Subclassifying atypical squamous cells in Thin-Prep cervical cytology correlates with detection of high-risk human papillomavirus DNA.

Abstract

Recent studies have proposed subclassifying ASCUS into "favor reactive" (ASFR), "not otherwise specified" (ASNOS), and "favor squamous intraepithelial lesion (SIL)" (ASFS). This study explored the reproducibility of these diagnoses with Thin-Prep cytology and their association with high-risk human papillomavirus DNA (HRHPV). Three pathologists and 1 cytotechnologist with 2 to 25 years of experience reviewed 144 Thin-Prep (Cytyc, Boxborough, MA) specimens previously diagnosed as normal, ASFR, ASNOS, ASFS, and SIL. Interobserver reproducibility was computed with the kappa statistic. The original laboratory diagnosis was compared with the presence of HRHPV types. Interobserver reproducibility for a normal or SIL diagnosis was very good (kappa = .68 and .63). Reproducibility for ASFR, ASNOS, and ASFS ranged from poor to fair (kappa = .21, .19, and .32). In a weighted analysis, kappa values for ASFR/ASNOS and ASFS/SIL were .36 and .62, respectively. HRHPV-positivity for preparations originally diagnosed as N, ASFR, ASNOS, ASFS, and SIL were 5.7%, 8.8%, 17.4%, 47.8%, and 54.5%, respectively. The difference in index of HRHPV for either N or ASFR and ASFS or SIL was significant (P < .001). Reproducibility for ASCUS is generally poor, but better reproducibility is obtained by combining ASFS with SIL and, to a lesser degree, ASNOS with ASFR. ASFS and SIL confer a similar index of HRHPV and merit similar management. ASFR may be managed with cytologic follow-up; but this may depend upon the individual laboratory. HPV testing, in conjunction with cytologic and biopsy follow-up, appears useful for estimating the significance of ASCUS subgroups in laboratory practice.