Patient-Level KRAS G12C Prevalence among Hispanic/Latino and Non-Hispanic White Patients in Real-world Clinicogenomic...

KRAS G12C is a treatment-defining biomarker in non-small cell lung cancer (NSCLC), particularly lung adenocarcinoma. Hispanic/Latino (H/L) patients remain undercharacterized in precision oncology datasets, and patient-level KRAS G12C prevalence has not been consistently quantified using strict ethnicity definitions. We analyzed MSK-CHORD as the initial reference cohort and American Association for Cancer Research (AACR) GENIE after excluding the Memorial Sloan Kattering Cancer Center contributing center. H/L required explicit H/L-origin annotation; non-Hispanic White (NH-W) required White race with explicit non-Hispanic ethnicity. Primary analyses compared patient-level KRAS G12C prevalence in mutually exclusive lung adenocarcinoma and non-lung adenocarcinoma NSCLC. In lung adenocarcinoma, KRAS G12C prevalence was lower in H/L than NH-W patients in MSK-CHORD [7.9% vs. 15.4%; odds ratio (OR), 0.47; 95% confidence interval (CI), 0.30-0.74; P < 0.001] and GENIE (7.2% vs. 15.1%; OR, 0.44; 95% CI, 0.33-0.58; P < 0.0001). In non-lung adenocarcinoma NSCLC, differences were smaller and not significant in MSK-CHORD (4.8% vs. 5.1%; OR, 0.94; P = 1) or GENIE (4.9% vs. 6.7%; OR, 0.73; P = 0.461). In MSK-CHORD lung adenocarcinoma, adjustment for smoking shifted the OR from 0.50 to 0.68. H/L patients had lower patient-level KRAS G12C prevalence than NH-W patients with lung adenocarcinoma across two real-world clinicogenomic cohorts. The finding was histology specific and attenuated after smoking adjustment, supporting a biomarker-yield interpretation.

H/L patients remain undercharacterized in real-world precision oncology datasets. This study provides patient-level, histology-specific estimates of KRAS G12C prevalence among H/L and NH-W patients using strict race/ethnicity definitions, coverage-aware denominators, and two large clinicogenomic cohorts. These findings inform biomarker screening yield in lung adenocarcinoma and support more rigorous, population-aware evaluation of actionable alterations in real-world oncology datasets.