Utility of the Thin Prep Imaging System® in the detection of squamous intraepithelial abnormalities on retrospective evaluation: can we trust the imager?

Prospective studies analyzing the ThinPrep Imaging System (TIS) have demonstrated a significant decrease in screening time and detection rates comparable or better than manual screening. We retrospectively analyzed the accuracy of the TIS in detecting cervical abnormalities. Our study included all new HSIL diagnoses in 2007 with previous negative (NIL) pap tests screened with TIS. The original 22 fields of view (FOV) were reviewed by 2 blinded screeners followed by manual screening of all slides. Any ASC-US or above was considered "abnormal." Of a total of 111,080 pap tests performed in 2007, 180 were reported as HSIL. Of these, 45 cases had a previous NIL pap diagnosed within the last year, screened with TIS. Following re-examination of the NIL pap, 31 diagnoses remained unchanged and 9 were reclassified as abnormal on the basis of cells present within the original FOV. When manually reviewed, all nine cases were confirmed as abnormal. Four cases were reclassified as abnormal on the basis of the manual screen (abnormal cells absent in the FOV). The sensitivity of TIS for the detection of abnormality was 99.95% (false-negative rate FNR: 0.05%) and the sensitivity for detection of HSIL was 99.07% (FNR: 0.92%). When analyzing the cytotechnologist interpretation of the FOV, the sensitivity for detection of abnormality and HSIL was 99.89% (FNR: 0.1%), and 99.53% (FNR: 0.4%), respectively. On retrospective analysis based on newly diagnosed HSIL cases, the sensitivity of TIS was comparable to that of manual screening with a slightly decreased rate of false negatives.