The purpose of this study was to assess the ability of cytology to pre
dict the results of colposcopically directed cervical biopsies in HIV-
infected women. We performed a case-control study of 52 HIV(+), 31% of
whom had AIDS, and 57 HIV(-) women referred to two tertiary care cent
ers for colposcopy from July 1991 to November 1993. All 57 HIV(-) cont
rols and 27 HIV (+) cases underwent colposcopy for evaluation of an ab
normal Pap smear. The remaining HIV(+) cases (n = 25) had colposcopy a
s part of their routine assessment. In women with abnormal Pap smears,
colposcopic biopsy agreed with the Pap smear results in 83% of 24 HIV
(+) women and 65% of 37 controls (chi2; P = 0.34). For patients with l
ow-grade SIL on Pap smear, 14% of HIV(+) and 11% of HIV(-) women had m
oderate or severe dysplasia on biopsy (P = 0.52). The positive predict
ive value of an abnormal Pap smear was 96% in HIV(+) women vs 78% in n
oninfected patients (P = 0.05). In the overall series of 52 HIV(+) wom
en, the Pap smear did not match the biopsy in 44% of patients and was
less severe than the cervical biopsy results in 91% of these mismatche
s. The Pap smear had a sensitivity of 57%, a specificity of 92%, a pos
itive predictive value of 96%, and a negative predictive value of 39%,
when compared to biopsy results in HIV-seropositive patients. Pap sme
ars missed 43% of biopsy-proven intraepithelial lesions in this series
of HIV (+) women. However, when abnormal, the Pap smear was no worse
in predicting the presence and degree of an intraepithelial lesion in
HIV(+) women than in noninfected women. These characteristics may just
ify immediate treatment of HIV(+) women at the time of colposcopy afte
r an abnormal Pap smear given its high positive predictive value. (C)
1995 Academic Press, Inc.