Smoking and other patient factors in HPV-mediated oropharynx cancer: A retrospective cohort study.

PURPOSE: To characterize the significance of patient-level influences, including smoking history, on oncologic outcomes in human papillomavirus (HPV)-mediated oropharyngeal cancer (OPC). MATERIALS AND METHODS: A bi-institutional retrospective cohort study of previously untreated, HPV+ OPC patients who underwent curative treatment from 1/1/2008 to 7/1/2018 was performed. The primary outcome was disease-free survival (DFS) and the primary exposure was </=10 versus >10-pack-year (PY)-smoking history. RESULTS: Among 953 OPC patients identified, 342 individuals with HPV+ OPC were included. The median patient age was 62 years, 33.0% had a > 10-PY-smoking history, 60.2% had AJCC8 stage I disease, and 35.0% underwent primary surgery. The median follow-up was 49 months (interquartile range [IQR] 32-75 months). Four-year DFS-estimates were similar among patients with </=10-PY-smoking history (78.0%, 95% CI:71.7%-83.1%) compared to >10-PYs (74.8%; 95% CI:65.2%-82.0%; log-rank:p = 0.53). On univariate analysis, >10-PY-smoking history did not correlate with DFS (hazard ratio[HR]:1.15;95% CI:0.74-1.79) and remained nonsignificant when forced into the multivariable model. On adjusted analyses, stage, treatment paradigm, and age predicted DFS. Neither >10-PYs, nor any other definition of tobacco use (e.g., current smoker or > 20-PYs) was predictive of DFS, overall survival, or disease-specific survival. Conversely, age nonsignificantly and significantly predicted adjusted DFS (adjusted HR[aHR]:1.02,95% CI:0.997-1.05, p = 0.08), overall survival (aHR 1.05; 95% CI: 1.02-1.08; p = 0.002) and disease-specific survival (aHR 1.04;95% CI: 0.99-1.09;p = 0.09). CONCLUSION: Other than age, patient-level influences may not be primary drivers of HPV+ OPC outcomes. Although limited by its modest sample size, our study suggests the significance of smoking has been overstated in this disease. These findings and the emerging literature collectively do not support risk-stratification employing the >10-PY threshold. LEVEL OF EVIDENCE: Level 4.