A methodological approach to multiple-type human papillomavirus infections

Introduction. Human papillomavirus (HPV), the universally accepted cause of cervical cancer, has been detected in up to 40% of women of reproductive age. Most HPV infections are transient or intermittent; only the small number that remain persistent appear to trigger carcinogenic developments. One possible cause of persistence is infection with multiple HPV types, which has itself been linked to disease. However, the issue of how to deal with the presence of multiple-type infections in analyses has not been addressed adequately.; Methodology. PCR HPV status traditionally has been categorized as positive or negative despite the fact that the test produces an estimate of relative viral load. Markov models were used to explore in particular the utility of a more complex definition of HPV 16 status that incorporates this information. A variety of models were fitted, beginning with a full five-state model. This model was compared to more parsimonious models, with a focus on three-state models defining HPV status as negative, low-positive, and high-positive. A three-state model was selected based on overall performance in terms of estimated transition intensities, BIC and likelihood statistics, proximity to empirical data, and ability to predict cytological outcomes. This model was then used to predict acquisition and clearance rates of co-infections with HPV types 31, 33, 35, 52 and 58, the five types most closely related phylogenetically to HPV 16.; Results. A three-state definition of infection status for HPV 16 clearly adds information beyond that which can be gained from a binary definition of viral infection. When used as a covariate in Markov models estimating transition rates for similar phylogenetic types, the expanded definition of HPV status suggests a trend towards a greater likelihood for clearance of these types in the presence of HPV 16 occurring at a high-positive level.; Conclusion. To date, studies have found clearance of phylogenetically related types to be independent of infection with HPV 16. Observed trends in our data suggest that the clearance of other viral types in fact may be facilitated by co-infection with HPV 16 at a high viral load.