Metformin use was associated with a decreased risk of pancreatic cancer in women, but not in men, in a large case-control study involving more than 19,300 U.K. primary care patients.
Additionally, use of insulin and sulfonylureas was found in the study to be associated with increased risk of pancreatic cancer in both men and women, Dr. Michael Bodmer of the University of Basel, Switzerland, and his colleagues reported online in the Jan. 31 issue of the American Journal of Gastroenterology.
Overall, metformin use was not found to be associated with pancreatic cancer in 2,763 patients with a first-time diagnosis of pancreatic cancer, compared with 16,578 matched controls with no such diagnosis – all from the U.K.-based General Practice Research Database, which includes data on more than 8 million individuals (adjusted odds ratios, 1.01, 0.92, and 0.87 for those who received 1-9, 10-29, and 30 or more metformin prescriptions, respectively), the investigators said.
The findings were similar when analysis was restricted to cases with pancreatic surgery, radio/chemotherapy, or with a specific oncology code, and also when analysis was restricted to diabetic patients. However, in an analysis by gender, long-term metformin use was associated with a decreased risk of pancreatic cancer in women (adjusted OR, 0.43), but not in men (adjusted OR, 1.59), they found (Am. J. Gastroenterol. 2012 Jan. 31 [doi10.1038/ajg.2011.483]).
As for long-term use of sulfonylureas and insulin, the overall analysis demonstrated a "materially increased risk of pancreatic cancer (adjusted OR, 1.90 and 2.29, respectively), and as with metformin, the findings were closely similar in those with pancreatic surgery, radio/chemotherapy, or an oncology code, while in the analysis restricted to patients with diabetes, the adjusted odds ratios for long-term use of sulfonylureas and insulin were 1.59 and 1.69, respectively.
In contrast with the findings regarding metformin, the increased risk of pancreatic cancer among long-term users of sulfonylureas was mainly restricted to women (adjusted OR, 3.05), the investigators said.
Cases in this study were individuals with a first-time diagnosis of pancreatic cancer before the age of 90 years between 1995 and 2009. Controls had no diagnosis of pancreatic cancer and were matched based on calendar time, age, sex, general practice, and number of years of active history in the database before the index date. The average age was 70 years, and 46% were men.
The analyses were conducted using the index date shifted back in time by 2 years to account for cancer latency, the investigators explained.
Using that same shifted index date, the investigators also looked for a link between diabetes history and the risk of pancreatic cancer, and found one. "Overall, the adjusted odds ratio of pancreatic cancer associated with a recorded diagnosis of diabetes mellitus was 1.22. When we stratified the risk of pancreatic cancer by diabetes duration, the odds ratio decreased with increasing duration of diabetes; the odds ratios for a diabetes history of less than 2, 2-5, 5-10, or greater than 10 years were 1.71, 1.21, 0.94, and 0.71, respectively," they said.
In the analysis using the recorded index date, rather than the shifted date, the overall odds ratio for pancreatic cancer in diabetic patients compared with nondiabetic patients was 1.60, and for those with disease duration of less than 2 years, 2-5 years, 5-10 years, and greater than 10 years were 3.30, 1.13, 0.72, and 0.56, respectively.
The finding of an effect modification by gender for the use of metformin was "somewhat unexpected and could not be explained by confounding by use of estrogens," the investigators said, noting that the finding should be interpreted with caution since it is based on a limited number of exposed cases and controls, and since there is no obvious pathophysiologic explanation.
Also, the findings regarding metformin differed somewhat from those of prior smaller studies showing an overall decrease in pancreatic cancer with long-term metformin use.
The findings regarding sulfonylureas and insulin, however, support those of prior studies, and as with metformin, effect modification by gender seemed to have a role: The increased risk with long-term use of sulfonylureas was mainly seen in women, and the increased risk with insulin was mainly seen in men.
"Of importance, in our study, neither short-term use of metformin or sulfonylureas, nor insulin was associated with altered relative risk estimates for pancreatic cancer, indicating that our study design effectively eliminated detection bias and/or bias by intensification of antidiabetic therapy due to undiagnosed pancreatic cancer," they said.
Also the finding of a link between diabetes mellitus and increased risk for pancreatic cancer, which was restricted to those with newly diagnosed diabetes, strongly supports the hypothesis suggested by other authors that short-term diabetes mellitus is likely caused by pancreatic cancer.