Cancer risk among insulin users: Comparing analogues with human insulin in the CARING five-country study

Background: Observational studies suggest an association between certain insulin analogues and increased cancer risk. However, methodological shortcomings, short follow-up, and small sample size, have hindered interpretation and generalization of findings. Objectives: The primary aim was to compare the risk of cancer at ten specific sites (prostate, breast, lung, colorectal, bladder, pancreas, liver, corpus uteri, melanoma of skin, and non-Hodgkin lymphoma) between the users of insulin glargine or detemir and human insulin. Methods: In this retrospective study of new insulin users, we used cancer and prescription data from the Norwegian, Swedish, Danish and Finnish National Health Registries and the Clinical Practice Research Datalink from the United Kingdom. We assessed the cumulative insulin exposures in a time-dependent manner. Effect of exposure on cancer incidence was examined by applying multivariate Poisson models to the semi-aggregate data on five cohorts. Results: During a mean follow-up of 4.6 years, a total of 1.45 million person-years accumulated, and 21,298 cancer cases occurred among 327,040 new insulin users. For 0-0.5 years of glargine exposure relative to that of human insulin, we found an increased risk of colorectal (RR=1.49, 95% CI: 1.02 -2.17), and endometrial (1.81, 1.09-3.00) cancers in women, for 2-3 years an increased risk of melanoma cancer (2.18, 1.05-4.52) in women, and a decreased risk of pancreas cancer (0.34, 0.17-0.67) in men, for 3-4 years a decreased risk of liver cancer (0.36, 0.14-0.93) in men, for 4-5 years an increased risk of lung cancer (2.18, 1.05-4.52) and melanoma of skin (3.50, 1.67-7.39) in women, for >6 years a decreased risk of liver cancer (0.22, 0.05-0.93) in men. Similarly, comparisons between detemir and human insulin revealed only a few random associations. Altogether, no trends with longer cumulative use were observed for any of the 10 studied cancer outcomes. Conclusions: Present large-scale study found no major differences in risk of any of the studied cancers between human insulin and different insulin analogues.