Using the brain P300 response to identify novel phenotypes reflecting genetic vulnerability for adolescent substance misuse

Abstract

We used a novel approach to identify candidate alternative phenotypes for investigating genetic influence underlying substance use disorders (SUDs) in adolescents. The existing literature suggests that P300 amplitude reduction (P3-AR) observed in brain event-related potentials is associated with risk for SUDs generally, not just alcoholism. Using data from a community-based sample of 17-year-old male and female twins, we fit biometric models to P3 amplitude data to show that it is strongly heritable, especially in boys. The extant evidence coupled with our findings strongly supports treating P3-AR as an endophenotype indexing SUD risk. We then examined a set of 15 potential alternative phenotypes (e.g., frequent use of cannabis) to determine whether they were associated with P3-AR. The results indicated that almost all of these alternative phenotypes were associated with P3-AR, with larger effect sizes observed for boys. Given the strong association of these use phenotypes with P3-AR, which is itself an index of genetic risk for SUDs, we conclude that these use phenotypes may provide tools for finding vulnerability genes in adolescents who have yet to pass through the age of risk for SUDs.

Introduction

Studies using genetic epidemiological designs suggest that there is substantial genetic influence for various substance use disorders (SUDs), including alcohol dependence (see McGue, 1999, for a review), nicotine dependence (Heath & Madden, 1995, Kendler et al., 1999), cannabis abuse and dependence (Kendler et al., 2002, Kendler & Prescott, 1998, Lynskey et al., 2002) and problems with other illicit substances (McGue et al., 2000, Tsuang et al., 1996). To date, however, progress in finding genes for these disorders has been slow. In general, failures to replicate well-publicized claims of linkage between genes and psychiatric disorders may have tempered enthusiasm for such endeavors (Risch & Merikangas, 1996) and certainly pose a challenge for the field of psychiatric genetics. Diagnostic systems such as the DSM-IV (American Psychiatric Association, 1994) provide clinicians with a shared language and clearly capture important functional consequences of substance disorders. They may nevertheless not be optimal for finding genes for SUDs, which are likely to be complex, multifactorial, and heterogeneous. Indeed, the DSM was not developed for this purpose. The search for susceptibility genes may thus profit from an approach which complements DSM diagnoses with other approaches to identifying genetic vulnerability.

In this report, we note the evidence supporting reduced amplitude of the P3 brain event-related potential as a marker of genetic risk for substance use disorders generally. We then examine the heritability of this endophenotype in a general population sample of adolescent twins, extending our earlier work with adolescent boys by examining a large community sample composed of both genders. Using P3 amplitude as a proxy measure of genetic risk, we examined its association with various measures of adolescent substance misuse to determine if, by virtue of their association with P3 amplitude, these misuse measures may also tap genetic risk for substance abuse.

Phenotypes that are related to a DSM disorder and that may be heritable themselves can aid in identifying genetic risk. Examples relevant to substance use might include phenotypes associated with severity or course of substance use (such as heavy use or early onset), specific symptoms of use (such as marked tolerance for a drug), or personality traits associated with the disorder (such as novelty seeking or behavioral lack of control) (Iacono, Malone, & McGue, 2003).

Such phenotypes are likely to be particularly salient for research on adolescents, the focus of the present investigation. Standard diagnostic criteria may not be appropriate for adolescents (Chung & Martin, 2002, Clark, 2004, Flory et al., 2004, Martin & Winters, 1998). In particular, criteria derived primarily from adult research may yield false negatives or ‘diagnostic orphans,’ because features such as tolerance or impaired control may show differential age-related manifestations (Clark, 2004). Initiation of substance use is greater during adolescence than at any other time during development (Johnston, O'Malley, & Bachman, 2003), but many adolescents have not yet developed manifest psychopathology. Phenotypes representing aspects of substance use prior to the development of SUDs may prove particularly fruitful with this age group. Although such phenotypes may reflect intermediate stages between genes and the clinical endpoint, they may also reflect problems in their own right (e.g., “binge drinking”; Wechsler & Nelson, 2001).

In the absence of many well-established susceptibility genes for SUDs it is difficult to know a priori which phenotypes will be most promising. In the present paper, we adopt a novel approach to identifying candidate substance use phenotypes, which themselves represent novel phenotypes reflecting genetic vulnerability for SUDs. This approach consists in using an endophenotype reflecting genetic risk for SUDs as a criterion variable for evaluating the usefulness of candidate phenotypes. Insofar as this criterion variable reflects genetic risk, associations between it and various candidate phenotypes may well help identify which of these phenotypes also reflect genetic risk. Amplitude of the P300 (or P3) component of event-related brain potentials (ERPs) represents an appropriate endophenotype, particularly for substance use-related phenotypes in adolescents.

An endophenotype for a disorder should be heritable, if in fact it is more directly related to the susceptibility genes for the disorder (Almasy, 2003, Gottesman & Gould, 2003, Iacono, 1998). Twin and family studies indicate that P3 amplitude satisfies this condition, with heritability estimates ranging from 39% to 79% (Almasy et al., 1999, Katsanis et al., 1997, O'Connor et al., 1994, van Beijsterveldt et al., 1998, van Beijsterveldt et al., 2001, Wright et al., 2001). A recent meta-analysis reported a pooled heritability estimate of 60% (van Beijsterveldt & van Baal, 2002). Furthermore, research from the Consortium on the Genetics of Alcoholism (COGA), summarized recently by Porjesz et al. (2005), yields evidence of linkage to P3 on several chromosomes. An intriguing finding involves possible pleiotropic loci on chromosome 4 near the alcohol dehydrogenase gene that may both affect P3 amplitude and relate to the heritability of alcoholism.

In addition to being heritable, P3 amplitude is reduced in preadolescent, alcohol-naïve sons of alcoholic fathers relative to boys without such a familial history (Begleiter, Porjesz, Bihari, & Kissin, 1984), a finding that appears to be generally replicable (Polich, Pollock, & Bloom, 1994). Because these children have not initiated substance use, these results suggest that P3 amplitude reduction (P3-AR) reflects genetic risk. Even among alcoholic adults, however, P3-AR appears to be associated with familial risk for alcoholism and not to substance exposure per se (Cohen et al., 1995, Pfefferbaum et al., 1991). More compelling evidence that P3-AR qualifies as an endophenotype comes from studies indicating that P3 amplitude in middle childhood or early adolescence predicts future problematic substance use (Berman, Whipple, Fitch, & Noble, 1993) and even SUDs (Habeych, Charles, Sclabassi, Kirisci, & Tarter, 2005). P3-AR in late adolescence also predicts SUDs 3 years later (Iacono, Carlson, Malone, & McGue, 2002), and is observed in the unaffected members of twin pairs discordant for alcohol-related disorders compared with twin pairs in which both are free of such diagnoses (Carlson, Iacono, & McGue, 2002).

That P3-AR predicts different types of substance problems suggests that it is an endophenotype associated with risk for SUDs in general and not just for alcohol dependence. Indeed a number of studies have indicated that P3 amplitude is reduced among individuals with other SUDs, including cocaine and heroin addiction (Bauer, 2001, Biggins et al., 1997, Branchey et al., 1993), long-term cannabis use (Solowij, Michie, & Fox, 1991) and nicotine dependence (Anokhin et al., 2000, Polich & Ochoa, 2004). P3-AR has also been observed in preadolescent sons of fathers with a history of substance dependence (Brigham, Herning, & Moss, 1995), a result paralleling the seminal finding of Begleiter et al. (1984) regarding boys at risk for alcoholism.

As a heritable phenotype reflecting risk for SUDs in general, P3-AR seems especially appropriate for evaluating candidate substance use phenotypes. Although a number of phenotypes have been proposed as complements to alcoholism diagnoses (Enoch et al., 1999, Lex et al., 1993, Malone et al., 2002, McGue et al., 2001a, McGue et al., 2001b, Rangaswamy et al., 2004, Schuckit et al., 2000, Tabakoff et al., 1988), we are unaware of any for other types of substance problems. One goal of the present study, therefore, was to develop phenotypes for substances other than alcohol that are appropriate for adolescents, using P3-AR as an index of genetic risk.

Another major goal was to investigate gender differences in relation to genetic risk for SUDs. Although rates of substance use, misuse, and subsequent negative consequences have historically been higher among males than females (Grant, 1996, Naimi et al., 2003, Wilsnack et al., 2000, Wilsnack & Wilsnack, 1991), rates of deviant substance use are increasing among females (Holdcraft & Iacono, 2002, Johnston et al., 2000). Moreover, genetic risk for SUDs appears to be comparable for males and females. For example, family studies of alcoholism yield cross-gender transmission rates that are similar to within-gender transmission rates, suggesting similar magnitude of influence of inherited factors. Behavior-genetic studies (see meta-analysis by McGue & Slutske, 1996) also yield roughly comparable heritability estimates for alcoholism for the two sexes, despite greater variability in estimates among females. Studies investigating P3-AR as an index of genetic susceptibility for SUDs have yielded conflicting results. Some have found P3-AR in female subjects at risk for alcoholism compared to those without such a family history (Polich et al., 1988, Suresh et al., 2003), whereas others have not (Hill & Steinhauer, 1993b, Socorro et al., 1998). The heritability of P3 amplitude may also be greater among males than among females (van Beijsterveldt et al., 1998, van Beijsterveldt et al., 2001), although this has not been uniformly observed (Wright et al., 2001).

In summary, in the current investigation we used a novel approach to identify substance use phenotypes that may reflect genetic risk for SUDs. We began by assessing the degree to which P3 amplitude is heritable in a community sample of adolescent boys and girls. This extends previous work based on a subset of the male twins in the present investigation (Katsanis et al., 1997) and helps to establish the degree to which P3 amplitude can serve as a risk marker in both genders. In light of our heritability findings and converging evidence that P3 amplitude is an endophenotype reflecting risk for SUDs in general, we used P3-AR as a criterion variable for assessing the utility of candidate phenotypes reflecting aspects of substance misuse in a large sample of adolescents. We determined whether these candidate phenotypes were associated with P3-AR, which served here as a measure of genetic risk. The candidate phenotypes examined reflect different aspects of misuse (early age of initiation, frequency of use, excessive one-time use, and so on) and include misuse of illicit substances as well as of alcohol and tobacco (cigarettes).