Racial/ethnic representation and disparities in preclinical cancer models.

Abstract

1601 Background: Patient-derived xenograft models (PDXs) recapitulate tumor characteristics credibly and have become a standard for preclinical inquiries that form the basis of clinical trials of novel therapies in oncology. While ample evidence reveals racial/ethnic disparities in cancer care delivery and clinical research, limited data exists regarding racial composition of available PDXs. We sought to define the extent of racial/ethnic representation and disparities among existing PDXs. Methods: Data regarding available PDXs was gathered from the publicly accessible CancerModels.org website ( https://www.cancermodels.org/overview ). Seven members of the research team were involved in data extraction. Information on race/ethnicity (White, Black, Hispanic, Asian), sex, age, and cancer type were recorded. The primary objective was to determine the racial/ethnic composition of PDX models and compare this to racial/ethnic demographics of cancer patients, for which we used US population-based cancer estimates calculated using National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER Incidence Data, 11/2022 Submission (1975 - 2020), SEER 22 registries). Descriptive statistics were used. Proportions were compared using Fischer’s exact test or Chi-squared tests with Yates' correction (odds-ratio [OR] and 95% confidence intervals [95%CI] or Woolf logit interval) were reported. Results: We reviewed 4597 unique PDXs across 33 SEER cancer sites spanning 11 oncology sub-specialties. Of these, 55% models were derived from males and age groups were (years): < 20: 6%; 20-70: 69% and ≥ 70: 25%. Most common cancer sites represented were colorectal (26%), lung (12%), breast (9%), melanoma (9%) and leukemia (7%). Race/ethnicity was not reported in 3395 (73.9%) cases. Racial/ethnic composition of the remaining models was Whites (80.9%), Blacks (7.3%), Hispanics (6.4%) and Asians (5.4%). Compared with their respective proportion of US cancer incidence (69.9%, 10.9%, 13.2% and 5.9%, respectively), these models were over-representative for Whites (OR: 1.82, 95%CI: 1.6-2.1, P < 0.001) and under-represented Blacks (OR: 0.64, 95%CI: 0.5-0.8, P < 0.001) and Hispanics (OR: 0.45, 95%CI: 0.4-0.6, P < 0.001) but not Asians (OR: 0.89, 95%CI: 0.7-1.2, P = 0.43). Similar trends were seen in subgroups focused by cancer site. Among colorectal (N = 1201) models, race/ethnicity was reported for 17.2% of cases and Blacks (OR: 0.52; 55.7% of expected; P = 0.014) and Hispanics (OR: 0.57; 60.8% of expected; P = 0.015) were underrepresented compared to Whites (OR: 1.82; 116% of expected; P < 0.001). Conclusions: Race/ethnicity are infrequently reported for PDX models. Minority races/ethnicities (Blacks and Hispanics) are underrepresented in preclinical models compared to their burden of cancer incidence. There is a need to develop a diverse repertoire of preclinical models to ensure inclusivity and guide equitable research.