A binational cancer registry analysis of Hispanic patients in Mexico and the United States

e22513 Background: A marked disparity in cancer burden exists between regions of the world. This global picture can only be obtained because of data obtained from population-based cancer registries, which allow estimates for different geographic areas. We investigate the data differences, clinical characteristics, tumor information, and patient outcomes of Hispanic patients in North America. Methods: We used data from Mexico’s National Cancer Registry and the United States Surveillance, Epidemiology, and End Results Program (SEER) program with four years of follow-up on each registry. of Mexican registry, we excluded outliers and records with more than 80% missing variable; of the United States registry, only patients identified as hispanic or latino were included. On the Mexican registry we re-coded all variables and the United States registry data on secondary neoplasm was excluded to match boths registry. Baseline characteristics and a geographical information system was constructed to map cancer registries and their relative frequencies. A survival analysis was done to understand the correlation between covariates. Lastly, we build machine learning models to predict 3-year overall survival. Results: The Hispanic cancer database (HCB) consists of 291,178 patients (19,904 from Mexico, and 271,274 from the United States). The top three most frequent cancer types were breast, prostate and hematological. The age of diagnosis was 55±17 years. Mexico has a slight skewness towards the earlier age of diagnosis of females. The registries with the highest burden of cancer were New Mexico (USA) and Baja California Sur (Mexico). Average survival months seem very stable across registries in the United States, but not in Mexico. Our linear regression model achieved a coefficient of determination (R-squared) of 0.49, while the logistic regression achieved an AUC of 0.82, with an F1-score of 0.88. Conclusions: Cancer registries are important tools for prevention and development of control programs. Hispanics are a traditionally neglected population in oncological clinical trials, with low enrollment of patients outside of the United States. Mexico enacted its National Cancer Registry Law in 2017, which alongside the Hispanic data on the SEER program in the United States offers enormous opportunities for continued collaboration and understanding of cancer. Both Mexico and the United States can strengthen their cancer prevention strategies and generate trans-border collaboration, research, and patient-support networks.