Abstract
IntroductionCancer clinical trials do not always represent the real-world cancer population, as older adults and marginalized racial groups are often underrepresented. This study assessed patterns of enrollment of underrepresented patients and how trial and site factors may influence enrollment.MethodsThis retrospective, pooled cross-sectional study used de-identified data from ECOG-ACRIN (EA)-sponsored breast cancer therapeutic clinical trials from 2002-2020. Patient- and trial-level data were extracted from EA trials and ClinicalTrials.gov. Site-level data were from de-identified Landscape Assessment surveys voluntarily completed by National Cancer Institute Community Oncology Research Program sites. Outcomes included the proportions of underrepresented patients enrolled on a trial. Fractional regression models evaluated associations between trial-level factors and enrollment proportions of underrepresented patients using adjusted means and 95% confidence intervals (CI). Weighted Kappa (Kw) statistics and corresponding 95% CI estimated the level of agreement between patient populations served by sites versus enrolled on a trial.ResultsOf 9,015 patients enrolled across 12 trials, 18% were aged ≥65 years old, 12% were Black, 15% were Medicare beneficiaries, and 15% were rural residents (unadjusted enrollment). Adjusted mean proportion enrollment of underrepresented patients was similar to unadjusted results. Over half of trials were randomized and 92% of studies had two or more drugs in the protocol, yet these did not appear to influence mean enrollment of underrepresented groups. Moderate levels of agreement were found between Black patients served versus enrolled (Kw 0.64; 95% CI 0.50-0.78), and low or no agreement for Medicare beneficiaries and patients aged ≥65 (Kw 0.18, 0.05-0.31; Kw 0.02, -0.06-0.11; respectively).ConclusionsEA-sponsored breast trials continue to enroll few individuals from underrepresented backgrounds. Trial design had a minimal impact on enrollment and patient populations sites served did not typically match the patients enrolled on trials. More research is needed to engage sites and test strategies for enrolling underrepresented patients.