Efficacy and safety of varenicline and bupropion, in combination and alone, for alcohol use disorder: a randomized, double-blind, placebo-controlled multicentre trial

Abstract

Background: Alcohol use disorder (AUD) is associated with an enormous burden of disease and cost to society. The dopamine deficiency hypothesis posits that negative reinforcement generated by a low brain dopamine state drives ethanol intake. Here, we evaluated the efficacy and safety of combined administration of two dopamine-enhancing drugs, varenicline (a partial nicotinic acetylcholine receptor agonist) and bupropion (a weak dopamine-reuptake inhibitor) on alcohol intake in AUD.

Methods: Participants aged 25-70 years with moderate-to-severe AUD (defined as ≥4/11 Diagnostic and Statistical Manual of Mental Disorders [DSM]-5 criteria) were enrolled in this randomized, double-blind, placebo-controlled trial, done at four outpatient clinics in Sweden. Participants were randomly assigned (block size 8) 1:1:1:1 to Placebo + Placebo, Varenicline + Bupropion, Varenicline + Placebo, or Placebo + Bupropion. After a 1-week titration period, Varenicline was taken as 1 mg orally twice per day and bupropion as 150 mg orally twice per day for 12 weeks. Participants, investigators, and all study personnel were unaware of treatment allocation. The two primary outcomes were phosphatidylethanol in blood (B-PEth) and self-reported percentage heavy drinking days (%HDD), assessed over a steady state 10-week-period (from start of week 2 to end of week 11). Modified intention-to-treat (mITT) and per protocol analyses (PP) were performed using a sequential hierarchical statistical method. This registered study (EudraCT 2018-000048-24; clinicaltrials.govNCT04167306) is completed.

Findings: Between March 4, 2019, and December 14, 2022, 384 participants were randomly assigned: Placebo + Placebo = 97, Varenicline + Bupropion = 100, Varenicline + Placebo = 96, Placebo + Bupropion = 91. 72% participants were male (277/384) and 28% female (107/384), median age 57 (13) years. In the mITT analyses, Varenicline + Bupropion reduced B-PEth (Cohen's d [d] = 0·39, p = 0·004) and %HDD (d = 0·31, p = 0·008) vs Placebo + Placebo. Varenicline + Placebo also reduced B-PEth (d = 0·30, p = 0·005) and %HDD (d = 0·36, p = 0·023) vs Placebo + Placebo. For both primary endpoints, differences between the Varenicline + Bupropion and Varenicline + Placebo groups were not statistically significant (B-PEth: d = 0·022, p = 0·97, %HDD: d = 0·027, p = 0·76), precluding further comparisons according to the statistical hierarchy. In PP analyses, both primary outcomes were reduced with Varenicline + Bupropion (d = 0·43 [B-PEth]; d = 0·41 [%HDD]) and Varenicline + Placebo (d = 0·29 [B-PEth]; d = 0·34 [%HDD]) compared with Placebo + Placebo. Nausea, the only safety concern, was more common in the Varenicline + Placebo group than in the Placebo + Placebo group (49/96 vs 11/97, p < 0·0001) and of longer median duration (45 (70) vs 10 (14·5) days, p = 0·001). Nausea incidence was lower in the Varenicline + Bupropion group vs Varenicline + Placebo (36/100 vs 49/96, p = 0·048) and of shorter median duration (16·5 (39·3) vs 45 (70) days, p = 0·010).

Interpretation: Two brain dopamine elevating treatments (Varenicline + Bupropion; Varenicline + Placebo) reduce alcohol consumption compared with placebo alone. Effect sizes were largest when Varenicline and Bupropion were combined and compliance was high (PP-population). Bupropion reduced Varenicline-induced nausea. Varenicline + Bupropion or other mild dopamine enhancers should be further explored for treatment of AUD.

Funding: This study was funded primarily by the Swedish Research Council.

Keywords: AUD; Bupropion; Combination treatment; RCT; Varenicline.

Fig. 1

Study flow of COMB participants (CONSORT) and analysis groups. Figure shows the final cohort included in the mITT analyses (i.e., all participants who completed the screening and randomisation visits and that reported having taken at least one dose of the randomised study drug). All mITT participants were analysed according to their randomised study drug. The PP analysis set constituted all participants from the mITT population considered to be a “completer” defined as a participant who had taken the study drug at least 80% of the planned number of days of the active treatment period and provided successful outcome measures of B-PEth and/or TFLB at screening and visits 2, 4, 6 and 8. All inclusion and exclusion criteria have been reported previously. ∗1 including B-PEth levels <0·5, abstinence between screening and randomisation, need of alcohol withdrawal treatment, and concomitant alcohol treatment; ∗2 including, for example, blood pressure >180/110 mmHg and high liver enzymes/physical signs of liver failure. ∗3 mITT criteria; completed screening and randomisation visit and reported having taken at least one full dose of study drug. ∗4 PP B-PEth criteria; reported having taken at least 80% of the study drug from Visit 2 through Visit 8 and provided successful B-PEth sampling from Visit 4, Visit 5 or 6, and Visit 7 or 8. ∗5 PP HDD criteria; reported having taken at least 80% of the study drug from Visit 2 through Visit 8 and provided successful TLFB data from Visit 4, Visit 5 or 6, and Visit 7 or 8. AE, adverse event; B-PEth, phosphatidylethanol in blood; Bup, bupropion; COVID-19, infection with SARS-Cov·2; HDD, heavy drinking days; mITT, modified intention-to-treat; PP, per-protocol; Pla, placebo; TFLB, timeline follow-back procedure; Var, varenicline.

Fig. 2

Mean change from baseline (±SEM) of primary outcome variables B-PEth and HDD share according to the modified intention-to-treat analysis. a (B-PEth) and c (HDD share) show the mean change at the respective visit (±SEM). b (B-PEth) and d (HDD share) show the mean reduction (±SEM) over visits 4–8, the pre-determined, steady-state period over which statistics were performed. Significance levels and effect sizes (Cohen's d) are presented in Tables 2 and in heat maps (Figure S3, Appendix). B-PEth, phosphatidylethanol in blood; Bup, bupropion; HDD, heavy drinking days; Pla, placebo; SEM, standard error mean; Var, varenicline; Scr., screening day; R, randomisation day; w, week. ∗p = 0·023 vs Pla + Pla, ∗∗p = 0·004 vs Pla + Pla for both Var + Bup and Var + Pla regarding B-PEth. ∗∗p = 0·008 vs Pla + Pla regarding HDD share.

Fig. 3

Mean change from baseline (±SEM) of primary outcome variables B-PEth and HDD share according to the per-protocol analysis. a (B-PEth) and c (HDD share) show the mean change at the respective visit (±SEM). b (B-PEth) and d (HDD share) show the mean reduction (±SEM) over visits 4–8, the pre-determined, steady-state period over which statistics were performed. Significance levels and effect sizes (Cohen's d) are presented in Tables 3 and in heat maps (Fig. 3, Supplement). B-PEth, phosphatidylethanol in blood; Bup, bupropion; HDD, heavy drinking days; Pla, placebo; SEM, standard error mean; Var, varenicline; Scr., screening day; R, randomisation day; w, week. ∗p = 0·029 vs Pla + Pla regarding B-PEth. ∗∗p = 0·007 vs Pla + Pla regarding B-PEth. ∗p = 0·012 vs Pla + Pla regarding HDD share for Var + Bup. ∗p = 0·044 vs Pla + Pla regarding HDD share for Var + Pla.

Fig. 4

Mean change from baseline (±SEM) of secondary outcome variable total consumption of alcohol (g/day) according to the modified intention-to-treat and per-protocol analyses. a (mITT) and c (PP) show the mean change at the respective visit (±SEM). b (mITT) and d (PP) show the mean reduction (±SEM) over visits 4–8, the pre-determined, steady-state period over which statistics were performed. Significance levels and effect sizes (Cohen's d) are presented in Tables 3 and 4 in the Supplement and in heat maps (Fig. 3, Supplement). B-PEth, phosphatidylethanol in blood; Bup, bupropion; HDD, heavy drinking days; mITT, modified intention-to-treat; Pla, placebo; PP, per-protocol; SEM, standard error mean; Var, varenicline; Scr., screening day; R, randomisation day; w, week. ∗∗p = 0·002 vs Pla + Pla regarding mITT. ∗∗p = 0·005 vs Pla + Pla regarding PP.

Fig. 5

Nausea load across the four treatment groups. Box plot shows the nausea load (median number of days with nausea in participants experiencing nausea) across the four treatment groups. Shown are the medians (fat horizontal bars), first and third quartiles (the lower and upper borders of the boxes), minimum and maximum values (ends of vertical bars) and outliers (dots) Var + Pla vs Pla + Pla, p = 0·001; Var + Pla vs Var + Bup, p = 0·010; Var + Bup vs Pla + Pla, p = 0·186. Bup, bupropion; Pla, placebo; Var, varenicline.

Figure S1

Figure S2

Figure S3