Influence of ABO blood group on the natural history of advanced pancreatic adenocarcinoma (PC).
307 Background: An association between blood group (ABO) and PC has been demonstrated at epidemiologic and genomic levels. Variations in ABO type may lead to higher pro-inflammatory cytokines levels with modifications in cellular adhesionand signalling promoting carcinogenesis. This study investigated the influence of ABO on the clinical behaviour of advanced PC in patients (pts) , treated with chemotherapy (ctx). Methods: Pts with confirmed PC were identified from 4 institutional databases. Inclusion criteria were unresectable (UPC) or metastatic disease (MPC), receipt of ctx and availability of ABO data. Clinicopathologic details were collected. 200 random anonymied non-cancer ABO samples were collected as control. Descriptive statistics and survival analyses were performed. Results: Between 2001 and 2012, 222 pts met inclusion criteria. Median age was 63 years (range: 33 – 83) 56% were males. 60% of pts had MPC and 27% received doublet ctx. ABO distribution was: A (40%), AB (5%), B (11%) and O (44%). The incidence of blood type A was higher in PC cohort than control (40 vs 30%, p=.03) but identical between UPC and MPC (41 vs 40%, p=.84). Overall survival between type A and non A were identical for the entire cohort (5.8 vs 6.6 mos, HR 1.04 95% CI 0.76 – 1.40, p=.82), UPC (7.6 vs 9.5 mos, HR 1.08 95% CI 0.65 – 1.76, p=.77) or MPC (5.4 vs 4.7 mos, HR 0.94 95% CI 0.66 – 1.34, p=.78). For UPC, 56 pts (64%) had radiographic documentation of the pattern of progression. Type A pts had lower propensity for developing distant metastasis (7/21) than non A (23/35), at 33 vs 66%, p=0.03. Amongst pts with MPC, the incidence of hepatic and pulmonary metastases for type A and non A were identical (77 vs 74%, p=.83; 21 vs 18%, p=.81). However, peritoneal dissemination was less common in type A pts (6 vs 23%, p=.01). Conclusions: Consistent with existing epidemiologic data, the incidence of blood type A is significantly higher in pts with PC, although this does not appear to be stage dependent. ABO did not appear to influence OS in this cohort. However, our data suggests that the pattern of disease spreadmay be related to the ABO blood type. ABO-related glycosylated products could be a target for disease modulation in further studies.