Variable detection of actionable alterations across racial groups and association with testing patterns.

3116 Background: Molecular landscape studies are critical to biomarker discovery and precision oncology research. However, non-White patients (pts) have historically been under-represented. In this study, we examine clinico-genomic data from a network of community oncology clinics to evaluate real-world mutational frequencies in Black, White, and Asian pts with cancer. Methods: We utilize Genospace, a precision medicine software, to harmonize clinical data from the Sarah Cannon research network of U.S. community oncology practices with genomic data from commercial NGS vendors. Pts with known race and NGS test results from September 2015 to November 2021 are assessed. Variants of uncertain significance (called by NGS vendor) are excluded. Asian pts are not evaluable in all instances because of low pt number. Statistical analyses include Chi squared and two sample test of proportions with Benjamini-Hochberg false discovery rate correction. Results: A total of 18,399 pts are assessed, with 7% Black (n = 1,319), 92% White (n = 16,903), and 1% Asian (n = 177) pt representation. The most common tumor types include non-small-cell lung (NSCLC; 27% of Black pts, 30% of White pts, 33% of Asian pts), colorectal (15% of Black pts, 12% of White pts, 14% of Asian pts), breast (14% of Black pts, 11% of White pts, 9% of Asian pts), and prostate (6% of Black pts, 5% of White pts, 3% of Asian pts) at roughly equal proportions across racial groups. In NSCLC, EGFR mutations and ALK fusions significantly enrich in Asian pts while KRAS G12C is detected more frequently in Black patients (see table for details; Black: n = 353, White: n = 5,127, Asian: n = 58). In both prostate and breast cancers, plasma-based NGS is less common in Black pts than White pts (plasma-to-tissue ratio = 1.1 (Black prostate: n = 82); 1.5 (White prostate: n = 863); 0.6 (Black breast: n = 178); 0.9 (White breast: n = 1,918)). In prostate, AR mutations are detected less frequently in Black pts. However, when considering plasma-NGS results alone the detection frequency is roughly equal. Similarly, the gap in ESR1 detection frequencies in breast decreases when considering plasma-NGS results alone. Conclusions: Actionable alterations are detected across racial groups at variable frequencies with potential biological and technical underpinnings. In scenarios where plasma-NGS is commonly used to monitor for resistance mutations, discrepancies in NGS ordering patterns likely affect detection frequencies. [Table: see text]