Abstract
An automated enzyme metallographic silver in situ hybridization method (SISH) has been reported to successfully determine human epidermal growth factor receptor 2 (HER2) gene amplification. We evaluated the staining and interpretative reproducibility of the HER2 SISH assay at five laboratories and compared SISH results with other in situ hybridization (ISH) methods. The HER2 gene status of 89 breast carcinomas was analyzed in parallel using manual dual-color fluorescence ISH, manual chromogenic ISH, and bright-field automated SISH. A total of 1098 SISH-stained slides were evaluated. For comparison, all specimens were stained by 4B5 immunohistochemistry for HER2 protein expression. Interpretation was performed by pathologists at five different laboratories using the algorithms provided by the manufacturers and the guidelines of American Society of Clinical Oncology/College of American Pathologists. Staining and interpretative reproducibility were measured through the computation of weighted kappa statistics. Following the optimization of SISH staining, 1077/1098 (98%) of slides were evaluable. Excellent reproducibility and efficacy of HER2 SISH staining, and interobserver interpretation (Kw = 0.91), were observed among five sites. For the 89 invasive breast cancer cases, the overall rate of concordance between consensus 4B5 and consensus SISH, fluorescence ISH, and chromogenic ISH was 96.6% (86/89), 97.8% (87/89), and 96.6% (86/89), respectively. Overall concordance between positive and negative SISH and fluorescence ISH results, as well as between individual and consensus positive and negative SISH results, was excellent (P < 0.001).