Meta-analysis of sex and racial subgroup participation rates and differential treatment effects for trials in solid tumor malignancies leading to US Food and Drug Administration registration between 2010 and 2021.

Abstract

The lack of sociodemographic diversity in clinical trials limits the generalizability of results. The authors examined participation rates and effect modification by sex and race in oncology trials. The authors extracted outcome data stratified by sex and race for registration trials supporting US Food and Drug Administration (FDA) approval (2010-2021). Effect modification by race and sex was examined using quantitative and qualitative methods. A random-effects meta-analysis and pairwise comparison of progression-free survival (PFS) and overall survival (OS) outcomes was conducted by sex and race. Ninety-five trials with 123 end points and 54,365 patients provided information on sex. Trial patients were more often male (n=35,482; 65% vs. 56% male patients in US Surveillance, Epidemiology, and End Results [SEER] data), although the proportion of male patients was similar after adjusting by tumor type (60% in FDA data vs. 58% in SEER data). There was no difference in pooled outcomes among male versus female patients (PFS: hazard ratio, 0.99; 95% confidence interval, 0.92-1.07; p=.89; OS: hazard ratio, 0.99; 95% confidence interval, 0.93-1.07; p=.90). In total, 111 trials including 74,217 patients provided information on race, and 68% of patients identified as White, compared with 72.3% in US SEER incidence data. Black patients were under-represented compared with US SEER incidence data, although ethnicity was poorly reported throughout the data set. In the authors' network meta-analysis by race, there were no statistically significant differences in PFS or OS outcomes. No significant differences in PFS or OS outcomes were identified when the analyses were stratified by sex or race. Certain racial minorities remain under-represented, and clearer reporting of race and ethnicity is needed. Representation of female patients in FDA trials is similar to that in SEER data after adjusting for tumor type.