A Matched Pair Analysis of Patients With HPV-Associated Carcinoma of Unknown Primary With T1-2 HPV-Associated Oropharynx Cancer: Implications for Clinical Trial Design

Abstract

Patients (pts) with human papillomavirus associated (HPV+) cancer of unknown primary (CUP) are generally excluded from clinical trials, despite surgical series that report detection rates of occult oropharynx primaries of >80%. We performed a matched pair analysis to compare outcomes between T0N+M0 HPV+ CUP and T1-2N+M0 HPV+ oropharynx cancer of known primary (OPX). Pts with cT1-2N1-3M0 HPV+ OPX or T0N1-3M0 HPV+ CUP treated with curative intent between 1999 and 2016 were identified in an IRB approved database. CUP was defined as biopsy proven HPV+ cervical lymph node metastasis without an identified primary tumor despite imaging and direct visualization/biopsy. HPV status was confirmed by p16 IHC staining and/or by ISH for HPV DNA. For a subgroup of CUP pts with an unknown HPV status, we performed multiple imputation analysis using a training and validation cohort of 447 pts with known oropharyngeal primary and HPV status and then applied the model to CUP pts. We then included CUP pts with a >80% probability of being HPV+. Matched pair analysis was performed using a nearest neighbor propensity technique (MatchIt package in R). Pts were matched based on age, gender, N-stage, performance status, smoking status, pack-years, chemotherapy use and neck dissection. Pts were also matched according to either a favorable (N1-2b and <10 pk-yrs) or unfavorable (N2c-3 or >10 pk-yrs) risk stratification designations per current NRG oncology clinical trial enrollment criteria. Disease-free survival (DFS) and overall survival (OS) were calculated using Kaplan-Meier analysis. Of 313 pts with T1-2N1-3 OPX, 48 were matched to 48 HPV+ CUP pts (32 confirmed and 16 imputed HPV-status). All pts received curative radiotherapy (RT) and all CUP pts had potential oropharynx mucosal sites targeted. Overall, 40% of CUP and 27% of OPX first underwent primary surgery and/or neck dissection followed by adjuvant RT, while the remaining pts were treated non-operatively. Chemotherapy was used in 63% and 71% of each group, respectively. Median follow-up for CUP (34.1 months) and OPX (27.8 months) pts were similar (P = 0.23). After matching, no significant differences were found between the two groups. The majority of pts fit the unfavorable HPV + risk category (73% CUP vs. 63% for OPX) while the remaining pts (27% CUP vs. 37% OPX) fit the favorable HPV+ category (P = 0.28). There were no significant differences between the CUP and OPX groups for 3yr DFS (89% vs. 81%; P = 0.94), and 3yr OS (91% vs. 91%; P = 0.11), respectively. Pts with T0N+M0 HPV-associated CUP have similar survival outcomes to matched pts with T1-2N+M0 HPV+ OPX. These pts should be included in clinical trials and risk stratified for de-intensification or intensification strategies based on their other known risk factors, such as nodal burden and smoking exposure.