Abstract
BACKGROUND: A persistent goal of oncologic histochemistry is to microscopically identify neoplasia tinctorially. Consequently, the newly developed CellDetect® staining technology, that appears to exhibit this property, warrants clinical evaluation. The objective of this study was to compare the diagnostic results using CellDetect® to the outcomes of standard microscopic examination based on hematoxylin and eosin (H&E) staining for the recognition of different squamous epithelial phenotypes of the uterine cervix. METHODS: Pairs of adjacent sections were made from 60 cervical biopsy cases that were diagnosed originally as either normal or neoplastic (CIN, SCC). One section of the pair was stained for H&E; the second section, with CellDetect®. Based on the examination of these pairs by two experienced pathologists, we investigated the following issues:(1) diagnostic agreement between the pathologists on each pair; (2) agreement between H&E and CellDetect® for each pair (3) tinctorial characteristics in micro-regions (n = 130) evaluated as either normal, reactive or neoplastic. RESULTS: Qualitatively, CellDetect®-stained preparations displayed cyto-morphological detail comparable to H&E images. Tinctorially, non-neoplastic cells appeared green/blue when stained withCellDetect®, contrasting with cytologically neoplastic foci, where cells of every grade were red/magenta in color. Due to these tinctorial characteristics, even small foci of neoplasia could be readily distinguished that were inconspicuous on H&E at low magnification. In some instances, this prompted re-examination of the H&E and revision of the diagnosis. Quantitatively, we found that despite diagnostic variation between pathologists, in about 3% of the cases, each pathologist made the same diagnosis regardless of whether CellDetect® or H&E was used, i.e. there was 100% self-agreement for each pathologist between stains. Particularly noteworthy was the finding of a 0% false negative rate, coupled with a 10-15% false positive rate. Regarding specificity, the performance in reactive squamous processes was similar to that observed for morphologically normal squamous epithelium. CONCLUSIONS: In this first order assessment of clinical applicability, CellDetect® staining technology was at least comparable to results using H&E, and perhaps surperior. CellDetect® provided a uniquely useful tinctorial clue for the detection of neoplasia, which exhibited an impressive 0% false negative rate. A more extensive, blinded study is needed to confirm these promising findings.