Use of Automated Primary Screening on Liquid-Based, Thin-Layer Preparations

Abstract

To evaluate the accuracy of the AutoPap System (TriPath Imaging, Inc., Burlington, North Carolina, U.S.A.) (TriPath) in screening AutoCyte PREP liquid-based, thin-layer preparations by comparing the final cytologic diagnoses with instrument slide classification results. A total of 9,665 AutoCyte PREP thin-layer slides were first independently screened to establish a final cytologic diagnosis (reference diagnosis). The slides were then processed on the AutoPap System. Each slide successfully processed was reported into result categories. The generated report gave a ranking score for each slide designated for "review." Slides designated "no further review" (NFR) were also listed in the report. The reported results were then compared to the reference cytologic diagnoses. Of 9,665 slides initially submitted to the AutoPap, 8,688 (90.8%) were qualified for scanning, and 884 (9.2%) were definitely classified as process review or rerun and excluded from the study. Of high grade squamous intraepithelial lesions and greater (HSIL+), 85.2% were ranked in the first rank, 12.7% in the second, one (2.1%) in the third, none in the fourth and fifth and none in the NFR category. Of low grade squamous intraepithelial lesions, 47.4% were ranked in the first rank, 20.8% in the second, 10.6% in the third, 10.1% in the fourth, 5.3% in the fifth and 5.8% in NFR. Of atypical squamous cells of undetermined significance and atypical glandular cells of undetermined significance, 53.6% were ranked in the first rank, 22.5% in the second, 12.4% in the third, 5.4% in the fourth, 3.8% in the fifth and 2.3% in NFR. Considering a cutoff value at < or = 3rd rank, 84% of cervical abnormalities (RR 6.52, 95% CI 4.96-8.66) and 100% of HSIL+ were identified. The AutoPap demonstrates a high capability for detecting cervical abnormalities on AutoCyte PREP thin-layer slides. HSIL+ was associated with the highest instrument scores.