Glucose-lowering agents and the patterns of risk for cancer: a study with the General Practice Research Database and secondary care data

Recent studies suggesting an increased cancer risk with glucose-lowering agents have received widespread publicity. The objectives of this study were to evaluate the comparability in underlying cancer risk and patterns of cancer risk over time with different glucose-lowering agents. The General Practice Research Database (GPRD) was used to identify cohorts of new users. Cancer outcomes were obtained from the GPRD, Hospital Episode Statistics and cancer registries. Relative rates of cancer comparing different glucose-lowering agents were estimated using Poisson regression. A total of 206,940 patients was identified. There was no difference in cancer risk and quartile for HbA(1c) value. There were differences in cancer incidence in the first 6 months after starting treatment (adjusted relative rate of 0.83 [95% CI 0.70, 0.99] with thiazolidinediones, 1.34 [95% CI 1.19, 1.51] with sulfonylureas and 1.79 [95% CI 1.53, 2.10] with insulin, compared with metformin). Insulin users had decreasing cancer incidence over time (adjusted relative rate of 0.58 [95% CI 0.50, 0.68] during months 6-24, relative rate of 0.50 [95% CI 0.42, 0.59] during months 25-60 and relative rate of 0.48 [95% CI 0.40, 0.59] during months 60+) compared with months 0-6 after starting insulin. Similar patterns were found with sulfonylureas and metformin. There were no increases over time with insulin glargine (A21Gly, B31Arg, B32Arg human insulin; relative rate of 0.70 [95% CI 0.52, 0.95], 0.77 [95% CI 0.56, 1.07] and 0.60 [95% CI 0.36, 1.02], respectively, for 6-24, 25-60 and > 60 months). These findings do not provide evidence of either beneficial or adverse effects of glucose-lowering agents on cancer risk and are consistent with changes in diabetes treatment in the few months prior to the diagnosis of cancer.