A major review of GLP-1 drugs takes a closer look at the risks and rewards

A umbrella review of non-cardiometabolic outcomes of GLP-1 drugs has shown a range of potential benefits and safety concerns, although the quality of data for most remains limited.

Across 60 meta-analyses that evaluated 116 noncardiometabolic outcomes, some of the most consistent indications were for gastrointestinal adverse events based on moderate or high-quality evidence. These included nausea (OR 2.47, 95% CI 1.84-3.34), vomiting (OR 2.78, 95% CI 1.91-4.06), and diarrhea (OR 1.94, 95% CI 1.52-2.49), Li, First Hospital of Yongzhi Medical University, China. Shenyang and colleagues reported The Jama network was launched.

High-quality, convincing evidence also suggests a possible protective association against infection-related outcomes, particularly for serious infections (OR 0.89, 95% CI 0.87–0.92).

Other indications have also emerged in fracture, respiratory, neurological, psychiatric, and endocrine outcomes. Despite reaching significance, most pooled results did not reach high reliability thresholds, exploring associations and generating hypotheses.

“Clinically, gastrointestinal adverse events are well known with GLP-1 receptor agonists, and our results confirm this as a very consistent and robust indication,” Lee said. MedPage Today.

“What was more striking was that for many other non-cardiometabolic outcomes, particularly those that have recently attracted attention – such as infections, respiratory diseases and cancer-related outcomes – the evidence is often suggestive based on limited reliability,” he continued. “This reinforces the gap between emerging signals and high-certainty evidence.”

Some of the investigated associations include dementia, incidence of respiratory disease, and all-cause dementia. Other indications are that the use of GLP-1 drugs is associated with a higher complication rate for the resolution of metabolic disorders associated with steatohepatitis, gastroesophageal reflux disease, gall bladder or biliary diseases, and thyroid disease.

Lee cautioned that not all statistically significant findings should be interpreted as clinically actionable.

“For GLP-1 receptor agonists, gastrointestinal adverse events are consistent and expected. However, for many other outcomes … current evidence should be interpreted with caution.” “Clinicians should avoid over-interpreting initial symptoms and rely primarily on well-established cardiometabolic benefits when making treatment decisions.”

“In particular, any identified safety signals should prompt careful, individualized risk-benefit discussions and targeted monitoring — especially for symptomatic patients or high-risk clinical profiles — rather than routine changes in GLP-1 receptor agonist prescription patterns or monitoring protocols,” the team added. “GLP-1 receptor agonists are the mainstay of antidiabetic therapy but should not be misinterpreted as a cure.”

Lee called for large-scale, long-term randomized trials and well-designed real-world studies to clarify whether these proposed associations translate into meaningful clinical benefits or risks.

The umbrella review included 1,751 randomized clinical trials with more than 3.5 million participants. Studies were identified by a systematic search of databases up to January 2026. Most of them were published in the last decade.

The majority of the study population (71.7%) included people with type 2 diabetes and a third included people with obesity. Follow-up ranged from 3 months to 5.4 years or longer.

Lee and co-authors graded the evidence by the Certainty Rating (Recommendation Evaluation, Development and Evaluation) framework. The strength of the evidence is classified as satisfactory (grade 1), very suggestive (grade 2), suggestive (grade 3), weak (grade 4), or based on a number of factors such as the number of trial participants and the threshold of statistical significance.

“This allowed us to distinguish between findings that are statistically significant and those that are really robust and clinically valid,” Lee explained. “In this sense, our work helps move the field from signal detection to evidence evaluation.”

Evidence on cancer risks was inconclusive. While preliminary data have raised concerns about thyroid cancer, the researchers noted that the relevance of these findings to humans is “unconfirmed.”

Since most noncardiometabolic outcomes were captured in studies as adverse events rather than primary endpoints, the data may be susceptible to misclassification and reporting bias. Other limitations included the lack of stratification by GLP-1 drug dose or treatment duration.