Figure 1

The impact of adding individual carcinogenic, probable carcinogenic, and possible carcinogenic human papillomavirus (HPV) genotypes to a clinical HPV assay. The impact of adding individual carcinogenic, probable carcinogenic, and possible carcinogenic human papillomavirus (HPV) genotypes to a clinical HPV assay was estimated using a receiver-operator curve (ROC)-like approach of stepwise adding individual HPV genotypes according to their prevalence, from most prevalent (HPV16) to least (HPV53), in cervical cancer [8, 17]. Relative sensitivity (true positive fraction) was estimated based on those prevalences using the following assumptions: 1) HPV genotypes act independently such that cancers with multiple HPV genotypes detected were proportionally attributable to the HPV genotypes according to their overall prevalence; and 2) all cervical cancer is caused by HPV, and false negative results were independent of HPV genotype. Therefore, it was assumed that all cancers would have tested HPV positive with a perfectly sensitive test, and the relative contribution of each HPV genotype to cervical was a constant (although the absolute number increased). The relative false positive fraction, 1-Specificity, was estimated from the prevalence of individual HPV genotypes in the cytologic normal population, ignoring the overlap between HPV genotypes due to multi-HPV genotype infections (~25% of HPV positives).