EUS and survival in patients with pancreatic cancer: a population-based study

Background: There is no direct evidence that EUS improves patient outcome. Objective: To study the association of undergoing EUS with survival in patients with pancreatic adenocarcinoma. Design: Population-based study. Patients: Persons aged 65 years and older with a diagnosis of pancreatic cancer who were captured in the linked Surveillance Epidemiology and End Results-Medicare database between 1994 and 2002 were identified. Interventions: Demographic, cancer-specific, and EUS procedural information was extracted, and survival curves were compared for patients who underwent EUS in the peridiagnostic period (1 month before the diagnosis to 3 months after the date of diagnosis: group I) with those who had not undergone EUS (group II). Main Outcome Measurements: Relative hazard ratios for survival. Results: A total of 8616 patients with pancreatic adenocarcinoma were identified. Only 610 (7.1%) patients underwent EUS evaluation. In patients with locoregional cancer, the median survival (interquartile range) in group I and II patients was 10 (5-17) and 6 (2-12) months, respectively, P < .0001. There were more patients with early-stage disease in group I than group II (69.3% vs 36.2%, P < .001). Curative-intent surgery, chemotherapy, and radiation therapy were also performed more frequently in the patients in group I. Undergoing EUS, adjusted for age, race, sex, tumor stage, curative-intent surgery, chemotherapy, radiation therapy, and comorbidity score, was an independent predictor of improved survival (relative hazard, 0.71; 95% CI, 0.63-0.79). Limitations: Retrospective design. Conclusions: EUS evaluation is independently associated with improved outcome in patients with locoregional pancreatic cancer, possibly because of detection of earlier cancers and improved stage-appropriate management including more selective performance of curative-intent surgery.