Correlating Papanicolaou Smear, Colposcopic Impression, and Biopsy: Results from the Women's Interagency HIV Study.

OBJECTIVE: To determine associations among cervical cytology, colposcopy, and biopsy in HIV-seropositive women. MATERIALS AND METHODS: HIV-seropositive women and uninfected comparison women in a multicenter prospective cohort study underwent colposcopy for protocol indications. Women were eligible if they had a cervix, satisfactory cytology, and colposcopy between October 1994 and September 1999. Cytology, colposcopic impression, and biopsy were compared using equivalent categorizations. Kappa statistics with bootstrap sampling assessed strength of associations. RESULTS: Colposcopy was performed in 978/1370 HIV-seropositive women and in 154/224 seronegative women. Biopsies were performed on 603 (44%) seropositive women at least once during 1015 colposcopy visits and on 82 (37%) seronegative women at 116 visits. The positive predictive value of cytology was 72% for seropositive women and 60% for seronegative women. The positive predictive value of colposcopy was 71% for seropositive women and 55% for seronegative women. CONCLUSION: The correlation between either cervical cytology or colposcopic impression and colposcopic biopsy was poor.

Interrater Variability in Diagnosis of Cervical Biopsies from Women with HIV-1: Results from the Women's Interagency HIV Study.

OBJECTIVES: To determine interrater variability in classifying cervical biopsies from women with human immunodeficiency virus (HIV). MATERIALS AND METHODS: Cervical biopsies performed on women participating in the Women's Interagency HIV Study (WIHS) were read at the six participating sites. A 10% random sample was retrieved and reviewed using standardized terminology by pathologists with a special interest in gynecologic pathology. Results were compared with kappa values and Mantel-Haentzel tests. RESULTS: Biopsies from 288 HIV-seropositive and 24 HIV-seronegative women were reviewed. The weighted kappa value of 0.67 indicated moderate to strong agreement between original and review diagnoses, with a range of 0.54 to 0.84 across sites. No cancers were identified. Significantly more specimens showing cervical intraepithelial neoplasia (CIN) grade 2 or 3 were identified by review pathologists (p = .02). CIN2 or CIN3 was graded less severely by local pathologists in 18 (51%) of 35 cases, all from HIV-seropositive women. Local pathologists' diagnoses of CIN2 or CIN3 were downgraded by reviewers in 4 of 21 cases (19%). Discrepancies were more common among women with lower CD4 lymphocyte counts. CONCLUSIONS: Although discrepancies occur, interrater correlation in the interpretation of cervical biopsies from women with HIV is moderate to strong.