Institute for Medical Informatics, Biometry and Epidemiology, University Hospital Essen, Germany.
Institute for Biometrics and Epidemiology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich Heine University Düsseldorf, Germany.
J Diabetes Complications. 2024 Oct;38(10):108851. doi: 10.1016/j.jdiacomp.2024.108851. Epub 2024 Aug 30.
Recently, a health-care database study showed that persons with type 2 diabetes taking GLP-1 receptor agonists (GLP-1 RA) had a significantly lower risk of 10 out of 13 obesity-related cancers than patients taking insulin (Wang L, et al. JAMA Netw Open. 2024 7: e2421305). For some cancers, hazard ratios <0.5 were reported. This is reminiscent of studies published >10 years ago showing that people with type 2 diabetes taking metformin had a lower risk of many types of cancer than those not taking metformin. In some studies, also risk reductions of >50 % were reported. The strong effects observed in the metformin studies were explained by time-related biases, in particular, immortal time bias. In the current GLP-1 RA study, it was striking that the curves for the cumulative incidence of several cancers in GLP-1 RA and insulin users diverged immediately after therapy onset. This indicates that there is most likely a time-related bias: insulin is given at much later stages of type 2 diabetes than GLP-1 RA. The current study suggests that one should be sceptical about database results when spectacular risk reductions are reported. Time-related bias should always be considered as an alternative explanation.
最近，一项医疗保健数据库研究表明，与使用胰岛素的患者相比，服用 GLP-1 受体激动剂（GLP-1RA）的 2 型糖尿病患者罹患 13 种肥胖相关癌症中的 10 种的风险显著降低（Wang L, et al. JAMA Netw Open. 2024 7: e2421305）。对于一些癌症，报告的风险比 <0.5。这让人想起了 10 多年前发表的研究，这些研究表明，服用二甲双胍的 2 型糖尿病患者罹患多种癌症的风险低于未服用二甲双胍的患者。在一些研究中，也报告了 >50%的风险降低。二甲双胍研究中观察到的强烈效果可以用与时间相关的偏倚来解释，特别是，无事件时间偏倚。在当前的 GLP-1RA 研究中，令人惊讶的是，GLP-1RA 和胰岛素使用者中几种癌症累积发病率的曲线在治疗开始后立即出现分歧。这表明很可能存在与时间相关的偏倚：胰岛素的使用时间远远晚于 GLP-1RA。本研究表明，当报告令人瞩目的风险降低时，人们应该对数据库结果持怀疑态度。应该始终将与时间相关的偏倚作为替代解释来考虑。
