No abstract

Background:Upfront surgery is a valuable treatment option for oropharyngeal squamous cell carcinoma (OPSCC) and risk stratification is emerging for treatment de-escalation in human papillomavirus (HPV)-related OPSCC. Available prognostic models are either based on selected, mainly non-surgically treated cohorts. Therefore, we investigated unselected OPSCC treated with predominantly upfront surgery.Methods:All patients diagnosed with OPSCC and treated with curative intent between 2000 and 2009 (n=359) were included. HPV association was determined by HPV-DNA detection and p16INK4a immunohistochemistry. Predictors with significant impact on overall survival (OS) in univariate analysis were included in recursive partitioning analysis.Results:Risk models generated from non-surgically treated patients showed low discrimination in our cohort. A new model developed for unselected patients predominantly treated with upfront surgery separates low-, intermediate- and high-risk patients with significant differences in 5-year OS (86%, 53% and 19%, P<0.001, respectively). HPV status is the most important parameter followed by T-stage in HPV-related and performance status in HPV-negative OPSCC. HPV status and ECOG remained important parameters in risk models for patients treated with or without surgery.Conclusions:Regardless of treatment strategies, HPV status is the strongest predictor of survival in unselected OPSCC patients. The proposed risk models are suitable to discriminate risk groups in unselected OPSCC patients treated with upfront surgery, which has substantial impact for design and interpretation of de-escalation trials.

Background A remarkably better prognosis is associated with oropharyngeal squamous cell carcinomas (OPSCC) driven by human papillomaviruses (HPV) compared with HPV-negative OPSCC. Consequently, de-escalation of standard treatment has been suggested. Due to modest specificity rates, debates are ongoing, whether p16INK4a, a surrogate marker for HPV-driven OPSCC, is sufficient to correctly identify those tumours and avoid substantial HPV misattribution and thus undertreatment of patients by de-escalation. Robust data estimating the proportion of potentially undertreated patients are missing. Methods We assessed a large-scale cohort of consecutively included OPSCC diagnosed between 2000 and 2017 for HPV–DNA, HPV genotypes, p16INK4a expression and multiple tumour- and patient-related risk factors, and investigated their impact on patients’ survival in comprehensive uni- and multivariate analyses. Results Aetiological relevance of HPV (p16INK4a- and high-risk HPV–DNA-positivity) was detected in 27.1% (n = 192) of OPSCC, with HPV16 being the most abundant HPV type (94.6%). In 5.5% patients (n = 39), p16INK4a overexpression but no HPV–DNA was detected. Principal component and survival analyses revealed that 60.6% of these p16INK4a-positive OPSCC lacking HPV–DNA did not resemble HPV16-driven but HPV-negative OPSCC regarding risk-factor profile and overall survival. Notably, this group represented 10.6% of all p16INK4a-overexpressing OPSCC. Conclusions p16INK4a as a single marker appears insufficient to indicate OPSCC patients suitable for treatment de-escalation.

No abstract