Understanding causes of racial/ethnic survival disparity in 47,178 patients with colorectal cancer: A quantitative evaluation of molecular, socioeconomic, and clinical covariates.

Abstract

20 Background: A disparity in overall survival (OS) of colorectal cancer (CRC) patients based on self-reported race has been observed. Separately, differing frequencies of driver gene mutations among racial groups have been observed. The contribution of these molecular differences to disparity in outcomes is not well understood. We conducted a systematic investigation of molecular, socioeconomic (SES) and clinical covariates to quantitate the degree to which each influences the OS disparity. Methods: The Foundry software platform was used to query electronic health records to identify CRC patients and extract clinical and molecular data. The relative contribution of each variable to OS disparity was determined using mediation analysis consisting of sequential multivariate Cox regression models. Results: 47,178 patients with CRC diagnosed between 1973 and 2023 with self-reported race were identified. Relative to Non-Hispanic white (NHW), African American (AA) had worse OS (HR=1.16, P= 1.5E-6), while Asians and Hispanics had better OS (HR=0.66, P=4.2E-13 and HR=0.86, P=3.4E-6, respectively). The magnitude of disparity between AA and NHW was greater in metastatic disease (HR 1.2, P=1.7E-4). Among 7,628 patients with clinical molecular testing, five genes differed significantly in mutation frequencies; APC (54% NHW, 62% AA, 57% Hispanics, 46% Asian, FDR= 0.01), KRAS (45%, 58%, 50%, 44%, FDR=8.4E-9), PIK3CA (17%, 23%, 17%, 12%, FDR=0.006) were higher in AA, while BRAF (8%, 3%, 4%, 7%, FDR=2.4E-8), KIT (4%, 1%, 2%, 3%, FDR=0.013) were higher in NHW. In multivariate analysis mutations of BRAF (HR= 1.9, P=1.2E-8), KRAS (HR= 1.39, P=1.7E-7), and APC (HR= 0.81, P= 5E-4) were significantly associated with OS as was national area deprivation index (HR= 1.07, P=0.004), marital status (HR= 1.16, P=0.03), performance status (HR= 1.42, P= 5.3E-29), disease stage (HR= 1.502, P=4.4E-26), right sided tumors (HR= 1.23, P=0.003), tumor differentiation (HR= 1.22, P=0.004) and MSI-high (HR= 0.37, P=1.8E-7). Mediation analysis identified neighborhood SES as the greatest contributor to disparity (38%), followed by primary tumor side (10%), and molecular characteristics (MSI-status, KRAS and BRAF mutations, 10%). After including all covariates, 39% of disparity remained unaccounted. Conclusions: Systematic evaluation of a large cohort from a single institution confirmed substantial racial disparity in survival and frequencies of driver mutations. SES factors have the largest contribution but account only for forty percent of the disparity, with tumor molecular features also contributing. Molecular characteristics should be incorporated in multicentric studies of racial survival disparity. Future clinical studies should ideally account for genetic ancestry and molecular features when evaluating chemotherapy outcomes.