Radiation therapy sequencing for resected pancreatic adenocarcinoma in the National Cancer Data Base: A multi-institutional comparative analysis.

Abstract

305 Background: Pancreatic adenocarcinoma (PAC) remains a devastating malignancy and the role for preoperative radiation therapy (prRT) remains uncertain. This analysis used the National Cancer Data Base (NCDB) to report outcomes for the largest known retrospective cohort of patients treated with prRT for resectable PAC to date. Methods: NCDB data were obtained for all patients who underwent resection and RT for PAC from 1998-2002. Patients with metastatic (M1) disease, combined prRT and poRT, missing OS or missing RT variables were excluded. The difference in patients’ characteristics was assessed by a Chi-square test and ANOVA. The Kaplan-Meier method was used to estimate overall survival (OS); survival differences were assessed using a log-rank test. A Cox proportional hazards model was used for unadjusted and multivariable (MV) survival analysis. Two MV logistic regression models also were used to assess for differences in margin status and lymph node (LN) status across treatment groups. Results: 5,414 analyzable patients were identified. 5.1% (277/5,414) received prRT and 94.9% (5,137/5,414) received poRT. All patients underwent resection and prRT or poRT +/- chemotherapy; 92.9% chemo. 793 patients were AJCC 5th edition stage I, 1,002 stage II, 2,990 were stage III, and 522 were stage IVa; 107 patients were M0 with unknown overall stage. There was no significant difference in OS between the two groups on unadjusted Kaplan-Meier analysis. In covariate-adjusted analysis, there was no significant difference in OS in the prRT (HR, 1.20 [95% CI .98-1.48]; p = 0.08) compared with poRT; prRT was associated with lower rates of LN positivity (OR, 0.45 [95% CI 0.31-0.65]; p < 0.001) and higher rates of negative margins (OR 1.71 [95% CI 1.17-2.52]; p = 0.006). Conclusions: In the largest published series of prRT patients to date, prRT was associated with negative margins and LN negativity at resection. There was no significant difference in OS between the groups. Limitations of this study include its retrospective nature, a small prRT cohort and lack of comorbidity data. These data may be useful in generating future prospective studies of prRT for PAC.