Back

GLP1R gene expression is associated with metabolic and mental health effects in a phenome-wide drug repurposing and safety analysis

Triozzi, J. L.; Chen, H.-C.; Mamak, F.; Yu, Z.; Wilson, O. D.; Ikizler, T. A.; Ferolito, B. R.; Cho, K.; Gaziano, J. M.; Tao, R.; Pereira, A. C.; Hung, A. M.

2025-10-15 pharmacology and therapeutics
10.1101/2025.10.13.25337903 medRxiv
Show abstract

Glucagon-like peptide-1 receptor agonists (GLP1RAs) are transformative therapies for diabetes and obesity, yet their long-term clinical effects are incompletely understood. To address this, we conducted a phenome-wide association study (PheWAS) to assess potential therapeutic and adverse effects of GLP1RAs using genetic drug proxies near the GLP1R locus. Fifteen variants associated with GLP1R expression in the Genotype-Tissue Expression (GTEx) project were tested against 1,204 phecodes in the Million Veteran Program (MVP, n = 464,626), with replication and meta-analysis within the UK Biobank (n = 449,349) and the Vanderbilt BioVU (n = 118,130). In the MVP, GLP1RA proxies showed expected metabolic benefits, including lower risk of type 2 diabetes (OR = 0.966, 95% CI [0.957-0.974], p = 4.55 x 10-14] and obesity (OR = 0.978, 95% CI [0.969-0.986], p = 6.94x10-7). Protective effects were also found for chronic venous insufficiency, obstructive sleep apnea, cancer of the esophagus, orthopnea, and diabetes complications (retinopathy, nephropathy, retinopathy). However, associations with mental health disorders were cohort-dependent (anxiety disorders in MVP OR = 1.02, 95% CI [1.012-1.0316], p = 1.07 x 10-5, BioVU OR = 0.960, 95% CI [0.938-0.981], p = 3.12 x 10-4). The results reinforce the therapeutic promise of GLP1RAs for metabolic diseases while underscoring the need for cautious monitoring of potential mental health effects.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.