Author(s): Shields TS, Brinton LA, Burk RD, Wang SS, Weinstein SJ
Oncogenic human papillomavirus (HPV) infections, the necessary cause of most cervical cancers, are common and usually clear within 1 to 2 years. Identifying cofactors that lead to cancer among HPV-infected women has depended mainly on case-control studies defining HPV by DNA testing. DNA testing assesses only current infection; thus, concerns about residual confounding remain. To assess cofactors, we used seropositivity to five oncogenic HPV types as a marker of past exposure and confined our analysis to seropositive controls compared with cancer cases. Study subjects had participated in a multicenter U.S. case-control study conducted in the early 1980s. The detailed questionnaire and stored sera for 235 cases of squamous carcinoma and 486 controls motivated the reanalysis. We measured antibodies to HPV types 16, 18, 31, 45, and 52. Independent, significant predictors of seropositivity among controls included numbers of sexual partners, Black race, and oral contraceptive use. Condom use was protective. Among HPV-exposed women, Papanicolaou screening, Black race, and yeast infection were significantly associated with reduced cancer risk. Current smoking was associated with a 2-fold increase in risk; there were independent, significant trends of increased risk with numbers of cigarettes smoked (P for trend = 0.003) and years of smoking (P for trend = 0.01). Other significant predictors of increased risk included low education and income and history of nonspecific genital infection. Unlike recent HPV DNA-based investigations, based on the use of HPV-seropositive controls in this study, oral contraceptive use was unrelated to the risk of cervical cancer and multiparity was only weakly related to risk. It is particularly worth considering further why studies of different designs are inconsistent regarding the effect of oral contraceptive use.