The main objective of this study is to test the hypothesis that detection of HPV in exfoliated cells collected in an oral rinse is a biomarker for subsequent development of HNSCC. To date, there are no prospective data that have evaluated this hypothesis, using available mouthwash specimens. We propose to pool together samples of ACS CPS-II and PLCO cohorts and conduct nested case-control studies among the participants with available mouthwash samples to test this hypothesis. Incident cases of HNSCC (n=183) identified during the follow-up period (as of 6/30/2010) in both the CPS-II and PLCO cohorts among subjects who provided mouthwash samples and were free of cancer at the time of sample collection will be selected for the study. Controls for these HNSCC cases will be selected from participants with available mouthwash samples in each respective cohort using incidence density sampling with matching on age, race, gender, and study center. The case to control ratio will be 1:3; yielding a total of 549 controls in both cohorts. We will extract and test the mouthwash samples from all cases and controls using 3 HPV tests established in the Burk lab to detect the complete spectrum of HPV types found in the oral cavity. Associations between oral HPV and risk of HNSCC will be analyzed using conditional logistic regression models after adjusting for known risk factors of these cancers including tobacco and ethanol use. This investigation should provide fundamental information about the utility of oral HPV detection in mouthwash samples as a biomarker for subsequent risk of HNSCC. To accomplish these goals, we propose the following specific aims: 1. To utilize mouthwash samples from all incident HNSCCs (n=183) that arose in the CPS-II and PLCO cohorts subsequent to providing the mouthwash sample and a control group (3 controls per case) from each respective cohort. We will purify DNA using proteinase K and phenol/chloroform extraction and test these samples using 3 different HPV PCR assays, specifically detecting (1) a broad spectrum of HPV types; (2) oncogenic alpha-HPV types; and (3) HPV 16 using a highly sensitive test. 2. To estimate the association of the biomarker (i.e., HPV detected in an oral rinse specimen prior to the diagnosis of HNSCC) and risk of HNSCC. We will analyze HPV classified into 3 categories: (1) any HPV type; (2) known oncogenic HPV types; (3) HPV16, after adjusting for known risk factors including age, smoking and alcohol use. Given our study design and available sample size, we will have 80% statistical power to detect at least odds ratios (OR) of 1.6 or 3.5 for an HPV prevalence range between 2% to 40% in the control group (p=0.05, 2-sided test).

Richard Hayes (New York University)
Robert D. Burk (Albert Einstein College of Medicine)
Susan Gapstur (American Cancer Society)
Tao Wang (Albert Einstein College of Medicine)