Tumor-informed molecular residual disease in resected mucosal and cutaneous squamous cell carcinomas of the head and neck

Patients with resected mucosal (HNSCC) or cutaneous (cSCC) squamous carcinomas of the head and neck undergo adjuvant therapy based on clinicopathologic prognosticators. Most patients are overtreated or undertreated, with substantial morbidity from locoregional recurrences and locoregional therapies. It is unknown whether postoperative ctDNA molecular residual disease (MRD) testing could improve patient selection. We conducted an institutional cohort study of 110 consecutive patients with non-metastatic cSCC or HPV-negative HNSCC who underwent prospectively-tested tumor-informed ctDNA testing after curative-intent resection. We hypothesized that MRD positivity before adjuvant therapy would be prognostic and predictive. After 34 months median follow-up, the sensitivity and specificity of a single detectable MRD ctDNA before adjuvant therapy were 71.0% and 85.1%, respectively. Detectable MRD was associated with inferior overall survival, locoregional recurrence, and distant metastasis. Among patients with detectable postoperative MRD, 37.5% cleared ctDNA after adjuvant therapy and had an intermediate prognosis. Detectable MRD outperformed clinicopathologic prognosticators and was predictive for adjuvant chemotherapy benefit. Patients with CCND1/CTTN amplification or CDKN2A/CDKN2B/MTAP deletions had high rates of MRD positivity (60%), were less likely to experience ctDNA clearance during adjuvant therapy (17% vs. 54%), and had inferior 24-month DFS (30.0% vs 61.7%). These findings inform the design of biomarker-adapted clinical trials.