Final Report Summary - DIABETES AND CANCER (The association between type 2 diabetes diagnosis and diabetes medications with risk of cancer)
The overall goal of the project was to evaluate the association between type 2 diabetes diagnosis and type 2 diabetes pharmacologic treatment with cancer risk. Two sets of aims were designed and conducted to satisfy the above goal. First, the literature on type 2 diabetes diagnosis and treatment on cancer risk was critically evaluated using cutting-edge synthesis methods. Second, two large observational studies were used to further shed light on the latter associations. Diabetes and cancer are two multifactorial, severe and chronic diseases with a large burden of morbidity and mortality for our society. Therefore, if diabetes diagnosis and/or diabetes medications are even weakly associated with the risk of cancer, this may have important public health consequences.
We first evaluated the literature on biomarker associations, including biomarkers related to the insulin system that is greatly involved in the pathophysiology of diabetes, with cancer risk. We systematically searched electronic databases for meta-analyses of non-genetic biomarkers and cancer risk, and finally included 98 meta-analyses with 847 studies. The proportion of meta-analyses with statistically significant effects was high for the insulin-like growth factor (IGF)/insulin system (52%) biomarkers. An excess of observed versus expected statistically significant results was observed in the studies of IGF/insulin and cancer risk (P≤0.04). In summary, our evaluation suggests that only a minority of these biomarker associations have statistically significant results and no suggestion of bias. We suggested in the manuscript several ways to improve this evidence in the future.
We also assessed critically the literature of type 2 diabetes and cancer risk. A total of 26 meta-analyses were identified in the literature associating diabetes with 26 different cancer sites. 20 (77%) meta-analyses observed a statistically significant summary effect, and most of which reported a positive association between diabetes and cancer risk. Substantial heterogeneity was present in the majority of these meta-analyses (77%), but an excess of statistically significant results was observed only in 5 meta-analyses (liver, bladder, kidney cancer, non-Hodgkin's lymphoma and multiple myeloma). This evaluation suggests that largely type 2 diabetes is a consistent risk factor for the development of several cancers.
Next, we evaluated the association of type 2 diabetes with risk of prostate cancer in a large European cohort. Among 139,131 eligible men, 4,531 were diagnosed with prostate cancer over an average follow-up of 12 years. Compared to men with no diabetes, men with diabetes had a 26% lower risk of prostate cancer. There was no evidence that the association differed by stage or grade of the disease.
Finally, we evaluated the association between antidiabetic medications and cancer risk using state-of-the-art methods to emulate the intention-to-treat analysis of a randomized trial in a large retrospective cohort study within the United Kingdom Clinical Practice Research Datalink. Compared to initiators of sulfonylurea, initiators of metformin had a similar incidence of total cancer (hazard ratio, 0.96; 95% confidence interval, 0.89 - 1.04). There was no evidence for an association for cancers of the colorectum, prostate, lung, postmenopausal breast or any other cancer. When we adjusted for non-adherence, the hazard ratios were very similar to the main findings. In summary, we found no evidence that metformin was associated with a decreased or increased risk of any cancer compared to sulfonylurea. Several large trials are currently planned or are in progress testing metformin for the primary or secondary prevention of several cancers following the evidence from the mechanistic studies and the results from some early epidemiological studies that suggested an inhibitory effect of metformin on cancer risk or progression. However, several of the latter observational studies were potentially biased and should be redone using rigorous designs and analyses to avoid the future initiation of unnecessary trials.