ESTRO 2020 Abstract book

S617 ESTRO 2020

Asia or Europe; therefore, large randomized trials have underrepresented Hispanic (HISP) patients. Our hypothesis is that gastric adenocarcinoma (GAC) is clinically distinct between HISP, Asian (AS), and non-Asian/non-Hispanic (NANH) populations with regard to demographics, treatment patterns, treatment response, and survival outcomes. Material and Methods The Surveillance, Epidemiology, and End Results (SEER) and National Cancer Database (NCDB) were used to identify patients with pathology-proven GAC from 2004- 2016 and 2010-2014, respectively. Given uncertainties regarding chemotherapy treatment in SEER which would impact survival, only NCDB data was used for survival analyses. Continuous and categorical variables were compared between racial/ethnic (RE) groups using ANOVA and chi-square. Survival analysis was conducted using Kaplan-Meier with log rank tests and Cox proportional hazards for overall survival (OS). To better evaluate the OS benefit of chemotherapy and radiotherapy (RT), only patients with stage 0-III disease were included on multivariate analysis (MVA). Results Results: 39,915 and 22,638 patients met criteria in SEER and NCDB, respectively. Table 1 depicts significant differences in demographic and treatment characteristics between RE groups in both databases. There was a significant difference in every characteristic at presentation including age, sex, and stage (p<0.0001). Using NCDB data, median OS for patients of all stages was 16.1, 24.7, and 14.5 months for the HISP, AS, and NANH groups, respectively (p<0.001). For the 13,375 NCDB patients with clinical Stage 0-III disease, variables significantly associated with OS on MVA included RE, insurance, rural residence, use of chemotherapy, treatment at an academic center, age, RT, comorbidity score, clinical and pathologic stages, grade, and rural county (Table 2). In a subset analysis of clinical Stage II/III patients, there was an OS difference between the three RE groups (p<0.0001) which was eliminated with the use of neoadjuvant chemotherapy (p=0.0664) or neoadjuvant chemoRT (p=0.7282). In contrast, when patients with pathologic stage II/III and ≥16 dissected lymph nodes were examined, there was a difference in OS with the use of adjuvant chemoRT (p=0.0220) (Table 3). T is bstract has been withdrawn