Anatomic Pathology / HER2 ANALYSIS BY HERCEPTEST AND PATHVYSION ASSAYS HER2 Assessment by Immunohistochemical Analysis and Fluorescence In Situ Hybridization Comparison of HercepTest and PathVysion Commercial Assays

We determined HER2 protein overexpression by immunohistochemical analysis and HER2 gene amplification by fluorescence in situ hybridization (FISH) in 215 formalin-fixed, paraffin-embedded breast tumors. Pathologist concordance for immunohistochemical scoring, and HER2 status concordance, as determined by immunohistochemistry and FISH, were high for immunohistochemical 3+, 1+, and 0 tumors but poor for 2+ tumors. Consensus immunohistochemical scores correlated with absolute and chromosome 17 (CEP17)–corrected HER2 gene copy number. Among HER2-nonamplified tumors, the immunohistochemical score and mean absolute chromosome 17 (CEP17) copy number were weakly correlated. Seventeen tumors were HER2-amplified using absolute HER2 gene criteria but nonamplified when corrected for chromosome 17 polysomy (8 of these were immnuohistochemical 2+). Assessment of benign epithelium within the immunohistochemical slides revealed either no staining or basolateral membrane staining, suggesting normal HER2 protein expression. Twenty tumors showing similar basolateral HER2 immunostaining were all low-moderate grade, tubule-forming, and HER2-nonamplifed (17) or borderline amplified (3). Additional studies relating changes in HER2 gene content due to amplification or chromosome 17 polysomy and HER2 protein expression may be helpful to pathologists who interpret HER2 immnuohistochemical slides. Breast tumors scored at 2+ should be analyzed by FISH, preferably using a dualprobe FISH assay capable of distinguishing HER2 gene amplification from chromosome 17 polysomy. The human HER2 gene (HER-2/neu, c-erbB-2) encodes 1 member of a family of 4 transmembrane tyrosine kinases (HER1-4), the prototype of which is the epidermal growth factor receptor (HER1, erbB-1) (reviewed in Yarden and Sliwkowski1). Various stromal-derived ligands, including epidermal growth factor, epidermal growth factor–like ligands, and neuregulins bind HER1, HER3, and HER4, inducing homodimerization and heterodimerization, phosphorylation of cytoplasmic tyrosine kinase moieties, and activation of complex signaling pathways essential for cell survival, differentiation, and proliferation.1-4 HER2, however, is an orphan receptor with no known high-affinity ligand. HER2 becomes activated by heterodimerization after direct ligand binding by HER1, HER3, or HER4. Thus, the role of HER2 in the network of membrane receptor kinases seems to be as an amplifying coreceptor for HER1, HER3, and HER4.2,3 A specific erbB-2 interacting protein (ERBIN) restricts the spatial distribution of the HER2 molecule to the basolateral membrane of epithelial cells.5 ERBIN binds HER2, but not HER1, HER3, or HER4, and may be involved in connecting HER2 to cytosolic and cytoskeletal-associated components.5 The HER2 gene is central to the oncogenesis and clinical behavior of 25% to 30% of human breast cancers.6,7 HER2 overexpression and/or gene amplification is prognostic for node-positive6,7 and node-negative breast cancers,8 predictive for some chemotherapeutic and hormonal agents, and is an indication for trastuzumab (Herceptin, Genentech, South San Francisco, CA) therapy in patients with metastatic breast carcinoma.9,10 The risk of cardiac toxic effects11 and the specter of patients with false-positive results accruing into breast cancer clinical trials12 have recently focused attention Am J Clin Pathol 2002;117:935-943 935 © American Society for Clinical Pathology McCormick et al / HER2 ANALYSIS BY HERCEPTEST AND PATHVYSION ASSAYS on the sensitivity and specificity of clinical HER2 assays. As the level of HER2 protein overexpression and gene amplification have prognostic and predictive relevance, HER2 assays must be semiquantitative and reproducible and discriminate normal from cancer-associated HER2 protein expression or gene content.13-15 For clinical HER2 determination, tissue-based methods, such as immunohistochemical analysis and fluorescence in situ hybridization (FISH), have replaced whole-tissue extraction methods, such as Southern blot analysis, enzyme-linked immunosorbent assay, and polymerase chain reaction, which may require fresh tissue or suffer dilution owing to admixing of tumor and normal cells.16 Frozen section immunohistochemical analysis, a “gold-standard” method for HER2 overexpression, is impractical in the current era of early cancer detection, in which tumor size often precludes ancillary testing of fresh tissue. Immunohistochemical analysis is an attractive method for clinical HER2 determination owing to its retrospective potential and specific targeting of tumor cells. The plethora of available antibodies, methods, and grading schemes,8,17 however, have made standardization impossible. HercepTest (DAKO, Carpinteria, CA), a formalin-fixed, paraffinembedded (FFPE)-suitable commercial immunohistochemical analysis assay, gained approval from the US Food and Drug Administration (FDA) in 1998 in part for its promise of standardization.16 Subsequent reports documented low specificity of the HercepTest 2+ category compared with gene amplification assays such as FISH,14,18,19 some authors even calling for elimination of the 2+ category as a criterion for trastuzumab therapy.20 Others have continued to advocate immunohistochemical analysis over FISH for its lower cost and greater availability, proposing manual or digital-assisted subtraction of background HER2 immunostaining using benign epithelium as a way of improving HercepTest specificity.18,21,22 Benign breast epithelium was available in only 54% of the HercepTest slides in one recent series, however.23 Finally, proposals to subtract background epithelial staining have not clearly distinguished membrane-associated immunostaining, which may represent normal HER2 expression,5,23,24 from cytoplasmic staining, widely regarded as nonspecific.6,7,16 One postulated advantage of immunohistochemical analysis over FISH for HER2 determination is the potential to detect protein overexpression in HER2-nonamplified tumors. Such overexpression-positive/amplification-negative tumors occur in 3% to 8% of breast cancers in most series,25-27 but were as high as 29% and 31% in comparisons of the PathVysion (Vysis, Downers Grove, IL) FISH assay with the HercepTest immunohistochemical23 assay and the Genentech clinical trials assays,28 respectively. In a study of 900 breast cancers, however, Pauletti et al14 found the immunohistochemical-FISH discrepant tumors behaved clinically as predicted by FISH. Finally, Tubbs et al20 found that the immunohistochemically 2+/FISH-negative tumors contained no detectable HER2 messenger RNA (mRNA), indicating that up-regulation of HER2 gene transcription without gene amplification is unlikely. These studies indicate that the majority of immunohistochemically positive/FISH-negative tumors are false-positive immunohistochemical results. Pauletti et al25 and Kallioniemi et al26 described the simultaneous enumeration of HER2 genes and chromosome 17 centromeres by FISH, defining HER2 gene amplification as the ratio of HER2 gene copies per chromosome 17 centromere. Two-color FISH is highly sensitive (96%-98%) and specific (100%) compared with Southern blot analysis. The FDA-approved FISH assay, PathVysion, also uses a dual-probe system for simultaneous enumeration of HER2 genes and chromosome 17 centromeres in FFPE breast tumors. Another FDA-approved commercial FISH assay, INFORM (Ventana Medical Systems, Tucson, AZ), defines HER2 gene amplification as a mean absolute HER2 gene copy number of more than 4 per tumor nucleus. A study by Pauletti et al14 using the PathVysion FISH system in 900 breast cancer patients found chromosome 17 correction essential for demonstration of HER2 gene amplification; others believe it is unnecessary.20 We determined the HER2 status of a diverse group of primary breast cancer specimens using the HercepTest immunohistochemical and PathVysion FISH assays. The concordance between 2 experienced surgical pathologists (S.R.M., T.J.L.) in assessing immunohistochemical scores was determined. We also compared the HER2 gene and chromosome 17 (CEP17) copy numbers in relation to immunohistochemical score and attempted background subtraction using benign epithelium as has been proposed to improve the specificity of the immunohistochemical 2+ group.18,21 Finally, we compared 2 definitions of HER2 gene amplification: absolute HER2 copies per tumor nucleus (threshold, 4.0) and HER2 copies per chromosome 17 centromere (threshold, 2.0). Materials and Methods