Sex Differences in the Genetic Risk for Alcoholism

Carol A. Prescott, Ph.D.

One of the characteristics influencing a person’s risk for alcoholism is his or her sex, and various factors may contribute to sex differences in risk. Adoption studies have provided some evidence of possible sex differences in the heritability of alcoholism, but overall the findings have been inconclusive. Twin studies have consistently supported the role of genetic risk factors in the heritability of alcoholism in men, and shared environmental factors also play a role in the familiality of alcoholism among women. In addition, sex differences exist in the patterns of transmission of alcoholism between family members. However, the genetic epidemiology research conducted to date on this issue has several limitations, some of which may be resolved by future molecular genetic studies. KEY WORDS: gender differences; genetic linkage; hereditary factors; risk factors; twin study; adoption study; epidemiological indicators; etiology; molecular genetics; AOD (alcohol and other drug) use susceptibility; AOD dependence potential; comorbidity; alcoholic beverage

ne of the factors associated sources of genetic influences (i.e., sex- accounting for about half of the popu­ with a person’s risk for alco- specific transmission of genetic risk fac­ lation variation in liability to develop Oholism1 is his or her sex. In tors). Studies in many cultures have alcoholism (Prescott 2001). However, this article, the term “sex” is used in both found that the prevalence of alcoholism as will be described in this article, the its biological sense (i.e., as a variable and heavy drinking generally is higher evidence regarding the role of genetic based on genetic differences between among men than among women. Both cultural and biological explanations males and females) and its cultural CAROL A. PRESCOTT, PH.D., is an associ­ have been invoked to account for this meaning (i.e., in the sense of gender ate professor of psychiatry at the Virginia roles). Sex is associated with both bio­ difference, but the mechanisms remain Institute for Psychiatric and Behavioral logical risk factors (e.g., sex-specific unclear. Moreover, differences in preva­ Genetics and co-director of the Virginia hormone systems) and cultural risk fac­ lence may arise even if the mechanisms Adult Twin Study of Psychiatric and tors (e.g., social expectations about underlying alcoholism development do Substance Use Disorders, both at the how men and women use alcohol). An not differ between the sexes. In this Medical College of Virginia of Virginia understanding of the mechanisms case, the same genetic factors could Commonwealth University, Richmond, influencing sex differences in risk can predispose men and women to alco­ Virginia. help illuminate not only the differences holism, but other sex-specific genetic in men’s and women’s drinking behav­ and/or environmental factors could Preparation of this manuscript was sup- ior and related problems but also the influence whether alcoholism develops ported by Research Scientist Development biological and cultural bases for vari­ in a given person. Award K01–AA–00236 and National ability within each sex. The magnitude of genetic influ­ Institute on Alcohol Abuse and Sex differences in the factors under- ences on alcoholism risk may also be Alcoholism grant R01–AA–11408. lying the development of alcoholism sex specific. Evidence from twin and (i.e., its etiology) may be manifested as adoption studies of alcoholism in males 1Unless otherwise noted, the term “alcoholism” here refers differences in prevalence, in the magni­ has consistently supported the exis­ to all types of problematic alcohol use, including the tude of genetic influences, and in the tence of moderate genetic influences, clinical diagnoses of alcohol abuse and dependence.

264 Alcohol Research & Health Sex Differences in Genetic Risk

factors in alcoholism in women has four times higher risk of being alco­ referred to reviews by McGue (1994) varied across studies. Such sex differ­ holics themselves than did relatives of and Prescott (2001). ences in the magnitude of genetic nonalcoholics (Cotton 1979). Studies influence could arise from the interac­ of intact families cannot determine, Adoption Studies tions between genes associated with however, whether this familiality arises alcoholism risk and other physiological because family members share genes or Adoption studies allow researchers to processes. For example, numerous because they share environments. This separate the effects of genetic and envi­ physiological differences between men question must be resolved using genetic ronmental factors because adoptees and women in the rate of alcohol epidemiology studies—studies using receive their genetic heritage from one absorption and metabolism are likely genetically informative samples, such as set of parents and their rearing environ­ to be genetic in origin and may influ­ twins or adoptees and their biological ment from another set. Given some ence the development of alcoholism. and adoptive relatives. assumptions (e.g., absence of intrauterine Another potential manifestation of The goal of genetic epidemiology effects on risk for developing alcoholism sex differences in alcoholism risk is the research is to understand sources of and no correlation between the adoptive presence of sex-specific etiological fac­ variability in the risk of developing a and biological family environments for tors. In this case, the risk that a relative factors that influence alcoholism risk), the of an alcoholic will develop alcoholism disorder such as alcoholism. It is usually assumed that alcoholism is a multifacto­ degree to which adoptees resemble their is greater when the relative and the biological relatives is a direct measure alcoholic are of the same sex than when rial disorder in which a combination of genetic and environmental factors con- of genetic influence. Conversely, the they are of different sexes. Sex-specific degree to which adoptees resemble differences in etiology could be caused tributes to the risk of developing the disorder. People differ in their level of their adoptive relatives is a measure of by genes that exhibit different levels of the influence of family environment. activity in men and women or which risk for the disorder, and individual risk probably arises from three sources: Five adoption studies of alcoholism are modified by other sex-specific in males and four in females have genetic or environmental factors to included adequate sample sizes to eval­ make male and female relatives less • Additive genetic variation (also called heritability), which is based on num­ uate genetic influences on alcoholism. similar. These sex-specific etiological Three of the studies were conducted in erous genes that exist in multiple differences also could arise from cul­ Europe and used government registries variants (i.e., alleles), each of which tural processes—for example, whether to identify adoptees whose biological girls are more likely to model their may have a different effect. A person’s parents had records of alcoholism treat­ drinking behavior after their mothers specific combination of alleles of ment or alcohol-related offenses. Two and boys after their fathers. those genes determines his or her other studies were conducted in Iowa This article explores the various genetic risk. The role of such among adoptees for whom alcoholism mechanisms that may contribute to sex genetic influences is usually not status was determined in adulthood differences in the genetic risk for alco­ measured directly but is inferred by using psychiatric evaluations as well as holism. After reviewing genetic epidemi­ comparing degree of similarity in a public records. In these two studies, the ology studies of alcoholism that assess characteristic or behavior among adoptees were classified as “family history genetic versus environmental contribu­ different classes of relatives. negative” if neither biological parent tions to alcoholism risk, the article eval­ showed evidence of alcoholism and as uates various explanations for apparent • Shared environment, which includes “family history positive” if at least one sex differences in heritability, summa­ all environmental factors that make biological parent showed evidence of rizes evidence for sex differences in pat- members of the same family similar alcoholism. terns of transmission, and discusses the to one another (and different from In all five studies of male adoptees, limitations of such research. The article members of other families), includ­ participants with a positive family his- then briefly addresses molecular genetic ing family environment; social class; tory were at significantly higher (i.e., studies before examining potential appli­ schools; and, for twins, the intra­ 1.6 to 3.6 times greater) risk for alco­ cations of research on sex differences. uterine environment. holism than were participants with a negative family history, suggesting a • Specific environment, which includes substantial genetic contribution to the Genetic Epidemiology all remaining environmental factors risk for alcoholism. The studies of Studies of Alcoholism not shared by family members. female adoptees, in contrast, obtained mixed results, with only two of the More than 50 family studies have The following sections discuss the four studies finding a significantly shown that alcoholism runs in families. findings of adoption and twin studies increased risk of alcoholism among Those studies found that first-degree with respect to sex differences. For females with a positive family history relatives (e.g., parents, children, or sib- more detailed descriptions of individ­ compared with females with a negative lings) of treated alcoholics had a two to ual studies reviewed here, the reader is family history. These results provided

Vol. 26, No. 4, 2002 265 some evidence of possible sex differ­ Four other studies matched twin ducted in Australia found evidence that ences in heritability, but are inconclu­ registries against archival records (e.g., the risk factors in males and females are sive because of the small numbers of government records of alcohol offenses, the same. alcoholic female adoptees in the studies. military records, and medical records) to categorize twin pairs as to whether The Virginia Adult Twin Study of both twins, only one twin, or neither Psychiatric and Substance Use Twin Studies twin had records suggesting alcoholism. Disorders. The VATSPSUD is a Studies comparing the resemblance of In these studies, the prevalence of alco­ longitudinal study of adult twin pairs identical and fraternal twins also allow holism generally was low, suggesting conducted at Virginia Commonwealth researchers to estimate genetic and that the samples were selected for University. It was initiated in 1988 as environmental contributions to the risk severe cases. This means that alcoholics a study of female twin pairs by Ken­ for alcoholism. Identical twins share all who were not hospitalized would have neth Kendler and colleagues (Kendler of their genetic factors and are equally gone undetected, leading to underesti­ et al. 1992) and has been funded by exposed to the shared environmental mates of how many twin pairs were the National Institute of Mental factors. Conversely, fraternal twins (like similar for alcoholism. Two of the four Health, the National Institute on ordinary siblings) share on average only studies included female twins; however, Alcohol Abuse and Alcoholism, and one-half of their genetic variation but in these samples the prevalence of alco­ the National Institute on Drug Abuse. are also equally exposed to the shared holism was less than 1 percent overall. To date, the original study of female environmental factors. Consequently, if As a result, these studies have a very twins has completed four rounds of identical twins are more similar to each limited ability to distinguish the extent interviews with the twin pairs as well other in their risk for alcoholism than to which genetic and environmental as interviews with their parents. A are fraternal twins, one can assume that factors caused familial resemblance and parallel study of male and opposite- genetic factors contribute to the vulner­ whether sex differences existed. sex twin pairs, begun in 1993, to date ability to alcoholism, with the extent of Three studies used population-based has completed two waves of inter- the resemblance of identical twins indi­ twin registries and recruited partici­ views with both groups, and a third cating the extent of the contribution of pants regardless of whether they were wave of interviews with male twin genetic factors. However, if identical alcoholic or not. All participants were pairs is scheduled to be completed in twins are no more similar to each other then assessed for history of alcoholism 2003. All three groups of twin pairs in their risk for alcoholism than are fra­ through personal interviews. One of are white twins born between 1934 ternal twins, this indicates that alco­ these studies, the Virginia Adult Twin and 1975 who were drawn from a holism runs in families because of envi­ Study of Psychiatric and Substance Use registry of multiple births in the ronmental factors shared by family Disorders (VATSPSUD), includes Commonwealth of Virginia. Overall, members. The extent of dissimilarity of male, female, and male–female twin the study comprises more than 8,700 identical twins indicates the extent of pairs from Virginia and is described in people, including more than 3,750 the influence exerted by specific envi­ more detail in the following para- complete twin pairs. Comparisons ronmental factors and personal experi­ graphs. A second population-based with national and regional census ences on the risk for alcoholism. twin study, which also included male, data indicate that the twin samples The results of 11 major twin studies female, and opposite-sex pairs, is based are broadly representative of the white of alcoholism have varied with the meth­ on a volunteer registry in Australia. population of native-born Virginians ods used to identify participants. Four The third study included only male in this age range. studies identified affected people (i.e., twin pairs who served in the U.S. mili­ Kendler and colleagues (1992) pub­ probands) through treatment settings or tary during the Vietnam War era. lished the results on twin resemblance archival records and then assessed the All three of these studies found that, for alcoholism based on the first wave clinical status of their co-twins through among male twin pairs, identical twin of interviews with the female twin pairs individual followup. Three of these pairs were more similar for alcoholism in the VATSPSUD. Heritability esti­ four studies obtained evidence for signif­ than fraternal pairs, indicating a signifi­ mates ranged from 0.50 to 0.61 for icant genetic influences on alcoholism in cant genetic contribution to risk for several definitions of alcoholism. This men. In the three treatment-based alcoholism in males. The two studies means that 50 to 60 percent of the vari­ studies that included women, the evi­ that included female twin pairs yielded ation in risk for alcoholism results from dence for genetic influences was weaker, heritability estimates for women similar genetic factors. In 1999, Prescott and and some evidence suggested that envi­ to those for men. The one difference colleagues published an extension of ronmental factors also contributed to between these studies was for the this work that included the new sam­ the familial similarity for alcoholism. results based on the opposite-sex pairs. ples of male and opposite-sex pairs and However, because of relatively small Whereas the VATSPSUD study con­ an extensive re-evaluation of the female numbers of female twin pairs partici­ cluded that males and females have sample. Heritability estimates ranged pating, these studies had limited ability only partially overlapping genetic risk from 0.51 to 0.66 for all definitions of to detect sex differences. factors for alcoholism, the study con­ alcoholism studied and were of similar

266 Alcohol Research & Health Sex Differences in Genetic Risk

magnitude for males and females. Little for the variability in findings are dis­ patterns of clinical features. In sum­ evidence suggested that common envi­ cussed in the following sections. mary, although it seems likely that there ronmental factors contribute to twin are different etiologies that lead to the pair similarity for alcoholism. clinical manifestations of alcoholism, Etiological Heterogeneity researchers do not yet have strong evi­ Summary of Twin Study Findings. Another possible explanation for sex dence that these etiologies are associ­ In summary, twin studies of males differences in genetic influences on ated with genetic differences between have consistently found significantly alcoholism is etiological heterogene­ subtypes. Identification of genetically more resemblance in identical twins ity—differences among subgroups distinct subtypes, and an understand­ with respect to alcoholism than in (e.g., men and women) in the genetic ing of the role of sex in their etiology, fraternal twins, implying an impor­ factors that influence the development may require the identification of specific tant role for genetic factors in the of alcoholism. Most current theories genes that predispose to alcoholism. vulnerability to alcoholism. The heri­ assume that multiple genes contribute tability estimates in most studies to the risk for alcoholism, and different suggest that genetic variability accounts people have different sets of predispos­ Ascertainment Method for 50 to 60 percent of the variation ing genes. However, the existence of A third possible explanation for appar­ in risk for alcoholism in males. Studies heterogeneity is difficult to detect using ent sex differences lies in the fact that of female twins, in contrast, are less genetic epidemiology studies because the means by which cases were identi­ consistent. Studies based on samples the predisposing genes are not yet fied (i.e., case ascertainment) varied identified through treatment settings known. Therefore, researchers cannot from study to study. As noted above, suggest that familiality of alcoholism study those genes directly but must the four twin studies that reported among women can be attributed to infer the presence of heterogeneity lower heritability of alcoholism in shared environments rather than (or when heritability differs in subgroups. women than in men all used treatment- in addition) to genetic factors. Studies For example, two twin studies of ascertained samples. Studies using such using population-based twin registries males found evidence for greater heri­ treatment samples or archival data of have found the genetic influences on tability among males with early-onset alcoholism-related records are likely to alcoholism to be of similar magnitude alcoholism than among males with be nonrepresentative in several ways. in males and females. late-onset alcoholism (McGue et al. For example, such studies may include 1992; Caldwell and Gottesman 1991). cases that are characterized by greater The researchers theorized that, of the severity or comorbidity (that is, co­ Explanations for Apparent two types, late-onset alcoholism was occurrence with other disorders) than Sex Differences in clinically and etiologically more similar alcoholism in the general population Heritability to alcoholism in women. However, because patients with these characteristics preliminary analyses from the Virginia may be most likely to seek treatment or The results from twin and adoption and Australian twin samples did not be included in other alcoholism-related studies fairly consistently support a support this distinction. Furthermore, records. This selectivity may be particu­ moderate to strong familiality of alco­ even if early-onset alcoholism was larly severe in women, because treatment- holism in both men and women. For associated with greater heritability and, based twin studies have reported that men, most of this familiality can be consequently, with an increased risk of more than 60 percent of alcoholic attributed to genetic factors, whereas alcoholism in relatives compared with women had a comorbid disorder, a the proposed sources of the familiality late-onset alcoholism, this would not much higher proportion than among among women vary across studies. One necessarily indicate distinct etiologies men in the same samples. Common possible explanation for this apparent for the two types of alcoholism. comorbid disorders include depression, sex difference in the sources of familial­ Rather, early onset of the disorder other drug abuse and dependence, and ity is that the studies have limited abil­ could reflect greater liability (e.g., a antisocial personality disorder. ity to distinguish between genetic and higher number of risk factors) in the If a person’s decision to enter treat­ shared environmental effects because of affected people, which also would be ment is not simply dependent on his their small sample sizes. Even within expected to be associated with an or her genetic vulnerability to alco­ studies, different heritability estimates increased risk in relatives. holism, drawing study participants can be obtained depending on the Other attempts to find etiologically from among alcoholics in treatment diagnostic definition used (e.g., alcohol distinct subtypes of alcoholism have could result in an unrepresentative dependence versus alcohol abuse). Con­ used patterns of clinical symptoms sample and yield incorrect estimates sistent with this explanation, Heath and (e.g., Babor et al. 1992). However, of genetic influences. Results from colleagues’ reanalyses (1998) of these typologies tested in genetically infor­ two U.S. twin studies suggest that studies suggest that some of the reported mative samples suggested that such treatment seeking is not simply deter- sex differences are not statistically sig­ “types” are distinguished primarily by mined by a person’s liability to have nificant. Other possible explanations disorder severity rather than by distinct alcoholism (True et al. 1996). For

Vol. 26, No. 4, 2002 267 example, having comorbid depression ferences in either severity or etiology was designed specifically to test for sex or having family members who have between alcoholic women who had differences and includes more than received treatment could influence a been in treatment and alcoholic women 1,400 pairs of opposite-sex twins, still person’s decision to seek treatment. in the general population. Across all has limited power for some questions. Consequently, the use of a treatment- studies, relatives of female alcoholics It is unlikely that future studies using ascertainment design may result in had an approximately 40 percent greater twin or adoption designs will have increased evidence for shared environ­ risk of alcoholism than relatives of male greater sample sizes. To address this lim­ mental effects among both male and alcoholics. itation, researchers may have to await female twins and could be one expla­ The second possible explanation for the identification of the specific genes nation for why studies of female twins sex differences is the existence of sex- and physiological mechanisms involved using this design obtain lower heri­ specific transmission of risk factors— in the development of alcoholism. tability estimates than studies using that is, that family members are at These processes could then be studied population-based samples (Prescott greater risk of developing alcoholism if to determine if they differ between men and Kendler 2000b). they are the same sex as the alcoholic and women and if this explains the dif­ relative. Researchers conducting the ferences observed clinically. previously mentioned VATSPSUD Analyses of sex differences are also Evidence for Sex Differences project found evidence consistent with limited by the fact that estimates of in Patterns of Transmission such a sex-specific transmission. Their genetic and environmental effects data suggested that the transmission of assume that all forms of alcoholism Several of the key questions about sex risk factors across sexes was only half as have the same etiology. However, dif­ differences in the etiology of alcoholism strong as within-sex transmission (Pres­ ferent genetic factors may lead to the can be addressed only by comparing cott et al. 1999). The evidence regarding development of different subtypes of the resemblance in male–female rela­ sex-specific transmission in other studies alcoholism. As described previously, tive pairs with that in same-sex relative is inconclusive, however, because most some researchers have addressed this pairs. Four twin studies, two adoption studies include too few alcoholic issue by investigating whether heritabil­ studies, and seven family studies have women to evaluate this hypothesis ity differs for alcoholism subtypes that performed systematic diagnostic assess­ thoroughly. are characterized by a specific age of ments and have included information Overall, the evidence from twin and onset or certain clinical features. The on resemblance for all types of pairs adoption studies suggests that the dif­ results of these analyses, however, have (i.e., male–male, female–female, and ferences in transmission patterns result been equivocal. male–female). The results of these from genetic rather than cultural trans- As mentioned before, other limita­ studies can be used to analyze mecha­ mission effects between family mem­ tions are associated with the method of nisms underlying sex differences. bers. Such genetic transmission effects ascertaining cases. Thus, studies using Prescott (2001) conducted a meta­ could be caused by differences in the subjects undergoing alcoholism treat­ analysis of the data from these studies activity of genes that are located on the ment may include only a small fraction to test two alternative explanations for sex chromosomes (i.e., the X and Y of alcoholics residing in the commu­ the development of sex differences. chromosomes) or in the activities of nity. This selectivity can be a problem The first explanation posits the exis­ other genes that are triggered by bio­ particularly if sex differences exist in tence of sex-specific thresholds—that logical differences between the sexes the detection of alcoholism (i.e., in is, that women on average need to cross (e.g., hormone levels). how researchers and treatment providers a higher threshold (i.e., have a higher identify female and male alcoholics) or level) of inherited risk than men to in treatment entry of alcoholics. Studies develop alcoholism. This hypothesis Limitations of Research on using clinical interviews in community would account for the lower frequency Sex Differences in samples identify a larger proportion of of alcoholism in women than in men. Alcoholism alcoholics than do treatment studies. Moreover, the hypothesis predicts that However, one can argue that the defi­ relatives of female alcoholics are at The ability to detect sex differences in nition of alcoholism in these interviews higher risk of developing alcoholism the etiology of alcoholism is limited by is too broad because it includes many than are the relatives of male alcoholics. several methodological problems. people who experience a single tran­ Data from family, adoption, and twin Probably the most serious problem sient episode of alcohol-related prob­ studies based on treatment samples all arises from the small numbers of lems and recover on their own. show evidence of a higher risk associ­ affected women in most studies. It is Another limitation is that studies ated with having an alcoholic female difficult to overcome this limitation often include relatives who are in their relative than with having an alcoholic because twin and adoption studies have twenties or thirties and still at risk for male relative. Twin studies based on already attempted to identify all affected developing alcoholism. Therefore, the population samples showed less in- people in entire countries or regions. results regarding similarities among rela­ crease in risk, possibly because of dif­ Even the VATSPSUD project, which tives might change if currently unaffected

268 Alcohol Research & Health Sex Differences in Genetic Risk

people were assessed at a later time. Studies likely to carry a gene that is associated standing of the causes of comorbidity examining different degrees in similarity with increased activity of the neuro­ may help researchers uncover the basis in different age groups typically have not transmitter serotonin in the brain (e.g., for sex differences in the genetic risk found that older relative pairs are more Preuss et al. 2001). Because this alco­ for alcoholism. similar, as would be expected if under- holism subtype is found mostly in men There are several explanations for diagnosis in younger age groups were a and only rarely in women, this could why two disorders may co-occur. One problem. However, the statistical power provide one explanation for why the explanation—known as phenotypic of these tests is limited. sexes differ in genetic risk factors for association—suggests that the presence Finally, most of the information alcoholism. of one disorder (e.g., alcoholism) available concerns only the etiology of increases the risk for another disorder alcoholism in white populations. Only (e.g., depression) because of physiologi­ one sample (Caldwell and Gottesman Applications of Research cal processes or as a reaction to the psy­ 1991) has included more than 10 on Sex Differences chosocial problems that accompany percent nonwhites. The heritability alcoholism (see figure 1A). (The reverse estimates from this study differed by One explanation for the existence of is also possible—that is, alcoholism can ethnicity but the differences did not sex differences in risk for alcoholism is result from depression because a person reach statistical significance because of that males and females differ on the may use alcohol in an attempt to deal the small sample sizes of different eth­ degree to which they experience the with the depression.) A second expla­ nic groups. various risk factors for alcoholism. This nation for comorbidity proposes the Despite all these limitations, how- means that sex is not a risk factor in presence of correlated causes (see figure ever, it is clear that sex is an important itself but is merely correlated with 1B). According to this model, alco­ factor to consider in the study of the some of the processes underlying alco­ holism and depression may occur etiology of alcoholism. holism. For example, the sexes may dif­ together because the risk factors for fer in the mechanisms underlying the both disorders are the same (i.e., are co-occurrence of alcoholism with other correlated). Molecular Genetic Studies disorders, and they may differ in fac­ Depression provides an example of tors that moderate or mediate genetic how genetic factors may be associated The goal of molecular genetic studies influences on alcoholism. These issues with sex differences in risk for a disorder. is to identify specific genes involved in are being addressed by several recent During adolescence through mid-adult- the development of alcoholism. Re- twin and adoption studies, as discussed hood, females have much higher rates of searchers have already conducted sev­ in the following sections. depression than males, whereas these eral large-scale genetic family studies differences are much less pronounced (e.g., Reich et al. 1998), and other Genetic Influences on the during other parts of the life span. This studies are under way to search the Occurrence of Comorbid observation, combined with results from entire human genetic material (i.e., the Disorders experimental research with animal mod­ genome) for genes that increase the risk els of depression, suggests that the level for alcoholism. Many other studies Alcohol use disorders are frequently of circulating estrogen (which is geneti­ have analyzed candidate genes— accompanied by other emotional disor­ cally influenced) plays a role in the regions known to influence alcohol ders (e.g., depression and anxiety) and higher incidence of depression in metabolism (e.g., genes encoding the other drug use disorders. Comorbid females (Halbreich and Kahn 2001). enzymes that break down alcohol) or disorders have been estimated to occur The causes of comorbidity between known to be involved in brain activity in as many as 80 percent of clinical alcoholism and depression have been (e.g., genes coding for various brain populations and 50 percent of commu­ investigated extensively using family chemicals [i.e., neurotransmitters])— nity samples (Kessler et al. 1997). The studies. The usual strategy in these for their possible association with alco­ etiology of this comorbidity is of theo­ studies is to identify probands affected holism in clinical populations. Thus retical and clinical importance because by one or both disorders and examine far, few reports suggest significant sex these coexisting disorders complicate the prevalences of depression and alco­ differences in the roles of specific genes. treatment and may alter the client’s holism alone or in combination among However, most studies have included prognosis. Male and female alcoholics the proband’s biological relatives. These relatively few alcoholic women, limit­ differ in their patterns of comorbidity, prevalences are then compared for dif­ ing the ability to test for sex effects. with women having higher rates of ferent types of probands (e.g., depressed, Some evidence from molecular comorbid anxiety and affective disor­ alcoholic, and depressed plus alcoholic genetic studies suggests that there may ders and males having higher rates of probands as well as unaffected control be etiologically distinct subtypes of comorbid abuse of other drugs, con- subjects) to determine their fit with alcoholism. For example, people who duct disorder, and antisocial personality possible models of comorbidity. For have alcoholism characterized by aggres­ disorder (Kessler et al. 1997). Because example, if alcoholism increases the sivity or antisocial personality are more of these differences, a better under- risk of depression as stated by the

Vol. 26, No. 4, 2002 269 for alcoholism only in the depressed person but not among nondepressed A B family members, similar to the model in figure 2A but with depression lead­ ing to alcoholism rather than vice versa. Other studies found increased rates of alcoholism among relatives of depressed probands; in some cases, however, this was true only for relatives of female probands. The family study literature has several limitations. One limitation is that probands recruited in treatment settings are likely to have more severe alcoholism and greater comorbidity and may have different underlying causes than probands recruited from the general population. In fact, stud­ Figure 1 Schematic models of possible mechanisms underlying comorbidity ies based on U.S. population samples between alcoholism and other psychiatric disorders. Circles indicate suggest that depression and alcohol- unobserved risk. Boxes indicate observed clinical outcomes. (A) The ism have correlated causes (Prescott model of phenotypic association posits that the presence of one disorder and Kendler 2000a). increases the risk of acquiring the second disorder. For example, a Another limitation is that family person with alcoholism could be at increased risk of developing depres­ studies cannot distinguish whether sion. (The reverse scenario—that a person with depression is at increased risk of alcoholism—is also possible but is not shown in this vulnerability is transmitted through figure.) (B) According to the model of correlated causes, the same risk genetic or environmental pathways. factors contribute to development of both disorders. This issue can be addressed using twin and adoption studies; however, the results of such studies have been equiv­ ocal. Of four adoption studies, two phenotypic association model, only methodological features (for a review, found weak evidence of increased risk factors for alcoholism would be see Swendsen and Merikangas 2000). In depression among adopted-away chil­ transmitted across family members, general, studies using alcoholic probands dren of alcoholic parents, one found no and these risk factors would only indi­ found that relatives were not at increased increased depression, and another rectly influence the risk for depression risk for depression unless the proband found increased alcoholism among bio­ (see figure 2A). In this case, if a proband also had depression. Moreover, studies of logical relatives of depressed adoptees. had both alcoholism and depression, probands treated for alcoholism with or Goodwin and colleagues (1977) a relative of the proband would be without comorbid depression found reported increased depression in the expected to be at increased risk for that the relatives of probands with both daughters of alcoholics only when they depression (compared with the relatives alcoholism and depression were not at were raised by alcoholic adoptive par­ of people without alcoholism or depres­ increased risk for alcoholism compared ents, suggesting a possible interaction sion) only if he or she also had alcoholism. with relatives of probands who had alco­ between genetic and environmental Conversely, according to the correlated holism alone. These findings are consis­ factors. Results from three twin studies causes model, the familial risk would be tent with a phenotypic association based on clinical samples suggest that nonspecific and could increase the risk between the two disorders (see figure depression and alcoholism are pheno­ for either alcoholism or depression (see 2A) where the presence of alcoholism typically associated. However, a recent figure 2B). In this case, if a proband had increases the risk for depression. molecular genetic study identified a both alcoholism and depression, a rela­ Family studies based on probands region on chromosome 1 that was asso­ tive would be at increased risk of either with primary depression (i.e., people in ciated with an increased risk for both depression or alcoholism. whom alcoholism is not present or is alcoholism and depression, consistent Although numerous studies of this considered the result of depression) with the correlated causes model in issue have been conducted, it is difficult have yielded mixed findings. Some which the same gene(s) contribute to to generalize about their results because studies found no increase in alcoholism both disorders (Nurnberger et al. 2001). of variation in the types of probands among relatives unless the probands Researchers in the VATSPSUD studied, in the method of data collec­ were also alcoholic. This finding is con­ project also addressed these issues by tion (e.g., by personal interview or from sistent with the phenotypic associa­ studying the causes of comorbidity of a family informant), and in other tion—that is, depression increased risk major depression and alcoholism (Pre-

270 Alcohol Research & Health Sex Differences in Genetic Risk

scott and Kendler 2000a). Their analy­ ses found that people with major depression were at significantly increased risk for alcoholism.2 Moreover, a his- A tory of major depression in one twin increased the co-twin’s risk for alco­ holism to a significant degree among identical twins, to a lesser degree among same-sex fraternal twins, and not at all among opposite-sex twins. The results of the statistical analyses were most consistent with the corre­ lated causes model, which posits that comorbidity occurs because the genetic and specific environmental sources of liability for depression overlap with those for alcoholism. This overlap was statistically significant only within a sex but not across sexes, indicating that the comorbidity appears to result from sex-specific genetic and environmental B risk factors. Unlike previous hypo- theses, these results suggest that the factors underlying depression in women appear to differ from the factors under- lying alcoholism in men. It is important to note that these findings do not point to one defini­ tive mechanism underlying comor­ bidity; in fact, they are what would be expected if all three processes were operating in different subsets of the population. It seems reasonable that for some people, depression is a conse­ quence of alcoholism; for others, alcoholism is caused by depression; and for still others, both disorders result from a shared, familially trans­ Figure 2 Schematic models of mechanisms underlying comorbidity between mitted vulnerability to both disorders. alcoholism and other psychiatric disorders among relatives. (A) According to the phenotypic association model, alcoholism develops because of a combination of individual and familial risk factors; moreover, alcoholism Factors That Modify Genetic increases the risk of depression. In this scenario, relatives of an alcoholic Liability (relative #1) are at increased risk for depression only if the relatives also Another promising application of have alcoholism. (B) Under the correlated causes model, individual and genetic epidemiology studies, including familial risk factors contribute to a general risk of developing alcoholism or studies of sex differences, is the identifi­ depression. In this case, relatives of an alcoholic would be at increased cation of factors that may influence the risk for either alcoholism or depression. expression of genetic liability for alco­ holism. Researchers are studying two types of influences: modifiers and alcoholism. For example, if people can Investigators have conducted exten­ mediators (discussed in the following be identified as being at risk for alco­ sive analyses among adoptees and their section). Modifiers are genetic or envi­ holism based on their genetic back- families in Iowa to identify environmental ronmental factors that interact with grounds (e.g., because they carry genes genetic risk factors to increase or de- known to increase risk or because they 2In this study, alcoholism was defined as the presence of crease the risk for alcoholism. Research have alcoholic family members), alcohol abuse and/or alcohol dependence as specified in the American Psychiatric Association’s Diagnostic and into these modifying factors has impor­ knowledge of modifying factors would Statistical Manual of Mental Disorders, Fourth Edition tant implications for the prevention of permit early interventions. (DSM–IV) (American Psychiatric Association [APA] 1994).

Vol. 26, No. 4, 2002 271 factors that may increase the risk for As with the studies of depression, the Recent research has changed from alcoholism, other drug abuse, and anti- results are generally consistent with the an emphasis on estimating the extent social personality disorder. These analy­ hypothesis that the same genetic risk of the heritability of alcoholism to ses found that having an adoptive fam­ factors contribute to the risk for alco­ studying the factors that influence or ily member with alcohol problems holism and the other disorders. can serve as markers of genetic liability. increased the risk for alcohol abuse However, the findings do not indicate By identifying specific genes and fac­ among adopted children whose biologi­ that these disorders directly mediate tors that increase or decrease risk in cal parents were alcoholics (Cadoret et the risk for alcoholism. people with a genetic risk for alco­ al. 1985). This was particularly true for Another approach is to use geneti­ holism, researchers may gain a better adoptees raised in rural communities, cally informative samples to identify understanding of the mechanisms suggesting shared environmental factors characteristics that are more closely through which genetic vulnerability may be less influential in more diverse, related to the biology of alcoholism leads to the actual development of urban settings. Additional analyses sug­ and may therefore be markers of the alcoholism. After specific genes are gested that environmental factors act genetic risk for alcoholism. Such char­ identified, this information can be differently in males and females to acteristics, also known as endopheno­ incorporated into epidemiologic stud­ influence the risk for alcoholism and types, include certain brain waves (i.e., ies to determine whether these genes related disorders (e.g., Cadoret et al. event-related potentials), a person’s account for the sex differences observed 1999). For example, in the presence of subjective response to alcohol, the in the prevalence of alcoholism and in environmental stresses (e.g., divorce of degree of impulsivity, and physiological the patterns of comorbidity. ■ adoptive parents or the presence of a reactivity (see Schuckit 2000). Sex psychiatric disorder in an adoptive par­ differences in these factors may provide ent), only women with a genetic liabil­ a clue to how the sexes differ in the References ity for alcoholism were at increased risk etiology of alcoholism. For example, American Psychiatric Association (APA). Diagnostic for major depression. patterns of event-related potentials in Hormonal status also may be an and Statistical Manual of Mental Disorders, Fourth adolescent boys (but not adolescent Edition. Washington, DC: APA, 1994. important modifier of the genetic risk girls) who are at high risk for alco­ for alcohol involvement, because re- BABOR, T.F.; HOFMANN, M.; DELBOCA, F.K.; ET holism because they have alcoholic rela­ AL searchers found that early alcohol use . Types of alcoholics, I. Evidence for an empiri­ tives differed from those of adolescents cally derived typology based on indicators of vul­ and level of alcohol consumption among who are not at high risk (Begleiter and nerability and severity. Archives of General female adolescents were associated with Psychiatry 49:599–608, 1992. the age at first menstruation (i.e., men­ Porjesz 1999). BEGLEITER, H. AND PORJESZ, B. What is inherited arche) (Dick et al. 2000). Menarche in the predisposition toward alcoholism? A pro- could represent either a biological or an Conclusions and Future posed model. Alcoholism: Clinical and Experi­ environmental modifier, because in Directions mental Research 23:1125–1135, 1999. addition to experiencing hormonal CADORET, R.J.; O’GORMAN, T.W.; TROUGHTON, changes, girls who mature early may E.; AND HEYWOOD, E. Alcoholism and antisocial have more contact with older peers and Twin and adoption studies have estab­ personality: Interrelationships, genetic and envi­ greater access to alcohol. lished an important role for genetic ronmental factors. Archives of General Psychiatry influences in the etiology of alcoholism 42:161–167, 1985.

in men. The evidence for genetic influ­ CADORET, R.; RIGGINS-CASPERS, K.; YATES, W.R.; Processes That Mediate Genetic Risk ence on alcoholism development in ET AL. Gender effects in gene-environmental When talking about genetic risk fac­ women is less consistent, but interpre­ interactions in substance abuse. In: Frank, E., ed. tation of the literature is complicated Gender and Its Effects on Psychopathology. Wash­ tors, it is important to realize that genes ington, DC: American Psychiatric Press, 1999. pp only carry information that dictates the by several methodological issues (e.g., 257–280. synthesis and structure of proteins. small sample sizes). More recent studies suggest a similar degree of genetic in­ CALDWELL, C.B., AND GOTTESMAN, I.I. Sex Proteins do not directly influence or differences in the risk for alcoholism: A twin control the behaviors that result in fluence for men and women. Some re- study. Behavior Genetics 21:563, 1991. alcoholism. Accordingly, much research search suggests that sex differences exist in the sets of genes that influence alco­ COTTON, N.S. The familial incidence of alco­ currently is focused on identifying the holism: A review. Journal of Studies on Alcohol processes that mediate between genes holism risk or perhaps in the interactions 40:89–116, 1979. and alcohol-related behaviors. of genes contributing to alcoholism risk Family, twin, and adoption studies with other genetic or cultural factors. DICK, D.M.; ROSE, R.J.; VIKEN, R.J.; AND KAPRIO, J. Pubertal timing and substance use: have been used to study whether familial The study of sex differences in factors Associations between and within families across resemblance for personality characteris­ that mediate and modify genetic risk late adolescence. Developmental Psychology 36: tics or certain disorders (e.g., anxiety, may provide clues that will aid the 180–189, 2000. conduct, and other drug use disorders) development of targeted prevention GOODWIN, D.W.; SCHULSINGER, F.; KNOP, J.; ET may mediate resemblance for alcoholism. and intervention efforts. AL. Alcoholism and depression in adopted-out

272 Alcohol Research & Health Sex Differences in Genetic Risk

daughters of alcoholics. Archives of General Psy­ MCGUE, M.; PICKENS, R.W.; AND SVIKIS, D.S. liability to alcohol abuse and dependence in a chiatry 34:751–755, 1977. Sex and age effects on the inheritance of alcohol population-based sample of U.S. twins. Alcoholism: problems: A twin study. Journal of Abnormal Clinical and Experimental Research 23:1136– HALBREICH, U., AND KAHN, L.S. Role of estrogen Psychology 101:3–17, 1992. 1144, 1999. in the aetiology and treatment of mood disorders. CNS Drugs 15:797–817, 2001. NURNBERGER, J.I.; FOROUD, T.; FLURY, L.; ET AL. PREUSS, U.W.; KOLLER, G.; SOYKA, M.; AND Evidence for a locus on chromosome 1 that in­ BONDY, B. Association between suicide attempts HEATH, A.C.; SLUTSKE, W.; AND MADDEN, fluences vulnerability to alcoholism and affective and 5–HTTLPR–S allele in alcohol-dependent P.A.F. Gender differences in the genetic contribu­ disorder. American Journal of Psychiatry 158:718- and control subjects: Further evidence from a tion to alcoholism risk and alcohol consumption 724, 2001. German alcohol-dependent inpatient sample. patterns. In: Wilsnack, R.W., and Wilsnack, Biological Psychiatry 50:636–639, 2001. S.C., eds. Gender and Alcohol: Individual and PRESCOTT, C.A. The genetic epidemiology of Social Perspectives. Rutgers, NJ: Rutgers alcoholism: Sex differences and future directions. REICH, T.; EDENBERG, H.J.; GOATE, A.; ET AL. University Press, 1998. pp. 114–149. In: Agarwal, D.P., and Seitz, H.K., eds. Alcohol in Genome-wide search for genes affecting risk for Health and Disease. New York: Marcel Dekker, alcohol dependence. American Journal of Medical KENDLER, K.S.; HEATH, A.C.; NEALE, M.C.; ET 2001. pp. 125–149. Genetics (Neuropsychiatric Genetics) 81:206–215, AL. A population-based twin study of alcoholism 1998. in women. JAMA: Journal of the American PRESCOTT, C.A., AND KENDLER, K.S. Sex-specific Medical Association 268:1877–1882, 1992. genetic influences on the comorbidity of alco­ SCHUCKIT, M.A. Genetics of the risk for alco­ holism and major depression in a population- holism. American Journal of Addictions 9:103– KESSLER, R.C.; CRUM, R.M.; WARNER, L.A.; ET based sample of U.S. twins. Archives of General 112, 2000. AL. Lifetime co-occurrence of DSM–III–R alcohol Psychiatry 57:803–811, 2000a. abuse and dependence with other psychiatric SWENDSEN, J.D., AND MERIKANGAS, K.R. The disorders in the National Comorbidity Survey. PRESCOTT, C.A., AND KENDLER, K.S. The influ­ comorbidity of depression and substance use Archives of General Psychiatry 54:313–320, 1997. ence of ascertainment strategy on finding sex disorders. Clinical Psychology Reviews 20:173– differences in genetic influences from twin studies 189, 2000. MCGUE, M. Genes, environment and the etiology of of alcoholism. Neuropsychiatric Genetics 96:754– alcoholism. In: Costello, C.G., ed. The Development 761, 2000b. TRUE, W.R.; HEATH, A.C.; BUCHOLZ, K.; ET AL. of Alcohol Problems: Exploring the Biopsychosocial Models of treatment seeking for alcoholism: The Matrix of Risk. Washington, DC: National Institute PRESCOTT, C.A.; AGGEN, S.H.; AND KENDLER, role of genes and environment. Alcoholism: Clinical on Alcohol Abuse and Alcoholism, 1994. pp. 1–40. K.S. Sex differences in the sources of genetic and Experimental Research 20:1577–1581, 1996.

New Brochures Available

Harmful Interactions: Mixing Alcohol with Medicines A Family History Describes the risks of using alcohol while of Alcoholism taking medications or Are You at Risk? herbal preparations and lists common medicines and their Offers easy-to-read possible reactions with information for anyone alcohol. who is concerned about a family history of alcoholism and lists organizations that can help the relatives of alcoholics.

To order, write to: National Institute on Alcohol Abuse and Alcoholism, Publications Distribution Center, P.O. Box 10686, Rockville, MD 20849–0686. Or fax: (703) 312–5230. Full text of these brochures is also available on the World Wide Web at http://www.niaaa.nih.gov

Vol. 26, No. 4, 2002 273