Ranking of Individual Laboratory Performance in Immunohistochemistry Proficiency Testing Using Digital Image Analysis
Heather Neufeld, Dragana Pilavdzic, John Garratt, Blake Gilks, Carol Cheung, Barry Ziola, Richard Berendt, John Krahn, Svein Carlsen, Emina Torlakovic. University of Saskatchewan, Saskatoon, SK, Canada; Jewish General Hospital, McGill University, Montreal, QC, Canada; Lions Gate Hospital, Vancouver, BC, Canada; Vancouver General Hospital, University of British Columbia, Vancouver, BC, Canada; University Health Network, University of Toronto, Toronto, ON, Canada; Cross Cancer Institute/University of Alberta, Edmonton, AB, Canada
Background: Various methods can be used for monitoring laboratory performance in proficiency testing (PT) in immunohistochemistry. Currently, Canadian Immunohistochemistry Quality Control (CIQC), PT results are evaluated by a panel of “expert assessors” including pathologists and technologists. The results of the assessments are used to calculate the sensitivity, specificity and kappa values in order to evaluate accuracy of IHC tests. In this study, we compared i) the rankings of participating laboratories using digital image analysis (IA)-based scoring vs. those derived by usual expert assessment, and ii) calibrated and standardized cell line microarrays (CLMAs) vs. human tissue-based TMAs.
Design: Both assessment methods evaluated two consecutive Canada-wide PT runs for estrogen receptor (ER), progesterone receptor (PR) and HER2. Newly calibrated and standardized cell line microarrays (CLMAs) were built from formalin-fixed paraffin-embedded cell pellets from nine human breast cancer cell lines. Laboratory Score Reference Score Ratios (LSRSR) were calculated using H-scores generated from both cell line- and tissue-based samples. Participating laboratories LSRSRs were compared to the results obtained by 6 reference laboratories, which range was used as acceptable standard.
Results: Laboratory ranking by IA was not significantly different than by ranking by expert assessment. CLMAs were better discriminators than tissue-based samples. Cell lines with intermediate H-scores were more sensitive in detecting interlaboratory variation than those with either very high or very low H-scores. The IA-based method also allowed for the monitoring of performance trends for each participating laboratory over multiple successive runs.
Conclusions: IA-based evaluation of PT provides useful information regarding monitoring of protocol calibration. This assessment method requires only a small number of cell line-based samples. This approach also significantly reduces the time and costs of PT evaluation.
Category: Quality Assurance
Tuesday, March 5, 2013 9:30 AM
Poster Session III # 247, Tuesday Morning