VA study finds liraglutide, semaglutide and dulaglutide similar on kidney and heart outcomes

A comparison of three GLP-1 receptor agonists in the Veterans Affairs health system found little difference in kidney failure or major cardiovascular events among veterans with type 2 diabetes who started liraglutide, semaglutide or dulaglutide, according to results published in JAMA Network Open ^[1].

To build the comparison, researchers used an active-comparator, new-user design that emulates a target trial, an observational method meant to approximate what a randomized study would look like. National records linked across the VA, Medicare and the US Renal Data System were used ^[1]. Everyone included had type 2 diabetes, was already taking metformin, had no history of GLP-1 receptor agonist use and did not have end-stage kidney disease. Treatment started between January 1, 2018, and December 31, 2021. Outcomes were followed through March 31, 2023. The analysis was conducted from September 2024 to June 2025 ^[1].

The cohort numbered 21,790 veterans. Their mean age was 63.5 years (SD 10.8), and 19,823 of them, or 91.0%, were male. Liraglutide was started by 5,425 (24.9%), semaglutide by 10,838 (49.7%) and dulaglutide by 5,527 (24.9%) ^[1].

Against semaglutide, liraglutide showed comparable hazards for kidney failure (hazard ratio 0.93; 95% CI, 0.60-1.44), for the combined cardiovascular and kidney metabolic outcome (HR 0.96; 95% CI, 0.84-1.10) and for major adverse cardiovascular events, which the authors defined as myocardial infarction, heart failure, or stroke or transient ischemic attack (HR 0.95; 95% CI, 0.83-1.09). The liraglutide-versus-dulaglutide and dulaglutide-versus-semaglutide comparisons pointed the same way ^[1].

Mortality was where the picture got murkier. Under an intent-to-treat analysis, liraglutide carried a lower hazard of death from any cause than semaglutide (HR 0.83; 95% CI, 0.69-0.99), but the difference was no longer statistically significant in per-protocol models ^[1]. Liraglutide also showed lower all-cause mortality than dulaglutide in both the intent-to-treat analysis (HR 0.69; 95% CI, 0.58-0.83) and the per-protocol analysis (HR 0.50; 95% CI, 0.31-0.82). Dulaglutide's mortality hazard exceeded semaglutide's only in the per-protocol model (HR 1.72; 95% CI, 1.20-2.47) ^[1].

Gastrointestinal side effects were mostly alike across the three drugs. The single exception was a lower risk of gallstones (HR 0.72; 95% CI, 0.54-0.95) and acute cholecystitis (HR 0.62; 95% CI, 0.39-0.99) with dulaglutide than with semaglutide ^[1].

Why it matters for patients

Protecting the kidneys and heart is one of the reasons GLP-1 medications are prescribed to people with type 2 diabetes. Within this group of veterans, none of the three drugs stood out on those measures, which suggests kidney or cardiovascular risk may not be a deciding factor when a clinician picks among them ^[1]. Observational work like this cannot prove cause and effect, and the authors call for randomized head-to-head trials to confirm the findings ^[1].

The mortality signals deserve caution. Liraglutide's advantage over semaglutide on death appeared in one analysis and vanished in another, and the dulaglutide-versus-semaglutide gap showed up in only one model ^[1]. Where confidence intervals cross 1.0, as several do here, the data cannot rule out no difference at all.

The findings also come from a narrow slice of patients: largely older men, all veterans, all taking metformin, all treated inside the VA system ^[1]. Whether they hold for women, for younger adults, or for people who take GLP-1 drugs mainly for weight loss has not been reported. Only these three medicines were compared; other GLP-1 drugs were not part of the analysis ^[1].

What happens next

The authors state that randomized head-to-head trials are needed to confirm the results ^[1]. No specific trials, sponsors or timelines are named in the paper. Because outcome tracking stopped on March 31, 2023, any shifts in how these drugs are prescribed since then are not captured ^[1].