HPV primary screening is being implemented in a number of countries around the world. However the optimal approach to determine which HPV positive women require colposcopy and which can safely be retested at some interval is unclear. In some countries genotyping for a select group of high-risk HPV types, either alone or in combination with cytology is being used. It is well known that genotype 16 presents the highest risk of significant disease, but genotyping for other specific high-risk genotypes may also be beneficial. A key piece in determining the genotypes to test is understanding the risk of cervical disease by individual genotype in a screening population. Subjects with high enough risk of CIN3+ should be referred to colposcopy, while those at lower risk may be retested at some interval to allow an opportunity for regression to occur. Determining exact risk thresholds is a subjective endeavor, and should be left to individual guideline groups. Informed decisions will require sufficiently strong evidence to stratify the risk of disease among the high-risk HPV types.
The population in this study is a subset of subjects enrolled from a screening population in an ongoing US registration trial. The subset includes Hybrid Capture 2 (HC2) (Qiagen, Germantown, MD) positive subjects with ASCUS (≥ 21 years) or NILM (≥ 30 years) cytology (n = 1933). These subjects underwent colposcopy and a standardized biopsy protocol. All biopsies had consensus panel review blinded to cytology and HPV status. For this substudy, specimens from subjects diagnosed with CIN2+ and a random selection of <CIN2 subjects were sequenced. The method included PCR amplification, denaturing high performance liquid chromatography separation and bi-directional Sanger sequencing (Transgenomic, Inc., Omaha, NE) to determine specific HPV genotype(s) (n = 921 subjects with HPV genotyping and histological diagnosis).
Analysis is ongoing, but we will present baseline risk of disease by HPV genotype analyzed using a multivariate Bayesian model. The risks of CIN2+ and CIN3+ will be calculated with 95% credible intervals. In addition, the probability that the baseline risk of disease from one genotype is greater than the baseline risk of disease from another genotype will be derived by Markov-Chain Monte Carlo (MCMC) methods.
The Bayesian multivariate model is a novel way of estimating the risk of disease by genotype, with intuitive treatment of subjects with multiple infections. This method also allows for a clear quantification of the risk ordering between two genotypes: the probability that the risk of CIN3+ for one genotype is greater than the risk of CIN3+ for another genotype.