Direct Detection of Herceptin/Trastuzumab Binding on Breast Tissue Sections

Abstract

The protooncogene product HER-2/neu is the target of the humanized monoclonal antibody trastuzumab (Herceptin). Several tests are used clinically to identify patients with HER-2/neu overexpression based on evaluation by pathologists of gene amplification by fluorescence in situ hybridization or protein expression using immunohistochemistry (IHC). A simple technique has been developed for staining formalin-fixed, paraffin-embedded breast cancer tissue using unmodified Herceptin/trastuzumab as the primary antibody. Results were compared with staining with the commercial kit, HercepTest, as well as with polyclonal anti-HER-2/neu antibodies and with biotinylated trastuzumab. These procedures were tested using four breast cancer microarrays. There were 854 cores that were stained with all four antibodies, representing 325 cases. A standard 4-point scoring system (0-3) was used. A total of 156 cases (48%) were scored as 0 by all the methods used and 31 (9.5%) were positive (3+) by all methods. Of interest, three cases scored negative using polyclonal anti-HER-2/neu antibodies but were positive using unmodified trastuzumab. To clarify this discrepancy, whole sections of tumors were examined with both antibodies using double labeling. There were some tumors that demonstrated a mosaic pattern of staining with neighboring cells or groups of cells stained exclusively with one antibody or the other. These results demonstrate that unmodified humanized or human therapeutic antibodies could be used for preclinical testing or in a clinical laboratory setting for IHC-based selection of patients for treatment, and results of such selection could be different from those obtained using polyclonal antibody-based IHC procedure.

The protooncogene HER-2/neu is amplified and overexpressed in 15-30% of breast cancers (Slamon et al. 1987). The humanized monoclonal antibody targeting HER-2/neu, trastuzumab (Herceptin; Genentech, South San Francisco, cA), was approved by the US Food and Drug Administration in 1998 for treatment of metastatic breast cancer. Success of this therapy depends on tumor cells overexpressing HER-2/neu. It is important to have a sensitive and predictive assay for selecting patients because treatment is expensive (Elkin et al. 2004), and there are associated cardiovascular toxicities (Seidman et al. 2002). Currently, there are two different methods used to determine HER-2/neu status, immunohistochemistry (IHC), which detects expression of the protein on cell membranes, and fluorescence in situ hybridization (FISH), which reveals gene amplification (Tubbs et al. 2001; van de Vijver 2001; Elkin et al. 2004).

IHC is faster and more economical than FISH. However, results from IHC are subjective and semiquantitative, at best. The method, therefore, is difficult to standardize and highly dependent on use of proper controls. Different antibodies and IHC techniques have been used for detection of HER-2/neu on breast cancer tissue sections (Birner et al. 2001), and there are a number of different anti-HER-2/neu antibodies available commercially, developed against the external or internal domains of the receptor (Press et al. 2002). All IHC tests use a semiquantitative grading scale ranging from 0 (absence of HER-2/neu protein overexpression) to 3+ (maximum HER-2/neu overexpression), determined by a pathologist by inspection of conventional immunostained sections. In many studies, results of IHC are compared with FISH. The FISH method appears to be reliable, quantitative, and sensitive, but with limitations. In addition to being expensive and time-consuming, FISH also requires specialized expertise and equipment. Currently, because IHC assays have relatively low specificity, FISH is recommended as a confirmatory test for cases having 2+ in an IHC assay, regardless of the antibody or methodology used. Because the two assays measure different molecules, some differences between FISH and IHC results are expected (Elkin et al. 2004).

Several studies have demonstrated that breast cancers positive for HER-2/neu oncogene are heterogeneous with respect to the final HER-2/neu protein product expressed (Christianson et al. 1998; Kwong and Hung 1998; Codony-Servat et al. 1999). Some deletion mutations in HER-2/neu have been described (Kwong and Hung 1998). Rapid proteolytic shedding of the HER-2/neu antigen from the surface of tumor cells both in vivo and in vitro has also been reported, leading to lack of HER-2/neu in lysates of cells expressing the gene, whereas the incubation medium had high levels (Codony-Servat et al. 1999). If this occurs in patients, their tumors may be positive by IHC or gene amplification yet may not express the target epitope for trastuzumab on the cells. In a clinical trial, the independent response evaluation committee identified an objective response rate of only 15% in an intent-to-treat population with median response duration of 9.1 months (Cobleigh et al. 1999) in patients identified by the previously described methods. This low response rate may be attributed in part to an ineffectiveness of assay techniques (Elkin et al. 2004), suggesting that selection of patients for Herceptin (Genentech) treatment based on these tests is unsatisfactory.

Theoretically, a more appealing approach to identify candidates for Herceptin (Genentech) treatment would be to use trastuzumab as the primary antibody in IHC. Trastuzumab is a hybrid molecule consisting of the antigen-binding fragment (Fab) of the murine monoclonal antibody 4D5 and the Fc fragment of human IgG (Press et al. 2002). Humanization of the Fc fragment minimizes host immune response against the drug and enhances recruitment of immune cells to the binding site (Tokuda et al. 1996). However, this modification also makes it impossible to use traditional IHC amplification systems because anti-human secondary antibodies bind to endogenous antibodies present in human tissue sections. This results in high background staining making unambiguous interpretation of the results impossible. To overcome this, researchers have used biotinylated trastuzumab with the streptavidin system for revealing antibody staining (Bussolati et al. 2005). This approach demonstrated the potential of using the therapeutic antibody for identification of patients for treatment.

We have developed a novel method to perform IHC for detection of HER-2/neu antigen on formalin-fixed, paraffin-embedded human breast cancer tissues using unmodified trastuzumab as the primary antibody and a secondary antibody amplification system that can be performed on an automated platform. The results obtained with this method were compared with those obtained with commercially available polyclonal rabbit anti-HER-2/neu antibodies from Dako (Carpinteria, CA) and Zymed (South San Francisco, CA), representative of many HER-2/neu IHC assay systems in use worldwide as well as to biotinylated trastuzumab.

Materials and Methods

Tissue Samples

All tissue samples were obtained from the fixed tissue biorepository of Asterand (Detroit, MI). All samples were collected from surplus surgical material with full informed patient consent under Institutional Review Board-approved protocols for research purposes.

Four breast cancer tissue microarrays (TMA) were prepared from formalin-fixed, paraffin-embedded tissues with 0.6-mm diameter cores using a TMA arrayer (Beecher Instruments; Silver Spring, MD). TMA 60 with 342 tissue cores from 58 patients (including tumor tissue cores and adjacent normal tissue cores from the same cases); TMA 188 with 231 tissue cores from 77 cases (including tumor tissue cores and corresponding normal tissue cores); TMA 126 with 360 tissue cores (including 10 cases of normal mammary tissues and 110 cases of breast adenocarcinoma); TMA 203 with 360 tissue cores from 120 cases. All tissues on each array were represented by at least three cores that were selected after diagnosis was confirmed and representative areas were marked by a pathologist. Various histological forms of the breast cancer were represented on the arrays. Four-μm sections were cut and placed on coated glass slides using a tape-transfer system (Instrumedics Inc.; Hackensack, NJ).

Biotinylation of Trastuzumab

Biotinylation of trastuzumab was performed using the Biotintag Microbiotinylation kit from Sigma (St Louis, Mo) according to manufacturer's instructions. Briefly, trastuzumab was diluted with PBS to 100 μl, to which 10 μl BAC-Sulfo NHS at 5 mg/ml in DMSO was added. The mixture was incubated at room temperature for 30 min. Biotinylated trastuzumab was purified using a G-50 column (supplied in the kit). Four fractions at 200 μl each containing biotinylated trastuzumab were collected, and protein concentrations were determined using the Bradford assay (Bradford 1976). Protein concentration of the biotinylated trastuzumab in the first fraction was 10 μg/μl. This fraction (200 μl) was used for staining, and the concentration is equivalent to that of the unmodified trastuzumab in this study.

Pilot Studies for Optimal IHC Staining Conditions

Optimal IHC staining conditions were tested on full sections from breast HER-2/neu-positive tumor tissues for unmodified trastuzumab, anti-HER-2/neu antibody (Zymed), and biotinylated trastuzumab. The length of epitope retrieval time at 97C, 20-min and 40-min, with antigen-retrieval solution (Dako) was assessed. Antibodies were titrated at various concentrations to define a condition that gives strongest staining with minimum nonspecific background. Optimal IHC conditions for each antibody were used for IHC stainings on tissue microarray sections.

IHC Manual Staining Procedures

IHC using antibodies of HercepTest (Dako) unmodified trastuzumab, ani-HER-2/neu antibody, and biotinylated trastuzumab was performed using tissue microarray serial sections as well as the control cell line slides from SK-BR-3, MDA-175, and MDA-231 cells provided with HercepTest. All sections were deparaffinized with xylene, rehydrated by washing first with absolute ethanol, followed by 95% ethanol, and finally with distilled water.

Manufacturer's instructions were followed for HercepTest staining. Briefly, epitope retrieval was performed at 97C for 40 min with the supplied epitope retrieval solution. Slides were cooled for 20 min at room temperature in the same solution. All subsequent steps were performed at room temperature. Peroxidase was blocked using the supplied reagent. Arrays were incubated with the primary antibodies for 30 min, then with the visualization reagent for 30 min, and finally with DAB. Arrays were counterstained with hematoxylin and mounted in permanent medium.

For staining with unmodified trastuzumab, sections were incubated in antigen-retrieval solution (Dako) for 20 min. The slides were cooled for 20 min at room temperature. All subsequent treatments were at room temperature. Slides were washed three times with PBS (pH 7.6) and incubated for 10 min in peroxidase blocking system (Dako). They were washed three more times in PBS and incubated for 5 min in Super Block (ScyTek Laboratories; Logan, UT) to block general nonspecific tissue reactivity. They were washed again with PBS and incubated with Human-to-Human blocking reagent (prepared for Asterand by ScyTek Laboratories) for 60 min. They were washed with PBS before incubation for 30 min with trastuzumab (Herceptin, Genentech; a generous gift from Dr. Weizen Wei, Barbara Ann Karmanos Cancer Institute, Wayne State University, Detroit, MI) diluted 1:100 in antibody diluent (Dako). Slides were washed in PBS and incubated in 1 μg/ml of mouse biotin-conjugated anti-human IgG (BD Pharmingen; San Diego, CA) diluted in antibody diluent, followed by PBS washing and 10-min incubation in prediluted streptavidin-peroxidase from LSAB2 system (Dako) according to the manufacturer's instructions. Slides were washed again in PBS, the color was developed by 5-min incubation in DAB/H₂O₂, and then counterstained with hematoxylin.

For staining with rabbit polyclonal antibody, deparaffinization and antigen retrieval were performed as described for unmodified trastuzumab. Slides were incubated for 20 min with the ani-HER-2/neu antibody (Zymed) solution diluted 1:85 in antibody diluent, washed in PBS, and stained with the LSAB2 system (Dako) according to the manufacturer's instructions. Slides were counterstained with hematoxylin.

The same deparaffinization and antigen-retrieval procedures were also used for staining with biotinylated trastuzumab. Slides were incubated in a final dilution of 1:50 of this reagent in antibody diluent for 1 hr at room temperature, followed by incubation with streptavidin-peroxidase from the LSAB2 system (Dako) for 15 min. Color was developed by 5-min incubation in DAB/H₂O₂. Slides were counterstained with hematoxylin.

IHC Automated Procedure

Following deparaffinization and antigen retrieval as described, the slides were immersed in Tris buffered saline with Tween 20 (TBST) for 15 min. They were placed on a Dako Autostainer and were rinsed again in TBST.

Staining protocol with trastuzumab was as follows: peroxidase block for 5 min, followed by TBST wash; Superblock (ScyTek Laboratories) for 5 min, excess blown off, no rinse; Human-to-Human blocking reagent (ScyTek Laboratories) for 70 min followed by TBST washes; trastuzumab 1:100 dilution for 30 min followed by TBST wash; biotinylated anti-human IgG (1 μg/ml) for 30 min followed by TBST wash; LSAB2 streptavidin-horseradish peroxidase (HRP) labeled ready-to-use (RTU) for 15 min, TBST wash; DAB substrate and chromogen for 5 min followed by TBST wash.

Staining protocol with polyclonal anti-HER-2/neu was as follows: peroxidase block for 5 min followed by TBST wash; Super Block (ScyTek Laboratories) for 5 min, excess blown off, no rinse; anti-HER-2/neu diluted 1:85 for 45 min followed by TBST wash; LSAB2 link (RTU) for 15 min, TBST wash; LSAB2 streptavidin-HRP-labeled (RTU) for 15 min, TBST wash; and DAB substrate and chromogen for 5 min followed by TBST wash.

Double-staining protocol was the same as for trastuzumab alone except trastuzumab and the rabbit polyclonal anti-HER-2/neu were added at final dilutions of 1:100 for trastuzumab and 1:85 for the rabbit antibodies; biotinylated anti-human IgG (1 μg/ml) was mixed with alkaline phosphatase-labeled anti-rabbit IgG (Jackson ImmunoResearch Laboratories; West Grove, PA) for 30 min followed by TBST rinses; LSAB2 streptavidin-HRP labeled (RTU) for 15 min, TBST rinse; DAB substrate and chromogen for 5 min followed by wash in TBST.

Slides were removed from the autostainer, washed twice for 5 min with PBS, and incubated in BCIP/NBT substrate until a bright blue color developed (∼2 hr), followed by washes in PBS. No counterstaining was performed.

Counterstaining, dehydration, and placement of coverslips were performed manually.

Scoring

Slides were evaluated by a single pathologist (AG) according to routine procedures (Birner et al. 2001). The four-tiered scoring system (0, 1⁺, 2⁺, 3⁺) was applied. On arrays, each case was represented by three spots. The highest of the scores was attributed to the case. Samples with scores of 0 or 1 can be classified as ‘negative’ and those with scores of 2 or 3 as ‘positive’ in the manner used for clinical testing. There were 325 cases for which results were obtained using all four staining methods.

Results

All antibodies except the standardized HercepTest (Dako) were titrated on full sections of breast HER-2/neu-positive tumors to establish optimal staining conditions. Unmodified trastuzumab demonstrated strong staining at dilutions up to 1:5000, and nonspecific staining was evident at the dilution of 1:50. Biotinylated trastuzumab gave strong staining at 1:50 and 1:100 with staining disappearing at the dilution of 1:200. To maximize sensitivity, the highest concentration at which there was no nonspecific staining was selected for each antibody (1:100 for unmodified trastuzumab, 1:50 for biotinylated trastuzumab, and 1:85 for the polyclonal anti-HER-2/neu from Zymed). Biotinylated trastuzumab gave strong cytoplasmic staining, and membrane staining was not obvious in some cases because it was masked by the cytoplasmic reaction. This degree of cytoplasmic reaction was not seen with the other methods and was considered nonspecific staining. Use of higher dilutions of biotinylated trastuzumab to avoid nonspecific binding was not possible because of the loss ofstaining at dilutions of 1:200 and greater.

All methods were tested on the control cell line slides provided with HercepTest (Dako). Results demonstrated valid staining by all four methods, with strong brown (3+) membrane staining on SK-BR-3 cells (Figures 1A-1D), weak staining (1+) on MDA-175 cells (results not shown), and no staining on MDA-231 cells (results not shown). Although all methods resulted in 3+ staining of the positive cell line, there was some variation in staining intensity among the different methods. Staining intensity was highest with HercepTest (Figure 1A), followed by the Zymed polyclonal anti-HER-2/neu (Figure 1B), unmodified trastuzumab (Figure 1C), and was lowest using the biotinylated trastuzumab (Figure 1D).

To compare the staining of the four antibodies over a large number of patient samples in a high-throughput fashion, four breast cancer TMAs were used. The results for staining of 824 individual cores representing 325 cases of breast cancer are shown in Table 1 and Table 2. In the majority of cases, similar staining patterns were seen with all four methods. There was prominent membrane staining on cancer cells and no significant staining of stromal cells or neighboring normal cells (Figures 1E-1H). None of the antibody preparations gave perceptible staining with normal samples present on the arrays. When serial sections stained with the different antibodies were compared, corresponding cores showed the most intense staining (2+ and 3+) with HercepTest (Dako) (Table 1). In some cases, cores reacted with all of the antibodies except biotinylated trastuzumab (Figures 1I-1L). This is also evident by the fact that only 16.5% of the cores were scored as 2+ and 3+ with this antibody. Staining by the Zymed polyclonal anti-HER-2/neu antibody was generally weaker than HercepTest but stronger than either unmodified or biotinylated trastuzumab, although there were some cases where unmodified trastuzumab gave more intense staining (Figures 1M-1P). The most common discrepancy between methods was when cases were 2+ or 3+ with HercepTest but failed to react with unmodified trastuzumab (Figures 1Q-1T). These represented 12.1% of the cases (Table 1). Nonspecific cytoplasmic staining with biotinylated trastuzumab made interpretation difficult in some cases.

Results for cases represented in TMA experiments based upon the highest score for any of the replicates are summarized in Table 2. The majority of cases were recognized as positive or negative by all methods used. There were 156 (48%) cases that were given a score of 0 with all four assays and 31 (9.85%) cases were scored 3+ by all methods. Intermediate scores showed little concordance between assays with only one case with a score of 1 and six cases with a score of 2 in all assays. There were six cases scored as 3+ using HercepTest (Dako) that were 0 or 1+ with both methods using trastuzumab. For cases scored as 2+ with HercepTest, there were 28 cases not recognized by the trastuzumab methods. Interestingly, there were three cases that scored 2+ or 3+ with unmodified trastuzumab, which were not recognized by HercepTest (score 0) and two of these had a score of 0 with the Zymed polyclonal. The remaining case was scored as 2+ with the Zymed reagent.

It is possible that the discrepancies between results obtained with the different antibodies are due in part to the small (0.6 mm) cores on these high-throughput TMAs. Therefore, experiments were performed using whole sections of breast cancer samples to compare staining with the Zymed polyclonal antibody and unmodified trastuzumab. An autostainer was used for these sections to standardize staining. There were no significant differences in staining with either antibody between the manual method and the autostainer when serial sections were compared (data not shown). All further experiments were performed using the autostainer. A total of 32 cases were examined using full serial sections. Of these, half (16 cases) were negative by both methods. There were nine cases that were positive with both trastuzumab and the anti-HER-2/neu polyclonal antibody and seven cases positive with the anti-HER-2/neu polyclonal antibody and negative with unmodified trastuzumab. No cases in this limited set were positive for trastuzumab but not for the polyclonal antibody.

Reactivity of the two antibodies was examined in more detail using a double-labeling procedure. Studies were performed on full sections from 14 cases, some negative for both antibodies, some positive for both, and several where there was staining discordance in the TMA study. TMA-negative cases were negative when full sections were examined (data not shown). Samples reactive with both antibodies displayed membrane staining that was usually uniform, giving a mixed color as depicted in Figure 2A, and other slides displayed wide areas of mosaic staining (Figures 2B and 2C).

Figure 1

Comparison of HercepTest, polyclonal anti-HER-2/neu, unmodified trastuzumab, and biotinylated trastuzumab staining on serial sections of cell lines and breast cancer tissue microarrays with hematoxylin counterstaining. (A,E,I,M,Q,S) Stained with HercepTest. (B,F,J,N) Stained with polyclonal anti-HER-2/neu. (C,G,K,O,R,T) Stained with unmodified trastuzumab. (D,H,L,P) Stained with biotinylated trastuzumab. The positive cell line, SK-BR-3, which is supplied in the Dako kit, is shown in A-D, and had similar staining with all of the antibodies used with strongest staining by HercepTest test (A) and weakest by biotinylated trastuzumab (D). (E-H) Breast cancer case stained similarly by all methods. (I-L) A case positive by all the methods except biotinylated trastuzumab (L). (M-P) Depiction of a case reacting with HercepTest (M) and unmodified trastuzumab (O) but not with anti-HER-2/neu polyclonal (N) or biotinylated trastuzumab (P). (Q-T) Cases positive by HercepTest with 3+ (Q) or 2+ (S) reactivity but negative by unmodified trastuzumab (R,T). Bars: A-D = 50 μm; E-T = 100 μm.

Table 1

Comparison of staining of 824 breast cancer core in tissue microarrays by HercepTest, anti-Her/neu polyclonal antibodies, unmodified trastuzumab, and biotinylated trastuzumab

		Score
Method	0	1	2	3
HercepTest	580 (67.9%)	21 (2.5%)	96 (11.2%)	157 (18.4%)
Anti-Her/neu (Zymed)	626 (73.3%)	45 (5.3%)	65 (7.6%)	118 (13.8%)
Trastuzumab	619 (72.5%)	78 (9.1%)	70 (8.2%)	87 (10.2%)
Trastuzumab-biotin	616 (72.1%)	97 (11.4%)	61 (7.1%)	80 (9.4%)

Cases that were positive with the polyclonal anti-HER-2/neu but not with trastuzumab on the TMAs demonstrated exclusively blue (polyclonal anti-HER-2/neu) staining of the tumor cells when stained with both antibodies (Figure 3A). This result was confirmed on slides stained with single antibodies, which gave strong staining with the polyclonal antibody (Figure 3B) and virtually no staining with trastuzumab (Figure 3C). Cases positive on the TMAs for trastuzumab, but not the polyclonal antibody, had a mosaic pattern of staining (Figure 4A) with large areas exclusively brown (trastuzumab-reactive cells). Neighboring areas had both colors or were predominantly blue (polyclonal anti-HER-2/neu). Interestingly, as can be seen in Figure 4A and Figure 2C, there were a few areas with mosaic staining where neighboring cells or clusters of cells were exclusively blue (polyclonal anti-HER-2/neu) or brown (trastuzumab). On serial sections stained with single antibodies, those areas that were brown (trastuzumab reactive) on the double-labeled slides corresponded to areas where staining with the polyclonal antibody (Figure 4B) was weaker than that with trastuzumab (Figure 4C).

Discussion

Targeted therapies in oncology such as trastuzumab require accurate selection of patients whose tumors express the target molecule. None of the current methods used for selecting responsive patients for trastuzumab therapy is completely satisfactory. The most widely used and cost-effective method is IHC, where there are a number of different antibodies in use worldwide (van de Vijver 2001). Each antibody performs differently depending on the fixative conditions and is associated with different parameters of sensitivity and specificity (Gancberg et al. 2000). One study demonstrated that monoclonal antibody A0485 and HercepTest (Dako) were the most sensitive antibodies, whereas TAB250 was the most specific (Ginestier et al. 2004). HercepTest was less sensitive than the R60 polyclonal antibody and the monoclonal 10H8 in another study (Press et al. 2002). In our hands, HercepTest also demonstrated the highest sensitivity among the methods used. The importance of the choice of antibody used to detect HER-2/neu overexpression should not be underestimated. There is significant variability of immunohistochemical results in inter-laboratory quality control studies, and it has been suggested that more than one antibody test be routinely used (Ginestier et al. 2004).

Table 2

Comparison of staining of 325 breast cancer cases in tissue microarrays by HercepTest, anti-Her/neu polyclonal antibodies, unmodified trastuzumab, and biotinylated trastuzumab a

		Score
Method	0	1	2	3
HercepTest	204 (62.8%)	11 (3.4%)	42 (12.9%)	68 (20.9%)
Anti-Her/neu (Zymed)	221 (68.0%)	18 (5.5%)	33 (10.2%)	53 (16.3%)
Trastuzumab	216 (66.5%)	33 (10.2%)	36 (11.1%)	40 (12.3%)
Trastuzumab-biotin	216 (66.5%)	39 (12.0%)	28 (8.6%)	42 (12.9%)

^a

Highest score for each case was used from the cores described in Table 1.

Figure 2

Double staining of a breast cancer case positive for both polyclonal anti-HER-2/neu (blue) and trastuzumab (brown) with no counterstain. Cells are equally stained by both reagents (A) or in a mosaic fashion with staining by one or another predominant antibody (B,C). (C) Higher magnification of the mosaic staining. Bars: A,B = 200 μm; C = 50 μm.

Figure 3

Breast cancer case positive for polyclonal anti-HER-2/neu and not for trastuzumab. (A) Section double stained with mixture of anti-HER-2/neu (blue) and trastuzumab (brown). (B) Section of the same case stained with anti-HER-2/neu only and DAB as the chromogen with hematoxylin counterstaining. (C) Section stained with trastuzumab only and DAB chromogen with hematoxylin counterstaining. Bar = 200 μm.

One reason for relatively low specificity and clinical predictability of the IHC tests for selecting candidates for Herceptin (Genentech) therapy is that the antibodies used for diagnosis may recognize different epitopes from that recognized by the drug. False positive results reported with polyclonal anti-HER-2/neu antibodies compared with gene amplification measured by FISH are likely due to this, especially because the protein product can be modified and shed from the cell membranes, leaving only the cytoplasmic tail on the cells (Kwong and Hung 1998; Codony-Servat et al. 1999).

We have established a procedure that minimizes nonspecific staining of the humanized therapeutic antibody, trastuzumab, to examine staining of human breast cancer tumors. A similar approach was recently reported using trastuzumab labeled with biotin (Bussolati et al. 2005), which showed good correlation with clinical outcome. However, there are drawbacks to the direct labeling method for routine diagnosis: (1) antibody biotinylation requires expertise that clinical pathology laboratories may not have; (2) it is time-consuming; (3) immunoreactivity was lost at 3 months after biotinylation (Bussolati et al. 2005), affecting reproducibility of the results after minimal storage; (4) biotinylation of primary antibodies can reduce their affinity and specificity to a considerable extent (Dale et al. 1994; Hoyer-Hansen et al. 2000); (5) direct biotinylation of primary antibodies permits only limited amplification of the staining, thereby decreasing sensitivity as compared with use of secondary antibodies. In the present study, use of biotinylated trastuzumab resulted in a high level of cytoplasmic nonspecific binding. The nonspecific binding limited the amount of antibody that could be used so that 2.2% of the cases were scored negative (0 or 1+) with biotinylated trastuzumab, although they were positive (2+ or 3+) using all the other antibodies. Clearly, biotinylation of trastuzumab decreases assay sensitivity. Our results demonstrate that therapeutic antibodies without further modification can be used as diagnostic tools using a standard secondary amplification system to give high sensitivity. The method is applicable with an automated stainer so it could be adapted for use in a clinical pathology laboratory.

Figure 4

Breast cancer case predominantly positive for trastuzumab. (A) Section double stained with mixture of anti-HER-2/neu (blue) and trastuzumab (brown). (B) Section of the same case stained with anti-HER-2/neu only and DAB as the chromogen with hematoxylin counterstaining. (C) Section stained with trastuzumab only and DAB chromogen with hematoxylin counterstaining. Bar = 200 μm.

We initially tested our method on breast cancer TMAs, which enabled us to screen a large number of cases in a single experiment. TMAs were constructed with triplicate 0.6-mm cores from each case, a strategy shown to provide ∼98% concordance for detection of HER-2/neu expression compared with staining whole sections (Camp et al. 2000). To examine discrepancies in binding of the polyclonal antibodies and unmodified trastuzumab seen in the TMA study, we performed double staining with the two types of antibodies on whole sections. We confirmed that some tumors reacted only with the polyclonal antibody, which is likely due to cleavage and/or modification of trastuzumab binding site. We speculate that the patients from whom these samples were obtained would be less likely to respond to trastuzumab therapy than those that bound trastuzumab. Surprisingly, we found that in some tumors that stained with both antibodies, there were areas where some cells stained exclusively with one antibody and neighboring cells that stained exclusively with the other antibody. We also identified a few cases where there were large areas with strong staining by trastuzumab and little or no reactivity with the polyclonal antibody. Further studies are required to understand this phenomenon and its clinical relevance. Our data indicate that there is discrepancy between cases identified using the highly sensitive HercepTest (Dako) compared with direct staining with the therapeutic antibody itself. The question of which approach is better for identifying patients who will respond to Herceptin (Genentech) therapy would need to be addressed in a clinical comparison. Such a comparison should include detection using FISH, because it is considered the current ‘gold standard,’ although it also has some drawbacks.

We have used this method to examine staining of other monoclonal antibodies of human origin (data not shown), demonstrating that the approach could be applied for other human antibody-based therapeutics as well. The applications could be beneficial for the preclinical stage of antibody drug development to examine specificity and cross-reactivity of new antibodies. It also could be used for screening patients in clinical trials. A number of human or ‘humanized’ antibodies have recently entered the market such as cetuximab (Erbitux; ImClone Systems Inc., New York, NY), which targets the epidermal growth factor receptor, or bevacizumab (Avastin; Genentech), which neutralizes the vascular endothelial growth factor (Holbro et al. 2003; Caponigro et al. 2005). The use of polyclonal antibody-based IHC appears to be unsatisfactory for selecting patients for cetuximab therapy (Dei Tos and Ellis 2005). Of relevance to our findings is the observation that patients who had tumors that were negative in this diagnostic IHC test in fact responded to therapy with the therapeutic antibody. Clinical studies will be required to determine if IHC assays using therapeutic antibodies can provide better selection of responsive patients for these drugs.