Impact of EGFR Overexpression and Imaging Biomarkers on Risk Stratification for Locoregional Failure in HPV-Related Oropharyngeal Cancer

HPV-related (+) oropharyngeal cancer (OPC) patients without heavy smoking or advanced stage have been proposed as candidates for radiation dose de-escalation. Though EGFR overexpression has been established as a marker of poor prognosis in HN cancers, its role in HPV+ OPC is unknown. We hypothesized that the addition of EGFR expression and imaging biomarkers to clinical factors could improve prediction of locoregional failure (LRF) and patient selection for de-intensification of locoregional therapy in HPV+ OPC. HPV status was determined by in situ hybridization for 198 patients with Stage III/IV OPC treated with definitive chemoradiation with weekly carboplatin/paclitaxel from 5/2003 to 10/2010. The impact of pre-therapy primary tumor EGFR expression, primary and nodal gross tumor volumes (GTV-T and GTV-N, respectively), and primary tumor and nodal maximum standardized uptake values on FDG-PET (SUVmax-T and SUVmax-N, respectively) on LRF (including residual tumor cells on consolidative neck dissection) was assessed. Primary tumors were HPV+ in 184 patients and HPV- in 14 patients. At 50 months median follow-up, LRF occurred in 27 (15%) HPV+ and 7 (50%) HPV- patients (including residual microscopic tumor cells on neck dissection). Almost all LRFs occurred within the GTVs EGFR overexpression was more frequently observed in HPV- tumors (77% vs 32%, p=0.003), and though associated with LRF in the overall cohort (p=0.01), was not predictive for LRF within the HPV+ population (p=0.21). GTV-T was similarly associated with LRF within the overall cohort (p=0.03), but not within the HPV+ population (p=0.3), while GTV-N was associated with LRF in both the HPV+ and overall cohorts (p=0.01). SUVmax-T and SUVmax-N both showed only borderline significant associations (p=0.07-0.12) with LRF in the overall and HPV+ cohorts. On multivariate analysis, neither EGFR overexpression nor imaging biomarkers were significantly associated with LRF in HPV+ patients after adjustment for T4 and N3 stages, which were the only independent predictors of LRF in HPV+ patients (3-year LRF: 27.5% for T4 or N3 vs 8.0% for T1-3 N0-2c; HR 4.3, p=0.001). Among HPV+ patients with non-T4 or N3 tumors, smoking history of 10 or more pack-years was the only significant discriminator of LRF risk (3-year LRF: 13.6% vs 3.3;, HR 4.4, p=0.04). In this cohort of uniformly treated Stage III/IV OPC patients, the prognostic impact of EGFR overexpression and imaging biomarkers on LRF was predominantly related to their association with HPV status and T or N stage, respectively. Only T4-stage, N3-stage, and smoking contributed to risk-stratification for LRF.