Objective To investigate the risk factors of pancreatic cancer based on the prospective cohort of UK Biobank (UKB).

Methods The study objects from 500, 000 participants were recruited by UKB from 2006 to 2010(the age and gender of pancreatic cancer patients were matched with controls without pancreatic cancer at a ratio of 1:10), and multivariate binary Logistic regression model was used to analyze the risk factors of pancreatic cancer. The population-attributable risk for each risk factor was estimated.

Results A total of 1, 538 cases of pancreatic cancer were included and were compared with 15, 380 controls. Multivariate binary Logistic regression analysis results showed that higher Townsend index (OR=1.020, 95%CI, 1.001 to 1.040), smoking history (OR=1.182, 95%CI, 1.048 to 1.333), history of type Ⅰ diabetes(OR=1.890, 95%CI, 1.228 to 2.910), type Ⅱ diabetes(OR=2.109, 95%CI, 1.794 to 2.481), acute pancreatitis (OR=5.266, 95%CI, 3.679 to 7.538), chronic pancreatitis(OR=17.345, 95%CI, 8.820 to 34.111) and a history of acute and chronic pancreatitis (OR=4.787, 95%CI, 1.989 to 11.522) were all significant independent risk factors for pancreatic cancer (P < 0.05). After multivariate adjustment, the population attributable risk of smoking history, type Ⅰ diabetes, type Ⅱ diabetes, acute pancreatitis, chronic pancreatitis, and both acute and chronic pancreatitis history were 8.40%, 19.86%, 12.37%, 6.50%, 7.30% and 7.13%, respectively.

Conclusion Higher Townsend index, smoking history, type Ⅰ and type Ⅱ diabetes history, acute pancreatitis history and chronic pancreatic history are independent risk factors for pancreatic cancer. Effective identification of risk factors is conducive to reducing the incidence of pancreatic cancer.