Risk of stroke with thiazolidinediones: a ten-year nationwide population-based cohort study.

BACKGROUND: Thiazolidinediones (TZDs) - rosiglitazone and pioglitazone - a class of insulin sensitizer for treating type 2 diabetes, have been reported to exhibit neuroprotective effects in preclinical studies and have good effects in the control of blood sugar for diabetic patients with insulin resistance. However, clinical trials and observational studies have raised the possibility of higher stroke risk in patients treated with rosiglitazone. Whether pioglitazone poses similar stroke risk remains uncertain. Most of the studies on cardiovascular effects of TZDs were based on studies in the USA and Europe. The present study aimed to compare the stroke risk among diabetic patients on TZD to those on non-TZD medications in an Asian population.

METHODS: The study cohort included 15,981 patients with a diagnosis of diabetes without prior stroke, acute myocardial infarction (AMI) or heart failure who were followed from 2001 to 2010. Patients were classified by their prescriptions into rosiglitazone, pioglitazone and non-TZD groups. The study end points included ischemic and hemorrhagic stroke. In view of the reported association of heart failure and AMI with rosiglitazone, these 2 end points were also included in the present study. Cox hazard proportional models were used to estimate the risk of developing the end points. Likelihood ratio test was used to examine the age-drug interactions. Dose-response effects were evaluated by comparing the incidence rates among patients with different cumulative exposures to TZD.

RESULTS: During the 10-year follow-up, the rosiglitazone group showed significantly higher risk of ischemic stroke (multivariate adjusted hazard ratio, HR = 1.39; 95% confidence interval, CI = 1.16-1.66) and heart failure (HR = 1.59; 95% CI = 1.18-2.14) than the non-TZD group. The pioglitazone group did not show significant difference from the non-TZD group in ischemic stroke (HR = 0.97; 95% CI = 0.75-1.26) and heart failure (HR = 0.94; 95% CI = 0.59-1.50). The results also showed a significant dose-dependent effect of higher risk of ischemic stroke with increasing dosage of rosiglitazone, while there was no increased risk at any level of pioglitazone dosage.

CONCLUSIONS: This population-based cohort study shows that rosiglitazone imposes a higher risk of developing stroke or heart failure in this Asian patient population, suggesting the adverse side effects of rosiglitazone across ethnic boundaries. Pioglitazone, on the other hand, does not increase cardiovascular or stroke risk compared to the non-TZD group among diabetic patients without a history of macrovascular disease.