Abstract
<title>Abstract</title> <p>Background A nationwide study found a large historical difference in operational biomarker positivity between liquid comprehensive genomic profiling platforms. Whether case mix, calendar time, or facility composition explained it was uncertain. Methods We analyzed 14,878 FoundationOne Liquid CDx (F1L) and 2,042 Guardant360 CDx (G360) records through March 31, 2025. The patient-level composite comprised seven prespecified biomarkers linked to approved therapies in Japan. Analyses included sequential adjustment, restriction to facilities using both platforms, and evaluation of tumor mutational burden-high (TMB-H). Results Positivity was 14.323% (2,131/14,878) with F1L and 4.652% (95/2,042) with G360 (risk difference [RD], − 9.671 percentage points; 95% confidence interval [CI], − 10.744 to − 8.598). Patient-background- and calendar-adjusted positivity was 14.351% versus 4.498% (RD, − 9.853; 95% CI, − 11.087 to − 8.619). Across 171 both-platform facilities, the RD was − 10.233 (95% CI, − 11.586 to − 8.880). During the study period, TMB was not reported by the Japanese G360 configuration, and all G360 C-CAT TMB fields were blank. TMB-H alone occurred in 1,398 F1L-positive cases (65.6%); excluding TMB-H yielded an RD of − 0.384 (95% CI, − 1.420 to 0.652). Conclusions Measured adjustment did not attenuate the historical contrast, which persisted in both-platform facilities. Its marked reduction after TMB-H exclusion identifies historical TMB reporting asymmetry as a major feature. Findings concern operationally available information, not comparative assay performance.</p>