Study on GLP-1s and Breast Cancer ‘Too Good to Be True’?

Use of GLP-1 receptor agonists in breast cancer patients was linked to lower mortality and recurrence risk in a retrospective study, but experts are questioning the quality of the underlying data.

At 10 years, the risk of all-cause mortality in those with concurrent diabetes was 91% lower with GLP-1 agonist use compared with insulin or metformin use (HR 0.09, 95% CI 0.06-0.15). And for those with obesity, the risk was 65% lower compared with no GLP-1 drug use (HR 0.35, 95% CI 0.21-0.58), reported Bernard Fuemmeler, PhD, MPH, of the Massey Comprehensive Cancer Center in Richmond, Virginia, and colleagues.

Interestingly, no such association between GLP-1 agonist use and mortality was observed when compared with SGLT2 inhibitors in an unadjusted analysis (HR 0.97, 95% CI 0.82-1.14), according to the findings in JAMA Network Open, though adjustment for covariates showed a modest association with reduced mortality that favored GLP-1 use (HR 0.77, 95% CI 0.68-0.88).

The researchers, who noted that patients with breast cancer and metabolic comorbidities have increased risks of cancer recurrence and reduced survival, touted the results as “promising” and called for prospective studies “to understand the potential risks and the optimal timing and tailoring of GLP-1 RA [receptor agonist] treatment for patients with breast cancer.”

However, in comments provided to the U.K. Science Media Centre, multiple experts raised questions about the study, pointing to outsized effects on mortality not seen in randomized trials of GLP-1 drugs and considerable gaps in the TriNetX database.

The results seem “too good to be true,” said Paul Pharoah, MD, PhD, of Cedars-Sinai Medical Center in Los Angeles.

“There are some immediate red flags to suggest that the results are not a reliable indication of causality,” he said. “The biggest red flag is that the effect sizes are enormous.”

Beyond the roughly 90% relative reduction observed in mortality in the GLP-1 users with diabetes — compared with insulin or metformin — the study also reported reductions in the risk of breast cancer recurrence ranging from 56% to 67%, he noted.

“To put this in perspective, the most effective chemotherapy regimens reduce relapse by 38%,” said Pharoah. “These effect sizes simply cannot be due to the drug and so the only reasonable explanation is confounding/bias.”

Mangesh Thorat, MBBS, of Queen Mary University of London, noted areas of incomplete data, such as estrogen receptor (ER) status and treatment history in the patients’ baseline characteristics.

“The ER status is known for around 20% of the population, whereas this figure should be near 100% in well-curated datasets,” said Thorat. In addition, surgery information was available for fewer than 10% of the patients, while information on radiotherapy or systemic therapy use was available for fewer than 5% and around 25%, respectively.

For this study, Fuemmeler and colleagues used data from the TriNetX database for women with breast cancer from 68 U.S. healthcare organizations who received a diagnosis from April 2006 to April 2023, which followed the 2005 approval of GLP-1 drugs for diabetes management and included the 2021 approval for chronic weight management.

The study included 841,831 eligible patients with GLP-1 receptor agonist use (2 or more prescriptions) during the 6 months before and any time after their index breast cancer diagnosis. Mean age was 69.1 years, 71.7% were white, 12.1% were Black or African American, 5.6% were Hispanic or Latino, and 4.5% were Asian.

After exclusions and 1:1 propensity score matching, the three cohorts included 1,610 patients for GLP-1 drug use versus nonuse among patients with obesity (body mass index ≥30), and 2,323 patients for GLP-1 drug use versus insulin or metformin and 4,052 patients for GLP-1 drug use versus SGLT2 inhibitors among patients with type 2 diabetes.

The authors acknowledged that the study had limitations, including its retrospective design and reliance on a U.S. database, which limits generalizability.