A large-scale investigator-sponsored field study of the test performance of the ThinPrep papanicolaou test in a low-risk routine outpatient setting

Abstract

The performance of the ThinPrep Papanicolaou test (TP)(CYTYC Corp., Boxborough, MA) was evaluated in a low- risk outpatient population in order to assess its efficacy in routine clinical use. A total of 18,246 cervical cytology samples were prepared using TP during a 6-month evaluation period from January through June 1997. Results of TP interpretation, assessed for specimen quality and diagnostic information as reported according to the Bethesda system terminology, were reviewed and compared with corresponding results from 2 historical control series of 35,347 and 35,409 conventional cervical smears (Papanicolaou smears). Archived conventional smears obtained during the first 6 months of 1995 and 1996 from the same general patient population by the same group of gynecologists and general practitioners were selected for analysis. Positive cytologic diagnoses in the TP series were reviewed and cytologic-histologic correlations were analyzed when biopsy material was available. The results demonstrated that the proportion of cytologic examinations classified as “satisfactory” was significantly greater in the TP series as compared to conventional controls (88.8% vs 71.1%, P < .0001). There was a significant increase in the number of TP examinations with an adequate endocervical component as compared to conventional controls (90.2% vs 86.9%, P < .005) and a significant decrease in the number of TP examinations that were classified as “satisfactory but limited by” either poor fixation or obscuring factors, such as blood, neutrophils, and contaminants (1.1% TP vs 14.1% nonTP, P < .0001). The overall sensitivity of detection for abnormal cells as indicated by the number of positive diagnoses was increased in the TP series (5.7% vs 2.4%, P < .01). The cytologic diagnoses of ASCUS, LSIL, and HSIL were respectively 1.6, 3.6, and 3 times more frequent in the TP series than in controls. Cytologic-histologic correlations were available for 110 (27%) of the 403 cytologic diagnoses of LSIL and for 90 (83%) of the 121 cytologic diagnoses of HSIL. The presence of a histologic lesion (LSIL or HSIL) was confirmed in 71 (65%) biopsies for LSIL and in 81 (90%) biopsies for HSIL. In conclusion, both sample quality and diagnostic sensitivity were significantly improved using the ThinPrep Papanicolaou test under routine conditions in a low-risk population.