Glucagon-like peptide-1 receptor agonists, but not dipeptidyl peptidase-4 inhibitors, reduce alcohol intake

Abstract

BACKGROUNDDespite growing preclinical evidence that glucagon-like peptide1 receptor agonists (GLP-1RAs) could be repurposed to treat alcohol use disorder (AUD), clinical evidence is scarce. Additionally, the potential impact of dipeptidyl peptidase-4 inhibitors (DPP-4Is) on alcohol intake is largely unknown.METHODSWe conducted a large cohort study using 2008-2023 electronic health records data from the U.S. Department of Veterans Affairs. Changes in Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) scores were compared between propensity-score-matched GLP-1RA recipients, DPP-4I recipients, and unexposed comparators. We further tested the effects of 2 DPP-4Is, linagliptin and omarigliptin, on binge-like alcohol drinking in mice and operant oral alcohol self administration in alcohol-dependent rats, models previously used to show a significant effect of the GLP-1RA semaglutide in reducing alcohol intake.RESULTSGLP-1RA recipients reported a greater reduction in AUDIT-C scores than unexposed individuals (difference-in-difference [DiD]: 0.09 [95% CI: 0.03, 0.14], P = 0.0025) and DPP-4I recipients (DiD: 0.11 [95% CI: 0.05,0.17], P = 0.0002). Reductions in drinking were more pronounced among individuals with baseline AUD (GLP-1RA versus unexposed: 0.51 [95% CI: 0.29,0.72], P < 0.0001; GLP-1RA versus DPP-4I: 0.65 [95% CI: 0.43,0.88], P < 0.0001) and baseline hazardous drinking (GLP-1RA versus unexposed: 1.38 [95% CI: 1.07,1.69], P < 0.0001; GLP-1RA versus DPP-4I: 1.00 [95% CI: 0.68,1.33], P < 0.0001). There were no differences between DPP-4I recipients and unexposed individuals. The latter results were confirmed via a reverse translational approach. Specifically, neither linagliptin nor omarigliptin reduced alcohol drinking in mice or rats. The rodent experiments also confirmed target engagemhent, as both DPP-4Is reduced blood glucose levels.CONCLUSIONConvergent findings across humans, mice, and rats indicated that GLP-1RAs, but not DPP-4Is, reduce alcohol consumption and may be efficacious in treating AUD.FUNDINGThis work was supported by the National Institutes of Health Intramural Research Program (ZIA DA000635, ZIA DA000644, ZIA DA000602), National Institute on Alcohol Abuse and Alcoholism extramural funding (R01 AA030041, P01 AA029545, U01 AA026224, U24 AA020794, U01 AA020790, U10 AA013566), the U.S. Department of Veterans Affairs (I01BX004820), and an Alkermes Pathways Research Award.

Keywords: Addiction; Endocrinology; Neuroscience; Peptides; Pharmacology.

Figure 1. Flow diagram of the human cohort study.

Numbers presented for excluded individuals are not mutually exclusive.

Figure 2. Association between receipt of GLP-1RAs or DPP-4Is and alcohol use in humans.

Difference-in-difference estimates and 95% confidence intervals of changes in AUDIT-C scores, overall (white) and stratified by baseline AUD diagnosis (green) and by baseline AUDIT-C score (blue). (A) GLP-1RA recipients versus unexposed individuals, (B) DPP-4I recipients versus unexposed individuals, (C) GLP-1RA recipients versus DPP-4I recipients. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 3. Effects of DPP-4Is on alcohol intake and blood glucose levels in mice.

(A) Linagliptin (2.5, 5, 10, and 20 mg/kg, s.c.), tested using a between-subjects design (see Figure 5A), had no effect on binge-like alcohol drinking, measured on the first 4-hour session (Tuesday) of each week, in mice (n = 16 males, 16 females). Drug (linagliptin) effect: F_1,28 = 1.90, P = 0.18; week effect: F_3,8 = 1.04, P = 0.38; sex effect: F_1,28 = 0.80, P = 0.38, drug × week interaction: F_3,84=0.16, P = 0.93; drug × sex interaction: F_1,28=0.05, P = 0.82; week × sex interaction: F_3,84 = 2.12, P = 0.10; drug × week × sex interaction: F_3,84 = 1.07, P = 0.37. (B) Linagliptin (20 mg/kg, i.p.), tested using a between-subjects design, lowered blood glucose levels following both glucose (n = 8 males, 8 females) and glucose-plus-alcohol (n = 8 males, 8 females) challenge tests in mice. Drug (linagliptin) effect: F_1,28 = 11.16, P = 0.002; alcohol effect: F_1,28 = 12.19, P = 0.002 (glucose alone > glucose plus alcohol); drug × alcohol interaction: F_1,28 = 0.52, P = 0.48. (C) Omarigliptin (10, 20 mg/kg, i.p.), tested using a within-subjects design (see Figure 5B), had no effect on binge-like alcohol drinking in mice (n = 4 males, 3 females). Drug (omarigliptin) effect: F_2,10 = 0.04, P = 0.96; sex effect: F_{1, 5}= 0.51, P = 0.51; drug × sex interaction: F_2,10 = 0.52, P = 0.61. (D) Omarigliptin (20 mg/kg, i.p.), tested using a between-subjects design, lowered blood glucose levels following both glucose (n = 8 males, 8 females) and glucose plus alcohol (n = 8 males, 8 females) challenge tests in mice. Drug (omarigliptin) effect: F_1,28 = 13.99, P = 0.0008; alcohol effect: F_1,28 = 0.12, P = 0.73; drug × alcohol interaction: F_1,28 = 0.16, P = 0.69. Individual data symbols are shown in black for males and in gray for females. Data are expressed as mean (SEM). **P < 0.01, ***P < 0.001.

Figure 4. Effects of DPP-4Is on alcohol intake in rats.

(A) Linagliptin (10, 20 mg/kg, i.p.), tested using a within-subjects design (see Figure 5C), had no effect on operant oral alcohol self administration in alcohol-dependent rats (n = 10 males, 9 females). Drug (linagliptin) effect: F_2,34 = 2.01, P = 0.15; sex effect: F_1,17 = 0.18, P = 0.68; drug × sex interaction: F_2,34 = 1.75, P = 0.19. (B) Omarigliptin (10, 20 mg/kg, i.p.), tested using a within-subjects design (see Figure 5C), had no effect on operant oral alcohol self administration in alcohol-dependent rats (n = 10 males, 9 females). Drug (omarigliptin) effect: F_2,34 = 1.73, P = 0.19; sex effect: F_1,17 = 6.99, P = 0.02 (female > male); drug × sex interaction: F_2,34 = 0.82, P = 0.45. Individual data symbols are shown in black for males and in gray for females. Data are expressed as mean (SEM).

Figure 5. Schematics of the main rodent experiments.

(A) Effect of linagliptin on drinking-in-the-dark in mice was tested using a between-subjects design. Mice were assigned to 1 of the 2 groups: vehicle or linagliptin. The vehicle group received vehicle once a week for 4 weeks, whereas the linagliptin group received escalating doses of linagliptin (2.5, 5, 10, and 20 mg/kg, s.c.), 1 injection per week (Tuesdays). Sweetened alcohol solution was given for 4 hours on Tuesdays (results in Figure 3A) and Fridays (results in Supplemental Figure 3) and for 2 hours on Mondays and Thursdays (data not shown). (B) Effect of omarigliptin on drinking-in-the-dark in mice was tested using a within-subjects design. Mice received vehicle and 2 doses of omarigliptin (10, 20 mg/kg, i.p.) in a randomized (Latin-square) order on each 4-hour drinking test day (Tuesday/Friday; results in Figure 3C). Sweetened alcohol solution was given for 2 hours on Mondays and Thursdays (data not shown). (C) Effects of linagliptin and omarigliptin on operant oral self administration in alcohol-dependent rats were tested using a within-subjects design. Rats were first made dependent using alcohol vapor exposure. They received daily, intermittent cycles of 14 hours of alcohol vapor exposure and 10 hours off (withdrawal). Operant oral alcohol self administration was performed 6–8 hours into withdrawal. Male rats were tested first with linagliptin then omarigliptin (as shown in the figure). Female rats were tested first with omarigliptin then linagliptin (opposite of the order shown in the figure). Linagliptin and omarigliptin testing was separated by at least 4 days (washout). Rats received vehicle and 2 doses of linagliptin (10, 20 mg/kg, i.p.; results in Figure 4A) or vehicle and 2 doses of omarigliptin (10, 20 mg/kg, i.p.; results in Figure 4B) in a randomized (Latin-square) order on each test day (Tuesday and Friday). Alcohol intake was measured after each 30-minute, fixed ratio 1, operant self-administration session.