Advancing Education, Research, and Quality of Care for the Head and Neck oncology patient.
Background/Objectives: Human papillomavirus positive (HPV+) oropharyngeal carcinoma (OPC) has a favorable survival, yet standard treatment carries substantial toxicity. Treatment de-escalation that optimizes survival while reducing toxicity is therefore a priority. Current biomarkers to identify candidates for treatment de-escalation are suboptimal. By contrast, circulating tumor HPV-DNA (ctHPV-DNA) is emerging as a sensitive and specific biomarker, however it has not been evaluated as a predictor of tumor response. In this prospective study, we leverage longitudinal ctHPV-DNA assessment with induction chemotherapy to understand how the dynamics of ctHPV-DNA in plasma predict patients’ response to treatment.
Materials/Methods: Patients with HPV+ OPC were treated with three cycles of induction chemotherapy followed by standard or de-escalated locoregional treatment stratified by response. ctHPV-DNA was measured at baseline, following each cycle of induction chemotherapy, and at time of radiographic disease assessment. Tumor volume was estimated by transforming 2-Dimenional RECIST v.1.1 criteria to calculate 3-Dimensional disease volume. Dynamic changes in ctHPV-DNA were correlated to outcomes measured by changes in tumor volume (volumetric response). Statistical analyses include linear regression, Spearman correlation, ROC curve analysis student’s t test.
Results: 45 patients completed induction and were evaluable for analysis. Most were male (86.7%) and had never smoked (60%). Our cohort had very high-risk disease defined by T4, N3, or >=20 pack year smoking history in 51% of patients. Overall and progression-free survival at 24 months were 95.3% and 88.4%, respectively. Baseline ctHPV-DNA was moderately correlated with estimated tumor volume overall (R2=0.204, p=0.002) with stronger correlation observed among patients with a 20 pack year smoking history (R2 = 0.721, p = 0.003). Dynamic changes in ctDNA load from baseline were significantly correlated to tumor response during induction chemotherapy at all-time points as early as following cycle 1: in an ordinal Spearman correlation, reductions in ctDNA were correlated with changes in estimated tumor volume as early as cycle 2 of chemotherapy (Rho = 0.47, p = 0.002) and remained significantly associated across induction. In ROC analysis, we found that changes in ctHPV-DNA between baseline and end of chemotherapy were predictive of deep tumor response, defined as >=50% tumor shrinkage by estimated tumor volume (AUC=0.83, p<0.001). Rapid clearance of ctHPV-DNA (defined as >95% ctHPV-DNA reduction from baseline) predicted radiographic response by both RECIST (64.8%±14.1 vs 49.0%±20.0, p = 0.008) and volumetric (91.8±9.18 vs 79.4±20.2, p = 0.016) criteria. Notably, patients with rapid clearance of ctHPV-DNA by cycle 2 of chemotherapy demonstrated a recurrence-free survival of 100%.
Conclusions: Circulating tumor HPV-DNA dynamic changes predicts tumor shrinkage with induction chemotherapy. Rapid clearance of plasma ctHPV-DNA best predicts volumetric tumor response over traditional RECIST criteria and is associated with excellent recurrence free-survival. Clinical trials leveraging ctHPV-DNA to select patients for de-escalated treatment are warranted.