Enhanced Accuracy and Reliability of HER-2/neu Immunohistochemical Scoring Using Digital Microscopy

Transcription

1 Anatomic Pathology / HER-2 IMMUNOHISTOCHEMICAL SCORING RELIABILITY Enhanced Accuracy and Reliability of HER-2/neu Immunohistochemical Scoring Using Digital Microscopy Kenneth Bloom, MD, 1 and Douglas Harrington, MD 2,3 Key Words: Image analysis; Breast cancer; HER-2/neu; Immunohistochemistry Abstract We evaluated the HER-2/neu status of 129 invasive breast cancer specimens for gene amplification by fluorescence in situ hybridization (FISH) and protein overexpression by immunohistochemical analysis. Each immunohistochemically stained slide was interpreted on a standard microscope independently by 10 pathologists. Separately, each pathologist reviewed the same slide set with the assistance of digital microscopy. A total of 1,258 manual immunohistochemical scores and 1,269 digital microscopy immunohistochemical scores were completed. When the same 10 pathologists scored the same immunohistochemical slides with the assistance of digital microscopy, each reviewer improved concordance with FISH, and overall concordance with immunohistochemical analysis improved significantly, to 93% (P <.001). The interrater κ was used to compare interobserver agreement in HER-2 immunohistochemical scoring for manual and digital microscopy interpretation. Significant improvement in interobserver agreement (κ = 0.51 vs 0.86; P <.001) was achieved when HER-2 immunohistochemical analysis was scored with the assistance of the digital microscope. The assistance of digital microscopy improves the accuracy and reliability of HER-2 immunohistochemical analysis. These data suggest that documented discrepancies between HER-2 immunohistochemical analysis and FISH reflect predominantly errors in manual immunohistochemical interpretation as opposed to immunohistochemical reagent limitations. The emergence of molecularly targeted therapeutics in oncology has placed increased importance on molecular diagnostic tests (pharmacodiagnostic assays) in guiding therapeutic intervention. For example, the measurement of p185 HER-2/neu expression has gained new importance based on the approval of trastuzumab (Herceptin, Genentech, South San Francisco, CA, and F. Hoffman-La Roche, Basel, Switzerland) for the treatment of patients with metastatic breast cancer. Trastuzumab is a humanized monoclonal antibody therapeutic that targets the extracellular domain of the p185 HER-2/neu protein product. 1 Treatment with trastuzumab alone or in combination with chemotherapy is indicated only for patients whose tumors overexpress p185 HER-2/neu or demonstrate HER-2/neu gene amplification based on fluorescence in situ hybridization (FISH). 2 In clinical trials, trastuzumab used alone or in combination with chemotherapy was shown to improve the clinical outcome for patients with breast cancer whose tumors overexpressed p185 HER-2/neu. 3-5 Trastuzumab plus first-line chemotherapy recently was documented 6 to reduce the relative risk of death from metastatic breast cancer by 20%, relative to chemotherapy alone in a study of 469 patients whose tumors overexpressed HER-2. In this study, the combination of trastuzumab with cytotoxic chemotherapy also improved other clinical endpoints, including response duration, time to tumor progression, and overall survival, compared with the chemotherapy-only arm. Earlier trastuzumab clinical trials screened tumors based only on immunohistochemical analysis, defining tumors with immunohistochemical staining of 2+ or 3+ as positive. More recently, the level of expression of p185 HER-2/neu was shown to affect responsiveness to trastuzumab. A higher response 620 Am J Clin Pathol 2004;121: Downloaded 620 from

2 Anatomic Pathology / ORIGINAL ARTICLE rate to trastuzumab plus chemotherapy has been demonstrated for patients with tumors scored immunohistochemically as 3+, relative to those whose tumors expressed lower amounts (2+) of p185 HER-2/neu. 2,6-8 Similarly, elevated response rates have been reported for tumors scored immunohistochemically as 3+ vs 2+ for women receiving first-line trastuzumab monotherapy 9 or trastuzumab monotherapy after disease progression following previous chemotherapy. 5 Recent studies using quantitative enzyme immunoassay 10 and radioimmunohistochemical 11 methods for measuring p185 HER-2/neu have demonstrated strong associations between the level of HER-2 protein expression and biologic features, including HER-2 gene amplification, estrogen receptor status, and prognosis. In contrast, there is considerable evidence of discordance between p185 HER-2/neu as measured by immunohistochemical analysis and FISH gene amplification in the peer-reviewed literature, particularly in tumors scored immunohistochemically as 2+. These results, along with the availability of an increasing variety of HER-2 assays, have led to considerable controversy about which assay is superior for selecting patients for trastuzumab therapy. Many technical factors can influence immunohistochemical and FISH assay results. These include the method of tissue fixation, 19,20 the specific probes or antibodies used, 21,22 the antigen retrieval method, 23 and specimen storage. 24,25 Scoring discrepancies also arise because of differences in stain interpretation by the human observer using a microscope, 20,26,27 reflecting lack of observer experience or observer bias. HER-2 scoring discrepancies between observers were highlighted recently in an article describing results of proficiency surveys performed under the auspices of the Cell Markers and Cytogenetics committees, College of American Pathologists. 28 Another report 20 suggests that more specific HER- 2 immunohistochemical results (ie, a 93% concordance of the DAKO immunohistochemical HercepTest [DAKO, Carpinteria, CA] with FISH and improved concordance with other immunohistochemical assays) can be achieved when interpretation is performed with a subtractive score that corrects for variability in background staining as evaluated in benign breast epithelium. Computerized image analysis (digital microscopy) has been shown to provide a more accurate and objective means for HER-2 immunohistochemical interpretation, relative to manual microscopy The goal of the present study was to investigate the impact of digital microscopy on interobserver HER-2 scoring reproducibility and accuracy. For this purpose, a single set of slides from 129 breast cancer cases was interpreted independently by 10 pathologists using manual microscopy. The same reviewers independently analyzed the same slides with the assistance of digital microscopy. Scoring results by both methods were compared with FISH, which was used as an index method. Materials and Methods Specimens Specimens for the present studies consisted of 100 consecutive cases of breast cancer accessioned at the Beth Israel Deaconess Medical Center (BIDMC), Boston, MA, between July 1997 and April An additional 54 cases, which had been accrued from a total of 7 community hospitals, were received from Quest Diagnostics (QD), San Juan Capistrano, CA. Specimens were included in the study only when they yielded sufficient invasive carcinoma for multiple assays that also resulted in interpretable immunohistochemical and FISH preparations. Of 100 cases evaluated from BIDMC, 88 satisfied this requirement, along with 42 of 54 cases from QD. Immunohistochemical Staining Specimens from the BIDMC were stained immunohistochemically for HER-2 protein expression at PhenoPath Laboratories, Seattle, WA. They were fixed initially in alcoholic formalin (Anatech, Battle Creek, MI), followed by postfixation in 10% neutral buffered formalin. Immunohistochemical staining for HER-2 for these specimen slides was evaluated using the same anti HER-2 polyclonal antibody provided in the US Food and Drug Administration approved kit (HercepTest) as previously detailed by Jacobs et al. 32 Specimens from QD were fixed using neutral buffered formalin. These specimens were evaluated following HER-2 immunohistochemical staining using reagents and techniques as specified in the product insert of the HercepTest kit, using an autostainer (DAKO). Ten pathologists were chosen to independently interpret the results of immunohistochemical analysis using manual microscopy and the Automated Cellular Imaging System (ACIS; ChromaVision Medical Systems, San Juan Capistrano, CA) for digital microscopy assisted analysis. The pathologists represented a range of experience from the most experienced (>50 HercepTest cases read per day) to moderately experienced (3-5 cases read per week) at reading HercepTest results manually. The manual immunohistochemical analysis was performed following rebarcoding the slides to create a blind set. This slide set was circulated individually to each pathologist who was asked to interpret the slides manually using the scoring method described on the DAKO HercepTest package insert. Downloaded from Am J Clin Pathol 2004;121:

3 Bloom and Harrington / HER-2 IMMUNOHISTOCHEMICAL SCORING RELIABILITY The ACIS digital microscopy assisted readings were done by first scanning all slides to collect complete images of each tissue sample and storing results on a data tape. Each pathologist received a tape with the images, along with an ACIS digital microscope computer review station. The pathologist selected representative regions of carcinoma for scoring the collected image. HER-2 FISH Staining and Analysis FISH for specimens from BIDMC was performed using the INFORM HER-2/neu Gene Detection System (Ventana Medical Systems, Tucson, AZ) as described in the guide accompanying the kit and detailed previously by Jacobs et al. 32 Staining and interpretation for FISH staining at QD were performed as specified in the PathVysion package insert (Vysis, Downers Grove, IL). Cases were scored as amplified by FISH when the mean number of fluorescent signals per nucleus was more than 4 for Oncor/Ventana and a FISH signal ratio of 2 or more for the Vysis PathVysion. Additional FISH studies were performed on unstained slides from total of 19 specimens. In 15 cases, the initial FISH and immunohistochemical results were discordant (n = 11) or borderline discordant (n = 4). The second FISH staining and analysis also was repeated in 4 additional cases for which no discrepancy between ACIS digital microscopy assisted immunohistochemical analysis and FISH was observed (to serve as control cases). Digital Microscopy Image Analysis Digital microscopy was performed with the ACIS digital microscope. Briefly, each immunohistochemically stained slide was digitized and the images were reviewed by the pathologist on a computer monitor. A minimum of 6 high-power fields were selected by the pathologist. The pathologists could scan the entire tissue section on low power before selecting the high-power field and were not aided in field selection. Once a field was selected, an ACISassisted score was generated by the system software. Following completion of the slide, an average ACIS score of all selected regions was generated. The software calculates a score on a 0 to 255 scale and provides a binned score that relates to conventional manual (eg, 1+, 2+, 3+) immunohistochemical scores. Details about the hardware and software available on the ACIS digital microscope system are available elsewhere. 31,33,34 Statistical Analysis Ten pathologists independently evaluated the immunohistochemical preparations by manual microscopy and ACIS digital microscopy assisted analysis in a test of observer reproducibility. Analysis of interobserver agreement was done using the κ statistic, which provides a conservative way of comparing how well reviewers agree with each other, taking into account chance agreement. 35 The P values for differences in concordance between immunohistochemical results and copy number were computed as a percentage of concordant readouts of the total number of readings. The significance of the difference between manual immunohistochemical result copy number concordance and ACIS digital microscopy assisted immunohistochemical result to copy number was computed using McNemar s procedure for paired binary outcomes. 36 Results In the present study, 130 breast cancer specimens were evaluated by 10 pathologists for HER-2 immunohistochemical analysis. The cases were analyzed independently by manual microscopy using the HercepTest scoring system devised by DAKO and by digital microscopy assisted analysis. For each specimen, a separate slide also was stained for HER-2 FISH and the gene copy determined (as described later in this section); amplified cases were classified as positive and nonamplified cases as negative. A total of 1,258 manual immunohistochemical scores were completed, along with 1,269 digital microscopy assisted scores. When manual and digital microscopy assisted immunohistochemical scores were compared, the digital microscopy assisted scores demonstrated better concordance with gene copy number overall (93.3% vs 71.1%; P =.001%) and for each of the 10 readers. In 9 (6.9%) of 130 specimens, there was a discrepancy between the average digital microscopy assisted immunohistochemical score of the 10 readers and the initial gene amplification result. For these 9 cases and for 5 additional specimens, for which a borderline result was observed, the gene amplification staining and analysis procedure was repeated. The staining and analysis procedure also was repeated for 4 additional cases for which no discrepancy between digital microscopy assisted analysis and the index method was observed (to serve as control cases). The results of the original and repeated FISH analyses are shown in Table 1. In 7 cases, the FISH results were reversed: 4 positive FISH results were classified as negative on the repeated test, and 3 negative results were subsequently classified as positive. One case was unreadable and was dropped from the study, and 2 includes immunohistochemical staining artifacts. There were 2 cases positive by digital microscopy assisted immunohistochemical analysis but negative by FISH and 2 cases just under the level of positivity by digital microscopy assisted immunohistochemical analysis but positive by confirmed FISH. In 5 cases, including 4 of 4 control cases and 1 borderline discrepant 622 Am J Clin Pathol 2004;121: Downloaded 622 from

4 Anatomic Pathology / ORIGINAL ARTICLE case, repeated FISH matched the original FISH and immunohistochemical results. All concordance statistics between immunohistochemical score and gene number used these gene copy numbers from the repeated FISH analysis. Results of immunohistochemical analysis obtained by manual microscopy and corresponding FISH data are given in Table 2. Of the 1,258 manual immunohistochemical determinations, 674 were classified as HER-2 positive ( 2+ staining); however, only 312 (46.3%) were classified as positive by gene copy number. Of the 1,269 digital microscopy assisted readings, 334 cases were positive by digital microscopy assisted immunohistochemical analysis and 51 of these (15.3%) were negative by the gene copy number. The overall agreement between manual immunohistochemical analysis and gene copy number was 71.1%. A significantly improved (P =.001%) overall concordance of 93.3% was found when digital microscopy assisted immunohistochemical analysis was compared with the gene copy number. The highest concordance for digital microscopy assisted immunohistochemical analysis and FISH was obtained when a digital microscopy immunohistochemical scoring threshold of 2.2 was used. To evaluate the interobserver agreement of the 10 reviewers for manual immunohistochemical analysis or digital microscopy assisted analysis, interrater κ analysis was performed. These calculations revealed an interrater κ of 0.51 for manual immunohistochemical analysis. For digital microscopy assisted immunohistochemical analysis, the interrater κ increased significantly (P =.0004%) to 0.86, relative to manual immunohistochemical analysis. In interpreting the interrater κ for the digital microscopy assisted immunohistochemical readings, the optimum digital microscopy assisted immunohistochemical threshold of 2.2 was used. If the same threshold as manual readings (2.0) was used, the results were essentially the same, with a κ for digital microscopy assisted readings of Figure 1 illustrates the summary results comparing the concordance of each reviewer s manual and digital microscopy assisted immunohistochemical analyses with gene copy number. By using digital microscopy assisted immunohistochemical analysis, each reviewer s immunohistochemical scores demonstrated improved agreement with the gene copy number assay, relative to the agreement between manual immunohistochemical analysis and the gene copy number. This difference was statistically significant (P <.05%) for 9 of 10 readers. In addition, whereas the agreement between the manual immunohistochemical result and gene copy number varied widely between observers (range, 41.9%-91.9% concordance), results were more consistent (range, 90.4%-95.3% concordance) when digital microscopy assisted immunohistochemical scores and the gene copy number were compared. Table 1 Original and Repeated FISH Analysis in Relation to Digital Microscopy Assisted Immunohistochemical Analysis * FISH Analysis Average ACIS-Assisted Immunohistochemical Original Repeated Specimen No. Score Result Study Positive Negative Positive Negative Positive Negative Positive Negative Negative Negative Negative Negative Negative Negative Negative Negative Positive Positive Positive Positive Negative Negative Negative Positive Negative Negative Negative Negative Negative Negative Negative Negative Negative Positive Negative Positive ACIS, Automated Cellular Imaging System (ChromaVision Medical Systems, San Juan Capistrano, CA); FISH, fluorescence in situ hybridization. * Results of original and repeated HER-2 FISH analysis on 18 breast cancer specimens, in relation to average digital microscopy assisted HER-2 immunohistochemical intensity score, following analysis by 10 pathologists. Further studies were undertaken to evaluate the possible impact of the immunohistochemical staining method on agreement with gene amplification. Of 42 cases in which DAKO HercepTest staining was performed according to the manufacturer s protocol, there was an 82.2% overall concordance of manual immunohistochemical results with gene copy number, compared with 90.2% concordance for digital microscopy assisted immunohistochemical results and the gene copy number (P <.1%). Of the 88 cases stained using the HercepTest primary antibody but visualized with an alternative chromogenic detection method, the overall concordance between manual microscopy and the gene copy number was 66.4% compared with 95.1% for digital microscopy assisted immunohistochemical analysis. Again, these data indicated significantly improved concordance (P <.001%) with the gene copy number when immunohistochemical analysis was performed with the assistance of the digital microscope. Interrater κ next was calculated to assess whether interobserver agreement varied with the immunohistochemical staining method used. For HercepTest, κ was 0.56 for manual microscopy and 0.81 for digital microscopy assisted analysis. For the alternative immunohistochemical staining procedure, which used the HercepTest primary antibody, the interrater κ ranged from 0.48 with manual immunohistochemical analysis to 0.89 for digital microscopy assisted immunohistochemical Downloaded from Am J Clin Pathol 2004;121:

5 Bloom and Harrington / HER-2 IMMUNOHISTOCHEMICAL SCORING RELIABILITY Table 2 Multiple Reviewer Comparison of Manual HER-2 Immunohistochemical Intensity Scores and Digital Microscopy Assisted Immunohistochemical Analysis With FISH for 129 Breast Cancer Specimens * Manual Score Pathologist Case No. A B C D E F G H I J NA 0 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA Am J Clin Pathol 2004;121: Downloaded 624 from

6 Anatomic Pathology / ORIGINAL ARTICLE Digital Microscopy Assisted Score Pathologist A B C D E F G H I J Mean FISH Result NA NA NA NA NA NA NA NA NA Downloaded from Am J Clin Pathol 2004;121:

7 Bloom and Harrington / HER-2 IMMUNOHISTOCHEMICAL SCORING RELIABILITY Table 2 (cont) Manual Score Pathologist Case No. A B C D E F G H I J NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA FISH, fluorescence in situ hybridization; NA, not available; +, positive;, negative. * Results are expressed in relation to results from FISH. A case was classified as positive if the manual immunohistochemical score was 2.0 or if the digital microscopy assisted immunohistochemical score was >2.2. Bold type indicates cases that were classified as positive after immunohistochemical analysis but were negative by FISH; italic type indicates cases that were classified as negative by manual or digital microscopy assisted analysis but were amplified by FISH. 626 Am J Clin Pathol 2004;121: Downloaded 626 from

8 Anatomic Pathology / ORIGINAL ARTICLE Digital Microscopy Assisted Score Pathologist A B C D E F G H I J Mean FISH Result NA NA NA NA NA NA 2.8 NA 1.9 NA NA NA 3.5 NA 5.0 NA Downloaded from Am J Clin Pathol 2004;121:

9 Bloom and Harrington / HER-2 IMMUNOHISTOCHEMICAL SCORING RELIABILITY Concordance With FISH (%) analysis. For both staining methods, these differences in interrater κ were statistically significant (P <.001%). Discussion A B C D E F G H I J Reviewer Figure 1 Multiple reviewer comparison study in 129 cases. Summary of results of 1,258 manual readings and 1,269 digital microscopy assisted readings of HER-2 immunohistochemical analysis by 10 pathologists in relation to fluorescence in situ hybridization (FISH). Gray bars, results of manual immunohistochemical scoring by each reviewer in relation to FISH; black bars, results of digital microscopy assisted immunohistochemical analysis in relation to FISH. A major source of error in manual HER-2 immunohistochemical quantification is the lack of absolute color recognition by the human eye, as opposed to the competence of the pathologist viewing the material or deficiencies in the staining protocol. The use of digital microscopy permits the observer to achieve absolute color recognition to 255 levels of intensity as opposed to 4 (0-3+) for the human eye aided only by conventional light microscopy. This study demonstrated a significantly improved concordance with HER-2 gene copy number for digital microscopy assisted HER-2 immunohistochemical analysis relative to manual immunohistochemical analysis. The highest concordance between digital microscopy assisted immunohistochemical analysis and gene copy number occurred when a digital microscopy assisted immunohistochemical threshold of 2.2 was used, but differences were statistically significant even at the level of positivity of manual immunohistochemical analysis (intensity score = 2.0). The higher concordance between digital microscopy assisted immunohistochemical analysis and FISH was independent of the immunohistochemical staining method used. In the present study, approximately 25% of breast cancer specimens were HER-2 amplified as assessed by FISH, consistent with other large breast cancer series. 37 In contrast, by manual immunohistochemical analysis, about 50% were determined to be positive. Discordance between manual immunohistochemical analysis and FISH also has been observed in other published series. For example, previous reports indicate that from 64% to 83% of cases read as 2+ by manual immunohistochemical analysis do not demonstrate gene amplification by FISH. In the present study, scores that were positive by manual immunohistochemical analysis but negative by digital microscopy assisted immunohistochemical analysis constituted 28% of the manual readings, suggesting that the use of digital microscopy assisted immunohistochemical analysis might eliminate the majority of the overcalls in immunohistochemical analysis. There were 5 cases in which the readings of the pathologists using digital microscopy were largely concordant (at least 8 of 10 agreed), but the immunohistochemical reading was discordant with gene copy number. Presumably this reflects true biologic discordance between gene amplification and protein overexpression. The remaining discordant results all were near the cutoff for immunohistochemical analysis and might reflect uncertainty of biologically borderline cases. There were 2 cases positive by FISH but negative by digital microscopy assisted immunohistochemical analysis that might reflect specimen handling problems that compromised immunohistochemical staining and could not be corrected by digital microscopy. In addition, there were 2 cases that were negative by FISH but positive by immunohistochemical analysis. These cases could represent biologic events such as frameshift mutations or protein overexpression in the absence of gene amplification. In the present study, we found a 93% concordance with FISH for digital microscopy assisted immunohistochemical readings compared with 71% for manual immunohistochemical analysis. In addition, significantly higher interrater reliability was observed for digital microscopy vs standard manual immunohistochemical analysis (κ = 0.86 vs 0.51). These results suggest that digital microscopy is capable of significantly improving the scoring accuracy and interobserver consistency of immunohistochemical analysis in HER-2/neu determination and might, therefore, more accurately identify patients who would benefit from HER- 2/neu directed therapies. The improved immunohistochemical scoring accuracy provided by digital microscopy might be of considerable value in reducing the number of cases that require FISH evaluation. However, true 2+ cases by immunohistochemical analysis remain that should be considered for FISH analysis before making an informed decision about trastuzumab therapy. 628 Am J Clin Pathol 2004;121: Downloaded 628 from

10 Anatomic Pathology / ORIGINAL ARTICLE The goals of the present study were to identify sources of error in immunohistochemical analyses and determine whether digital microscopy could mitigate the error modes. We conclude that the digital microscope not only enhances the accuracy of immunohistochemical determinations compared with other methods such as FISH but also neutralizes the experience factor in determinations of this nature, elevating all observers to approximately the same level of proficiency. Based on the results of this study, it would be appropriate to prospectively address digital microscopy assisted immunohistochemical analysis as a predictor of response in patients undergoing screening for HER-2 directed therapies. In addition, more standardized immunohistochemical interpretation would seem likely to benefit pharmacodiagnostic assays for stratifying patients for other emerging molecularly targeted therapies. From the 1 Department of Pathology, Rush-Presbyterian St Luke s Medical Center, Chicago, IL; 2 Specialty Laboratories, Santa Monica, CA; and the 3 Department of Pathology, University of Southern California, Los Angeles. Address reprint requests to Dr Harrington: Specialty Laboratories, 2211 Michigan Ave, Santa Monica, CA Acknowledgments: We are grateful to the following people for their substantial help with the present studies: Allen Gown, MD, and Hadi Yaziji, MD, PhenoPath Laboratories, Seattle, WA, who were responsible for immunohistochemical staining of specimens accrued from the BIDMC; Froilan Espinosa, MD, who was responsible for immunohistochemical staining of specimens provided by Quest Diagnostics; Stuart Schnitt, MD, and Timothy Jacobs, MD, who were responsible for FISH staining of specimens accrued from BIDMC. References 1. Carter P, Presta L, Gorman CM, et al. Humanization of an anti-p185her2 antibody for human cancer therapy. Proc Natl Acad Sci U S A. 1992;89: Trastuzumab/Herceptin [package insert]. South San Francisco, CA: Genentech; Baselga J, Tripathy D, Mendelsohn J, et al. Phase II study of weekly intravenous recombinant humanized anti-p185 HER2 monoclonal antibody in patients with HER2/neuoverexpressing metastatic breast cancer. J Clin Oncol. 1996;14: Pegram MD, Lipton A, Hayes DF, et al. Phase II study of receptor-enhanced chemosensitivity using recombinant humanized anti-p185her2/neu monoclonal antibody plus cisplatin in patients with HER2/neu-overexpressing metastatic breast cancer refractory to chemotherapy treatment. J Clin Oncol. 1998;16: Cobleigh MA, Vogel CL, Tripathy D, et al. Multinational study of the efficacy and safety of humanized anti-her2 monoclonal antibody in women who have HER2- overexpressing metastatic breast cancer that has progressed after chemotherapy for metastatic disease. J Clin Oncol. 1999;17: Slamon DJ, Leyland-Jones B, Shak S, et al. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med. 2001;344: Burstein HJ, Kuter I, Campos SM, et al. Clinical activity of trastuzumab and vinorelbine in women with HER2- overexpressing metastatic breast cancer. J Clin Oncol. 2001;19: Uber KA, Nicholson BP, Thor AD, et al. A phase II trial of weekly docetaxel (D) and Herceptin (H) as first- or secondline treatment in HER2 over-expressing metastatic breast cancer [abstract]. Proc Am Soc Clin Oncol. 2001;20:50b. 9. Vogel C, Cobleigh MA, Tripathy D, et al. First-line, singleagent Herceptin (trastuzumab) in metastatic breast cancer: a preliminary report. Eur J Cancer. 2001;37(suppl 1):S25-S Eppenberger-Castori S, Kueng W, Benz C, et al. Prognostic and predictive significance of erbb-2 breast tumor levels measured by enzyme immunoassay. J Clin Oncol. 2001;19: Cooke T, Reeves J, Lannigan A, et al. The value of human epidermal growth factor receptor-2 (HER2) as a prognostic marker. Eur J Cancer. 2001;37(suppl 1):S3-S Ridolfi RL, Jamehdor MR, Arber JM. HER-2/neu testing in breast carcinoma: a combined immunohistochemical and fluorescence in situ hybridization approach. Mod Pathol. 2000;8: Lebeau A, Deimling D, Kaltz C, et al. HER-2/neu analysis in archival tissue samples of human breast cancer: comparison of immunohistochemistry and fluorescence in situ hybridization. J Clin Oncol. 2001;2: Hoang MP, Sahin AA, Ordonez NG, et al. HER-2/neu gene amplification compared with HER-2/neu protein overexpression and interobserver reproducibility in invasive breast carcinoma. Am J Clin Pathol. 2000;113: Bankfalvi A, Simon R, Brandt B, et al. Comparative methodical analysis of ERBB2/HER-2 gene dosage, chromosomal copy number and protein overexpression in breast carcinoma tissues for diagnostic use. Histopathology. 2000;37: Tanner M, Gancberg D, DiLeo A, et al. Chromogenic in situ hybridization: a practical alternative for fluorescence in situ hybridization to detect HER-2/neu oncogene amplification in archival breast cancer samples. Am J Pathol. 2000;157: Hayes DF, Yamauchi H, Broadwater G, et al. Circulating HER-2/erbB-2/c-neu (HER-2) extracellular domain as a prognostic factor in patients with metastatic breast cancer: Cancer and Leukemia Group B Study Clin Cancer Res. 2001;7: Konigshoff M, Wilhelm J, Bohle RM, et al. HER-2/neu gene copy number quantified by real-time PCR: comparison of gene amplification, heterozygosity, and immunohistochemical status in breast cancer tissue. Clin Chem. 2003;49: Gusterson BA, Gullick WJ, Venter DJ, et al. Immunohistochemical localization of c-erbb-2 in human breast carcinomas. Mol Cell Probes. 1998;2: Jacobs TW, Gown AM, Yaziji H, et al. Specificity of HercepTest in determining HER-2/neu status of breast cancers using the United States Food and Drug Administration approved scoring system. J Clin Oncol. 1999;7: Press MF, Hung G, Godolphin W, et al. Sensitivity of HER- 2/neu antibodies in archival tissue samples: potential source of error in immunohistochemical studies of oncogene expression. Cancer Res. 1994;54: Downloaded from Am J Clin Pathol 2004;121:

11 Bloom and Harrington / HER-2 IMMUNOHISTOCHEMICAL SCORING RELIABILITY 22. Seidman AD, Fornier MN, Esteva FJ, et al. Weekly trastuzumab and paclitaxel therapy for metastatic breast cancer with analysis of efficacy by HER2 immunophenotype and gene amplification. J Clin Oncol. 2001;19: Fitzgibbons PL, Page DL, Weaver D, et al. Prognostic factors in breast cancer: College of American Pathologists Consensus Statement Arch Pathol Lab Med. 2000;124: Thor AD, Berry DA, Budman DR, et al. erbb-2, p53, and efficacy of adjuvant therapy in lymph node positive breast cancer. J Natl Cancer Inst. 1998;90: Jacobs TW, Prioleau JE, Stillman IE, et al. Loss of tumor marker immunostaining intensity on stored paraffin slides of breast cancer. J Natl Cancer Inst. 1996;88: Kay ER, Walsh CJ, Cassidy M, et al. C-erbB-2 immunostaining: problems with interpretation. J Clin Pathol. 1994;47: Tetu B, Brisson J. Prognostic significance of HER-2/neu expression in node-positive breast cancer: the influence of the pattern of immunostaining and adjuvant therapy. Cancer. 1994;73: Nagle RB, Tubbs RR, Roche PC, et al. Clinical laboratory assays for HER2/neu amplification and overexpression: quality assurance, standardization, and proficiency testing. Arch Pathol Lab Med. 2002;126: Press MF, Pike MC, Chazin VR, et al. HER-2/neu expression in node-negative breast cancer: direct tissue quantitation by computerized image analysis and association of overexpression with increased risk of recurrent disease. Cancer Res. 1993;53: Hatanaka Y, Hashizume K, Kamihara Y, et al. Quantitative immunohistochemical evaluation of HER2/neu expression with HercepTest in breast carcinoma by image analysis. Pathol Int. 2001;51: Wang S, Saboorian H, Frenkel EP, et al. Assessment of HER- 2/neu status in breast cancer: Automated Cellular Imaging System (ACIS)-assisted quantitation of immunohistochemical assay achieves high accuracy in comparison with fluorescence in situ hybridization assay as the standard. Am J Clin Pathol. 2001;116: Jacobs TW, Gown AM, Yaziji H, et al. Comparison of fluorescence in situ hybridization and immunohistochemistry for the evaluation of HER-2/neu in breast cancer. J Clin Oncol. 1999;17: Bauer KD, de la Torre-Bueno J, Diel IJ, et al. Reliable and sensitive analysis of occult bone marrow metastases using automated cellular imaging. Clin Cancer Res. 2000;6: Sun W, Zhang PL, Herrera G. p53 protein and Ki-67 overexpression in urothelial dysplasia of bladder. Appl Immunohistochem Molecul Morphol. 2002;10: Bradley RA, Hunter JS, Kendall DG, et al, eds. Statistical Methods for Rates and Proportions. 2nd ed. New York, NY: John Wiley & Sons; Harlow HF, ed. Nonparametric Statistics for the Behavioral Sciences. New York, NY: McGraw-Hill; Slamon DJ, Clark GM, Wong SG, et al. Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science. 1987;235: Am J Clin Pathol 2004;121: Downloaded 630 from