Radiation Utilization and Overall Survival Trends for Resected Pancreatic Cancer in the United States: An Analysis of the Surveillance, Epidemiology, and End Results (SEER) Database

Abstract

The efficacy of radiation therapy (RT) for patients with resected pancreatic cancer remains controversial since the publication of several phase III trials published in the last several decades with conflicting results. One of these studies, ESPAC-1, which was published in 2004 showed chemotherapy plus RT (CRT) had worse overall survival (OS) vs. chemotherapy alone. We examined the SEER database to determine if radiation utilization and OS trends for resected pancreatic cancer has changed since the ESPAC-1 publication. The SEER 18 registries database was queried to identify patients age 18 and older with pathologically confirmed pancreatic cancer diagnosed from 1998 to 2016 that were staged as either localized or regional, and completed surgical resection. Patients without staging were excluded, as well as those with metastatic disease, or <3 months of follow up. The chi square test was used to calculate significance of temporal trends. Patients were separated into three groups (all received chemotherapy): neoadjuvant RT (NART), post-operative RT (PORT), or chemo alone to analyze OS between groups using the Kaplan-Meier method and log-rank test. Multivariate analysis using a Cox’s proportional hazard model was utilized with the co-variables: RT use, age, sex, year of diagnosis, N stage, and T stage. A total of 27,258 patients met the inclusion criteria and were included in the study. Prior to the ESPAC-1 publication, the utilization rate of PORT in 2003 peaked at 43.2% followed by a decline on average of 2.0% per year starting in 2004 to a rate of 16.7% in 2016 (p<0.001). The use of NART increased from 1.4% in 1998 to 5.3% in 2015 (p = 0.01). In this same time period, the median and 2-year OS increased from 15 months and 30% in 1998 to 30 months and 56% in 2014. For all years, the median OS for the NART group was 25 months compared to 22 months and 21 months in the chemo alone and PORT groups, respectively (log rank p = 0.011). NART use correlated with improved OS for patients with node positive disease (median OS 22 months vs 20 months and 19 months for PORT and chemo alone respectively, p = 0.001) or T3-T4 disease compared to chemo alone (p = 0.005). NART patients were significantly less likely to have N+ disease (28.8% vs 65.7% and 61.1%, PORT and chemo alone respectively, p<0.001). Conversely, NART or PORT use correlated with worse OS for patients with node negative patients (chemo alone median OS 32 months vs. 28 months and 27 months for PORT and NART respectively, p<0.01). Margin status could not be assessed from the SEER database. After multivariate adjustment, NART remained highly significant for improved OS compared to PORT or chemo alone. (p<0.001). Trends in PORT utilization are declining for resected pancreatic cancer since 2004 based on this SEER database analysis which coincides with the publication of the ESPAC-1 trial. However, NART utilization increased over the study period and was associated with improved OS compared to PORT or chemo alone.