Both serology-based and genetic studies have reported an association between pancreatic cancer risk and ABO blood groups. We have investigated this relationship in a cohort of pancreatic cancer patients from Western Norway (n =237) and two control materials (healthy blood donors, n = 379; unselected hospitalized patients, n =6149). When comparing patient and blood donor ABO allele frequencies, we found only the A1 allele to be associated with significantly higher risk for pancreatic ductal adenocarcinoma (PDAC) (23.8% vs. 17.9%; OR = 1.43, P =0.018). Analyzing phenotypes, blood group A was more frequent among PDAC cases than blood donors (50.8% vs. 40.6%; OR = 1.51, P =0.021), an enrichment fully explained by the A1 subgroup. Blood group O frequency was lower in cases than in blood donors (33.8% vs. 42.7%; OR = 0.69, P =0.039). This lower frequency was confirmed when cases were compared to hospitalized patients (33.8% vs. 42.9%; OR = 0.68, P =0.012). Results for blood group B varied according to which control cohort was used for comparison. When patients were classified according to surgical treatment, the enrichment of blood group A was most prominent among unresected cases (54.0%), who also had the lowest prevalence of O (28.7%). There was a statistically significant better survival (P =0.04) for blood group O cases than non-O cases among unresected but not among resected patients. Secretor status did not show an association with PDAC or survival. Our study demonstrates that pancreatic cancer risk is influenced by ABO status, in particular blood groups O and A1, and that this association may reflect also in tumor resectability and survival.
The relationship between ABO blood groups and pancreatic ductal adenocarcinoma was investigated in patients from Western Norway. There was an over-representation of blood group A among patients when compared with healthy blood donors, and this association could be fully attributed to the A1 subgroup. Blood group O was found to be protective, both when blood donors and random hospitalized patients were used as controls. The frequency of blood group O was particularly low among unresected patients, in whom this blood group also associated with better survival.