BMJ 2016 Mar 30;352:i1541. Epub 2016 Mar 30.
Centre for Clinical Epidemiology, Lady Davis Institute, Jewish General Hospital, Montreal, Quebec, Canada Department of Oncology, McGill University
Objective: To determine whether pioglitazone compared with other antidiabetic drugs is associated with an increased risk of bladder cancer in people with type 2 diabetes.
Design: Population based cohort study.
Setting: General practices contributing data to the United Kingdom Clinical Practice Research Datalink.
Participants: A cohort of 145,806 patients newly treated with antidiabetic drugs between 1 January 2000 and 31 July 2013, with follow-up until 31 July 2014.
Main Outcome Measures: The use of pioglitazone was treated as a time varying variable, with use lagged by one year for latency purposes. Cox proportional hazards models were used to estimate adjusted hazard ratios with 95% confidence intervals of incident bladder cancer associated with pioglitazone overall and by both cumulative duration of use and cumulative dose. Similar analyses were conducted for rosiglitazone, a thiazolidinedione not previously associated with an increased risk of bladder cancer.
Results: The cohort generated 689,616 person years of follow-up, during which 622 patients were newly diagnosed as having bladder cancer (crude incidence 90.2 per 100,000 person years). Compared with other antidiabetic drugs, pioglitazone was associated with an increased risk of bladder cancer (121.0 v 88.9 per 100,000 person years; hazard ratio 1.63, 95% confidence interval 1.22 to 2.19). Conversely, rosiglitazone was not associated with an increased risk of bladder cancer (86.2 v 88.9 per 100,000 person years; 1.10, 0.83 to 1.47). Duration-response and dose-response relations were observed for pioglitazone but not for rosiglitazone.
Conclusion: The results of this large population based study indicate that pioglitazone is associated with an increased risk of bladder cancer. The absence of an association with rosiglitazone suggests that the increased risk is drug specific and not a class effect.