False positive diagnosis in conventional and liquid-based cervical specimens

OBJECTIVE: To examine conventional and liquid-based cervical smears falsely diagnosed as malignant at our institution and to investigate, through cytologic-histologic correlation, factors influencing false positive diagnoses. STUDY DESIGN: Cervical cytologic diagnoses of malignancy from May 1, 1995, to April 30, 2001, were retrieved through a computer search. A retrospective review of hospital records and pathology reports was performed. Cases identified as false positives were reviewed and correlated with histologic follow-up specimens. RESULTS: A group of 68 patients with malignancy reported on cervical smears and with histologic follow-up was identified. Conventional smears numbered 32 (47%); the remaining 36 (53%) were liquid-based samples. Of the total, 7 false positive cases (10.3%) were identified in 4 conventional and 3 liquid-based preparations. Cytologic diagnosis in these cases was squamous cell carcinoma in 5 and adenocarcinoma in 2. On histologic follow-up, all 7 patients were ultimately found to have high grade squamous intraepithelial lesions (HSILs) without invasion. Review of the original slides confirmed most, or all, of the following features in all cases: major cellular pleomorphism, extensive cytoplasmic keratinization, intense nuclear pyknosis, background necrosis and severe atrophy. CONCLUSION: There was no significant difference in rates of false positive diagnoses between conventional (12.5%) and liquid-based (8.3%) samples. The chief reason for overdiagnosis in this series was the capacity of HSIL to exfoliate cells mimicking invasive malignancy, particularly when keratinized and especially in an atrophic milieu. The other cause of false positivity was superimposition of inflammation and atypical reparative change on a background of HSIL, which then suggested invasion.