Association of Glycated Hemoglobin Levels with Risk of Pancreatic Cancer

Abstract

Importance: New-onset diabetes after the age of 50 years is a potential indicator of pancreatic cancer. Understanding the associations between hyperglycemia, diabetes, and pancreatic cancer, including pancreatic ductal adenocarcinoma, is key to developing an approach to early detection. Objective: To assess the association of elevation in glycated hemoglobin (HbA1c) with the risk of pancreatic cancer. Design, Setting, and Participants: This cohort study was conducted using data collected from an integrated health care system in California. A total of 851402 patients aged 50 to 84 years who had HbA1cmeasurements taken between 2010 and 2014 were identified as the base cohort, with 12 contemporaneous cohorts created based on varying HbA1cthresholds (ie, 6.1%, 6.3%, 6.5%, and 6.7%) and prior diabetes status. Data analysis was conducted from August 2018 to September 2019. Main Outcomes and Measures: New cases of pancreatic cancer identified through cancer registry and California death files during a 3-year period. Three-year risk, incidence rate, sensitivity, number of patients needed to screen to detect 1 case, timing, and stage at diagnosis were determined. Results: Among 851402 patients in the base cohort, 447502 (52.5%) were women, 255441 (30.0%) were Hispanic participants, 383685 (45.1%) were non-Hispanic white participants, 100477 (11.8%) were Asian participants, and 88969 (10.4%) were non-Hispanic black participants, with a median (interquartile range) age of 62 (56-69) years and a median (interquartile range) HbA1clevel of 6.0% (5.7%-6.6%). The incidence rate of pancreatic cancer was 0.45 (95% CI, 0.43-0.49) per 1000 person-years. After excluding prior diabetes as well as confirmation of new-onset hyperglycemia based on an HbA1clevel of 6.5%, a total of 20012 patients remained, with 74 of 1041 pancreatic ductal adenocarcinoma cases (7.1%) from the base cohort included. The rate of pancreatic cancer was 0.72 (95% CI, 0.32-1.42) per 1000 person-years among Asian patients, 0.83 (95% CI, 0.35-1.71) per 1000 person-years among non-Hispanic black patients, 0.84 (95% CI, 0.48-1.37) per 1000 person-years among Hispanic patients, and 2.37 (95% CI, 1.75-3.14) per 1000 person-years among non-Hispanic white patients. Overall, 42 of 74 cancers (56.8%) were diagnosed within 1 year of the index laboratory test. Among 1041 total cases, 708 (68.0%) had staging information available, of whom 465 (65.7%) had stage III or IV disease at diagnosis. In the base cohort, the number needed to undergo evaluation to identify a single case of pancreatic ductal adenocarcinoma was 818 (95% CI, 770-869), with estimates ranging from 206 (95% CI, 160-264) to 600 (95% CI, 540-666) in the subcohorts. Conclusions and Relevance: The findings of this study suggest that screening patients for pancreatic cancer based solely on elevation in HbA1clevel is unlikely to represent an effective strategy. Future efforts to identify a high-risk population based on changes in glycemic parameters should account for racial/ethnic differences.