HPV testing. Visible expectations and hidden realities.

In this issue of the Journal, de Cremoux et al1 report a study examining the efficiency of human papillomavirus (HPV) testing using the Hybrid Capture 2 (HC2) in cervical cancer screening. In conducting this study, they focused on several factors that are emerging as important determinants of accuracy in the use of HC2 in any setting, including the following: (1) sensitivity, (2) positive predictive value, (3) false-positive results, and (4) crossreactivity between the probe set used in HC2 and viral types not included in the probe cocktail. Two groups of subjects were studied, including those referred for colposcopy because of a smear abnormality and consecutively screened women. All subjects underwent colposcopic evaluation.1 HPV testing by HC2 is becoming widely accepted in the United States because of the recent recommendations for its use in managing nondiagnostic squamous atypia (atypical squamous cells of undetermined significance [ASCUS]) and the ease with which patients are triaged into follow-up and intervention (colposcopic triage) categories.2,3 The role of HPV testing in the management of women with no previous Papanicolaou smear abnormality is early in its development, but the notion that testing could replace the Papanicolaou smear in screening women of reproductive age is not farfetched.4 Obstacles to such a transition include the potential of missing other abnormalities in the smear, an inherent small but definite false-negative rate with a single HPV test for highgrade squamous intraepithelial lesion (HSIL), a high rate of HPV-positive results in women younger than 30 years, and a medical profession that is unprepared to counsel a large number of women likely to have a positive result for a sexually transmitted virus even in the absence of a cytologic abnormality. These issues are likely to remain in the forefront for the next several years as proponents of the “DNA Pap” evaluate the capacity of the medical and lay populations for changes in practice and outlook. Two issues arise in the study that are important and remain to be sorted out in practice. The first is the sensitivity of the assay. Estimations of sensitivity can be applied to both adjudicated Papanicolaou smear and biopsy diagnoses. Both have some degree of false negativity depending on what is defined as a “gold standard.” It is reasonable to assume that sensitivity will be slightly lower when the end point is a cytologic diagnosis, inasmuch as not every HSIL can be distinguished from immature metaplasia, atypical atrophy, and lower uterine segment on cytologic grounds. In contrast, biopsy specimens are considered more evaluable, notwithstanding problems in interobserver reproducibility. By using histologic results as their outcome variable, Clavel et al5 reported a sensitivity of 100% for HSIL biopsy outcome in screened samples using HC2. This is similar to our own experience with women who have had a previous biopsy result of HSIL and were followed up with cone biopsy (C.P. Crum and R. Urban, unpublished data, 2003). In contrast, in this report the sensitivity of HC2 for biopsy proven HSIL was just 83%.1 The logical question that arises is whether this low sensitivity reflects histologic misclassification, ie, HPV-negative conditions are overclassified as HSIL on the biopsy specimen. Lonky et al6 found that 25% of histologically verified HSILs scored negative by HC2, yet HPV was verified by polymerase chain reaction (PCR) analysis in the histologic material of many of these cases. Moreover, histologic misclassification is highly unlikely to occur in diagnoses of invasive squamous carcinoma. Thus, the classification of 17% of HSILs and 14% of invasive squamous cell carcinomas as HPV-negative in the