Association of metformin exposure with oncological outcomes in BCG-treated non-muscle-invasive bladder cancer: a three-arm retrospective cohort study

To evaluate the association between metformin exposure and oncological outcomes in bacillus Calmette–Guérin (BCG)-treated non-muscle-invasive bladder cancer (NMIBC) using a three-arm cohort design. Because all metformin-exposed patients were diabetic, the study was not designed to establish a direct antitumor effect of metformin. This retrospective cohort included 624 patients with NMIBC treated with transurethral resection and intravesical BCG. Patients were classified as diabetic metformin users ( n =147), diabetic non-metformin users ( n =89), or non-diabetic patients ( n =388). Diabetes was defined as a previously documented diagnosis of type 2 diabetes mellitus and/or current antidiabetic therapy; patients with type 1 diabetes were excluded. Metformin exposure was defined as use concurrent with the BCG treatment period. The BCG regimen comprised 6 weekly induction instillations with maintenance instillations in eligible patients; no concomitant intravesical chemotherapy was administered. Recurrence-free survival (RFS), progression-free survival (PFS), and overall survival (OS) were assessed using Kaplan–Meier analysis, log-rank tests, and multivariable Cox regression adjusted for clinicopathological and treatment-related covariates. A propensity-score matched sensitivity analysis comparing diabetic metformin users with diabetic non-metformin users was also performed. The median follow-up was 64.8 months. Diabetic patients were older and had higher body mass index and Charlson comorbidity index values than non-diabetic patients. Among diabetic patients with available data, median HbA1c was 7.85% (interquartile range 6.9–8.5) in metformin users and 7.45% (6.7–8.5) in non-metformin users; the median metformin dose was 1500 mg/day (range 500–2550). Tumor-related variables were broadly comparable, although T1 disease was more frequent in non-diabetic patients. During follow-up, 166 recurrences, 52 progressions, and 224 deaths occurred. Recurrence rates were 25.9%, 37.1%, and 24.5% in the metformin, diabetic non-metformin, and non-diabetic groups, respectively; corresponding 5-year RFS rates were 79.4%, 65.6%, and 79.1% ( p =0.066). Progression rates were 6.8%, 16.9%, and 7.0%, with 5-year PFS rates of 97.6%, 92.7%, and 96.5% ( p =0.007). OS differed significantly, favoring non-diabetic patients ( p <0.001). Multivariable Cox analysis did not confirm metformin exposure as an independent protective factor. In the propensity-matched analysis (78 pairs), metformin use was associated with lower progression risk (HR 0.32, 95% CI 0.12–0.88; p =0.028) and a non-significant trend toward lower recurrence (HR 0.62, 95% CI 0.35–1.09; p =0.095), with no difference in overall survival (HR 1.09, 95% CI 0.73–1.62; p =0.685). In BCG-treated NMIBC, metformin-exposed diabetic patients showed progression rates comparable to non-diabetic patients and lower than diabetic non-metformin users, an association that persisted in a propensity-matched sensitivity analysis. Because all metformin-exposed patients were diabetic, a direct antitumor effect of metformin cannot be established. Prospective studies incorporating diabetes severity, glycemic control, treatment duration, and adherence are warranted.