Poor correlation of high-grade cervical dysplasia with referral cytology: clinical implications

Abstract

Objective: We questioned the diagnostic utility of cervical cytology by studying whether women with colposcopically guided biopsy proven high-grade cervical dysplasia or cancer had referral cytology results that correlated highly with their disease. Methods: All women with any cytologic abnormality or visible lower tract lesion(s) are referred for colposcopic examination. Ten thousand visits to our colposcopy clinic were prospectively collected, tracked, and reviewed, which revealed 566 cases of high-grade dysplasia and 8 cancers on biopsy. The proportion of high grade disease that was discovered by high grade cytology was measured. Results: The diagnosis of high grade cervical disease was not preceded by high-grade cytology (HGSIL) or “rule out cancer” cytology result ( n = 95, 16.8%) in the majority of cases ( P < .0001, χ 2). High-grade disease was more often preceded by low-grade cytology (LGSIL) ( n = 224, 39.5%), followed by atypical Papanicolaou smears (ASCUS) ( n = 220, 38.9%) and visible cervical or vaginal lesion(s) ( n = 27, 4.8%). A very high percentage of cancer cases had preceding HGSIL cytology (6 of 8 cases at least HGSIL, 75%). Conclusion: High-grade cervical neoplasia was more likely to be discovered following colposcopic triage of “minor” Papanicolaou smear abnormalities (ASCUS or LGSIL). The false negative rate of a single Papanicolaou smear to detect dysplasia may be as high as 65% (with specificity at the 90% range) (Fahey MT, Irwig L, Macaskill P. Meta analysis of Pap test accuracy. Am J Epidemiol 1995;141:7:680–9), making cytology an inaccurate diagnostic replacement for colposcopy. We should rethink the significance of minimally abnormal Papanicolaou smears when designing triage protocols that delay or eliminate colposcopy based on the Bethesda Grading System alone.