Quantification of Tumor-Associated Cell-Free HPV DNA in Surgical Drain Fluid Equates to Molecular Residual Disease and Correlates with High-Risk Pathology

Abstract

<h3>Purpose/Objective(s)</h3> Multi-institutional efforts aimed at safely de-escalating treatment in human papillomavirus (HPV) associated oropharyngeal squamous cell carcinoma (OPC) patients rely on subjective pathology for risk stratification. While plasma HPV assays are emerging for post-treatment disease surveillance, they have poor sensitivity to detect postoperative molecular residual disease (MRD). Here we present a liquid biopsy assay of the serosanguinous fluid removed from surgical drains, proximal to the surgical site, to sensitively measure postoperative MRD. We hypothesize that HPV quantified in surgical drain fluid (SDF), 1) represents post-operative regional MRD, and 2) could improve patient risk-stratification to guide precision radiotherapy. <h3>Materials/Methods</h3> We collected SDF at 24-48 hours post-surgery for cell-free (cf) DNA extraction. Our 87-patient head and neck cancer cohort included 74 HPV(+) and 8 HPV(-) OPCs, and 5 benign tumors. Paired plasma and tumor samples were collected from approximately half of all patients. We used the HPV16 E6T2aE7 plasmid to compare our Taqman quantitative PCR (qPCR) to digital droplet PCR (ddPCR). We validated HPV(+) OPC PCR results using next-generation HPV sequencing (NGS) in 9 patients. Then we correlated SDF HPV copies to extranodal extension (ENE), AJCC 8^th edition pathological stage, smoking pack-years, recurrence-free survival (RFS), and composite risk scores derived from OPC de-escalation clinical trials. Statistical methods included Fisher's Exact test, Mann-Whitney (MW) test, Kruskal-Wallis (KW) test, and Kaplan-Meier (KM) analysis. <h3>Results</h3> Our HPV qPCR assay's limit of detection was 9.7 copies, similar to what we saw with ddPCR. HPV genotyping was concordant between NGS and PCR, and all SDF cfDNA samples with nondetectable (ND) HPV via PCR were ND via NGS. HPV was detected in 63 of 74 (85%) SDF samples compared to only 7 of 48 (15%) paired plasma samples. Median HPV copies in ENE+ patients' SDF was 15-fold higher than SDF from ENE- patients (P = 0.002). HPV copies also increased with increasing nodal (P = 0.01) but not primary tumor stage (P = 0.4), suggesting post-operative SDF HPV load reflects invasive nodal disease and is a proxy for regional MRD. When we stratified our cohort according to high-risk and low-risk scores drawn from four clinical trials (RTOG 9501, EORTC 22931, ECOG 3311, MC1273), SDF HPV load was consistently higher in patients with trial-defined high-risk pathology (mean AUC = 0.73). Lastly, KM RFS analysis of HPV-high and HPV-low patients, defined by an RTOG 9501-derived SDF classifier, showed HPV-high patients trended toward worse RFS (HR = 3.1; P = 0.1). <h3>Conclusion</h3> Liquid biopsy assessment of SDF, the biofluid most proximal to the nodal environment, enables postoperative regional MRD detection in HPV(+) OPC, and correlates with high-risk nodal pathology and worse RFS. If confirmed, SDF cfDNA analysis could aid traditional pathology and imaging to personalize adjuvant radiotherapy for surgically treated HPV(+) OPC.