Potential hazards of following atypical and low-grade cervical cytology without colposcopy

Abstract

Objective: To determine the frequency of high-grade cervical intraepithelial neoplasia in patients with atypical and low-grade cervical cytology and to assess the optimal evaluation and follow-up. Methods: Prospective observational study of 367 of 7,651 private patients who had cytologic, virologic, or colposcopic changes suggesting cervical intraepithelial neoplasia or cervical carcinoma. The study was performed to determine the frequency of cervical intraepithelial neoplasia in the various cytologic groups and to assess the effect of testing for human papillomavirus on the sensitivity, specificity, and positive predictive value of these tests. Results: Papanicolaou smears that included all non-negative tests (high- and low-grade squamous intraepithelial lesions and atypical squamous or glandular epithelial cells of undetermined significance) had the maximal sensitivity (89%) for high-grade cervical intraepithelial neoplasia and cancer. The combined cytologic categories of high- and low-grade squamous intraepithelial lesions had a sensitivity of 58%, this was reduced to 24% if only high-grade squamous intraepithelial lesions were considered relevant for additional evaluation. If we had not evaluated the patients with atypical squamous and glandular cells of undetermined significance, we would have missed diagnosing one third of high-grade and one half of low-grade cervical intraepithelial neoplasia. Cervical cytology was false negative in 8% of patients with high-grade and in 14% of those with low-grade cervical intraepithelial neoplasias. High-risk human papillomavirus deoxyribonucleic acid was detected in 40% of women with high-grade and in 24% of those with low-grade grade cervical intraepithelial neoplasias. The positive predictive value of cytology with atypical squamous cells of undetermined significance increased from 5% to 38% for high-grade and from 30 to 85% for high- and low-grade cervical intraepithelial neoplasias in patients with detectable high-risk human papillomavirus deoxyribonucleic acid. Virologic studies produced no significant improvement on these diagnoses in women with high- or low-grade cytology. Conclusions: Because of the poor sensitivity of cytology suggesting high-grade squamous intraepithelial lesions, we recommend that all women with atypical or low-grade cytology be recalled for colposcopy and high-risk human papillomavirus deoxyribonucleic acid testing, if available. Decisions whether to perform a biopsy should be based on the result of colposcopic examination. Performing colposcopy only on those patients who have cytologic high-grade squamous intraepithelial lesions and following those with lower grade cytologic anomalies without colposcopic diagnosis appears inadequate to rule out high-grade cervical intraepithelial neoplasia.