Survival analysis of veteran patients with pancreatic cancer

Objective: For patients with pancreatic cancer, the identification of reliable predictors of their outcomes could be invaluable for directing the managements. This study aimed to identify clinical and laboratory factors that could be used to predict early (≤6 months) or late (>6 months) mortality. Methods: Medical records of patients diagnosed with pancreatic cancer in the VA North Texas Health Care System from 2005 to 2010 were retrospectively reviewed. Univariate and multivariate analyses (MVA) were performed and the utility of cancer antigen 19-9 (CA19-9) test was explored. Results: Altogether 109 patients with pancreatic cancer, 89.0% of whom were with adenocarcinoma, were divided into early (n = 62) and late (n = 47) mortality groups. Kaplan–Meier analysis revealed a median survival of 154 days [95% confidence interval (CI) 93–194 days]. On MVA, abdominal pain (OR = 10.6, P = 0.009) and large tumor size (OR = 2.4, P = 0.028) were significantly associated with early mortality, while palliative chemotherapy (OR = 0.048, P = 0.001) and neuroendocrine tumor (OR = 0.009, P = 0.024) were significantly associated with late mortality. Subgroup analyses of adenocarcinoma and late-stage patients revealed similar results. Serum CA19-9 performed poorly as a prognostic indicator in both groups (P = 0.43), in metastatic disease at diagnosis (P = 0.32) and after treatment (P = 0.65). Conclusions: Abdominal pain and large tumor size portends a poor prognosis in patients with pancreatic cancer. Palliative chemotherapy and surgical intervention may prolong the patient's survival. CA19-9 is not universally reliable for predicting metastasis, survival, or the responses to chemotherapy.