Type 2 Diabetes and Cancer: An Umbrella Review of Observational and Mendelian Randomization Studies

Abstract

Background: Type 2 diabetes mellitus (T2DM) has been associated with an increased risk of developing several common cancers, but it is unclear whether this association is causal. We aimed to summarize the evidence on T2DM and cancer and evaluate the validity of associations from both observational and Mendelian randomization (MR) studies.

Methods: We performed an umbrella review of the evidence across meta-analyses of observational studies that examined associations of T2DM with risk of developing or dying from site-specific cancers, and MR studies that explored the potential causal association of T2DM and associated biomarkers with cancer risk.

Results: We identified eligible observational meta-analyses that assessed associations between T2DM and cancer incidence for 18 cancer sites, cancer mortality for seven sites, and cancer incidence or mortality for four sites. Positive associations between T2DM and six cancers reached strong or highly suggestive evidence. We found eight MR studies assessing the association of genetically predicted T2DM and seven and eight studies assessing the association of genetically predicted fasting insulin or fasting glucose concentrations, respectively, upon site-specific cancers. Positive associations were found between genetically predicted T2DM and fasting insulin and risk of six cancers. There was no association between genetically predicted fasting plasma glucose and cancer except for squamous cell lung carcinoma.

Conclusions: We found robust observational evidence for the association between T2DM and colorectal, hepatocellular, gallbladder, breast, endometrial, and pancreatic cancers.

Impact: Potential causal associations were identified for genetically predicted T2DM and fasting insulin concentrations and risk of endometrial, pancreas, kidney, breast, lung, and cervical cancers.

Figure 1.

Flow diagram of selection process of meta-analyses of type 2 diabetes and cancer in observational studies.

Figure 2.

Summary random effects estimates with 95% confidence and prediction intervals from 29 meta-analyses of type 2 diabetes and cancer incidence, mortality, or both.

Figure 3.

Flow diagram of selection process of Mendelian randomization studies.

Figure 4.

Triangulation of evidence from observational and Mendelian randomization studies assessing association between type 2 diabetes and site-specific cancers. Bubble size corresponds to the number of cases in the corresponding meta-analysis (more cases→larger bubble). If no meta-analysis was available but an MR analysis was, then bubble size represents the number of cases in the MR analysis. Unless stated as incidence/mortality (i.e., both), is incidence, *, mortality.

Figure 5.

Triangulation of evidence from observational and Mendelian randomization studies assessing association between fasting insulin and site-specific cancers. Bubble size corresponds to the number of cases in the corresponding meta-analysis (more cases→larger bubble). If no meta-analysis was available but an MR analysis was, then bubble size represents the number of cases in the MR analysis. Unless stated as incidence/mortality (i.e., both), is incidence, *, mortality.

Figure 6.

Triangulation of evidence from observational and Mendelian randomization studies assessing association between fasting glucose and site-specific cancers. Bubble size corresponds to the number of cases in the corresponding meta-analysis (more cases→larger bubble). If no meta-analysis was available but an MR analysis was, then bubble size represents the number of cases in the MR analysis. Unless stated as incidence/mortality (i.e., both), is incidence, *, mortality.