Accurate HER2 status testing is important for identifying breast cancer patients who may benefit from receiving trastuzumab therapy. Currently, in the United States, HER2 IHC methods are most commonly used for primary screening for HER2 status, and borderline cases are subjected to dual FISH for HER2 and CEN 17 to determine the HER2/CEN 17 ratio. Because the discordance rate between local and central/reference HER2 status testing with IHC and FISH is significantly high [14, 31–33], the standardization of diagnosing breast cancer cases is recognized as a very important task for improving personalized cancer patient care [3, 34]. The American Society of Clinical Oncology and the College of American Pathologists has published a guideline recommendation for testing HER2 status in breast cancer  and the Canadian National Consensus has updated the Canadian HER2/neu testing guideline . Two potential solutions for improving the standardization of HER2 status testing include: 1) automating the entire process for slide staining  and slide reading [36–39] and 2) consolidating the HER2 testing process within experienced laboratories and pathologists that perform large numbers of HER2 tests .
One way to improve the accuracy of HER2 status testing is to automate the assay procedure for HER2 IHC and HER2 FISH assays so that human errors can be diminished. HER2 IHC assays can be performed using an automated slide staining system, but HER2 FISH assays remain technically challenging and time consuming manual molecular diagnostic assays in most laboratories. An evaluator of FISH slides must have access to specialized fluorescence microscopy in a dark room. Because of unstable FISH staining characteristics, the signals of FISH slides can be bleached easily, even while reviewing and enumerating signals. Furthermore, digital images of the FISH slide need to be captured with a sensitive camera system for each patient case for the HER2 gene status record. Therefore, it is desirable to automate a tissue-based HER2 gene status test that can be observed with a regular brightfield microscope and that produces stained slides that can be archived.
While the concept of multi-color brightfield ISH applications was published in 1990's [40, 41], it was a recent achievement to visualize HER2 and CEN 17 targets within the same nuclei of tissue sections with a manual dual brightfield ISH application . This dual ISH application utilized TMB chromogen for HER2 gene staining. However, based on published images [28, 42], TMB staining does not provide discrete signals when compared to the SISH application. The advantages of the BDISH application for HER2 gene and CEN 17 presented in the current study are: 1) the automation of the ISH application; 2) the visualization of both HER2 gene and CEN 17 targets in the nuclei of the same cell; 3) the generation of discrete HER2 gene signals; 4) the ability to reproducibly detect endogenous HER2 and CEN 17 signals in the stromal tissues and lymphocytes as a reliable internal assay control; 5) the ability to visualise signal with brightfield microscopy with non-oil immersion lenses; and 6) the capability to permanently archive the slides.
HER2 and CEN 17 probes are co-hybridized for dual color HER2 FISH. However, for the BDISH assay, because the stringency conditions for the nick-translated HER2 probe and the CEN 17 oligoprobe were different, it was necessary to conduct sequential ISH staining steps for HER2 gene and CEN 17 targets. For CEN 17 ISH, we have optimized a new detection system with an alkaline phosphatase-conjugated antibody and fast red chromogen and naphthol phosphate substrate reaction. The fast red-based detection was selected to obtain a good contrast of CEN 17 ISH signal against the discrete black dots of HER2 SISH signal. DAB, BCIP/NBT, and TMB detection systems did not provide sufficient contrast against HER2 ISH black signal (data not shown). Because fast red precipitate is soluble in organic solvents, in general, aqueous mounting medium is used for coverslipping. However, the standard coverslipping method with aqueous mounting medium on wet tissue sections did not produce tissue sections with high resolution and therefore detailed tissue structure could not be observed (data not shown). A successful method to preserve fast red staining for CEN 17 and high resolution tissue morphology was, after completely dry the slides, to apply a toluene-based tissue mounting medium (Cytoseal 60) for coverslipping with cover glass or to use a film coverslipper (Tissue-Tek® film coverslipper). Incomplete drying resulted in faint red background staining particularly around the fast red precipitate sites with this method. Interestingly, the use of aqueous mounting medium onto the dried tissue slides produced yellowish background staining on tissue sections and this method did not produce satisfactory results (data not shown).
The specificity of single ISH for HER2 gene or CEN 17 and BDISH for both targets was evaluated with xenograft tumors. HER2 and CEN 17 copy numbers have been documented previously using the FISH assay . MCF-7 cells are characterized as non-amplified HER2 and chromosome 17 polysomy (3 copies of chromosome 17 per nucleus) while one of chromosome 17 with HER2 deletion (2 HER2 copies per nucleus) . BT-474 cell line presents HER2 amplification with 50–60 copies of HER2 genes and 4–6 copies of CEN 17 per nucleus . Amplified HER2 genes are located not only on chromosome 17, but also are translocated on other chromosomes . HER2 and CEN 17 copy numbers produced with the single target ISH and BDISH methods matched with previously reported results. As both probes are labeled with the same DNP hapten, our first concern was to determine if detecting specific signal for each probe was feasible. We confirmed that the fast red chromogen detection reagents did not produce red signal when the fast red ISH was performed without the CEN 17 probe after detection of HER2 by SISH (data not shown). Thus, the SISH detection and the fast red detection can be combined to perform a sequential double ISH assay with 2 probes labeled with the same hapten. Because the sequential BDISH application uses 2 specific stringency conditions based on the length and sequences of 2 probes, it is not necessary to design 2 probes that require the same stringency for co-hybridization, like double color FISH assays.
Concordance rates between a set of gold standard dual color HER2 FISH scores and HER2 and CEN 17 BDISH scores by 4 observers were calculated for assessing the performance of the BDISH assay. There were 9 discordant cases (9.6% of the total cases) based on BDISH score disagreement with FISH scores, even by one observer. We have found that the number of equivocal cases influences the concordance rate with the ASCO/CAP scoring method. There were 4 equivocal cases based on FISH scores and all cases showed the BDISH score disagreement by at least 2 observers. A similar observation was reported with an international HER2 testing proficiency study . In their study, the discordant cases (20%) were caused by the specimen having FISH HER2/CEN 17 ratios between 1.7 and 2.3 that are close to the 'equivocal' defined by ASCO/CAP HER2 scoring method (1.8 – 2.2). They also stated "equivocal cases are difficult to interpret, even highly experienced and validated laboratories" . In one study, when the FISH assay was used as the primary test for HER2 status assessment of breast carcinoma cases, heterogeneity of HER2 gene status was observed in 40 of 742 cases (5%) . It has been speculated that genomic and phenotypic heterogeneity of tumor cells is the main reason for the inconsistency of HER2 testing results . With current study, all of our discordant cases (9/9 or 100%) displayed the tumor cell population heterogeneity: three samples showed significant segregated tumor cell population heterogeneity (Figures 4B&C) and other cases showed subtle heterogeneity of tumor cell populations that are seen among equivocal cases (4/4 or 100%). Nonetheless, since consecutive tissue sections were not used for the FISH and BDISH analyses, one can speculate that the tissue sections for the FISH and BDISH tissue sections contained tumor cell populations with different HER2 status. Further clinical evaluations of HER2 and CEN 17 BDISH application with patient treatment outcome data are required for more accurate HER2 status assessment of breast cancer patients to be obtained.