4	Substance Use** Drank alcohol, 3+ drinks at one time, and smoked marijuana	Social Network: Intervention status - Strong African American Families program
Brody et al., 2009a	641	Youth	SLC6A4	Risk Behavior Initiation** Ever drank alcohol, used marijuana, or had sexual intercourse	Social Network: Intervention status - Strong African American Families program
Brody et al., 2009b	253	Youth	SLC6A4	Substance Use** Smoked cigarettes, drank alcohol, 3+ drinks at one time, and smoked marijuana	Social Network: Supportive parenting, involvement, and good communication
Brody et al., 2013	963	Youth and adolescents	DRD2, DRD4, ANKK1, GABRG1, GABRA2	Alcohol Use** Drank alcohol and 4+ drinks at one time	Social Network: Intervention status - Strong African American Families program
Brody et al., 2015	291	Adolescents	DRD4	Substance Use** Cigarette, alcohol, marijuana, and other drug use in past month	Social Network: Intervention status - Adults In the Making program	Social Network: Family risk - Parent-child conflict/communication, chaos in home, caregiver depression, and parental support
Enoch et al., 2010	563	Adult men	GABRA2	1) Alcohol; 2) Heroin; and 3) Cocaine dependence		Social Network: Childhood abuse or neglect
Goyal et al., 2016	541	Adolescents	CRHBP, NR3C1	Alcohol use** Frequency and quantity of any alcohol use, and frequency of 5+ drinks in a couple of hours	Social Network: Stressful life events§ - Life transitions (e.g., marriage, baby)	Social Network & Community Level: Stressful life events§ - Property damage, disruption in home, illness or injury, family/friend death or illness, victim of crime, witnessed violence, exposure to neighborhood violence
Kranzler et al., 2012	564 (African, Afro- Caribbean, and African American)	College students	SLC6A4	1) Any drinking; and 2) Heavy drinking frequency (males 5+ drinks, females 4+ drinks)		Social Network: Stressful life events - Relationship, academic, family health, and financial problems
Asian American
Bujarski et al., 2015	97 (Chinese, Korean, and Japanese)	College students	ALDH2, ADH1B	Alcohol use** Frequency and quantity of any alcohol use, heavy episodic drinking, and alcohol- related consequences	Social Network: Parental history of alcoholism§§ Community Level: 1) Ethnicity; and 2) Acculturation	Social Network: Parental history of alcoholism§§
Hendershot et al., 2005	428 (Chinese and Korean)	College students	ALDH2	1) Frequency and quantity of any alcohol use, and 2) Heavy episodic drinking (males 5+ drinks, females 4+ drinks)	Social Network: Parental alcohol use Community Level: Ethnicity
Irons et al., 2007	34 (Korean adoptees)	Adolescents and young adults	ALDH2	1) Past-year drinking; and 2) Past-year drunkenness	Social Network: 1) Parental history of alcoholism§§; and 2) Sibling alcohol use	Social Network: Parental history of alcoholism§§
Irons et al., 2012	356 (Korean adoptees)	Adolescents and young adults	ALDH2	1) Alcohol Use** Drinking frequency and quantity, maximum drinks consumed on one occasion, frequency of intoxication; 2) Alcohol abuse & dependence symptom count; and 3) Alcohol abuse or dependence diagnosis	Social Network: 1) Deviant peer behaviors; 2) Parental alcohol use problems§§; and 3) Elder sibling alcohol use problems	Social Network: Parental alcohol use problems§§
Luczak et al., 2003	347 (Chinese and Korean)	College students	ALDH2	Heavy episodic drinking (males 5+ drinks, females 4+ drinks)	Community Level: 1) Religious services attendance; and 2) Ethnicity
Latino/Hispanic
Du et al., 2009	703 (Mexican American)	Adults	SLC6A4, OPRM1, A118G, DRD2	1) Maximum drinks consumed in a 24-hour period; and 2) Alcohol abuse or dependence	Social Network: Marital status Community Level: Educational attainment
American Indian
Ducci et al., 2008	291 (Southwest American Indian)	Adult women	MAOA	Alcohol use disorder		Social Network: Childhood sexual abuse
Enoch et al., 2006	342 (Plains American Indian)	Adults	COMT	Alcohol use disorder and nicotine addiction**	Community Level: Gender
Multiple Population Sample
Chartier et al., 2016	7,716 (Combined African, European, and Hispanic American sample)	Adults	ADH1B, ADH1C, ADH4	1) Maximum drinks consumed in 24-hours; and 2) Alcohol dependence symptoms	Community Level: Religious services attendance
Daw et al., 2013	14,560 (Combined Black, White, Hispanic, Asian, and other sample)	Adolescents and young adults	SLC6A4	1) Drinking frequency, and 2) Drinking quantity	Community Level: School-level drinking
Dick et al., 2006	1,916 (Combined African and European American sample)	Adults	GABRA2	Alcohol dependence	Social Network: Marital status
Kaufman et al., 2007	127 (Combined African, European, and Hispanic American, and biracial sample)	Youth and adolescents	SLC6A4	1) Alcohol use yes/no; 2) Age of first use; and 3) Episodes of intoxication		Social Network: 1) Childhood maltreatment; and 2) severity of childhood maltreatment
Lieberman et al., 2016	1,845 (Combined African, European, and Hispanic descent sample)	College students	FKPB5	Heavy drinking (males 5+ drinks; females 4+ drinks) averaged from daily report across 29 days		Social Network: Past- year life stress§ - relationship, health, and financial problems Social Network & Community: 1) Early life trauma§ and 2) past-year trauma§ - accident, illness, disaster, war, violence/victimization, loss/separation
Luczak et al., 2001	108 (Chinese); 124 (Korean); and 96 (White)	College students	ALDH2	Binge drinking	Community Level: Ethnicity
Luczak et al., 2004	190 (Chinese); 214 (Korean); and 200 (European heritage)	College students	ALDH2, ADH1B	Alcohol dependence	Community Level: 1) Gender; and 2) Ethnicity
Olfson et al., 2014	1,550 (Combined African and European American sample)	Adolescents	ADH1B	1) Age of first intoxication; and 2) Age of first alcohol use disorder symptom	Social Network: Peer drinking
Sartor et al., 2014	2,617 (African American); and 1,436 (European American)	Adults	ADH1B	1) Maximum drinks consumed in 24-hours; and 2) Alcohol use disorder symptom count		Social Network & Community: Childhood adversity§ - Parental death, sexual or physical abuse, and witnessed or experienced violent
Vaske et al., 2009	402 (African American); and 1,552 (Caucasian)	Adolescent and young adult	DAT1	Problems (because of drinking)** At school or work, with friends or partner, hung over, sick, physical fights, and risky sexual behavior	Social Network: Paternal history of alcoholism§§	Social Network: Paternal history of alcoholism§§

Notes:

^*Numbered phenotypes reflect studies that examined multiple outcome measures;

^**Phenotype was a composite measure of more than one alcohol or drug use or risk behavior outcome;

^§Items in composite measure of environment could be individually classified as constraints or triggers or as social network or community level exposures; and

^§§Parental alcoholism is associated with known environmental risk factors for AUD that could be classified as constraints or triggers.

Population groups, phenotypes, and genetic variants covered

There were a limited number of studies addressing this topic, but even among those identified some population groups were more represented than others. Predominantly, studies involved African American^26–33 or Asian American^34–38 samples. There was a very limited focus on Latinos/Hispanics (one study on Mexican Americans³⁹) and American Indians (two studies identified^40,41). Additionally, across population groups, most samples included younger subjects, for example, older children and adolescents. Studies including adults fell into two categories, with most being conducted with young adult/college student samples and fewer involving samples representing a wide range of adult ages.

Due to the relative paucity of research in this area, the scope of the literature review was expanded to include studies with some degree of racial or ethnic diversity. Some of these studies reported G–E results separately for each population included^21,42,43, while others reflected the results of a combined sample^44–49. These multi-population studies primarily included European Americans and African Americans, although some did include subjects of Latino/Hispanic or Asian descent. The search was also expanded to studies that examined both alcohol and other substance use phenotypes^{26,27,29–31,40}. Therefore, the included studies either examined alcohol use discretely or as a part of a composite measure of substance use or risky behavior.

Studies of GxE relationships face a number of limitations to which the identified studies are vulnerable. These include challenges with small sample sizes, related concerns about false positive findings and the replicability of results, statistical and conceptual issues about the nature of the interaction, and publishing biases^5,6,50,51. Therefore, excluding several large studies that used combined samples of more than one population, study sample sizes for racial/ethnic minority groups were as small as 34 and 97 subjects and otherwise ranged from 108 to 703, with two larger samples including 963 and 2,617 subjects. Ideally, sample sizes greater than 1,000 participants are recommended to help protect against potentially spurious GxE effects⁵, although the realities of limited resources and restricted access to some population groups can provide a challenge to robust sample sizes. To address this concern, Argawal et al.⁵² promote greater consistency in the phenotypic measures utilized across studies, which would allow for the harmonization of results and meta-analysis across studies to provide more confidence in the findings drawn. However, even in the small number of studies reviewed here, one can note the wide variation in phenotypes that were examined, which ranged from different measures of alcohol consumption and alcohol use diagnoses to complex composite measures of more than one alcohol or drug use outcome.

Additionally, these studies relied almost exclusively on a candidate gene approach. Examples of the genes studied are SLC6A4 (i.e., the short, long 5-HTTLPR variant); DRD4; GABRA2; ADH1B; and ALDH2. Unfortunately, the replication of findings from candidate GxE studies can be challenging; most often noted are inconsistencies reported for the relationship between the 5-HTTLPR variant and stressful life events (see Duncan and Keller⁶ and Dick et al.⁵ for an expanded discussion of this topic). Some studies did examine candidate genes with relatively well-understood biological mechanisms related to alcohol consumption and AUD (e.g., ADH1B and ALDH2; see Hurley et al.⁵³), and a few tested multiple markers or a sum score of risk variants across multiple genes^28,45. However, other techniques, including polygenic risk scores that aggregate the effect of a number of gene markers, offer researchers newer alternatives that could help to increase confidence in the genetic associations that are being studied^54,55. Studies using a polygenic score approach were not represented in the literature we reviewed.

Environmental influences

The environmental influences examined across these studies were considered, based on our organizational framework in Figure 1, for their representation of the mechanisms of social controls and triggers and our three levels of environmental influence (social network, community, and societal). In the sections below, we provide an overview of the environmental influences covered. Additionally, Table 2 presents relevant key findings for the genetic and environment effects studied and their inter-relationships. These studies point to some inconsistencies and mixed findings that underscore the need for further replication and other empirical support. There may be many reasons for these differences¹¹, which are perhaps unsurprising based on the earlier challenges discussed (e.g., sample size, variations in phenotypic measures). Alcohol use in some population groups could also be less affected by certain genetic or environmental factors, including when samples involve younger subjects early in their development of drinking behaviors³⁵. Regardless, this research does provide an important foundation for evolving the dialogue in this area. These initial studies hopefully will serve to generate other investigations of G–E relationships in diverse racial/ethnic population groups.

Table 2.

Key findings from studies of G–E relationships by environmental feature and level

Feature	Social Network Level
Parental alcoholism	Interaction between parental alcohol problems and protective genetic variant against drinking in Asians observed in some studies (Bujarski et al.34; Irons et al.35). Additive and interactive G–E relationships for parental alcoholism may be different across gender and population groups (Vaske et al.43).
Parenting skills intervention status	Genetic variants were associated with differential responsiveness to interventions targeting more communicative and involved parenting for African American youth (Brody et al.27–30; Beach et al.26).
Sibling alcohol use	Sibling alcohol use was a significant predictor of reduced ALDH2*2 protection against drinking (Irons et al.35,36), but non-significant findings were encountered when examining phenotypes for AUD35.
Marital status	Evidence of both G–E correlation and interaction effect in Dick et al.46, but no G–E relationships in Du & Wan39.
Peer alcohol use	Reduced protective ADH1B effect when reporting that most or all peers drink in African and European American adolescents (Olfson et al.44). ALDH2 was not correlated with peer deviance; more consistent evidence for a G–E additive relationship with alcohol-related phenotypes (Irons et al.35).
Childhood maltreatment	Evidence for an interaction between childhood abuse and genetic risk by some (MAOA, Ducci et al.41; SLC6A4, Kaufman et al.47), but not all studies (GABRA2, Enoch et al.31). GABRA2 did not correlate with childhood trauma.
Social Network & Community Level*
Adverse life events	Mixed support for joint effects between genes and adverse life events based on genetic variant, gender, and race/ethnicity, respectively, demonstrating a moderating effect for CRHBP (Goyal et al.32), African American females (Kranzler et al.33), and European American males (Sartor et al.21) only. Early life trauma, but not past-year trauma or stress, interacted with FKBP5 on heavy drinking in college students (Lieberman et al.49).
Community Level
Religious involvement	Evidence of reduced genetic risk for drinking with greater religious services attendance (Chartier et al.45; Luczak et al.37)
Social drinking norm indexes	Higher school-level alcohol use interacted with SLC6A4 genotype to predict greater alcohol consumption (Daw et al.48). Acculturation predicted more drinking with no interactive effect (Bujarski et al.34). Studies of ethnicity, gender, and educational level reported mixed additive (Luczak et al.57) and interactive (Du & Wan39; Enoch et al.40) effects as well as no G–E relationships (Hendershot et al.38).

Notes: G–E = gene-environment;

^*adverse life events were classified as social network and community level exposures due to the items included in the composite variable; and none of the studies reviewed examined environmental features at the societal level.

Social networks

There was considerable attention given to the most proximate social network influences, as both constraints and triggers. Environmental variables at this level were characterized by parental influences^{34–36,38,43}, including parental alcoholism, which, framed within the context of our model, presents both potentially controlling and triggering influences. Parental alcoholism, for example, may foster conditions where alcohol is more accessible and drinking is more acceptable or that contribute to a turbulent and stress-inducing home environment. However, this environmental influence is strongly confounded with genetic effects, unless examined using highly robust designs to isolate genetic and environmental factors, e.g., an adoption paradigm (see Irons et al.³⁵). Also examining the role of parental influence, there was a subset of randomized studies that examined interventions targeting parenting skills^26–30, including communication skills around risks and expectations for substance use and providing emotional support and instrumental assistance to children.

Social-network level variables also focused on other family relationships^35,36,39,46, including marriage and sibling relationships, and peer influences^35,44. Genetic factors can inter-relate with marital status or peer groups in influencing alcohol use and AUD through both gene-environment correlations and GxE effects^46,56. These factors all potentially contribute to varying degrees of increased or decreased constraint related to access to alcohol and perceived permissiveness for drinking in the immediate environment.

Adverse life events were a prominent environmental (triggering) variable across many of the studies^{21,31–33,41,47,49}. As a category, they represented a number of different experiences: childhood maltreatment; early life trauma; life transitions; experiences of illness and death; disruptions in the home; and witnessing/exposure to family violence. These risk factors sometimes extended beyond social network influences to include exposure to violence and crime at the community or neighborhood level (e.g., Goyal et al.³², Sartor et al.²¹). Generally, composite measures of stressful life events incorporated a wide variety of circumstances, which can complicate the understanding and isolation of environmental exposures that contribute to a GxE effect¹².

Community

Only a handful of studies examined how broader social influences on drinking may influence the relationship between genetic risk or protection and alcohol use. These studies included the use of variables related to religious involvement and school-level (e.g., high school) drinking^37,45,48 to examine the effects of instituational social control and group-level social drinking norms. At this level, we also included several studies that used membership in various social groups^{34,38–40,42,57}, like ethnicity, gender, and level of education as indirect measures indexing group differences in social norms about drinking or vulnerability to environmental stressors. However, these variables can be difficult to assess as cultural indicators in G–E associations. Ethnicity may be confounded by population stratification, as the pattern of association between alleles (linkage disequilibrium) and genotype frequencies can vary across ancestry groups^42,45, and gender may indicate other biological features (e.g., sex hormone levels)⁴⁰. Level of education also has a complex relationship with alcohol-related phenotypes, both as a moderator of genetic effects and through shared genetic influences with alcohol problems⁵⁸. One study examined acculturation³⁴, which also may gauge changes in social drinking norms tied to exposure to mainstream American culture.

Societal

Even more striking was the absence of higher-order societal variables, for both social control and trigger mechanisms. This runs contrary to historic understandings that structural influences and differential exposures to risks and access to resources are related to the health differences experienced by racial/ethnic minority groups⁵⁹.

4. Some considerations when selecting an environment

The above review of the current literature identified a critical gap in the study of environments: There is a need to include exposures at community and societal levels⁸. Some environmental conditions are also potentially more relevant to racial/ethnic minority populations. We start here by offering a few key considerations for selecting an environmental factor for inclusion in a G–E relationship study, and then highlight research findings from alcohol epidemiologic studies to introduce some potential environmental variables, at these higher levels of influence, that have not been previously studied in this manner.

Fit with a theoretical mechanism

There has been some debate as to whether it is important to establish a link between an environmental exposure and the neurobiological changes associated with the psychiatric disorder being studied⁶⁰. We do not weigh in on that here; existing literature provides an examination of proposed pathways of environmental exposure to neurobiological functioning and the association with alcohol use, relapse, and recovery (e.g., Koob & Volkow, 2016⁶¹; Seo & Sinha, 2015⁶²). Research on neurobiological changes is more consistently linked to social triggers like stress and trauma, but social control influences, such as those studied by Brody and colleagues^26–30, may offer actionable targets for intervention. Here, we contend that for G–E studies of AUD, environmental conditions should be logically associated with changes in both exposure to alcohol and levels of drinking (see Figure 1). This could mean that an environmental feature increases risk for greater alcohol exposure and a resultant escalation of drinking, or it may be protective and decreases drinking. Such conditions can occur at the social network, community, and societal levels, but identification of plausible mediating mechanisms from higher-order (community or societal) factors to changes in alcohol consumption is needed. Social control mechanisms, compared to social triggers, may offer a strong theoretical link to changes in opportunities to use alcohol. Studies of adolescent psychiatric symptoms⁶³ and alcohol use⁶⁴, for example, point to intermediate social control influences (parental supervision and peer delinquency) that help explain the triggering effects of poverty and neighborhood socioeconomic disadvantage.

Ability to rule out alternative explanations

One end goal for this area of research is the identification and understanding of causal environmental influences on genetic effects as targets for prevention and intervention efforts. The selection of environmental features for G–E studies should consider fundamental conditions for causality including statistical association, time order of relationships, and whether there are other plausible explanations⁶⁵. Particularly relevant to this topic are gene-environment correlations as alternative explanations for a GxE effect⁵. A classic example is the correlation between genetic susceptibility for alcohol use and the selection into a deviant peer group⁵⁶. At the community level, a related phenomenon is the selection of people who drink heavily into more disadvantaged areas over time⁶⁶, which is likely to happen in addition to neighborhood causal effects on heavy drinking and AUD. To rule out alternative explanations in GxE studies, it is therefore advantageous to select environmental features that are exogenous to the individual and that exclude individual choice whenever possible.

Ability to predict phenotype variability

The selected environmental exposure must be predictive of phenotype variability⁶⁰. This point is straightforward, but should be extended to take into consideration the specific subgroup being studied. In order to examine the inter-relationships between genetic and environment effects, a predisposing or protective environment must be selected that varies in the targeted population group. For instance, while protective effects for religion are widely documented for African Americans^67–69, as well as Latinos/Hispanics and Whites⁷⁰, genetic studies of AUD that include highly-religious subgroups, such as rural and Southern African Americans or immigrant and first-generation Latinas, should carefully assess whether respondents have had enough exposure to alcohol to test G–E relationships. Certain religious affiliations (such as Southern/American Baptist, conservative Protestant, fundamentalist Christian, Mormon, Muslim, and others) are strongly associated with abstinence from alcohol⁷¹.

The consideration of variability can also be extended to provide an explanation for why GxE effects might be observed in one racial/ethnic group but not another. Social control and social trigger mechanisms, according to Dick and Kendler¹⁰ and Boardman et al.⁸, predict reduced genetic influences under low-risk environmental conditions and, conversely, increased genetic influences under more adverse conditions (illustrated in Figure 2, top panel). GxE effects are more likely observed in population groups when members are represented along the continuum of low- to high-risk environments, and less likely when members are overly represented at any single point along the continuum, whether more low-risk or adverse (illustrated in Figure 2, bottom left and right panels, respectively). Alike, selected genetic marker(s) must vary in the targeted population, with GxE effects being more likely observed when, in the case of biallelic markers, the frequencies of the common and minor alleles are more balanced. See Duncan and Keller⁶ for a more detailed illustration of the relationships between sample size, variance, and statistical power to detect GxE effects.

Figure 2.

Hypothetical fan-shaped GxE relationship for full distribution of environmental risk (top), compared to truncated distributions for low-risk (bottom left) and high-risk environments (bottom right)

Developmental timing of exposure

Developmental timing of environmental exposures is another important consideration for G–E studies, and evidence suggests that sensitive developmental periods may differ for U.S. racial/ethnic subgroups due to differences in alcohol use over the lifespan. One longitudinal study of a U.S. national sample⁷² showed an upward trend in heavy drinking by African Americans and, to a lesser extent, Latinos/Hispanics occurring in their 30s. This trend was accompanied by an increase in alcohol problems among African Americans starting in their mid-20s into their late 30s, which eventually exceeded those of Whites and Latinos/Hispanics. Thus, when assessing G–E relationships in groups with later onset of drinking, timing of the environmental exposure assessment should coincide with the periods in which the individuals may be starting to drink or escalating their alcohol use.

5. Special consideration of select environments for U.S. racial/ethnic minority groups

When we consider racial/ethnic minority groups, additional indicators of environmental social control and trigger mechanisms may be potential moderators of genetic influences. In this section we will consider selected environments for which there is empirical precedent, including evidence of strong main effects for greater risk of heavy drinking and AUD in racial/ethnic minority groups, or evidence that the environmental influence may be more important for racial/ethnic minority subgroups than for Whites. These may be strong candidates for study in G–E relationships. Additionally, social contexts that constrain drinking also are important to consider in analyses of G–E effects on AUD.

Social control

Social norms about drinking

Many studies suggest that social norms and attitudes about drinking have fairly uniform associations with alcohol use and problems across gender and race/ethnicity^73,74. When considering racial/ethnic minority groups, additional norm-related cultural elements may be relevant as environmental modifiers of genetic risk. Among Asian Americans, ethnic drinking culture tied to the country of origin, measured by per capita alcohol consumption, is associated with heavy drinking and drunkenness for certain subgroups such as Korean and Japanese adolescents and adults⁷⁵. Acculturation and birthplace also are predictors of heavy drinking and alcohol problems among some Latino/Hispanic⁷⁶ (particularly women) and Asian⁷⁷ subgroups living in the U.S. Among African Americans, elements of ethnic identity, or the affiliation and identification with people of common ancestry, language and religion in different social and civic contexts, also are associated with religiosity, drinking norms and alcohol consumption⁷⁸. Protective effects on AUD of ethnic and cultural identity (or enculturation) also have been documented for some American Indian groups⁷⁹.

At the neighborhood level, co-ethnic density may be a protective factor or a risk factor, depending on the specific racial/ethnic group in question. A higher density of African Americans may be a risk factor contributing to heavy drinking⁸⁰, but a higher density of Latinos/Hispanics may be a protective factor⁸¹. In addition to the social context, physical drinking contexts may contribute to heavy drinking, perhaps through influences on perceptions about behaviors that are normative in that setting. As examples, drinking at bars and parties may encourage drinkers to drink more heavily or more quickly than they might do during a quiet evening at home. Compared to Whites, Latinos/Hispanics and Blacks tend to drink in public more frequently and to consume higher volumes in these settings⁸². It may be useful to assess contexts where people drink to supplement information on social norms about drinking for G–E studies.

Alcohol availability

As alcohol availability increases, alcohol consumption also increases. Alcohol availability can be operationalized based on respondent-reported perceptions or using objective administrative data. Studies using objective measures suggest neighborhood alcohol outlet density is associated with greater alcohol consumption⁸³ and alcohol problems⁸⁴. Similarly, greater neighborhood drug availability can increase substance use⁸⁵, alcohol dependence⁸⁶, and drug abuse relapse⁸⁷. Some evidence suggests African Americans may be at even higher risk than other groups for alcohol-related health problems in areas with higher densities of off-premise alcohol outlets⁸⁴. In addition to increasing alcohol availability, a high density of alcohol outlets also may operate as a trigger for heavy drinking or as a stressor, as some evidence suggests a proliferation of alcohol outlets can increase violence in the surrounding area^84,88.

Other factors influencing alcohol availability are state licensure and tax structures, as well as alcohol advertising. Higher alcohol prices are associated with reductions in heavy drinking and alcohol problems, particularly among adolescents and young adults⁸⁹. Low-income youth may be especially sensitive to alcohol prices; this may manifest in stronger deterrent impacts of higher alcohol taxes on drinking of racial/ethnic minority youth and adults. Regarding advertising, evidence suggests that youth delay drinking and drink less frequently in areas with more restrictive policies about alcohol marketing⁹⁰, and some studies also have found greater exposure to alcohol advertising is associated with increased consumption, particularly by youth⁸³. Neighborhoods with a high density of African Americans (which often are economically disadvantaged) are selectively targeted for marketing of high-alcohol content beverages such as malt liquor beer and spirits^91,92. Indicators of local alcohol availability, including tax policies and advertisements for alcohol, are good candidates for studies of G–E relationships in racial/ethnic minority populations as they are likely to be exogenous exposures, particularly for youth and young adults.

Social triggers

Socioeconomic disadvantage

Socioeconomic disadvantage^93,94, long-term poverty^95,96 and cumulative disadvantage⁹⁷ from multiple sources are associated with heavy drinking and alcohol problems, and experiencing multiple and persistent forms of socioeconomic disadvantage across childhood and into adulthood is associated with heavy drinking and alcohol problems in adulthood⁹⁸. In the U.S., African Americans have much higher rates of poverty than Whites, and they also are more likely to experience long-term poverty⁹⁹. These socioeconomic risk factors for alcohol problems may be stronger for Blacks and Latinos/Hispanics than Whites^100,101.

Neighborhood poverty also can be a stressor that increases alcohol use and problems (or, as an alternative explanation, heavy drinkers may select into certain neighborhoods). In one longitudinal study of young adults, there was a strong effect of cumulative exposure to neighborhood poverty over 20 years on heavy alcohol use, but acute exposure to neighborhood poverty also increased heavy alcohol use¹⁰². African Americans are more likely than Whites to live in poor neighborhoods^103,104. Stronger associations of neighborhood disadvantage with alcohol use and alcohol problems have been noted for African Americans than for Whites or Latinos/Hispanics^1,101, with neighborhood disadvantage associated with less heavy drinking by White adult drinkers, but more heavy drinking by African American drinkers and marginally more heavy drinking by Latino/Hispanic drinkers¹. In addition to neighborhood socioeconomic disadvantage, neighborhood disorder is associated with many substance use outcomes¹⁰⁵, including heavy drinking^106,107. Thus, indicators of socioeconomic status encompass multiple levels of exposure across the lifecourse, including both acute and chronic disadvantage.

Racial discrimination and stigma

Racial discrimination is a stressor that contributes to adverse mental health and risk behaviors in adults and youth¹⁰⁸. Racial discrimination and perceived racial/ethnic stigma can increase heavy consumption, alcohol abuse, and drinking problems among Blacks and Latinos/Hispanics^109,110. Mulia et al.⁹⁷ and Chae et al.¹¹¹ examined the related constructs of perceived ethnic group stigma and unfair treatment, and found positive relationships for these variables in predicting problem drinking among Blacks, Latinos/Hispanics, and Asians. Among American Indians, discrimination may be associated with feelings of historical loss associated with long-term, intergenerational impacts of genocide and cultural “reformation” of American Indians in the U.S.; these stressors are associated with adverse alcohol outcomes including AUD^79,112.

Co-occurrence of environmental triggers

Socioeconomic status, racial discrimination, and neighborhood disadvantage often are closely related, with race/ethnicity impacting all three^113,114. The Mulia et al.⁹⁷ study detailed above found that experiencing multiple forms of social adversities (racial/ethnic stigma, unfair treatment, and poverty) increased risk for problem drinking across ethnic groups. Zemore et al.³ showed that adversities accumulate in their association with alcohol problems: Specifically, unfair treatment for Blacks and perceived ethnic stigma for Latinos/Hispanics each were associated with alcohol problems among those living below the poverty level but not above. At the neighborhood level, there also is co-occurrence with certain environmental constraints. Disadvantaged neighborhoods often contain an excess of alcohol outlets^115,116. We also note that low-income minority communities may be more likely to have more liquor stores, while low-income white communities may be more likely to have more bars¹¹⁷; thus, choosing the most appropriate measure(s) of alcohol availability is essential.

6. Conclusions and next steps

Through this survey of published, peer-reviewed studies, we sought to document the current state of the research on the joint effects of genetic and environmental influences on alcohol consumption and related phenotypes in U.S. racial/ethnic minority population groups. Although the reviewed studies made important contributions to science, we identified additional advances that are warranted. First, and most generally, there is a need for an overall increase in the inclusion of more diverse U.S. populations. Our focus here was on African Americans, Latinos/Hispanics, Asians, and American Indians. Studies are especially needed that include Latino/Hispanic and American Indian samples, as well as studies including adults from a range of age groups to coincide with key developmental periods for alcohol use initiation and escalation in different racial/ethnic groups. While important, we do recognize that it may not always be possible to compare differences in G–E relationships across racial/ethnic groups, because the frequency of alleles at specific markers, the associations between alleles, and the distribution of exposure to specific environmental conditions can be different by population group. Regardless, it is important to recognize and address the relevant ethical and cultural issues that will emerge in this effort to be more inclusive and to provide equal access to this area of research (a topic covered by other articles in this special issue). Second, methodological advancements will strengthen our confidence in study findings, including larger sample sizes, greater consistency in phenotypic measures to facilitate meta-analysis, and careful selection of genetic effects and rigorous testing of G–E relationships^5,50–52. Third, we focused on selection of environmental factors that have both theoretical and empirical precedent for inclusion in the study of G–E relationships. This revealed a relative dearth of studies that examined higher-order social control and trigger mechanisms, e.g., at the community- and societal-levels, which have broadly been shown to influence racial/ethnic differences in a variety of health outcomes. New investigations need to consider the challenge of how to more accurately capture and reflect the role of the environment in using a multi-level and longitudinal framework, as suggested by Boardman et al.⁸, and to conceptualize the mechanisms through which these influences shape GxE interactions^8,10–12. We proposed a number of community and societal variables (under both social control and triggering mechanisms) that are associated with alcohol use behaviors in U.S. racial/ethnic minority groups as potential environmental indicators to study alongside genetic effects. It will be an important next step, as has been recommended for phenotypic measures, that we test and identify a core of validated and standardized environmental measures for use in studies of G–E relationships.