209 - Does Detection of Tumor-Associated Mutations in Proximal Lymph Predict Relapse in HPV-Negative Head and Neck Cancer Patients?

W. Winckler¹, Z. Gu¹, D. Whitfield¹, N. Earland², A. Harmon¹, M. Long¹, M. Hurley¹, P. K. Harris³, Z. Xu⁴, R. J. Ramirez³, S. P. Gerndt³, M. Pacula¹, M. Francis¹, J. P. Zevallos⁵, and A. A. Chaudhuri⁶; ¹Droplet Biosciences, Cambridge, MA, ²Department of Radiation Oncology, Washington University School of Medicine, St. Louis, MO, ³Washington University School of Medicine, St. Louis, MO, ⁴University of Pittsburgh Medical Center, Pittsburgh, PA, ⁵University of Pittsburgh Medical Center, Department of Otolaryngology, Pittsburgh, PA, ⁶Mayo Clinic, Rochester, MN

Purpose/Objective(s): Radiotherapy or chemoradiation are recommended after surgery for head and neck squamous cell carcinoma (HNSCC) in patients with high-risk clinical/pathological features based on the landmark EORTC and RTOG studies from 20 years ago. Still, these features may not be precise enough, as locoregional failure rates remain unacceptably high in patients with HPV-negative HNSCC. To address this critical issue, we present a novel proximal assay for sensitively measuring circulating tumor DNA using lymphatic exudate collected via surgical drains (“lymph”). Lymph minimal residual disease (MRD) testing could significantly enhance conventional pathology methods and offer more tailored adjuvant radiotherapy and chemoradiation for patients with HPV-negative HNSCC. Materials/

Methods: Lymph, plasma, and blood samples were obtained from 46 HPV-negative HNSCC patients after surgery, 24 hours postoperatively, along with resected tumor tissue. Cell-free DNA was isolated from lymph and plasma samples and subjected to sequencing using the TruSeq Oncology 500 panel, achieving a sequencing depth of over 100 million reads. Somatic mutations were identified through 200x whole exome sequencing of tumor and blood samples. Five patients with <2 tumor-associated mutations were excluded, and 2 patients were removed due to incomplete clinical data. Variants were then identified in lymph and plasma samples utilizing a custom pipeline. Patients were classified as MRD positive if the mean variant allele fraction (mVAF) exceeded 0.015%, which was the estimated limit of detection. Survival analyses were conducted using the Kaplan-Meier (KM) estimator with log-rank testing and Cox proportional-hazards modeling. A logistic regression model was employed with 5-fold cross-validation.

Results: Eighteen patients relapsed, while 21 patients showed no evidence of disease with >1 year of follow-up. KM analysis revealed that lymph samples accurately predict recurrence (sensitivity (SN)=78%, specificity (SP)=67%, p=0.003, Hazard ratio (HR)=4.7), while plasma did not (p=0.92). Improved performance was seen for locoregional relapse (SN=92%, SP=67%, p=0.001, HR=13.9; N=33). Even in 19 N0 patients, lymph MRD predicted recurrence accurately (SN=83%, SP=69%, p=0.03, HR=7.7). Strikingly, individual high-risk pathological features (extranodal extension, perineural invasion, lymphovascular invasion, and nodal disease status) were inferior to lymph MRD (HR=3.4, 0.88, 2.5, 2.4, respectively). However, a logistic regression model combining lymph MRD with these pathological features showed superior performance (SN=89%, SP=67%, p=0.0007, HR=8.6).

Conclusion: Postoperative lymph significantly outperforms plasma for prediction of relapse. Accurate identification of relapsing patients with lower-risk pathologic features suggests that lymph MRD can potentially augment traditional pathology and provide more personalized adjuvant RT decision-making in patients with HPV-negative HNSCC.