Cytologic predictors of cervical intraepithelial neoplasia in women with an ASCUS Pap smear

OBJECTIVE: To identify cytologic parameters on Pnp smears of women with an atypical squamous cells of undetermined significance (ASCUS) diagnosis that could help cytologists to indicate whether a particular ASCUS case is most likely related to cervical intraepithelial neoplasia (CIN) grade 1 or 2/3. STUDY DESIGN: A total of 360 eligible women diagnosed with ASCUS and referr ed to the colposcopy clinic of Saint-Sacrement Hospital participated in the study. Eligible women were those aged 18-50 years, newly diagnosed with AS CUS, with no history of cervical biopsies or treatment, and not pregnant at the time qi the visit. Colposcopically directed biopsies of lesions were o btained. All Pay smears were reviewed according to 36 different cytomorphol ogic criteria. The regression logistic model was used to estimate the odds ratios (ORs) for the associations between cytologic criteria observed in sm ears and the diagnosis of CIN made on biopsies. All cytologic criteria sign ificantly (P < .05) associated with CIN were entered in the models, and a b ackward selection was none to determine independent cytologic predictors of CIN 1 and 2/3. RESULTS: Biopsies revealed that 22.2% of the study population had concurren t GIN. CIN 1 and 2/3 were identified in 61 (16.9%) and 19 women (5.3%), res pectively. Clear perinuclear spaces (OR = 2.5, P = .002) and moderate nucle ar atypia (OR = 4.4, P = .02) were two cytologic criteria independently ass ociated with CIN 1. Four independent predictors of CIN 2/3 were identified: the presence of clear perinuclear spaces (OR = 5.9, P = .004), hyperchroma sia (OR = 3.9, P = .04), moderate anisokaryosis (OR = 13.1, P = .01) and in creased nuclear volume of metaplastic cells (OR = 5.1, P = .007). CONCLUSION: These observations may help cytologists to better categorize AS CUS lesions Its intraepithelial ones and will also contribute to improving the Bethesda definition of ASCUS. Further studies are planned to validate t hese observations.