Tissue-Informed ctDNA Assay Performance in Head and Neck Squamous Cell Carcinoma

Importance Liquid biopsy is emerging as a strong biomarker for detecting and monitoring head and neck squamous cell carcinoma (HNSCC). The strongest evidence for circulating tumor DNA (ctDNA) in HNSCC comes from studies of tissue-agnostic assays detecting human papillomavirus (HPV)−associated HNSCC. In HPV-negative HNSCC, the lack of broadly shared variational targets precludes the use of tissue-agnostic ctDNA approaches. An emerging alternative, tissue-informed assays, leverage patient-specific tumor variant profiles to enable individualized ctDNA detection in the blood. Objective To measure the detection characteristics of a commercially available tissue-informed assay when compared to the standard criterion for tissue biopsies, both in the pretreatment and surveillance settings of HPV-positive and HPV-negative HNSCC. Design, Setting, and Participants This was a retrospective cohort study of adults with HNSCC who received a tissue-informed ctDNA assay as part of their oncologic treatment at a large, urban tertiary care center from December 2024 to March 2026. Data analysis was conducted from March to April 2026. Exposure Tumor tissue specimen, collected by fine-needle aspiration or surgical biopsy, used to create an individualized ctDNA assay for subsequent ctDNA blood tests. Main Outcomes and Measures For the pretreatment analysis: test detection rate, miss rate, and insufficient tissue rate. In the surveillance setting: sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) on a per-test and per-patient basis. Results The pretreatment cohort comprised 77 patients (mean [SD] age, 65.0 [12.6] years; 28 female [36%] and 49 male [64%] individuals) and the surveillance cohort, 107 patients (mean [SD] age, 65.0 [12.1] years; 28 female [26%] and 79 male [74%] male individuals). The median (IQR) time since treatment conclusion for patients in the surveillance cohort was 18 (11-39) months. Tissue-informed ctDNA was detected in 71 of 77 patients (92.2%) in the initial pretreatment setting. In the surveillance setting, the assay demonstrated a per-test sensitivity of 95.2% (20 of 21 patients; 95% CI, 83.3%-100%), specificity of 100% (157 of 157 patients; 95% CI, 100%-100%), PPV of 100% (20 of 20 patients; 95% CI, 100%-100%), and a NPV of 99.4% (157 of 158 patients; 95% CI, 97.9%-100%). On a per-patient level, the sensitivity was 94.4% (17 of 18 patients; 95% CI, 72.7%-99.9%), specificity was 100% (89 of 89 patients; 95% CI, 95.9%-100%), PPV was 100% (17 of 17 patients; 95% CI, 80.5%-100%), and NPV was 98.9% (89 of 90 patients; 95% CI, 93.0%-99.8%). Conclusions and Relevance In this cohort study, tissue-informed ctDNA assay showed excellent test creation ability in the pretreatment setting and performance metrics in the surveillance setting. Reasons for a failure rate of 7.8% in the pretreatment cohort remain unclear. These metrics support the feasibility of using this assay as a promising ctDNA approach for disease monitoring in HNSCC.