GLP 1 prescribing in cancer patients with diabetes follows complications more than the cancer

GLP 1 prescribing in cancer patients with type 2 diabetes rose steeply across a decade of Medicare records, a cohort study in JAMA Network Open reports. People with cancer were slightly more likely than matched people without cancer to receive the drugs, but that gap largely vanished once obesity, kidney disease and other conditions were accounted for. Use was lower with advanced stage, older age and pancreatic cancer, and among Black, Hispanic and Asian beneficiaries. The study counted prescriptions only.

What the Medicare records covered

The team, led by Xin Hu of the Winship Cancer Institute and the Rollins School of Public Health at Emory University, used the 2024 linkage of SEER cancer registry data with Medicare claims. SEER registries cover about 35% of the US population. The authors found 373,278 beneficiaries aged 66 or older who already had type 2 diabetes when they were diagnosed with one of eight cancers between 2010 and 2021. Of these, 232,924 were in traditional Medicare and 140,354 in Medicare Advantage. Each person was paired with a beneficiary without cancer from a 5% random sample, matched on birth year, state, plan type, sex, and race and ethnicity. The outcome was simple: whether someone filled any GLP 1 receptor agonist prescription in a given year, according to Part D drug files.

Breast cancer made up 23.4% of the cancer group, prostate 21.2%, non small cell lung cancer 17.3% and colorectal 14.6%, with bladder, gynecologic, pancreatic and small cell lung cancer making up the rest. Mean age was 76.1 years and 50.6% were women. Traditional Medicare claims ran through 2022, while Medicare Advantage encounter data covered only 2016 to 2021, so the two plan types were analysed separately. Diabetes, cancer and other conditions were all defined from billing codes. That approach reaches very large numbers of people at low cost, but it records what was billed, not why a doctor chose a drug or whether a patient took it. The analysis ran from December 2025 to July 2026, and the paper appeared online on 27 October.

How GLP 1 use grew from 2010 to 2022

In traditional Medicare, the share of the cancer cohort using a GLP 1 drug went from 0.5% in 2010 to 8.2% in 2022, which the authors describe as a more than 16 fold rise. In the matched group without cancer, use climbed from 0.9% to 6.2% over the same years. Medicare Advantage showed a similar curve over a shorter window, from 1.5% to 6.7% among people with cancer and from 1.8% to 5.9% among those without. Dulaglutide took off from 2014 and semaglutide from 2017, and by 2022 the two were almost level in the traditional Medicare cancer cohort, at 3.6% and 3.5%. Before any adjustment, people with cancer were 1.19 percentage points more likely to use the drugs in traditional Medicare and 0.45 points more likely in Medicare Advantage.

Once obesity, neuropathy, kidney disease and other conditions were taken into account for 2018 to 2022, the traditional Medicare gap shrank to 0.16 points and was no longer statistically significant.

Those raw gaps mostly reflect how sick the two groups were. Once obesity, neuropathy, kidney disease and other conditions were taken into account for 2018 to 2022, the traditional Medicare gap shrank to 0.16 points and was no longer statistically significant. In Medicare Advantage it reversed, with people with cancer 0.52 points less likely to use a GLP 1 drug than their matches, a result the authors call hypothesis generating. The conditions that pushed use up are the ones these drugs are meant to treat. Neuropathy went with 3.91 points higher use in traditional Medicare, obesity with 3.35, dyslipidemia with 2.26, retinopathy with 2.02 and chronic kidney disease with 2.01. Put simply, the severity of diabetes explained more of the GLP 1 prescribing in cancer survivors than the cancer diagnosis did.

Which patients were less likely to get the drugs

Within the cancer cohort the differences were larger. Compared with colorectal cancer, people with prostate cancer were 1.30 points more likely to use a GLP 1 drug in traditional Medicare and people with breast cancer 1.05 points more likely, while those with pancreatic cancer were 2.55 points less likely. Use fell as stage advanced, and distant disease went with 1.99 points lower use than in situ disease. Age had the largest single effect: people aged 85 or older were 5.75 points below those aged 66 to 74. The authors read the pattern by type and stage as a possible effect of the cancer context on treatment choices. Records of this kind, however, cannot show the reasoning behind any single prescription, or whether a drug was offered and declined.

Race, ethnicity and neighbourhood income also mattered. Compared with non Hispanic White beneficiaries, use was 1.37 points lower for non Hispanic Black, 0.72 points lower for Hispanic and 1.64 points lower for Asian or Pacific Islander beneficiaries in traditional Medicare, with similar gaps in Medicare Advantage. People in the most advantaged neighbourhoods were 1.32 points more likely to use the drugs than those in the least advantaged. Dual eligibility for Medicare and Medicaid went with 2.51 points higher use, a pattern the paper reports without testing its cause. The authors see these differences as possible signs of structural barriers to access. Several sensitivity checks, including one limited to new users and one limited to people who survived at least a year, gave similar results.

What the study cannot tell doctors

Claims data have known blind spots, and the authors list several. Obesity is often under recorded in billing codes, so part of the adjustment may be incomplete. The study could not separate people in active cancer treatment from long term survivors, did not know which clinician wrote each prescription, and did not track whether patients stayed on the drug. The results apply to adults aged 66 or older living in SEER regions, in the years up to 2022. Above all, this is a paper about GLP 1 prescribing in cancer care, not about outcomes. It says nothing about whether the drugs change recurrence, survival or treatment side effects, questions that broad records work such as the veterans data on 175 outcomes has only started to probe.

Funding came from the National Institute of Arthritis and Musculoskeletal and Skin Diseases and the National Cancer Institute, with registry support from state and federal cancer programmes, and the paper says the funders had no role in the work. Several authors report grants or personal fees from drug makers, including Pfizer, Johnson and Johnson, Novo Nordisk, AstraZeneca and Eli Lilly. The full JAMA Network Open paper sets out the models and every sensitivity check. For readers weighing these drugs during or after cancer care, the safety evidence is still uneven, and a recent review found firm support for gallbladder harm alone among seven feared serious risks. Trials built for people with cancer, not billing records, would be needed to show benefit or harm in this group.

Conclusion

This Medicare study shows GLP 1 prescribing in cancer patients with diabetes climbed sharply from 2010 to 2022 and mostly followed diabetes complications rather than the cancer itself. Lower use with advanced stage, older age, pancreatic cancer and minority race or ethnicity raises questions about access and clinical caution without answering them. Because the records track prescriptions only, the study cannot say whether the drugs help or harm people living with cancer.

Source Hu X, Jiang C, Han X and colleagues. GLP-1 RA Prescribing in Patients With Type 2 Diabetes and Cancer. JAMA Network Open 2026;9(10):e2640781. DOI 10.1001/jamanetworkopen.2026.40781

Amino Weekly reports research. It is not medical advice and never gives doses.