Tissue Microarray (TMA)-Based IHC Study Can Significantly Underestimate the Expression of HER2
J Hatem, Y Lin, JM Wang, DG Hicks, P Tang. University of Rochester Medical Center, Rochester, NY; RTI Health Solution, Research Triangle Park, NC
Background: The use of TMA has been shown to be a very cost effective tool in translational research. The accuracy of TMA-based studies is largely dependent on the uniformity of the staining pattern for a given antibody, which is frequently not the case for breast markers. This project aimed to study the non-concordance rates of ER, PR and HER2 expression in TMA and whole sections of breast cancer.
Design: Seventy-five cases of pure DCIS (ductal caicinoma in situ, 0.5cm - 4.8cm in size) were compared for the expression of ER, PR and HER2 in TMA and whole sections. The TMA was constructed with 3 cores (1mm) per case. ER and PR were recorded as Allred scores with 3 and above as positive, 2 and under as negative. HER2 were recorded as positive if >30% of tumor cells showed 3+ membrane staining. Concordance was defined as the same staining result for one or more cores from TMA with whole section result, and non-concordance includes cases with opposite staining result in at least one of the three cores compared to whole section (dis-concordance) and non-informative defined as no tissue or no tumor cell present in all three cores of a given case.
Results: Among the 75 DCIS cases, the rates of under estimation for ER, PR and HER2 expression were 10% (P=0.491), 24% (P=0.101), and 73% (P=0.001), respectively. The non-concordant rates (including discordant and non-informative cases) were 10.7%, 24%, and 28% for ER, PR and HER2, respectively (p=0.012). The non-concordant rates were inversely related to core number, with 46.67%, 22.67%, and 11.56% for one core, two cores, and three cores per marker per case (p<0.001). However, the non-concordant rates were not associated with tumor sizes.
Conclusions: While TMA's have been shown to be an effective tool for translational research, the current study calls attention to potential concerns for the evaluation of markers that do not demonstrate uniform expression throughout tumors. Significant under estimation and non-concordance by TMA analysis for breast cancer warrants caution in the interpretation of results. Increased numbers of tissue cores per case may help to improve the accuracy and concordance rates with whole sections for such studies.
Tuesday, March 10, 2009 1:00 PM
Poster Session IV # 40, Tuesday Afternoon