In breast cancer, the assays routinely used in clinical practice are those that address a specific management decision. Hormonal therapy is based on estrogen receptor (ESR) and progesterone receptor (PGR) status. Trastuzumab (Herceptin™, Genentech, Inc, San Francisco, CA, USA) therapy, a humanized monoclonal antibody, is based on HER-2 status.

ERBB2/HER-2 (HER-2/neu, NEU, NGL, HER2, TKR1, CD340) encodes a membrane receptor protein in the growth factor receptor gene family presenting tyrosine kinase activity [1, 2]. HER-2 plays a role in the pathogenesis of a significant number of human tumors. It is altered in approximately 20–30% of breast carcinomas and this is manifested as gene amplification and/or protein overexpression [3–5]. These alterations are associated with a shorter disease free period and overall survival and with resistance to tamoxifen antiestrogen therapy and other chemotherapy regimens, regardless of the nodal or hormone receptor status [3, 4, 6, 7]. Moreover, breast carcinoma patients presenting HER-2 amplification or overexpression can benefit from anthracycline-based regimens, as well as trastuzumab [8].

The therapy choice for breast cancer patients depends on the discrimination of HER-2 status. Reliable laboratory data in evaluating HER-2 status is essential, because the treatment is beneficial for advanced breast cancer and avoids potential cardiotoxic effects in women not showing amplification and overexpression [9]. The most commonly used assay in the clinical setting for evaluating HER-2 status is immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH), both approved by the FDA (U.S. Food and Drug Administration). More recently, the CISH (chromogenic in situ hybridization) methodology, also approved by FDA, has emerged as a potential alternative to FISH. When compared with FISH, CISH has been described as having several advantages. CISH does not require an expensive fluorescence microscope with multi-band-pass filters, it produces a permanent staining and samples can be archived indefinitely, thus avoiding archival recording with an expensive CDD camera. The morphology is easier to analyze, particularly for distinguishing invasive cancer cells and in situ components. Moreover, tumor heterogeneity is promptly identified, even at low magnification (20×) [10–13]. In FISH analysis, tissue morphology and gene amplification are primarily disconnected because of tumor cells for signal evaluation are based on nuclear DAPI (4', 6-Diamidino-2-phenylindole) or propidium iodide staining, which does not always permit adequate histopathological evaluation of the cells [14]. CISH is a useful methodology for confirming ambiguous IHC results [11]. In addition, polymerase chain reaction (PCR) based technology has been demonstrated to successfully evaluate specific mRNAs, especially those present in low copy numbers in a small number of cells or in small quantities of tissue, and mRNAs expressed in mixed-cell populations. Quantitative real time reverse transcript PCR (qRT-PCR) is a quantitative method easily amenable to standardization. However, qRT-PCR suffers from the same drawback as other PCR-based methods. The tumor cell population within the tissue under evaluation must be isolated, the template quality, operator variability, subjectivity in data analysis and reporting are technical aspects that must be considered [14, 15]. This procedure is an alternative for scoring HER-2 status in human breast cancer. Limitations based on tumor heterogeneity and amplification of HER-2 in noninvasive cancer can be eliminated by the use of laser microdissection, although this seems to be impracticable for routine diagnosis [14].

The purpose of the current study was to assess agreement between gene amplification detection by CISH and transcript (qRT-PCR) and protein (IHC) expression, as well as to evaluate their relevance for determining HER-2 status in breast carcinomas. In addition, the data were correlated with clinicopathological features, such as tumor size, lymph node status, histological grade and Ki-67 status.

Breast cancer samples were assessed by HER-2 gene amplification and protein expression in histological samples using CISH and IHC, respectively (Figure 1). Transcript expression was evaluated by qRT-PCR in fresh samples after microdissection. HER-2 data obtained in all the procedures are shown in Tables 2 and 3.

CISH was performed on 43 cytological samples; in three of these cases the spots presented inconclusive results; the brown spots were undistinguished. Twenty-four out of 37 tumors (64.9%) were not amplified by CISH, while 13 samples (35.1%) showed high-level HER-2 amplification. Different areas of the tumor were evaluated and the final analysis showed agreement, except in three cases (Figure 2A–C). Case A presented an equivalent number of nonamplified cells (2 copies and 3–5 copies) in area 1 and more than 50% of the cells with high-level amplification in area 2 (this case presented 0 score by IHC and normal expression level by qRT-PCR). Case B showed nonamplified cells in both areas, but in area 2, low-level and high-level amplification cells were also observed (IHC, 0 score; qRT-PCR, overexpression). Case C showed preferentially nonamplified cells; however, in area 1, cells with 3–5 copies were predominant and in area 2, a prevalence of two HER-2 copies was detected and the presence of sporadic cells with high-level amplification was also found (IHC, 3+ score; qRT-PCR, overexpression). More than 400 cells were evaluated for each of these cases. In the other samples (37 cases), the nuclear features were maintained, the morphological details were apparent and large gene copy clusters were easily detected at low magnification. Normal epithelial cells and lymphocytes showed one or two HER-2 signals per nucleus.

HER-2 amplification status by FISH was evaluated in 11 out of 43 cases investigated by CISH. The comparison between CISH and FISH revealed agreement in eight cases that presented high-level HER-2 amplification. These cases presented scores 2+ (two cases) and 3+ (six cases) by IHC. Only one of them, identified as amplified by CISH, showed sporadic cells (<10% of the cells) with polyploidy. The three cases that presented heterogeneity in two different areas of the tumor by CISH were also evaluated by FISH (Figure 2A–C). Case A presented similar results to that observed in the area 1 detected by CISH; amplification was observed in only 10% of the cells. Cases B and C showed principally no amplification by FISH, comparable to that detected by CISH (case B: 90% of the tumoral cells were not amplified and 10% presented polyploidy; case C: no amplification in 90% of the cells and amplification in 10%). An overall concordance between CISH and FISH results was found.

HER-2 relative expression level was evaluated in 75 samples by qRT-PCR in comparison with four healthy breast tissue and ranged between RQ = 0.05 to 20.43. Overexpression (RQ>2.00) was observed in 29 out of 75 cases (38.7%).

Among the 75 cases analyzed by IHC, 50 (66.7%) presented a score of 0 or 1+. The 2+ immunostaining cases comprised 13 samples (17.3%) and 12 cases (16.0%) presented 3+ scoring.

qRT-PCR compared to IHC results

HER-2 transcript levels were significantly lower in cases presenting low protein expression (0 or 1+) than in cases presenting high expression (3+) (Figure 3A). Ten out of 50 cases (20%) presenting a score of 0 or 1+ showed overexpression by qRT-PCR. Seven out of 12 samples (58.3%) comprising 2+ immunostaining showed overexpression by qRT-PCR. Twelve out of 13 cases (92.3%) which presented a score of 3+ overexpressed HER-2 transcript (Table 3). Concordance between qRT-PCR and IHC was observed in 59 cases (78.7%).

Clinico-pathological data in comparison with HER-2 status

HER-2 data were also compared to clinicopathological features (Table 1). No statistical correlation was observed with age, tumor size, clinical stage, histological grade, Ki-67 status or familial history of cancer. A marginally significant correlation with lymph node status was detected: 82.6% of the cases presenting RQ ≤ 2.00 showed less than four positive lymph nodes (P = 0.0915). Comparison of individual HER-2 relative quantification values between the two classes of lymph node status (<4 and ≥ 4) showed a significant correlation (P = 0.0350), confirming the association between cases presenting RQ ≤ 2.00 and the involvement of less than four positive lymph nodes (Figure 4).

In the four lobular carcinomas evaluated by all three methodologies (CISH, qRT-PCR, and IHC), two presented discordant results: one case showed high-level amplification by CISH, HER-2 overexpression by qRT-PCR and negative immunostaining; the other sample presented nonamplified by CISH, RQ ≤ 2.00 by qRT-PCR and 2+ immunostaining by IHC analysis. The two remaining cases were nonamplified, RQ ≤ 2.00 and negative immunostaining. The case evaluated exclusively by qRT-PCR and IHC presented RQ ≤ 2.00 and negative immunostaining. Individual values of HER-2 expression by qRT-PCR were compared between ductal and lobular carcinomas and no significant association was observed (median RQ = 1.34 ± 3.83 and RQ = 1.7 ± 1.02, respectively; P = 0.7375, data not shown).

HER-2 gene has been extensively studied as a prognostic and predictive marker in clinical breast cancer, making this receptor a valuable target for the treatment of human breast cancer [22]. HER-2 status is predominantly evaluated by IHC staining, because it is easy to perform and presents a relatively low cost. However, a wide range of sensitivity and specificity was observed among various commercially available antibodies [23]. In addition, FISH is used for those cases that are scored as 2+. The advantage of FISH testing is that the quantitative interpretation of results with experience is relatively straightforward and concordance rates among observers are higher than with IHC in some studies [for review, [19]]. CISH is an alternative method to evaluate amplifications that requires a conventional light microscopy, permits a more rapid interpretation time and a review of the morphological details. A further advantage of CISH is that the probe signals are permanent and the slides can therefore be archived for long periods of time [24]. Data do not clearly demonstrate the superiority of either IHC or in situ hybridization (ISH) as a predictor of beneficial effects from anti-HER-2 therapy after validated testing has been carefully performed [25]. Quantitative real time PCR has the potential to become standard in terms of its performance, accuracy, sensitivity, broad dynamic range and high throughput capacity [26, 27].

Concordance between FISH and CISH was 100% for the eight cases analyzed. This high concordance was also found by several other studies [11, 14, 28]. One of the eight cases displayed a small frequency of cells showing chromosome 17 polysomy by FISH and showed a score 2+ by IHC. Peiró et al. [29] showed that all of the polyploidy tumors analyzed presented 2+ immunostaining. Although a limited number of cases were evaluated by FISH in the present study, the data are in agreement with other studies which indicate that the chromogenic ISH technique seems to be sensitive and specific for the detection of HER-2 amplification in human archival tumor samples [14, 28].

CISH analysis revealed intratumoral heterogeneity in three cases. In case A, 50% of cells showed high-level amplification and nonamplified cells. In addition, this case showed discordant data between IHC and qRT-PCR methodologies, probably due to qRT-PCR false-positive results. The major question is whether cells showing different levels of amplification make any difference or whether a threshold (or its value) percentage of amplified tumor cells is required to define nonamplified and amplified tumors. Regardless of the CISH results, these three patients were treated with tamoxifen and radiotherapy (patients A and B) and radiotherapy and CMF (patient C), according to the IHC results. The outcome was favorable for more than 45 months. Further studies should be performed to clarify the tumoral heterogeneity involving HER-2 amplifications in breast cancer. These three cases were evaluated by FISH and similar results were observed between CISH and FISH analysis for all the cases. When using FISH, it was not possible to determine the two areas observed by CISH, probably due to the restricted number of cells evaluated. Case A presented a higher frequency of amplified cells by CISH in its area 2 than that observed by FISH. Specifically in this case, the paraffin sections used for CISH and FISH methods were not successive and, most likely, different patterns of tumor heterogeneity were evaluated by both methodologies.

Comparison between CISH and IHC results revealed six discordant cases (16%). These same cases showed concordance when comparing CISH and qRT-PCR data. Many studies have shown high concordance between IHC and CISH; frequently, the discordant cases were 2+ immunostained [27, 30–32]. In the present study, all of the samples scored as 3+ by IHC presented HER-2 amplification by CISH. In fact, >90% of HER-2 IHC 3+ tumors present HER-2 gene amplification [19, 33]. The concordance between both methodologies was similar to that observed in other studies, varying from 85% to 95.3% [24, 27, 34]. Differences exist in tissue screening between these two techniques. While the IHC test requires that a minimum of 10% of tumor cells are reactive to be considered positive, CISH scoring requires that more than 50% of tumor cells show an increase in gene copies to be considered amplified; according the criteria used in the present study. It is conceivable that this difference could account for some of the discordant results observed between IHC and CISH. In the present study, eight cases presenting 2+ and 3+ immunostaining showed HER-2 amplification by FISH. The two cases detected as 2+ by IHC and amplified by FISH also presented high-level amplification using CISH. Cases scored as 2+ and amplified by FISH were also observed in other studies [10, 35, 36].

A significant correlation was detected between gene and protein expression levels. These data are in agreement with several reports that showed good correlation between transcript and protein data [37–40]. However, ten samples scored as 0 or 1+ by IHC showed HER-2 overexpression by qRT-PCR. The transcript quantitative analysis revealed two classes of HER-2 overexpression: five cases showed a median RQ = 4.09 ± 4.33 and four of these showed gene amplification by CISH; and five other cases presented a median RQ = 2.19 ± 0.21, a value very close to the cut-off value selected (2.00). Tse et al. [39] used a cut-off of RQ>2.2 as positive for HER-2 overexpression by qRT-PCR when it was compared to Elisa, IHC, and FISH on sections obtained from paraffin-embedded breast carcinomas.

Currently, the most widely used assay to evaluate gene status in cases scored as 2+ immunostaining is FISH. However, qRT-PCR has emerged as a potential alternative technique for assessing HER-2 status. The present results demonstrated that 58.3% of the samples presenting a score of 2+ overexpressed the transcript. Among the current samples, four out of five patients that presented discordant results (2+ immunostaining and RQ ≤ 2.00) showed a favorable outcome, indicating that HER-2 status by qRT-PCR could be performed on 2+ staining tumors with potential value regarding the management of these patients. Of the patients presenting 2+ staining and RQ>2.00 (eight cases), three presented metastasis, one revealing spreading to the bone and two to the lung.

A higher concordance rate was observed between 3+ score by IHC and overexpression by qRT-PCR. Only one sample that presented 3+ status and HER-2 amplification showed transcript downexpression. This case, showing a qRT-PCR false-negative result, probably resulted from the dilution of cells carrying amplified genes among nontumor cells [41], although the sample was submitted to microdissection.

Concordance between the IHC and qRT-PCR results was 78.9%. The discordance between these methodologies could include differences in the specimen used in the experiments (paraffin-embedded and fresh tumor tissue, respectively). Two other possible causes for discrepancy exist: intraobserver error, due to subjectivity of the IHC interpretation; and qRT-PCR analysis, which can cause discrepancies particularly in the initial cycles, which depend not only on the melting temperature of the amplicon, but also on the behavior of the genomic vicinity of the amplicon [42, 43]. Using PCR-based methods, the expression of tumor- or tissue-specific genes and the presence of genetic abnormalities can be detected in a clinical specimen with higher sensitivity (one malignant cell out of 10⁶–10⁷ normal cells) than that of other techniques such as light microscopy (one malignant cell out of 10²–10³ normal cells). Using RT-PCR the nucleic acid molecules can be amplified 10¹⁰-fold [for review, [44]]. Moreover, HER-2 overexpression can be detected in 0.1 cell equivalent spiked into 8 mL of peripheral blood using qRT-PCR and the detection limited increases to 10 and 50 cell equivalent per 8 mL in cell lines expressing intermediate and low levels of HER-2 [45].

Comparison between qRT-PCR data and CISH results showed that the gene expression median was correlated with gene copy number, a finding also observed by Bergqvist et al. [46]. HER-2 gene amplification is the most prevalent genetic mechanism driving HER-2 overexpression. The discordant cases (8.1%) showed 2+ and 3+ immunostaining, confirming the CISH results. In this study, the concordance between qRT-PCR and FISH was 100%. High concordances were observed by other studies [23, 47].

In the present study, a high correlation rate among the three procedures used to score HER-2 status in breast carcinomas was observed. The correlation between CISH and qRT-PCR was higher than CISH and IHC, which was higher than qRT-PCR and IHC for the samples evaluated by all these procedures. In fact, CISH and qRT-PCR are complementary methodologies for evaluating HER-2 status.

No correlation was found between HER-2 expression by IHC or gene copy number and clinicopathological data. However, a significant association was observed between lymph node status and HER-2 transcript expression by qRT-PCR. Peiró et al. [48] analyzed HER-2 status by IHC and CISH and observed a correlation with histological grade and lymph-vascular invasion, but no association was found with age, tumor size and Ki-67 status. Similarly, the absence of correlation between HER-2 status and clinical and pathological features has been reported in other studies [32, 43].

No statistically significant difference was observed between ductal and lobular carcinomas evaluated by qRT-PCR. Among the lobular carcinomas, amplification by CISH was observed in one case, followed by HER-2 overexpression and negative IHC staining. In a previous study, this case was confirmed as negative for e-cadherin protein expression and presented CDH1 promoter hypermethylation [49]. Li-Ning-T et al. [35] evaluated five invasive lobular tumors and none showed amplification by CISH, however two cases presented 2+ immunostaining.

In conclusion, the present results suggest that HER-2 status can be performed by CISH and qRT-PCR analysis. CISH combines the advantages of IHC and FISH and is a promising practical alternative to FISH, while qRT-PCR is reliable, semiautomated and fast; and both methodologies can be performed in most pathology laboratories. However, there are several limitations to apply qRT-PCR as a routine method for clinical application, including the use of fresh frozen tissue, microdissection procedures and the incapacity to address cell-to-cell variations. In addition, the data confirm that HER-2 overexpression is associated with a worst prognostic in human breast tumors.