Laboratory/FDA-Approved Kit Score Ratio (LFSR) Is a Useful Tool in Measuring Laboratory Performance in Immunohistochemistry Proficiency Testing; CIQc Experience
EE Torlakovic, LAG Lining, D Pilavdzic, L Quenneville, J Garratt, N Makretsov, B Gilks. Royal University Hospital, University of Saskatchewan, Saskatoon, SK, Canada; Sir Mortimer B. Davis Jewish General Hospital, Montreal, QC, Canada; Lions Gate Hospital, Vacouver, BC, Canada; Faculty of Medicine, Memorial University of Newfoundland, St. John's, NFL, Canada; Vancouver General Hospital, University of British Columbia, Vancouver, BC, Canada
Background: Majority of participants in the Canadian Immunohistochemistry Quality Control (cIQc) proficiency testing (PT) produce concordance of ≥ 90% for ER and PR testing based on “positive” versus “negative” scoring. Image analysis may have greater discriminatory power and could potentially provide more accurate information on laboratory performance.
Design: Selected tissue cores with different expression of ER from TMAs of two different cIQc runs were scanned and analyzed by image analysis. H-scores were calculated from measured Percent 3+, Percent 2+, and Percent 1+ Positive Nuclei parameters. Each laboratories' H-scores were divided by H-scores obtained by FDA-approved kit to obtain Laboratory/FDA-Approved Kit Score Ratio (LFSR). Tissue cores with low expression (CLE) and high expression (CHE) were compared for their ability to discriminate between the sensitivity of staining of participants.
Results: LFSR varied between different cores of the same TMA and for the same cores between participants. CLE had largest variation of LFSR between laboratories. CHE LFSR had much lower variation (Figure 1). LFSRs of the five reference laboratories were consistently close to 1; their results were very similar to the FDA-approved kit although none used the kit for testing. Reference laboratories preformed better than other participants as a group (p<0.05, Mann-Whitney U Test) only with CLE samples.
Conclusions: Image analysis identifies subtle problems in sensitivity of ER testing, which are not detected when a simple “positive” versus “negative” assessment is used. CHE samples are not useful in PT since they poorly discriminate between laboratories with highly reproducible results and laboratories with lesser reproducibility. LFSR is a useful tool to measure laboratory performance in immunohistochemistry PT.
Category: Quality Assurance
Monday, March 22, 2010 1:00 PM
Poster Session II # 209, Monday Afternoon