The PROactive study investigated pioglitazone for secondary prevention of macrovascular events in type 2 diabetes mellitus. Pioglitazone showed a 10% (non‐significant) relative risk (RR) reduction for the primary composite endpoint and a significant 16% reduction for the main secondary endpoint (death, myocardial infarction, stroke) after a mean 34.5 months. There was no difference in cumulative malignancy incidence, but an imbalance in bladder malignancies (pioglitazone 14, placebo 5). We present a pre‐specified 6‐year interim analysis of a 10‐year observational follow‐up.

Any patient completing PROactive was eligible. No study treatments were provided. A Cox proportional hazard model compared non‐adjudicated macrovascular events (same endpoints as PROactive excluding acute coronary syndrome) based on original randomization. Malignancies were compared using conventional RR ratios.

Of 5238 randomized patients, 3599 (74%) entered the follow‐up. For the follow‐up (mean 5.8 years) or combined double‐blind and follow‐up periods (≤9.5 years, mean 8.7), there were no statistically significant differences in primary or main secondary endpoints. For the combined period, a similar percentage of patients had any diagnosed malignancy (RR = 1.05, 95% CI [0.89, 1.24]) or bladder malignancy (RR = 1.06, 95% CI [0.59, 1.89]) in the pioglitazone and placebo groups. There were fewer cases of bladder malignancy with pioglitazone (15 [0.6%] vs. 19 [0.7%] for placebo) for the combined period when events diagnosed in the first 365 days were excluded, and fewer cases for the follow‐up period alone (10 [0.5%] vs. 17 [1.0%] for placebo). Further analyses of pioglitazone use (including use during follow‐up) found no significant difference in bladder malignancies between any and no pioglitazone use for the combined period.

These data suggest that improved macrovascular outcomes seen with pioglitazone subside without continued pioglitazone treatment. The double‐blind period bladder cancer imbalance did not persist in follow‐up.