A historical cohort study on glycemic-control and cancer-risk among patients with diabetes

Aims: This population-based historical cohort study examined whether poor glycemic-control (i.e., high glucose and HbA1c blood levels) in patients with diabetes is associated with cancer-risk. Methods: From a large healthcare database, patients aged 21-89 years, diagnosed with diabetes before January 2002 (prevalent) or during 2002-2010 (incident), were followed for cancer during 2004-2012 (excluding cancers diagnosed within the first 2 years since diabetes diagnosis). Risks of selected cancers (all-sites, colon, breast, lung, prostate, pancreas and liver) were estimated according to glycemic-control in a Cox regression model with time-dependent covariates, adjusted for age, sex, ethnic origin, socioeconomic status, smoking and parity. Missing glucose or HbA1c values were imputed. Results: Among 440,000 patients included in our analysis, cancer was detected more than 2 years after diabetes diagnosis in 26,887 patients (6%) during the follow-up period. Associations of poor glycemic-control with all-sites cancer and most specific cancers were either null or only weak (hazard ratios (HRs) for a 1% HbA1c or a 30 mg/dl glucose increase between 0.94 and 1.09). Exceptions were pancreatic cancer, for which there was a strong positive association (HRs: 1.26-1.51), and prostate cancer, for which there was a moderate negative association (HRs: 0.85-0.96). Conclusion: Overall, poor glycemic-control appears to be only weakly associated with cancer-risk, if at all. A substantial part of the positive association with pancreatic cancer is attributable to reverse causation, with the cancer causing poorer glycemic-control prior to its diagnosis. The negative association with prostate cancer may be related to lower PSA levels in those with poor control.