HER2 status for all samples was determined by conventional FISH and nuclei microarray FISH. Signals for both HER2 and chromosome 17 centromere were clearly detected using the nuclei microarray method (Figure 1). Separate signals could be counted without difficulty for non-amplified, equivocal, and amplified HER2 samples. Of the 152 cases analyzed, 42 were found to be amplified both by conventional FISH and Nuclei Microarray FISH (NMFISH) (ratio of HER2/CEP17 more than 2.2), giving an amplification rate of 27.6%.

Of the 152 cases, the NMFISH method detected HER2 amplification (HER2 ratio above 2.2) in 44 cases, equivocal (ratios from 1.8 to 2.2) in 22 cases and 86 cases showed non-amplified (ratios less than 1.8), the conventional FISH method detected HER2 amplification in 42 cases, equivocal in 23 cases and 87 cases showed non-amplified HER2. The agreement between NMFISH and conventional FISH was almost perfect: 95.4% (145/152), κ coefficient = 0.920 (Table 1).

There was no significant difference between the two methods using the McNemar–Bowker Test (P = 0.333).

There were two discrepant cases that showed non-amplification by nuclei microarray FISH but equivocal amplification by conventional FISH. Three cases showed equivocal by nuclei microarray FISH but non-amplified by conventional FISH and two showed amplified by nuclei microarray FISH but equivocal amplification by conventional FISH. The HER2/CEP17 ratios of these seven cases are detailed in Table 2. HER2/CEP17 ratios detected with the two diagnostic methods were very similar and close to the cut-off value. This may have caused the differences in diagnoses.

Trastuzumab therapy has been used for the treatment of metastatic breast cancer for some years. Due to the development of adjuvant trastuzumab therapy, pathological laboratories are currently requested to perform HER2 status assays for all newly diagnosed breast cancers. The ASCO/CAP guidelines recommend the HER2 status to be optionally determined by IHC or FISH. In most laboratories, IHC is done first. In cases where the primary immunohistochemical result is equivocal at the protein level, such as an IHC score of 2+, faint staining or other artifacts, an additional test is strongly recommended to clarify the HER2 status at the genomic level to determine amplification or nonamplification. Thus, the need for reliable diagnostic and cost-effective HER2 gene tests is rapidly increasing.

Tissue microarrays, also called tissue chips, are a novel technology invented by Konenen et al. [19], based on cDNA microarrays in 1998. It is a high throughput and resource conserving technology where tens of thousands of typical minute cylindrical tissue samples or cells from hundreds of different tumors are transferred to a new paraffin block. Tissue microarrays can be used in the detection of DNA, RNA or protein in various clinical or basic research areas [20,21].

Christopher et al. [17] introduced methods to make cultured cells into microarrays. This inspired us to make nuclei microarray with nuclei extracted from the paraffin embedded tissues for the FISH detection. The present study proved this method was feasible for the detection of HER2 gene amplification in breast cancer.

During the process of making the nuclei microarray, 30-μm thick paraffin sections were cut for the extraction of nuclei instead of using the 4-μm thick sections used during the conventional FISH detection. This ensured that the nuclei were intact without being sectioned and the genetic DNA material was retained.

In the majority of the cases, the FISH method was performed successfully on nuclei microarray and almost perfect agreement with the FDA-approved HER2 FISH pharmDx method was revealed. Similar reagents and handling in the pretreatment, denaturation, hybridization and stringent washing steps were used in the two methods. One hundred samples can be detected simultaneously using this method with the same amount of reagents previously needed for one entire specimen, greatly reducing the cost.

The nuclei microarray FISH technology is advantageous for its high efficiency and low background compared to the conventional FISH using paraffin embedded tissue sections. The entire tissue section was used for the extraction of nuclei from multiple areas of the tumor. Each microarray disk contains more than five hundred cells [18], which are sufficient for the FISH detection.

During the conventional FISH detection using paraffin embedded tissue sections, variations can occur due to loss of the genetic DNA material occurring when the tissue blocks were sectioned. In the present study, we did not find this discrepancy since the ratios between HER2 signal and CEP17 were used to evaluate the HER2 gene amplification in breast cancer. The probability of the signal losses for HER2 and CEP17 due to sectioning is the same in this experiment. Thus, the nuclei microarray FISH and conventional FISH are both applicable for detecting HER2 gene amplification in breast cancer.

The present study compared nuclei microarray FISH and conventional FISH in detecting HER2 gene amplification in breast cancer. The McNemar–Bowker test revealed that there was no significant difference between the two methods (P = 0.333). κ test revealed there was almost perfect agreement between the two methods (κ coefficient = 0.920). There are seven cases with discrepant results while the ratios are close to the diagnostic cut-off value. This may be caused by the heterogeneity of the tissue specimens.

Methods for constructing nuclei microarrays (NMAs) and extraction of nuclei from paraffin-embedded tissue blocks have been described elsewhere [18,22]. Briefly, a blank cell microarray paraffin block was made in a 10 × 10 matrix using a manual arrayer (Beecher Instruments, Sun Prairie, WI, USA, MTA-1). Thirty-micrometer-thick sections were cut from the paraffin block and placed on a glass slide as a mold for holding nuclei suspensions. For each specimen, four pieces of 30-μm thick paraffin sections were cut and the tissue sections were placed into a 1.5 mL microcentrifuge tube. Nuclei were extracted after deparaffination and enzyme digestion [18]. During the process of nuclei extraction, tissue debris was separated from the nuclei and removed. This ensured a clear background in fluorescence in situ hybridization (FISH) testing. Cell density was adjusted to 1 × 10⁴ cells/μL with phosphate buffered saline (PBS, pH 7.4). Pretreated nuclei were injected into each well of the paraffin mold using a sample injector. This step is crucial to achieve the precise injection of 0.3 μL cell suspension into each well. The slides with cells were heated at 65 °C for one hour and then dewaxed in xylene twice for 40 and 20 minutes, respectively. The cell microarray was ready after open-air drying.

NMFISH analysis was evaluated using the PathVysion HER2 DNA Probe Kit (Abbott-Vysis, Downers Grove, IL, USA). The hybridization mixture included a centromere 17 specific Spectrum Green-labeled DNA probe and HER2/neu specific Spectrum Orange-labeled DNA probe.

The NMA slides were dried in a 65 °C oven for one hour, fixed with methanol-glacial acetic acid (3:1) for one hour. After air drying, slides were placed in citrate buffer (pH 6.0) and incubated for 10 minutes in a microwave oven, then transferred to freshly prepared 0.4% pepsin solution (0.16 g pepsin, 2850 U/mg solid, in 40 mL of 0.9% sodium chloride, pH 1.5) and dehydrated through a series of graded ethanol. Ten microliters of probe solution, as recommended by the manufacturer, was applied to the slide, protected with a coverslip, and sealed with a continuous bead of rubber cement. Slides were processed with a HYBrite instrument (Vysis, Downers Grove, IL, USA) programmed with a melt temperature of 82 °C for 10 minutes and hybridization temperature of 37 °C for 16 hours. After hybridization, the coverslip was removed and the slides were washed in stringent wash buffer first briefly at room temperature then in a 65 °C hot solution for 10 minutes. The slides were counterstained with diamidinophenyl-indole. The samples were analyzed under a 100× oil immersion objective using an Olympus BX-61 fluorescence microscope with appropriate filters.

FISH was performed on 4-μm sections using the same probe as NM FISH. Tissue pretreatment was performed using the paraffin pretreatment kit I (Gene Tech Company 02J02-032) as per the manufacturers’ instructions. For each case, thirty invasive tumor cells were counted by two different scorers (XB and HJ).

HER2 gene amplification status was classified according to the following criteria (ASCO/CAP) [11]. FISH signals were assessed by two independent assessors examining 30 non-overlapping nuclei for each tissue. The number of red signals (HER2) and green signals (CEP17) for each cell was recorded. Calculation of the HER2/CEP17 ratio was performed.

The results were graded as HER2/CEP17 ratio: Negative HER2 gene amplification was defined as a HER2/CEP17 ratio of less than 1.8. Equivocal HER2 gene amplification was defined as a HER2/CEP17 ratio between 1.8 and 2.2. Positive HER2 gene amplification was defined as a HER2 CEP17 ratio of more than 2.2.

Comparisons between FISH using conventional paraffin embedded tissue sections and nuclei microarray FISH were analyzed using McNemar–Bowker Test. Concordance data obtained from both the conventional FISH and nuclei microarray FISH were determined. The κ statistic method was used to measure the agreement of HER2/CEP17 ratio between the two assays. The κ statistic evaluates the level of agreement after adjustment for agreement expected to occur by chance alone with a κ coefficient >0.80 indicating near-perfect agreement, values of 0.61 to 0.80 substantial agreement, values of 0.41 to 0.60 moderate agreement, values of 0.21 to 0.40 fair agreement, values >0 to 0.20 slight agreement, and values of 0 no agreement or a random association [23]. SPSS (version 13.0; SPSS Inc.: Chicago, IL, USA, 2004) was used for all statistical analysis. A P value less than 0.05 was regarded as statistically significant.