Reproducibility in Assessing Ovarian Carcinoma Cell Types A Transcanadian Study
M Kobel, J Arseneau, P Baker, CA Ewanowich, D Fontaine, R Parker, DG Huntsman, CB Gilks. University of British Columbia, Vancouver, BC, Canada; McGill University, Montreal, QC, Canada; University of Manitoba, Winnipeg, MB, Canada; University of Alberta, Edmonton, AB, Canada; Memorial University, St. John's, NL, Canada; University of Calgary, Calgary, AB, Canada
Background: Reproducible histopathological cell type assignment for ovarian carcinoma becomes critical if different cell types are to be treated differently, as has been proposed. The purpose of this study was to test the concordance of cell type diagnosis between gynecological pathologists from across Canada.
Design: After an on site training session using a training set of 40 different ovarian carcinomas (available for review at http://spectrum.med.ubc.ca/pathology/ (username: ovcare, password: diamond then click on Search all Digital Slides and enter Folder name ovcare), six gynecological pathologists reviewed one representative slide of each case of the test set. The test set consisted of 40 cases from a population based review series were selected consecutively to ensure a representative distribution of cell types as follows 22 high-grade serous, 10 clear cell, 7 endometrioid, 5 mucinous, 4 low-grade serous. In a second round, pathologists received in tabular form the results of immunostains of the test set cases with 10 selected biomarkers, including WT1, ER, TP53 etc. and had the opportunity to change their diagnosis. Diagnosis of cell type had to be from one of the following six categories: high-grade serous, clear cell, endometrioid, mucinous, low-grade serous, unclassified. Concordance between pathologist pairs and overall concordance was calculated.
Results: The overall concordance between pathologists in assignment of cell type was 92.3% (range 85.0 97.5%), which was only slightly increased after consideration of immunohistochemistry staining data to 92.6% (range 87.5% - 97.5%).
Conclusions: Using a limited number of diagnostic categories, gynecological pathologists have a high concordance in diagnosis cell type, after a short training period. This is a large improvement over earlier studies of reproducibility of cell type diagnosis in ovarian carcinoma and has implications for clinical trials of subtype-specific treatments of ovarian carcinoma.
Wednesday, March 11, 2009 9:30 AM
Poster Session V # 134, Wednesday Morning