Variation in Assessment of ER and PR Expression of the Same Tumor Block with Repeated IHC Stainings by Computer Assisted Imaging Analysis and Manual Analysis
Huijiao Chen, Jianmin Wang, Loralee McMahon, Qi Yang, Hong Bu, David G Hicks, Ping Tang. University of Rochester Medical Center, Rochester, NY; RTI Health Solution, Research Triangle Park, NC; West China Hospital of Sichuan University, Chengdu, Sichuan, China
Background: The accurate assessment of estrogen receptor (ER) and progesterone receptor (PR) in breast cancer significantly impacts the decision of adjuvant therapy. Multiple pre-analytic, analytic and post-analytic factors have been implicated in affecting the accuracy of ER and PR assessment, which may impact patient outcome. Here, we sought to investigate the affect of repeating immunohistochemical (IHC) stainings of the same tissue block on ER and PR interpretation, comparing manual readings to computer assisted image analysis.
Design: We identified 48 cases of breast cancer with reported negative, weak, or strong staining for ER and PR from our 2009 departmental file. One representative block from each case was selected for IHC analyses for ER and PR. The IHC stainings were first scored manually with both percentage (0-100%) and intensity as (0-3). Two more IHC stainings for ER and PR were performed one week apart. Automated computer analysis with percentage (0-100%) and intensity (0-255) for ER and PR was then conducted on 10 representative tumor areas selected by a pathologist for each case. Direct comparison among three automated readings for percentage and intensity, and comparison between one automated and manual reading for ER and PR were performed for these cases.
Results: We found that 1) there were significant difference in percentage and intensity of ER and PR readings among the three different IHC stainings; 2) between any two automated readings for percentage, there was one of the six groups showing significant difference for ER (0.0028), and none for PR; 3) between automated and manual readings for percentage, there was significant difference for PR (0.0468) but not for ER (0.6848); 4) among 3 groups of automated reading for intensity, 2/3 groups showed significant difference when compared two groups at a time for ER or PR. No disagreement between automated and manual assessment case using 1% cut off for positive ER and PR as recommended by 2010 ASCO/CAP guidelines.
Conclusions: Although significant difference are present ER and PR interpretation, there is no difference in classification of ER and PR as positive or negative regardless of the different repeats of IHC stainings or the methods of assessment (automated or manual).
Wednesday, March 21, 2012 9:30 AM
Poster Session V # 34, Wednesday Morning