A Method for Decreasing Interobserver Variability in Quantitative HER2 Immunohistochemistry
David P Ng, Larry J Dumont, Gregory J Tsongalis, Vincent A Memoli. Dartmouth Hitchcock Medical Center, Lebanon, NH
Background: Over-expression of the HER2 receptor is associated with aggressive breast cancer development. The highly specific immunotheraputic targeting of HER2 by mABs and TKIs, as well as, the high cost of these treatments make accurate assessment of HER2 status essential. Currently, the main techniques used to assess HER2 status are FISH and IHC. FISH is limited to laboratories with fluorescent microscopy capabilities while IHC is the most prevalent technique for quantifying HER2 status. However, quantitative immunohistochemistry (qIHC) shows poor reproducibility between observers. Here we report a method to decrease inter-observer variability in HER2 qIHC through the inclusion of reproducible standard controls.
Design: Four immortalized cancer cell lines with consistently differing levels of HER2 expression (MCF-7, BT-20, MDA-MB-453 and SKOV-3) were grown, pelleted, formalin fixed and paraffin embedded. Control strips were made from 2 mm cores of these cell lines. Eight cases of breast carcinoma known to express HER2 as previously assessed by FISH were cut on a slide with the control strips. The optimum incubation and concentration of a mAB (CB11) for HER2 was found such that an intensity breakpoint was seen between 2+ and 3+ staining patterns corresponding to a HER2/CEP17 ratio of 1.8. Next, an additional 19 cases of breast carcinoma with HER2/CEP17 ratios ranging between 1.0 and 7.2 were evaluated by 5 pathologists using the control strip in addition to ASCO guidelines. Five additional pathologists assessed HER2 status without the control strip. The data was analyzed using Fleiss' correlation and chi-square analysis.
Results: The analysis of HER2 expression by use of control strips and ASCO guidelines significantly decreased inter-observer variability compared to analysis using only ASCO guidelines, (k=0.5686 vs k=0.4337) with a mean difference of 0.1349 (95% CI 0.0591 to 0.2107). Receiver Operating Characteristic curves for the experimental group showed an area-under-curve (AUC) of 0.888 while the control group showed an AUC of 0.856 (not significant).
Conclusions: We describe a new method for decreasing inter-observer variability using qIHC to score HER2 expression. These control strips are easily made from cultured cells and provide sufficient control material to run a large number of quantitative tests. qIHC may serve as a useful and important tool in determining HER2 status in small laboratories lacking the ability to assess HER2 status by FISH or in specimens with small tumor foci where FISH may fail.
Category: Quality Assurance
Monday, March 19, 2012 1:00 PM
Poster Session II # 277, Monday Afternoon