Preliminary findings on topiramate for addressing health behaviors in comorbid bipolar disorder and alcohol use disorder

Dear editors

Increased weight and cigarette smoking represent negative health behaviors that are prevalent in people with serious mental illnesses such as bipolar disorder; thus, treatments that can effectively target these behaviors have important clinical implications. Topiramate, an anticonvulsant that enhances the activity of gamma-aminobutyric acid and inhibits the activity of glutamate, has been associated with decreased weight among patients with bipolar disorder (Hiluy et al., 2019). Studies have also shown the benefit of topiramate for reducing cigarette smoking in patients with alcohol dependence (Johnson et al., 2005), an important finding as smoking quit rates are low in bipolar disorder relative to the general population (Heffner et al., 2011). To our knowledge, studies have not evaluated the effects of topiramate on changes in weight and tobacco use among patients with comorbid bipolar disorder and alcohol use disorder. We evaluated mean weight and waist circumference, as well as tobacco use status, among patients with DSM-IV bipolar disorder and alcohol dependence (n = 12) who participated in a clinical trial of topiramate for reducing alcohol consumption (Sylvia et al, 2016).

Participants were randomly assigned to 12 weeks of double-blind treatment with topiramate or placebo, with study medication titrated over a duration of 5 weeks to 150mg twice/day. Study inclusion criteria were 8 heavy drinking days (≥5 drinks per day for men, ≥4 drinks per day for women) in the 4 weeks prior to study screening and at least 4 weeks of accepted maintenance pharmacotherapy for bipolar disorder. Study exclusion criteria were substance dependence for a drug other than cannabis or nicotine, a history of topiramate use, current psychotic disorders, and current pregnancy. Participants attended weekly visits for the first 8 weeks of treatment followed by biweekly visits through week 12. At study sessions, participants completed clinical assessments evaluating a range of mood, medical, and substance-related outcomes including participants’ weight, waist circumference, and self-report of tobacco use (Sylvia et al., 2016).

Study patients reported an average age of 43.6 years (SD = 9.7). Out of study participants (n = 12), 5 participants were randomly assigned to receive topiramate and 7 participants were randomly assigned to receive placebo medication. Seven participants (58%; 2 randomized to topiramate and 5 randomized to placebo) completed treatment. Consistent with our main outcome paper, we imputed data using last observation carried forward (Sylvia et al., 2016). We calculated Cohen’s d to examine magnitude of the change from pre- to post-treatment (post-treatment minus pre-treatment) as opposed to conducting inference tests given the small sample size (Hamer and Simpson, 2009). Pre-treatment data was taken from participant’s screening visit, which occurred one week before the baseline/study randomization visit.

From pre- to post-treatment, participants had a mean weight change of −3.6 pounds (SD = 5.0) in the topiramate group, reflecting a large effect size (Cohen’s d Δ = 0.7), and a mean weight change of +2.1 pounds (SD = 6.5) in the placebo group, reflecting a small to medium effect size (Cohen’s d Δ = 0.3). Further, participants had a mean waist circumference change of −1.2 inches (SD = 2.5) in the topiramate group, consistent with a medium effect size (Cohen’s d Δ = 0.5), and a mean waist circumference change of +0.6 inches (SD = 0.9) in the placebo group, consistent with a large effect size (Cohen’s d Δ =0.7). One participant in the placebo group did not have pre- and post-treatment waist circumference data, so they were not included in the waist circumference calculations.

We also identified participants who met criteria for an elevated waist circumference that can increase risk for heart disease or type II diabetes (defined as greater than 35 inches for women and greater than 40 inches for men). We found that 2 participants in the topiramate group and 3 participants in the placebo group met criteria for an elevated waist circumference. Among these participants, 4 were male (1 in the topiramate group, 3 in the placebo group) and 1 was female (and in the topiramate group). There was no change in this status at post-treatment, though in the topiramate group one of these participants had weight loss.

At pre-treatment, 7 participants reported using tobacco (n = 3 in the topiramate group, n = 4 in the placebo group). At post-treatment, 5 participants reported using tobacco (n = 2 in the topiramate group, n = 3 in the placebo group). Small sample sizes of tobacco users in the topiramate group versus the placebo group prevented further statistical analyses of tobacco use.

We found that participants receiving topiramate had changes in weight and waist circumference from pre- to post-treatment, with changes in weight yielding a large effect size. However, it is worth noting that more participants met criteria for an increased waist circumference (one marker of an elevated body mass index) in the placebo group compared to the topiramate group at both pre- and post-treatment. Regarding tobacco use, as in the topiramate group, one participant in the placebo group who had been using tobacco at pre-treatment stopped tobacco use at post-treatment. These data are limited and should not be taken as evidence of the effectiveness of topiramate for improved health behaviors but, rather, as an indication that future clinical trials in this domain might be warranted. For example, these data are limited by a small sample size; despite a planned enrollment of 80 subjects (effect size of .64 with 80% power), we were only able to randomize 12 participants in this study. To that end, given the small cell sizes in each group, our statistical analyses relied on pre- to post-treatment mean changes and we could thus not account for variables that could theoretically confound our main results such as age, sex, age of disorder onset (for both bipolar disorder and alcohol use disorder), and bipolar symptom severity. We also did not have data on variables that could have informed outcomes (e.g., height to provide information on body mass indices). In addition, given that only a subset of participants in this study completed treatment, it is not possible to attribute any improvements in outcomes to topiramate itself. Further, these data were taken from the screening visit, as opposed to the baseline visit, to maximize consistency and completeness of data and thus these pre-treatment data are not taken from right before participants’ study randomization. Lastly, this study was not designed for patients who were overweight or obese, so it is not possible to know whether these findings would be upheld in an overweight or obese sample.

However, these preliminary data may support further study of topiramate for promoting health behaviors in bipolar patients with comorbid alcohol use disorder. Ideally, an upcoming clinical trial will involve a comparable study design (see Sylvia et al., 2016), such as using a similar dosing and titration schedule of topiramate, assessments (e.g., weight, waist circumference, and tobacco use), and inclusion/exclusion criteria, with the exception that participants be overweight (body mass index greater than 24.9 kg/m²) and using tobacco. We would also suggest that a future trial include a more comprehensive assessment of lifestyle factors (e.g., exercise/activity levels, smoking history, diet) and incorporate stakeholder involvement to improve messaging of the study so as to facilitate better engagement. Finally, including weekly phone or telehealth visits and/or online assessments in future trials could improve study retention.

Highlights.

• Many patients with bipolar disorder suffer from co-occurring alcohol use disorder.

• This study evaluated topiramate in bipolar patients with alcohol use disorder.

• We studied changes in weight, waist circumference, and tobacco use.

• Further study of topiramate in this population is warranted.