First GWAS of Metformin Blood Levels Reveals Age, Kidney Function, and Ancestry Shape Drug Exposure

Metformin is the world's most prescribed diabetes medication and a serious candidate for extending healthy lifespan, yet why some people achieve much higher blood levels than others on the same dose has remained poorly understood. Clarifying this matters because drug exposure often drives both efficacy and side effects, and because metformin is actively studied for cancer prevention, weight management, and aging itself.

Researchers at Massachusetts General Hospital and the Broad Institute analyzed 745 participants from SUGAR-MGH, an ancestrally diverse cohort given a standardized single-dose metformin challenge. Plasma concentrations were measured and linked to clinical data and genome-wide genotyping, making this the first GWAS of metformin pharmacokinetics.

Three clinical factors independently predicted higher metformin exposure: older age (about 2.5 ng/mL per additional year), lower estimated glomerular filtration rate reflecting reduced kidney clearance, and lower BMI. African ancestry participants had markedly lower concentrations than European ancestry participants — a difference of roughly 73 ng/mL — suggesting population-specific biology in metformin handling.

The GWAS uncovered four African ancestry-specific variants that reached genome-wide significance for higher metformin levels, along with suggestive signals near USP36 and DGKB, genes implicated in glucose metabolism. Critically, none of the top pharmacokinetic variants translated into differences in glycemic response — fasting glucose, HOMA-IR, or fasting insulin — after the challenge. Classic metformin transporter variants that previous smaller studies had implicated also failed to replicate.

These findings suggest that while genetics and clinical factors shape how much metformin circulates in the blood, the drug's glucose-lowering effect may be decoupled from simple plasma concentration. For clinicians prescribing metformin in aging, diabetes, and longevity contexts, age and kidney function remain the most actionable predictors of drug exposure, while ancestry-specific genetic architecture warrants further investigation.