GLP-1 Cancer Risk Associations Vary by Diabetes Drug Comparator
Key Takeaways
- Among adults with type 2 diabetes in one health system, GLP-1 receptor agonist use was associated with lower estimated three-year obesity-related cancer incidence than insulin use.
- A separate weighted comparison associated GLP-1 receptor agonist use with lower incidence than sulfonylurea use.
- Differences between GLP-1 receptor agonists and metformin, SGLT2 inhibitors, or DPP-4 inhibitors were not statistically significant overall, including at five years.
- The significant associations versus insulin and sulfonylureas appeared overall and in the low-severity stratum, which was the largest group; estimates in the moderate- and high-severity strata were imprecise and not statistically significant.
In a JAMA Network Open research letter, investigators retrospectively analyzed PCORnet electronic health records from a source population of 359,595 adults who received diagnoses or clinical care at UT Southwestern Medical Center, a single health system, between January 2010 and July 2025. From this population, adults with type 2 diabetes and no prior obesity-related cancer who started a glucagon-like peptide-1 (GLP-1) receptor agonist or a comparator drug after a 365-day washout were selected separately for each of five new-user, active-comparator target trial emulations. The comparators were insulin, sulfonylureas, metformin, sodium-glucose transport 2 (SGLT2) inhibitors, and dipeptidyl peptidase-4 (DPP-4) inhibitors. Investigators balanced groups using propensity-score weighting. They stratified diabetes severity using the most severe complication documented before drug initiation and assessed obesity-related cancer incidence after three and five years. Each comparison concerned a distinct cohort rather than one shared control group.
In the insulin comparison, estimated three-year obesity-related cancer risk was 2.01% among GLP-1 receptor agonist users versus 3.26% among insulin users (risk difference −1.25 percentage points; 95% confidence interval [CI] −1.74 to −0.83). The overall hazard ratio (HR) was 0.62 (95% CI 0.49–0.77). The lower hazard was statistically significant overall and among adults with low-severity diabetes, but not in the other severity strata.
In the sulfonylurea comparison, estimated three-year risk was 2.85% with GLP-1 receptor agonists versus 3.98% with sulfonylureas (risk difference −1.14 percentage points; 95% CI −1.85 to −0.51). The overall HR was 0.67 (95% CI 0.54–0.84). This lower hazard was statistically significant overall and in the low-severity group, but not in the other severity strata. Significant associations versus insulin and sulfonylureas persisted at five years overall and among low-severity patients; for sulfonylureas, the five-year differences were significant by hazard ratio (overall HR 0.75; 95% CI 0.61–0.92) but not by risk difference. Comparisons with metformin, SGLT2 inhibitors, and DPP-4 inhibitors showed no statistically significant differences overall, including at five years; for metformin, the HR was 0.78 (95% CI 0.58–1.05) at three years and 0.91 (95% CI 0.69–1.19) at five years.
In high-severity diabetes, the three-year risk difference versus SGLT2 inhibitors favored SGLT2 inhibitors (risk 1.61 percentage points higher with GLP-1 receptor agonists; 95% CI 0.17–2.89) and was statistically significant, as was the five-year risk difference (3.10 percentage points; 95% CI 1.01–5.09), although the corresponding hazard ratios (2.07 and 2.17) were not. This stratum included only 237 SGLT2 inhibitor initiators and fewer than 11 events. Moderate-severity estimates versus SGLT2 inhibitors favored SGLT2 inhibitors but were not statistically significant.
According to the investigators, glucose-lowering drugs may be prescribed at different stages of type 2 diabetes and for different clinical reasons. Insulin initiators generally had more advanced complications, insulin is often started later in the disease course, and SGLT2 inhibitor prescribing is strongly shaped by cardiorenal indications, so underlying health, earlier treatment, surveillance, and prognosis could contribute to the observed differences despite weighting. The severity-stratified comparisons were exploratory and not adjusted for multiple comparisons, and events were sparse in some strata. The associations establish neither that GLP-1 receptor agonists prevent cancer nor that insulin or sulfonylureas cause it.
The investigators described the findings as broadly reassuring regarding oncologic safety but not as supporting the selection of glucose-lowering therapy for cancer prevention; they said established metabolic, cardiovascular, and kidney benefits should continue to guide treatment. They concluded that the association between GLP-1 receptor agonist use and obesity-related cancer incidence depended on the comparison drug and the severity of diabetes before treatment.
Clinician Questions
Which complications defined baseline diabetes severity in GLP-1 obesity-related cancer comparisons?
Investigators assigned severity hierarchically using the highest-severity complication recorded before treatment initiation, adapting a validated index. High severity meant macrovascular disease or chronic kidney disease stages 4 to 5; moderate meant microvascular complications or chronic kidney disease stages 1 to 3 without a high-severity condition; and low meant none of these conditions.
What further research did the authors call for on GLP-1 receptor agonists and obesity-related cancer?
The authors called for longer multisystem follow-up. They noted that the data were retrospective and from a single system, that residual confounding, incomplete overlap between groups, misclassification of exposure and outcome, broad severity categories, and sparse events limit interpretation, and that five-year estimates relied disproportionately on earlier initiators because many GLP-1 receptor agonist users started recently.
Recommended Reading
- For more on GLP-1 therapies used in diabetes: How GLP-1s Provide New Options to Manage Heart Disease
This content is intended for healthcare professionals for educational purposes. The article and its image were developed with AI assistance and reviewed by members of the ReachMD team prior to publication. Please refer to the original source linked for the complete and most up-to-date information.