Health Disparities among Patients with Cancer Who Received Molecular Testing for Biomarker-Directed Therapy

Abstract

Health disparities present a barrier to successful oncology treatment. The potential for precision oncology to reduce health disparities has not previously been analyzed. We performed a retrospective analysis of 12,627 patients from six major cancer centers whose tumors underwent molecular testing at Caris Life Sciences between 2010 and 2020. Kaplan-Meier and Cox regression were used to describe and analyze overall survival. The molecular and demographic features of the cohort were analyzed by χ2 and ANOVA tests. Black patients composed 25% of the cohort and White patients 63%. Among this molecularly-tested cohort, there were minimal outcome differences based on race, geographic location, or poverty level. When analyzing the interaction of age, race, and sex, racial-based disparities were noted primarily for young non-White women in the study cohort but were more pronounced for men and women of all ages in the broader patient population within the Surveillance, Epidemiology, and End Results database. Mutations in five genes-APC, EGFR, STK11, TP53, and KRAS-were found to affect overall survival among our cohort, and their prevalence varied by race in specific tumor types. Real-world outcomes data in mutation-defined cohorts also provided additional context to previously reported therapeutic response trends. Our study shows that patients who undergo molecular testing display reduced racial health disparities compared with the general population, whereas persistent racial disparities are influenced by age and sex. Genomic-driven racial disparities should be examined at a tumor lineage-specific level. Increased access to molecular testing for all eligible patients may play a role in improving health equity. Significance: This study is the largest of its kind to analyze health disparities and genomic features among a diverse multiinstitutional cohort of patients who underwent molecular testing. Continuing to increase awareness of and access to molecular testing approaches may help to reduce cancer health disparities and improve outcomes for all patients.

Figure 1

Tumor lineages of the cohort and analysis of OS by race. A, Total numbers of each tumor type broken down by study site. B, Percentage of the total made up by each study site for the top 10 most represented tumor types. C, Kaplan–Meier curve for OS of White patients (black line) and Black patients (red line) among molecularly-tested cohort. D, COXPH analyzing the effect of race on OS in individual tumor types. *, P < 0.05. E and F, Median survival among the Caris cohort (E) or 5-year survival among the SEER cohort (F) was plotted for male and female White and non-White patients in three different age groups (<50 years; 50 ≤ years < 65; and ≥65).

Figure 2

Genomic features and clinico-demographic associations of cohort. A, Total number of tumors positive for each mutated gene. B, Percentage of total tumors from each study site for the top 20 most commonly mutated genes. C, Prevalence of co-mutations among APC, EGFR, STK11, TP53, and KRAS. Red indicates strong co-mutation whereas blue indicates mutual exclusivity. D, Percentage of indicated tumor types with mutations in APC, EGFR, STK11, TP53, and KRAS. E, Age of patients among (−) and (+) gene mutation cohorts. The red line indicates the median age. F, Ratio of males to females in (−) and (+) gene mutation cohorts. G, Percentage of total (−) and (+) cohorts composed by indicated racial groups. *, P < 0.05; **, P < 0.01; #, P < 0.001; ns, not significant.

Figure 3

Gene mutations by race. A, Percentage of tumors from Black versus White patients that are positive for the indicated gene mutations. #, P < 0.001; ns, not significant. B–E, Probability of Black patients having a TP53 (B), APC (C), KRAS (D), or PIK3CA (E) mutation relative to White patients for each individual tumor type. *, P < 0.05; #, P < 0.001. When data are not shown, we did not produce confidence intervals because of low gene mutation prevalence for the indicated cancer type.

Figure 4

Analysis of treatment-associated outcomes and effect of mutation status among patients in Caris CODEai clinico-genomic database. A and B, Kaplan–Meier analysis comparing posttreatment OS between patients with STK11-mutated (mut, blue lines) and STK11-wild type (wt, red lines) NSCLC treated with (A) carboplatin (no pembrolizumab) and (B) pembrolizumab (no carboplatin). C and D, Kaplan–Meier analysis comparing (C) posttreatment OS and (D) TOT between patients treated with pemetrexed (blue lines) and osimertinib (red lines) for EGFR-mut NSCLC. E and F, Kaplan–Meier analysis comparing posttreatment OS between patients with EGFR-mut (blue lines) and EGFR-wt (red lines) NSCLC treated with (E) osimertinib or (F) pemetrexed. G and H, Kaplan–Meier analysis comparing (G) posttreatment OS and (H) TOT between patients treated with fluorouracil (blue lines) and capecitabine (red lines) for colorectal cancer.