MON-008 Use of National Death Index to Assess Cancer Mortality and Diabetes Relationship

Background: We have previously examined the link between diabetes mellitus (DM) and malignancy with respect to glycemic control and survival. Given that mortality data derived from our own records is limited, and to better determine if DM changes mortality risk in cancer patients with co-existing DM, we linked our clinical records with the National Death Index (NDI), housed and maintained by the Centers for Disease Control and Prevention. Objective: The primary objective of this study was to evaluate the risk DM imposes on mortality in cancer patients via the National Death Index. The secondary objective was to ascertain the cause of death in cancer patients with and without DM. Methods: We identified 1565 patients with newly diagnosed prostate, breast, lung, colorectal, and pancreatic cancer from the institutional cancer registry from 2010-2015. This data was forwarded to the CDC/NDI for input on whether each patient was alive or deceased, and cause of death (if applicable). NDI records were reviewed and verified with a resulting data set containing 1404 patients, matched 2:1; cancer without DM (n=936) and cancer and concurrent DM patients (n=468). Patients were matched by year of diagnosis, age, gender, and cancer type. Primary cause of death was categorized into cancer-related, vascular, non-cancer/non-vascular, and unknown. Five-year survival was estimated with the Kaplan-Meier method and Cox proportional hazards regression analysis was used with the matched pair as the strata variable. Results: Distribution of cancer types was: prostate (43%), lung (19%), breast (15%), colorectal (12%) and pancreatic (11%). Mean (SD) age at diagnosis was 68 (10) years; 70% of patients were male. There were 435 deaths with 168 deaths (36%) in the DM cohort and 267 (29%) in the non-DM cohort. Median follow-up was 47.3 (0.5-94.1) months across all cancer types. Overall 5-year survival was estimated at 59% (95% CI: 54-64%) for DM patients versus 67% (95% CI: 64-71%) in non-DM patients. Matched pairs Cox regression was HR= 1.35; 95% CI 1.02-1.79; p =0.04. There were 134 cancer related deaths (79%) in the DM cohort, and 233 (87%) in the non-DM cohort. There were 26 non-cancer, non-vascular deaths (15% in the DM cohort, and 13 (5%) in the non-DM cohort (p<0.001). Conclusions: Overall there is an increased risk of death for DM vs non-DM patients who have cancer. There is a higher proportion of non-cancer, non-vascular deaths in DM patients than in non-DM patients.