Highly Accurate Detection of Circulating Tumor DNA for Monitoring Minimal Residual Disease in Solid Tumors:...

Advances in cell-free DNA processing and error-suppressed next-generation sequencing have enabled detection of circulating tumor DNA (ctDNA) for assessment of minimal residual disease (MRD). MRD testing requires high analytical specificity near the stochastic limit of rare molecule sampling amid wild-type cell-free DNA (ie, the "needle-in-a-haystack" challenge). This study reports the analytical and clinical validation of Haystack MRD, a tumor-informed ctDNA MRD assay that tracks patient-specific somatic mutations with background noise suppression to improve discrimination of ctDNA signal from technical and biological interferences. Analytical validation used reference materials and clinical specimens across 23 tumor types, followed by clinical validation in patients with stage II to III colorectal cancer and real-world evaluation across 27 cancer types. Haystack MRD exhibited a 95% limit of detection of 0.063 mean ctDNA molecules/mL and an analytical measurement range of 0.25 to 1000 mean ctDNA molecules/mL. Concordance of tested samples with expected MRD status yielded 99% agreement (95% CI, 94.2%-99.9%). Quantitative mean ctDNA molecules/mL measurements were highly concordant with an orthogonal method (R² = 0.9998), supporting accuracy and linearity. In patients with early-stage colorectal cancer, posttreatment MRD status was highly associated with recurrence, displaying 100% sensitivity (95% CI, 72.3%-100.0%) and 100% specificity (95% CI, 92.9%-100.0%) at 3-year follow-up, and 83.3% sensitivity (95% CI, 55.2%-97.0%) and 100% specificity (95% CI, 92.6%-100.0%) at 5-year follow-up, outperforming carcinoembryonic antigen measurements.