OBJECTIVE

New commercial HPV RNA assays require further validation studies in population-based cervical cancer screening settings. To assess the performance of (FDA-approved) APTIMA® HPV Assay (AHPV), Hybrid Capture 2 (HC2), in-house PCR genotyping, and ThinPrep LBC in population-based screening, stratified by three histological gold standards.

STUDY DESIGN

A multi-center trial in 5006 women undergoing routine screening in France was designed to compare the absolute and relative risks of diagnosing CIN3+ and CIN2+ lesions by different diagnostic tests.

RESULTS

Reproducibility between the primary and second pathology reading was excellent for CIN3+ and CIN2+ endpoints (Cohen's kappa 0.948 and 0.854). Absolute risks (PPV) of different tests (AHPV, HC2, PCR genotyping, LBC) in diagnosing CIN2+ (15-20%) and CIN3+ (4-6%) were similar for the first, second, and consensus pathology readings. The relative risks of diagnosing these lesions by the four tests were also similar when the first, second or third pathology readings were employed. AHPV had the highest absolute risk of both histological endpoints, and detects 5% to 15% more CIN3+ and CIN2+ lesions, respectively, than LBC. Compared with HC2 assay, the relative risk of AHPV is 24% to 29% higher, with a significant difference in CIN2+ detection. With LBC as reference, AHPV had the best sensitivity/specificity balance measured by AUC (area under ROC curve) comparison test (significant for CIN2+), and the colposcopy referral rate (9.2%) comparable to that of LBC (8.7%).

CONCLUSIONS

These data corroborate the suitability of AHPV for the primary cervical cancer screening.