Combined Analysis of the Moderating Effect of a GRIK1 Polymorphism on the Effects of Topiramate for Treating Alcohol Use Disorder

Abstract

Background:

In an initial study, we reported that topiramate reduced heavy drinking among individuals who sought to reduce their drinking and that the effect was moderated by a single nucleotide polymorphism (SNP; rs2832407) in GRIK1, which encodes the kainate GluK1 receptor subunit (Kranzler et al. 2014). In a subsequent study that prospectively randomized patients to medication group based on their rs2832407 genotype, we replicated the main effect of topiramate but not the moderating effect of the SNP (Kranzler et al. 2021). Given the similar design of the two studies, we combined the findings to provide greater statistical power to test the pharmacogenetic effect.

Material and methods:

This secondary analysis of two 12-week, randomized controlled trials of topiramate included a total of 292 European-ancestry individuals (67.1% male; topiramate: 48.3%, placebo: 51.7%) with problematic alcohol use. Using MANOVA, we examined changes in self-reported alcohol consumption, problems resulting from alcohol use, and quality of life, and the biomarker γ-glutamyltransferase. To test the pharmacogenetic hypothesis, all patients were genotyped for rs2832407.

Results:

There was a significant overall effect of topiramate on the alcohol-related outcomes (partial η²=0.134, p<0.001), with follow-up analyses showing significant reductions in percent heavy drinking days (Cohen’s d=0.49), percent days abstinent (d=0.23), drinks/day (d=0.29) and alcohol-related problems (d=0.45). Overall, the moderating effect of the SNP was non-significant (partial η²=0.026, p=0.37).

Conclusions:

Although topiramate is an efficacious pharmacotherapy for reducing drinking and alcohol-related problems among patients with AUD, rs2832407 does not appear to moderate its therapeutic effects.

1. Introduction

Heavy drinking and alcohol use disorder (AUD) are prevalent worldwide and associated with adverse physical, psychological, economic, and social outcomes (SAMHSA, 2019) and increased mortality risk (CDC, 2013). Despite AUD’s high prevalence (Grant et al., 2015), only a minority of individuals with the disorder seek treatment and fewer receive evidence-based care (Cohen et al., 2007). One of the factors contributing to this low rate of treatment utilization is the common belief among both individuals with AUD and healthcare providers that abstinence is the only legitimate goal of treatment (Wallhed Finn et al., 2014). This stringent definition of treatment success may discourage individuals from seeking care because they believe that the goal is not achievable or find it undesirable (Witkiewitz et al., 2016). Thus, there is growing interest in non-abstinent goals of alcohol treatment (Aubin and Daeppen, 2013), supported by evidence that reduced drinking is associated with improvements in both overall functioning and health (Kline-Simon et al., 2013; Laramée et al., 2015).

Topiramate, first approved by the FDA in 1996 as an anticonvulsant, is efficacious for treating AUD and yields small-to-medium effects (Hedges g > 0.4) on abstinence and heavy drinking and smaller effects (Hedges g > 0.3) on γ-glutamyltransferase (GGT) concentration and alcohol craving (Blodgett et al., 2014). Recent efforts have sought to use pharmacogenetics to identify patients who are most likely to benefit from the medication (Hartwell and Kranzler, 2019).

Topiramate antagonizes kainate receptors (Gibbs et al., 2000; Skradski and White, 2000) and is most potent and selective for those containing the GluK1 subunit (encoded by GRIK1) (Gryder and Rogawski, 2003; Kaminski et al., 2004). Based on evidence of an association with alcohol dependence of a single nucleotide polymorphism (SNP; rs2832407) in GRIK1 (Kranzler et al., 2009), we conducted two 12-week, randomized, placebo-controlled clinical trials (RCTs) (Kranzler et al., 2014, 2021) of topiramate to examine: 1) its efficacy and tolerability in treating heavy drinking and 2) the moderating effect of rs2832407 on its therapeutic effects.

In the first study (N=138 heavy drinkers, 92% of whom met DSM-IV criteria for current alcohol dependence) (Kranzler et al., 2014), topiramate-treated patients reported fewer drinking days (DDs) and heavy drinking days (HDDs) than the placebo group. Among the European-ancestry (EA) subsample (n=122), rs2832407*C-allele homozygotes treated with topiramate reduced the number of HDDs more than those treated with placebo, while rs2832407*A-allele carriers showed no drug-placebo difference. In a subsequent 12-week trial (Kranzler et al. 2021), 170 patients with AUD were assigned to receive topiramate or placebo using a stratified randomization based on rs2832407 genotype. Results showed that topiramate reduced HDDs significantly more than placebo, but rs2832407 did not moderate the therapeutic effect.

1.1. Hypotheses.

We report here the results of a secondary analysis of these two topiramate trials. We hypothesized that topiramate would be superior to placebo in the combined analysis of alcohol-related outcomes and that the greater statistical power provided by the combined analysis would yield a genotype-by-medication condition interaction, such that among topiramate-treated patients, rs2832407*C-allele homozygotes would show greater improvement in alcohol-related outcomes than rs2832407*A-allele carriers.

2. Materials and methods

2.1. Patients and Procedures

The procedures for both studies are described fully in the primary publications, along with CONSORT diagrams (Kranzler et al., 2014, 2021). In brief, 308 heavy drinking individuals were recruited to participate in one of two, 12-week RCTs in which they received up to 200 mg/day of topiramate or placebo. Patients were enrolled and treated at the University of Connecticut Health Center (n=76), the University of Pennsylvania Perelman School of Medicine (n=226), or the Corporal Michael J. Crescenz Veterans Affairs Medical Center (n=6). Because of population genetic differences, we limited the pharmacogenetic analysis to the 292 self-reported EA patients. Trial protocols were approved by the institutional review boards at participating sites and patients gave written informed consent to participate.

2.2. Study criteria

Inclusion criteria were age 18–65 (study 1) or 18–70 (study 2) and current heavy drinking (≥24 drinks/week for men, ≥18 drinks/week for women), ability to read English at an 8^th grade level, and, if a woman of childbearing potential, use of a reliable method of birth control. The majority of EA patients in study 1 (n=113 or 92.6%) had a DSM-IV diagnosis of current alcohol dependence and all had a goal of reduced drinking. All patients in study 2 had a DSM-5 diagnosis of current AUD and a goal of either reduced drinking or abstinence. In both studies, we excluded patients with significant physical or psychiatric comorbidities or a current DSM-IV diagnosis of drug dependence (excluding nicotine), or any participant who in the judgment of the study physician posed unacceptable risks for study entry.

2.3. Procedures

Patients in both studies were recruited through advertisements or clinical referrals and, if eligible after a telephone screening interview, were invited for an in-person visit, where they gave informed consent, provided a medical and psychiatric history, and underwent a physical examination and clinical laboratory testing. In study 1, patients were urn randomized to treatment condition based on age, sex, and alcohol use variables. In study 2, block randomization was stratified by treatment goal (abstinence vs. reduction) and genotype group (rs2832407*CC vs. AA/AC).

The same treatment protocol was used in the two studies. Medication was initially dosed at 25 mg/day at bedtime and gradually increased to a maximum of 200 mg/day in two divided doses. For the first 6 weeks, patients attended weekly medication visits, after which there were three bi-weekly visits. At each treatment visit, patients received medical management,²⁷ a brief intervention focused on medication adherence and counseling to reduce drinking and increase abstinent days.

2.4. Measures

The Timeline Follow-Back (TLFB) (Sobell and Sobell, 1992) was used to assess the quantity and frequency of alcohol consumption at each visit. The Short Index of Problems (SIP) (Miller and Tonigan, 1995) a 15-item assessment, was used to measure alcohol-related problems over the preceding 3 months. Using the Short Form Health Survey (SF-12) (Ware et al., 1996), a 36-item measure, patients rated their overall health and quality of life. Finally, γ-glutamyltransferase (GGT), an objective measure of alcohol intake, was assessed at baseline and at study completion.

PureGene kits were used to extract DNA from whole blood. A TaqMan SNP assay was used to genotype rs2832407.

2.5. Statistical Analysis

All EAs (88.4% of the sample in study 1 and 100% of the sample in study 2) were included in the analyses. Three drinking measures were calculated from the TLFB across the 12 weeks of treatment: the percent HDDs (defined as 4+/5+ drinks for women/men), percent days abstinent, and mean drinks per day. Outcome measures also included the SIP, the SF-12, and GGT concentration (log-transformed).

We used MANOVA to avoid inflating the type 1 error when testing 1) an overall medication group effect and 2) a moderating (pharmacogenetic) effect of rs2832407 across all six alcohol-related outcomes. We report effect size as partial η². A follow-up multivariate mixed model tested the effects on each outcome while modeling the correlation among outcomes. We report the observed effect sizes as Cohen’s d for the individual between-treatment and interaction effects. For the interaction effects, d was calculated as the difference in treatment effect (topiramate versus placebo) within each genotype followed by the difference of these differences between the genotype groups (C-homozygote versus A-allele carrier), which was divided by the square root of the pooled standard error (Olejnik and Algina, 2000). To test whether there was a moderation effect by study we included in the multivariate model interaction terms for study by treatment group and study by treatment group by genotype.

3. Results

The majority of patients were male (67.1%) and middle aged (mean=51.5 years, SD=10.1). Over one-third (38.4%) of patients were homozygous for the rs2832407*C allele and genotype distributions were consistent with Hardy-Weinberg equilibrium expectations. There were no significant differences between treatment groups on any demographic variable or genotype frequency (Table 1).

Table 1.

Demographics and rs2832407 Genotype by Treatment Group (N=292)

Demographics	Placebo (n=151)	Topiramate (n=141)	Total Sample (n=292)
Study
Kranzler et al. (2014)	66 (43.7%)	56 (39.7%)	122 (41.8%)
Kranzler et al. (2021)	85 (56.3%)	85 (60.3%)	170 (58.2%)
Sex
Male	98 (64.9%)	98 (69.5%)	196 (67.1%)
Age
Mean ± SD	51.3 ± 10.8	51.6 ± 9.4	51.5 ± 10.1
Rs2832407 Genotype
C-allele homozygotes	61 (40.4%)	51 (36.2%)	112 (38.4%)

^*Prevalence of the rs2832407*C allele in individuals of European ancestry is 61.5% (https://www.snpedia.com/index.php/Rs2832407. Accessed December 29, 2020).

The main effect of topiramate treatment was significant across the alcohol-related measures (partial η²=0.134, p<0.001). Follow-up analyses (Table 2) showed a medium effect size for the percentage of heavy drinking days (d=0.49, p<.001) and SIP score (d=0.45, p<.001). Smaller, though statistically significant, effects were seen for percent days abstinent (d=0.23, p=0.031) and drinks per day (d=0.29, p=0.026), while the small effects for both the SF-12 score (d=0.15, p=0.12) and GGT concentration (d=0.13, p=0.11) were non-significant.

Table 2.

Treatment Effects

Outcome Measures	Placebo	Topiramate	Effect Size (95% CI)	P-Value
Main Effects	Mean ± SD	Mean ± SD
Drinking Outcomes
% Heavy Drinking Days	47.9 ± 31.3	33.3 ± 28.4	0.49 (0.25, 0.73)	<0.001
% Days Abstinent	20.1 ± 23.9	26.2 ± 28.0	0.23 (0.00, 0.47)	0.031
Drinks per Day	4.2 ± 2.2	3.5 ± 2.4	0.29 (0.05, 0.53)	0.026
Other Outcomes
SIP Score	10.4 ± 8.4	6.9 ± 7.0	0.45 (0.21, 0.70)	<0.001
SF-12 Score	78.1 ± 16.7	80.5 ± 15.0	0.15 (−0.09, 0.39)	0.12
GGT Concentration	3.50 ± 0.74	3.40 ± 0.75	0.13 (−0.11, 0.38)	0.11
Interaction Effects	Coefficient	Std Err
Drinking Outcomes
% Heavy Drinking Days	−0.14	0.07	0.48 (−0.01, 0.97)	0.051
% Days Abstinent	0.12	0.06	0.44 (−0.04, 0.93)	0.070
Drinks per Day	−1.35	0.57	0.52 (0.03, 1.01)	0.018
Other Outcomes
SIP Score	−0.03	1.94	0.07 (−0.42, 0.56)	0.99
SF-12 Score	−2.21	3.98	0.06 (−0.44, 0.55)	0.60
GGT Concentration	−0.19	0.19	0.30 (−0.20, 0.79)	0.31

Effect Size=Cohen’s d; 95% CI=95% confidence interval; SIP=Short Index of Problems; SF-12=Short Form Health Survey; GGT=γ-glutamyl transferase

The moderating effect of rs2832407 genotype was not significant in the MANOVA (partial η²=0.026, p=0.37). Follow-up analyses (Table 2) showed a nominally significant interaction effect for drinks per day (d=0.52, p=0.018), reflecting a larger difference between the topiramate and placebo groups among C-allele homozygotes than among A-allele carriers. The effects on all other outcomes (percent HDDs, percent days abstinent, SIP score, SF-12 score and GGT concentration) were non-significant.

Adding the study by treatment group interaction yielded a small effect that was not statistically significant (partial eta squared=0.016, p=0.65). Adding the three-way interaction effect of study by genotype by treatment group also yielded a small effect that was not significant (partial eta squared=.029, p=0.32). These findings show that study membership did not moderate treatment effects or treatment by genotype effects.

4.0. Discussion

This secondary analysis of two RCTs of topiramate for treating problematic alcohol use examined the moderating effect of rs2832407 on treatment response. We found a robust overall effect of topiramate treatment on the alcohol-related outcomes, but no overall moderating effect of the SNP on this response.

Although in the initial RCT (Kranzler et al., 2014) there was a robust moderating effect of rs2832407 on HDDs, a prospective RCT (Kranzler et al., 2021) did not replicate that finding. To address the discrepancy between these results we jointly analyzed the data from the two studies, thereby increasing the power to detect a pharmacogenetic effect. The combined analysis showed a robust main effect of medication across the alcohol-related measures and follow-up analyses showed significant effects of topiramate on the three drinking measures and alcohol-related problems score. In the absence of a significant overall multivariate effect, the findings failed to support a pharmacogenetic effect on treatment response, despite a nominally significant effect on drinks per day. It is possible that a much larger study would demonstrate a significant overall moderating effect, but if so, it would, by definition, not be clinically significant.

There are study limitations that should be acknowledged. Combining the two RCTs could have introduced unmeasured confounding, despite the inclusion of study as a covariate in the analyses. We found no evidence that the parameters differed across the two studies (i.e., the moderation effect by study was not supported). Neither study required patients to be abstinent or have abstinence as a goal of treatment, thus the observed effects may underestimate the therapeutic benefit that might obtain with topiramate in an abstinence-oriented trial (Bujarski et al., 2013). Primary measures of alcohol use and resultant problems were self-reported retrospectively and subject to recall bias and underreporting. Because GGT concentration is not sensitive to changes in drinking behavior, subsequent trials should use a more sensitive biomarker such as phosphatidylethanol (Wurst et al., 2015). The use of a single SNP as a genetic moderator of treatment outcome limits the potential to demonstrate moderation. Although consistent with the available methodology and supported by a preliminary study showing an association of this SNP with alcohol dependence (Kranzler et al., 2009), the use of polygenic risk scores based on genome-wide genotyping would be a more powerful approach to characterize this complex pharmacogenetic trait (Torkamani et al., 2018). As patients included in this analysis were exclusively EAs and the majority were male, the findings may not generalize to other population groups or females.

The study’s strengths include its comparatively large sample in which to evaluate both the main effects of topiramate and the moderating effect of rs2832407. In addition to robust study designs, there was good treatment retention and medication adherence in the two RCTs, all of which supports the validity of the findings.

5.0. Conclusions

These results add to the growing literature on the efficacy of topiramate for reducing drinking and alcohol-related problems among individuals with problematic alcohol use. Whereas reduced drinking may be a more desirable and feasible goal than abstinence for most individuals who drink excessively (Mann et al., 2017), it could increase the proportion of patients who seek treatment. Reduced drinking is associated with improved overall functioning and health, including fewer psychiatric symptoms (Knox et al., 2019, 2020; Witkiewitz et al., 2018). In conclusion, the data presented here support the use of topiramate to help patients with problematic alcohol use reduce their drinking, but not the use of rs2832407 for treatment matching.

Highlights.

• In a prior study, topiramate reduced heavy drinking and a genetic variant moderated the effect.

• A subsequent study replicated the topiramate effect, but not the pharmacogenetic finding.

• We conducted a combined analysis of the two published studies to increase statistical power.

• The analysis showed a robust treatment effect for topiramate, but no pharmacogenetic effect.