Early Postoperative Minimal Residual Disease Detection with MAESTRO Is Associated with Recurrence and Worse Survival...

Although ctDNA is a promising biomarker for minimal residual disease (MRD) detection in head and neck squamous cell carcinoma (HNSCC), more sensitive assays are needed for accurate MRD detection at clinically relevant time points. Ultrasensitive MRD detection early after surgery could guide adjuvant therapy decisions, but early ctDNA dynamics are poorly understood.

We applied minor allele-enriched sequencing through recognition oligonucleotides (MAESTRO), a whole-genome, tumor-informed, mutation enrichment sequencing assay, in a pooled testing format called MAESTRO-Pool, to plasma samples from patients with HNSCC collected shortly after surgery and during surveillance. We evaluated whether early MRD detection could predict outcomes.

Among 24 patients with predominantly human papillomavirus-independent (95.8%) HNSCC, rapid ctDNA clearance occurred by the first postoperative sample (1-3 days postoperatively) in nine patients without an event (recurrence or death). Thirteen of fifteen patients with an event were MRD-positive (positive predictive value = 92.9%; negative predictive value = 80%) with a median tumor fraction (TFx) of 54 parts per million (ppm; range 6-1,177 ppm). In the first and last samples of the early postoperative window, 8/13 and 10/13 patients, respectively, had TFx below 100 ppm, the detection limit of leading commercial assays. Early MRD detection correlated with worse overall survival (HR, 8.3; 95% confidence interval, 1.1-66.1; P = 0.02) and event-free survival (HR, 27.4; 95% confidence interval, 3.5-214.5; P < 0.0001) independent of high-risk pathology.

Early postoperative MRD detection by MAESTRO was associated with recurrence and worse survival. Given the ultralow TFxs observed, ultrasensitive assays will be essential for reliable MRD detection during early postoperative time points to enable personalized adjuvant therapy decision-making in HNSCC. See related article by Bryan et al., p. 3483.