Metformin use and outcomes among breast cancer patients on immunotherapy: Insights from a real-world cohort.

Abstract

e23086 Background: An obesity-related cancer, 1 in 3 adults with breast cancer have preexisting type 2 diabetes (T2D) at diagnosis. Preclinical studies suggests that metformin, the most widely prescribed first-line therapy for T2D, may enhance response to PD-1 immune checkpoint blockade (ICB) by reducing immunosuppression, inhibiting tumor metabolism through AMPK signaling, and improving CD8⁺ T-cell and NK cell function within the tumor microenvironment. However, real-world evidence on anti-diabetic medication use in breast cancer patients with T2D receiving immunotherapy is limited. We sought to determine if metformin use is associated with improved overall survival (OS) in adults with T2D and breast cancer receiving ICB. Methods: TriNetX is a federated data network that aggregates electronic health record data across 44 healthcare organizations (HCOs) and links to insurance claims and mortality registries. Within the TriNetX Linked Network, we identified females with breast cancer and T2D who received ICB (pembrolizumab) for their breast cancer between 1/01/2021-12/01/2025. Our primary outcome was OS defined as time from pembrolizumab initiation to death and our key explanatory variable was metformin use (yes/no). We employed Cox proportional hazards models, including age, comorbid conditions, prior and concurrent cancer and diabetes treatments, to evaluate the association between metformin use and OS, overall, and for stage I-III vs. metastatic breast cancer (mBC) patients, and with time to next line of therapy (LOT) for mBC patients. Hazard ratios (HR) and 95% confidence intervals (95% CI) were calculated for all estimates. Results: Our analytic cohort included 1,101 females with breast cancer and T2D who received pembrolizumab. Mean age was 58.2 (SD 11.5) years, 40.0% were obese (BMI > 30), 51.7% had hyperlipidemia, 65.7% had hypertension, and 47.7% had metastatic disease at pembrolizumab initiation. Overall, 23.6% (260/1101) were taking metformin with a mean 5.2-year (SD 3.9) history of metformin use at baseline. Over a median follow-up of 542 days (IQR = 249.0-915.0), we observed 203 deaths (Figure 1). In multivariable models, metformin was suggestively associated with improved OS (aHR 0.71; 95% CI 0.50-1.02, p = 0.059) in the total cohort, and among mBC patients (aHR 0.64; 95% CI 0.40-1.02, p = 0.058). Within the mBC group, 72/525 progressed to another LOT after pembrolizumab, and patients taking metformin had 57% lower risk of progressing to a new LOT during follow-up (aHR = 0.43, 95% CI 0.20-0.90, p = 0.015). Conclusions: In this real-world cohort of 1101 adults with T2D and breast cancer treated with pembrolizumab, metformin use was associated with improved outcomes among individuals with metastatic breast cancer. Prospective studies with larger sample sizes and longer follow-up periods are needed to confirm our findings.