A detailed survival analysis questions the rationale behind use of the diabetes drug metformin to improve pancreatic cancer survival.
Several epidemiologic studies have shown that metformin use reduces cancer mortality, leading the diabetes drug to be included in the treatment arm of 20 open clinical trials in recalcitrant cancers, Dr. Roongruedee Chaiteerakij reported at the annual meeting of the American Association for Cancer Research.
The problem is that the epidemiologic studies commonly classified metformin use as simply “ever or never,” which may have introduced unintended biases.
© 2015 AACR/Todd Buchanan
Dr. Roongruedee Chaiteerakij
To address these potential biases, Dr. Chaiteerakij and her colleagues at the Mayo Clinic in Rochester, Minn., analyzed 1,360 patients diagnosed with pancreatic ductal adenocarcinoma (PDAC) and diabetes between 2000 and 2011 in the database of the Mayo Clinic Specialized Programs of Research Excellence (SPORE) in Pancreatic Cancer. More than half (59%) were male; the average age was 67 years.
A total of 380 patients were excluded for surgically induced diabetes, unknown diabetes duration, and PDAC diagnosis more than 90 days prior to the first Mayo visit. This left 980 patients in the final cohort.
An initial analysis using the ever vs. never classification suggested that metformin use was associated with marginally improved survival in patients with PDAC (median 9.9 months ever use vs. 8.9 months never use; unadjusted hazard ratio, 0.9; P = .08), Dr. Chaiteerakij said.
The association was most significant in locally advanced PDAC patients (10.2 months ever use vs. 8.1 months never use; unadjusted HR 0.7; P = .006).
The investigators then performed a subanalysis of locally advanced PDAC patients, this time stratified by timing of metformin initiation: never used (reference group), started more than 1 year before PDAC diagnosis, started within 1 year before PDAC diagnosis, started less than 30 days post PDAC diagnosis, and started more than 30 days post PDAC diagnosis.
Median survival was 8.1, 10.1, and 9.9 months in the first three groups, increasing to 11.4 months and 13.7 months in the two groups that started metformin after PDAC diagnosis, Dr. Chaiteerakij reported.
Hazard ratios for the four metformin groups were 0.7, 0.6, 0.9, and 0.5, after adjustment for age, sex, disease stage, body mass index, and diagnosis year.
The increased survival in patients who started metformin after PDAC diagnosis demonstrates the inherent survival bias in ever/never classification because these patients had lived long enough to receive the drug, she said.
The ever/never classification is commonly used because it can be difficult to extract detailed information on drug use, dose, or timing from retrospective medical records, she noted in a press briefing.
“Epidemiologic studies of medication exposure and cancer survival warrant very careful and detailed data collection and analysis to minimize biases,” Dr. Chaiteerakij concluded. “Researchers should exercise caution when initiating clinical trials based on retrospective epidemiologic studies.”
That said, 11 pancreatic cancer trials are currently listed on www.clinicaltrials.gov, she told reporters.
A simple search of the site for “cancer and metformin” yields no fewer than 230 trials.
It isn’t possible to say at this time whether the Mayo results are applicable to other nonrecalcitrant cancers, Dr. Chaiteerakij said.
On Twitter @pwendl