APPLIED EPIDEMIOLOGICAL STUDIES OF HIGH RISK HUMAN PAPILLOMAVIRUS INFECTIONS AND RISK OF CERVICAL PRECANCER

Abstract

Persistent infection with human papillomavirus (HPV) is a leading cause of cancer among women worldwide. Health care providers face a growing number of competing cervical cancer screening approaches and tests. HPV testing is very sensitive but a secondary test is needed to identify infections with sufficient risk of cervical precancer/cancer. This dissertation aims to address three questions in the management of HPV infections: (1) to compare the first HPV screening test seeking FDA approval that identifies many individual HPV types (BD Onclarity™) to two FDA approved assays (Roche cobas™ and Qiagen HC2™); (2) to clarify how HPV type influences cumulative risk of clearance, progression or persistence of HPV infections; and (3) to assess whether established etiologic co-factors for cervical precancer, given HPV infection, represent clinically useful, actionable factors that clinicians “need to know” when considering how to manage the HPV infected woman. All manuscripts in this dissertation used data from the HPV Persistence and Progression Cohort, conducted by Kaiser Permanente Northern California and the National Cancer Institute. The study population is a group of 33,295 women, ages 30 or older, who are HPV positive at baseline and have results for cytology. Contingency table methods, Kappa statistics and McNemar’s test were used to assess agreement between HPV DNA tests (manuscript 1). Competing risk proportional hazards models were used to estimate eight-year cumulative risks of HPV clearance, progression to precancer, or persistence (manuscript 2). Absolute risks from Logistic-Cox models were used to study whether co-factors acted as clinically relevant risk stratifiers (manuscript 3). Results from this dissertation suggest that: (1) Onclarity agreement was good to excellent compared with cobas and HC2, and clinical accuracy was high for detection of precancer; (2) cumulative risk of clearance varied little by HPV type, cumulative risk of progression was substantially higher for HPV16, and long-term persistence was uncommon; and (3) the most important predictors of progression from HPV infection to precancer were HPV type and cytologic result. By clarifying these aspects of methods, and management of HPV-positive women, it is hoped that this dissertation will contribute to the improvement of cervical cancer screening incorporating HPV testing.