Evaluation of ctDNA-guided adjuvant therapy de-escalation in head and neck squamous cell carcinoma: a comparative cohort study.

Background: While circulating tumor DNA (ctDNA) assessment after surgery has emerged as a promising biomarker for minimal residual disease detection in solid tumors, its clinical utility for guiding the selection between postoperative radiotherapy (PORT) and chemoradiotherapy (POCRT) in head and neck squamous cell carcinoma (HNSCC) remains poorly characterized. We evaluated whether ctDNA-directed stratification could optimize locoregional control in HNSCC patients following neoadjuvant chemoimmunotherapy. Methods: In this comparative cohort study, consecutive HNSCC patients treated with neoadjuvant chemoimmunotherapy were stratified into two management groups: a ctDNA-guided cohort where tumor-informed ctDNA testing determined POCRT administration given only for detectable ctDNA, and a traditional cohort where all patients received PORT, with postoperative chemotherapy decisions made by multidisciplinary team review based on pathologic response status and pretreatment imaging findings. The primary endpoint was 3-year locoregional control, with secondary analysis of POCRT utilization rates. Results: Among 257 patients who completed neoadjuvant chemoimmunotherapy, 209 (81.3%) underwent surgery with 187 (72.8% of treated patients) achieving major pathological response and comprising our study population. Of these, 69 (36.9%) received ctDNA-guided management, while 118 (63.1%) followed traditional protocols. POCRT utilization was significantly lower in the ctDNA-guided group (27.5% [19/69] vs 42.4% [50/118]; absolute difference -14.9%, p=0.042). ctDNA positivity rates were comparable between groups (ctDNA-guided: 27.5% [19/69] vs traditional: 29.6% [35/118], p=0.867). ctDNA-guided management demonstrated superior outcomes, with a 15% reduction in locoregional recurrence risk (adjusted HR 0.85, 95%CI 0.70-0.94; p=0.013) versus traditional management. Among ctDNA-positive patients, POCRT benefit was significantly greater in the ctDNA-guided cohort (HR 0.73, 95%CI 0.57-0.83; p=0.026) compared to ctDNA-positive patients receiving traditional management (HR 0.87, 95%CI 0.73-0.93; p=0.047; interaction p=0.039). Conclusion: Postoperative ctDNA analysis identifies HNSCC patients who benefit most from POCRT, enabling a 41% relative reduction in treatment utilization while maintaining superior locoregional control. The enhanced therapeutic effect observed in ctDNA-guided patients supports ctDNA's role as a decision-modifying biomarker for personalization management following neoadjuvant chemoimmunotherapy.