Abstract
ABSTRACT: Although human papillomavirus (HPV)–positive oropharyngeal squamous cell carcinoma (OPSCC) is associated with better prognosis than HPV-negative disease, ∼30% of cases relapse despite curative-intent radiotherapy (±chemotherapy). We aimed to develop a proteomic signature associated with risk of recurrence within HPV+OPSCC. We analyzed tumor specimens from 124 patients with T1–4N0–3M0 HPV+OPSCC: 50 patients with residual or recurrent disease within 5 years of treatment and 74 age and performance status–matched patients with no recurrence. Proteomic analysis was performed on archival formalin-fixed, paraffin-embedded primary tumor core biopsy specimens and matched normal adjacent tissues using quantitative data-independent acquisition mass spectrometry. Recurrence-free survival (RFS), both locoregional and distant, was the primary endpoint. Univariate Cox regression analysis identified peptides associated with RFS, from which a risk score was established to generate a peptide-based signature. A total of 7,597 protein groups were identified across the cohort, 1,565 of which were differentially abundant between tumor and normal adjacent tissues, with 1,218 being significantly increased in tumors. Improved 5-year RFS (q-value < 0.5) was associated with 405 differentially abundant peptides (from 233 unique proteins) within the 124 tumors. Among them a 26-peptide signature encompassing 26 protein groups was associated with RFS and was able to stratify patients into low, intermediate, or high risk of recurrence (concordance index = 0.941, P < 0.0001). Data available via ProteomeXchange PXD036891. Overall, a 26-peptide signature can be used to risk-stratify HPV+OPSCC. Validation of this proteomic prognostic signature in an independent cohort is required to assess its potential use in future clinical trials to better tailor initial therapy. SIGNIFICANCE: HPV+OPSCC incidence is increasing, with heterogeneous treatment outcomes despite favorable prognosis. Current de-escalation strategies show inferior results, highlighting the need for precise risk stratification. Using data-independent acquisition mass spectrometry proteomics, we identified a 26-peptide signature that stratifies patients into risk categories, potentially enabling personalized treatment decisions and optimal patient selection for de-escalation trials.