Cancer with diabetes: prevalence, metabolic control, and survival in an academic oncology practice

To determine the prevalence of diabetes mellitus, glycemic control, and impact of diabetes on overall survival in an academic oncology practice.

Data on cancer patients (1999 to 2008) were retrieved from the institutional cancer registry and linked to electronic files to obtain diabetes status and hemoglobin A1c (A1C) values within the first 6 months of cancer diagnosis. Overall survival by cancer type with and without diabetes was compared using Cox regression.

Excluding skin and hematologic malignancies, 15,951 cancer cases were identified. Overall diabetes prevalence was 6.8% (n = 1,090), declining over time (P<0.001). Diabetes was common among patients with pancreatic (9.8% [61 of 624]), colorectal (7.7% [89 of 1,151]), or bladder cancers (7.6% [68 of 899]). Patients with diabetes were older (mean age, 70 versus 66 years; P<0.001) and more likely to be male (66.3% [723 of 1,090] versus 60.2% [8,949 of 14,858]; P<0.001). The mean A1C among diabetic cancer patients was 6.8% and did not differ across cancer types (P = 0.80). Only 58.6% (331 of 565) of diabetic cancer patients had all A1C <7.0% during the first 6 months following cancer diagnosis. Pancreatic cancer patients with coexisting diabetes had better overall survival than pancreatic cancer patients without diabetes (hazard ratio, 0.60; 95% confidence interval 0.44 to 0.80; P<0.001). Conversely, diabetic prostate cancer patients had worse overall survival than prostate cancer patients without diabetes (hazard ratio, 1.36; 95% confidence interval 1.05 to 1.76; P = 0.02).

In this academic oncology practice, diabetes was common, glycemic control often was suboptimal, and survival varied by cancer type. Additional study is needed to optimize glucose management and investigate mechanisms underlying age, sex, and survival differences.