Development of automated brightfield double In Situ hybridization (BDISH) application for HER2 g ene and chromosome 17 centromere (CEN 17) for breast carcinomas and an assay performance comparison to manual dual color HER2 fluorescence In Situ hybridization (FISH)

MCF7 and BT-474 xenograft tumors were utilized for optimizing the BDISH assay. MCF7 is a breast adenocarcinoma cell line with non-amplified HER2 status and BT-474 is a breast ductal carcinoma with amplified HER2 status (50–60 copies of HER2) and chromosome 17 polysomy [29]. Paraffin sections (4 μm) containing tissue cores of formalin-fixed, paraffin-embedded MCF7 and BT-474 xenograft tumors were placed onto Superfrost^® Plus glass slides (Erie Scientific Company, Portsmouth, New Hampshire).

Ninety-four (94) breast cancer cases were used from the Cleveland Clinic Foundation and the Cleveland Clinic Lerner College of Medicine, Cleveland, OH, USA under IRB approved protocol. Tissue samples were routinely processed for paraffin-embedding after fixing with an alcoholic formalin fixative. All breast cancer cases had been previously tested for HER2 status by FISH using the PathVysion^® HER-2 DNA Probe Kit (Abbott Molecular, Des Plaines, Illinois) at the Cleveland Clinic Foundation. However, it should be noted that non-consecutive tissue sections were used for FISH and BDISH analyses.

The BenchMark^® XT automated slide processing system (Ventana Medical Systems, Inc., Tucson, Arizona) was used for the optimization and performance evaluation of the BDISH assay for HER2 and CEN 17 DNA targets. A protocol was established so that the entire assay procedure consisting of baking, deparaffinization, pretreatment, hybridization, stringency wash, signal detection, and counterstaining was completed as a one-step fully automated assay. Paraffin tissue sections on glass slides were baked at 65°C for 20 minutes prior to the deparaffinization step with EZ Prep™ (Ventana) at 75°C for 16 minutes. Deparaffinized tissue sections were pretreated with a combination of heat treatment with Reaction Buffer (Tris-based pH 7.6 solution, Ventana) and ISH Protease 2 or ISH Protease 3 (Ventana) to unmask DNA targets. Pretreatment conditions were chosen for optimal signal to noise ratio and tissue morphology preservation for the xenograft control slides as well as clinical case tissue slides.

Sequential ISH procedures for HER2 and CEN 17 signal detection were conducted for a complete BDISH assay (Figure 1). Reaction Buffer was used for the washing steps during immunological detection. Liquid Coverslip™ (LCS, a hydrophobic reagent, Ventana) was used for controlling liquid evaporation throughout the assay. For HER2 gene detection, the INFORM^® HER2 DNA Probe (Ventana), a dinitrophenyl (DNP)-labeled, nick-translated repeat deleted DNA probe was applied to the glass slide for co-denaturing the probe and target at 95°C. Then, the hybridization step was conducted at 52°C for 2 hours. After 3 stringency wash steps were performed at 72°C with 2× SCC (Ventana), tissue sections were incubated with monoclonal rabbit anti-DNP antibody (Ventana) for 20 minutes and then with HRP-conjugated anti-rabbit antibody for 16 minutes at 37°C. The metallic silver deposit for HER2 ISH signal was developed using silver acetate, hydroquinone, and H₂O₂ reaction in the presence of HRP using the ultra View™ SISH Detection Kit (Ventana). For CEN 17 detection, the INFORM Chromosome 17 Probe (Ventana), a DNP-labeled oligoprobe, was applied to the tissue sections, denatured at 95°C and hybridized at 44°C for 2 hours. Then, after 3 stringency wash steps at 59°C with 2× SSC, tissues were incubated with rabbit monoclonal anti-DNP antibody for 20 minutes and then with an alkaline phosphatase (AP)-conjugated anti-rabbit antibody for 12 minutes at 37°C. Finally, the signal for CEN 17 was visualized with a fast red and naphthol phosphate reaction using ultra View Red ISH Detection Kit. Diaminobenzidine (DAB) chromogen and H₂O₂ substrate reagents from the ultra View Universal DAB Detection Kit (Ventana), 5-bromo-4 chloro-3-indolyl phosphate (BCIP) substrate and nitro blue tetrazolium (NBT) oxidant reagents from the ISH i VIEW™ Blue Detection Kit (Ventana), and a ready-to-use tetramethyl benzidine (TMB) solution (Fitzgerald Industries International, Concord, Massachusetts) were also evaluated for CEN 17 signal detection of the BDISH application. Because both HER2 and CEN 17 probes were labeled with the same DNP hapten, CEN 17 signal detection was completed without DNP-labeled CEN 17 probe after HER2 signal detection to ensure that the anti-DNP antibody of CEN 17 detection didn't recognize the DNP hapten of HER2 probe.

Single or double stained tissue sections for HER2 and/or CEN 17 targets were counterstained with Hematoxylin II (Ventana) for 4 or 8 minutes and Bluing Reagent (Ventana) for 4 minutes. Counterstained slides were first rinsed with distilled water containing DAWN^® (Proctor & Gamble Company, Cincinnati, Ohio) for removing LCS from slides and then rinsed with distilled water until soap was removed completely from the slide. Slides were blotted very gently with paper towels and completely dried at 45°C or 65°C in the oven for at least 15 minutes. One drop of Cytoseal™ 60 (Richard-Allen Scientific) was applied onto a dried slide and a glass coverslip was carefully placed onto the slide. Excess mounting media was removed from the slides by gently pressing the slides against paper towels. Different coverslipping methods were also evaluated for preserving the fast red staining during the assay development. BDISH results were observed with a Nikon ECLIPSE 90i microscope (Nikon Instruments Inc., Melville, New York) equipped with Nikon digital camera DXM1200F (Nikon) without oil immersion objective lenses, up to 60×. However, for presentation purposes, mainly comparing to FISH images taken with a 100× objective lens, brightfield photographs contained in this report were obtained using a 100× oil immersion objective lens.

PathVysion HER-2 DNA Probe Kit was used for the FISH test for HER2 and CEN 17 targets of xenograft tumor controls as previously described [30]. Photographs of FISH images were taken with Zeiss Axioplan 2 microscope (Carl Zeiss MicroImaging, Inc., Thornwood, New York) with Metasystems JAIM4+ CCD1 Charge Coupling Imaging Camera (MetaSystems Group Inc., Watertown, Massachusetts) at 100× using an oil-immersion lens.

Performance of the BDISH assay was compared to FISH results as the reference standard using the historical criteria for HER2 amplification with the PathVysion assay (Negative: HER2/CEN 17 < 2.0 and Positive: HER2/CEN 17 ≥ 2.0) and using the ASCO/CAP guideline criteria (Negative: HER2/CEN 17 > 1.8, Equivocal: 1.8 ≤ HER2/CEN 17 ≤ 2.2, and Positive: HER2/CEN 17 > 2.2) with or without the equivocal cases. Scoring BDISH slides was conducted by 4 observers (MK, MD, FPL, and RRT), who were experienced with scoring HER2 FISH slides. Scoring occurred at different sites and at different occasions using different microscopes. Each individual observer evaluated the set of slides at their own pace and judgement. No scores were provided by the observers when the staining quality was deemed not adequate. There was no communication among observers regarding their scoring experience of the BDISH slides. Concordance data of FISH scores vs. consensus BDISH scores among 4 observers and FISH scores vs. individual BDISH scores by 4 observers were determined using SAS 9.1 (SAS Institute Inc, Cary, North Carolina) in calculating frequency tables and Kappa statistics. The consensus among observers was defined as the agreement of three or more observers on a given observation. Scoring of BDISH assays was also analyzed for the sensitivity and specificity against FISH scores with the historical scoring method and the ASCO/CAP scoring method without the equivocal cases. Discordant cases were investigated by a non-observer (HN) for possible causes using BDISH slides.

Images of HER2 single ISH, CEN 17 single ISH, and HER2 and CEN 17 BDISH results with formalin-fixed, paraffin-embedded xenograft tumor sections are presented in Figure 2. Single copies of HER2 signal were recognized as black discrete dots in the nuclei with MCF7 xenograft tumor (Figure 2A) while amplified HER2 gene signals were visualized as either an increased number of HER2 signals, clusters of black dots with BT-474 tumor, and/or both (Figure 2B). Single CEN 17 copies were observed as red dots that were slightly larger than the black dots for HER2 genes in the nuclei with MCF7 tumor (Figure 2C) and BT-474 tumor (Figure 2D). After single staining for HER2 gene or CEN 17 was optimized, the BDISH application with sequential detection for HER2 targets followed by CEN 17 targets was tested on xenograft tumors. Single copies of HER2 genes and CEN 17 were stained in the nuclei of MCF7 tumor cells (Figure 2E) and amplified HER2 genes and single copies of CEN 17 were visualized in the nuclei of BT-474 tumor cells (Figure 2F). Because of the size difference and color contrast of black dots for HER2 gene and red dots for CEN 17, they could be visually separated even when red and black signals were co-localized in the nuclei of MCF7 tumor cells (arrowheads, Figure 2E). When CEN 17 probe was omitted from the complete BDISH assay, there was no fast red staining on xenograft tumor sections (data not shown). Thus, the anti-DNP antibody used for CEN 17 signal detection (the second ISH detection) didn't recognize the DNP-hapten of HER2 probe signal (the first ISH detection) even though the same hapten was used for the sequential hybridization method. For image comparison of BDISH and FISH for HER2 gene and CEN 17, FISH images with MCF7 tumor and BT-474 tumor are presented in Figure 2G and Figure 2H, respectively. HER2 genes are seen as red-orange dots and CEN 17 targets are seen as green dots.

One successful way to preserve the fast red staining was the use of a toluene-based Cytoseal 60 mounting medium placed onto completely dried tissue sections prior to coverslipping with glass coverslips. We also confirmed that the red signal was successfully preserved with the Tissue-Tek^® film coverslipper method (Sakura Finetek Japan, Tokyo, Japan) after air-drying slides (data not shown). The most common method of coverslipping tissue sections stained with fast red is the use of an aqueous mounting medium. However, this method did not produce crisp fast red staining for quantitative analyses of BDISH signals (data not shown). For the second color for the BDISH application, DAB, BCIP/NBT, and TMB detection systems that produce brown, blue, and green to blue final product, respectively, were evaluated. However, they did not provide sufficient contrast against the HER2 ISH black signal (data not shown).

After optimizing the BDISH assay for HER2 gene and CEN 17 with formalin-fixed, paraffin-embedded xenograft tumor sections, we applied the assay to 94 breast carcinoma cases and scoring BDISH slides was conducted by 4 observers (MK, MD, FPL, and RRT). The consensus among observers was defined as the agreement of three or more observers on a given observation. With the historical scoring method (Negative: HER2/CEN 17 < 2 and Positive: HER2/CEN 17 ≥ 2.0) (Table 1), the consensus concordance rate was 98.9% (Simple Kappa = 0.9736, 95% CI = 0.9222 – 1.0000), the sensitivity was 96.3%, and the specificity was 100%. Individual concordance ranges were between 97.8% (Simple Kappa = 0.9466, 95% CI = 0.8736 – 1.0000) and 100% (Simple Kappa = 01.0000, 95% CI = 1.0000 – 1.0000). With the ASCO/CAP scoring method (Negative: HER2/CEN 17 > 1.8, Equivocal: 1.8 ≤ HER2/CEN 17 ≤ 2.2, and Positive: HER2/CEN 17 > 2.2) (Table 2), the consensus concordance rate was 95.7% (Simple Kappa = 0.8993%, 95% CI = 0.8068 – 0.9919). Individual concordance ranges were between 92.5% (Simple Kappa = 0.8275, 95% CI = 0.7102 – 0.9448) and 95.7% (Simple Kappa = 0.9069, 95% CI = 0.8206 – 0.9933). With the ASCO/CAP scoring method without the FISH equivocal cases (Table 3), the consensus concordance rate was 100% (Simple Kappa = 1.0000%, 95% CI = 1.0000 – 1.0000). The sensitivity was 100% and the specificity was also 100%. Individual concordance ranges were between 97.7% (Simple Kappa = 0.9442, 95% CI = 0.8678 – 1.0000) and 100% (Simple Kappa = 1.0000, 95% CI = 1.0000 – 1.0000).

Representative images of BDISH staining on clinical samples are presented in Figure 3. Cancer cells were easily identified based on the tissue morphology and assessments of HER2/CEN 17 ratios could be readily conducted. Non-amplified HER2 gene cases showed 0–4 copies of HER2 genes and 0–4 copies of CEN 17 depending on cell cycle stage and how each cell was cut within a tissue section (Figure 3A), while amplified HER2 gene cases showed multiple copies or clusters of HER2 genes and a few copies of CEN 17 (Figure 3B). Besides non-amplified and amplified HER2 cases, cases with a single copy of HER2 gene due to centromere 17 monosomy or a monoallelic deletion of HER2 gene (Figure 3C) and multiple copies of CEN 17 due to chromosome 17 polysomy (Figure 3D) were also observed.

All discordant cases, that we defined it even by one observer disagreement between the BDISH and FISH scores, were re-examined for possible causes of conflicting results and there were 9 discordant cases. All nine (9) discordant cases of the BDISH slides presented at least some degree of the genotypic heterogeneity of tumor cell populations. In general, there were two types of the tumor cell heterogeneity with HER2 gene status within the same tissue section: 1) variegated different genotype tumor cell populations in the same area of tissue section (Figure 4A) and 2) segregated tumor populations in different areas of tissue section (Figures 4B&C). Breast cancers with obvious tumor cell heterogeneity are shown as examples in Figure 4. However, the subtle genotype heterogeneity of tumor cell populations is often seen among the equivocal cases, and it also can be seen in Figures 3D &4B which show less obvious variegated tumor cell heterogeneity. Three (3) of 9 discordant cases demonstrated the segregated tumor cell heterogeneity while the other 6 cases showed various degrees of the variegated tumor cell heterogeneity.