Abstract
BACKGROUND: For resected early stage pancreatic cancer, RTOG 9704 evaluated the outcome of 3 weeks of postoperative chemotherapy (C) followed by chemoradiation (CRT) and further C. For unresectable locally advanced pancreatic cancer, a recent literature review of prospective studies showed that the duration of induction C prior to CRT can impact survival. However, the ideal duration of C prior to CRT remains unclear for these patient cohorts. This National Cancer Database (NCDB) study was performed to compare the outcome of various durations of C prior to CRT. METHODS: The NCDB was queried for resected primary stage I-II, cT1-3N0-1M0, and unresected stage III, cT4N0-1M0 pancreatic adenocarcinoma treated with C + CRT (2004-2015). Cohorts I-II and III included stage I-II and stage III cases, respectively. Patients were stratified by short (short C) and long duration (long C) of chemotherapy based on their median durations. Baseline patient, tumor, and treatment characteristics were examined. The primary endpoint was overall survival (OS). Kaplan-Meier analysis, multivariable Cox proportional hazards method, and propensity score matching were used. RESULTS: Among 1577 patients, cohort I-II had 839 patients and cohort III had 738 patients. The longer duration of chemotherapy prior to CRT showed improved OS in the multivariate analysis in both cohort I-II (hazards ratio [HR] 0.72, P < 0.001) and cohort III (HR 0.83, P = 0.03). Using 1:1 propensity score matching, 610 patients for cohort I-II and 542 patients for cohort III were matched. After matching, long C remained statistically significant for improved OS compared with short C in both cohort I-II (median OS 26.1 vs 21.9 months; P = 0.003) and cohort III (median OS 16.7 vs 14.2; P = 0.02). CONCLUSION: Our NCDB study using propensity score-matched analysis showed a survival benefit for using the longer duration of chemotherapy compared to the shorter duration for both resected stage I-II and unresected stage III pancreatic cancer.