HER2/neu Testing in Gastric Cancer by Immunohistochemistry Assessment of Interlaboratory Variation Brandon S. Sheffield, MD; John Garratt, RT; Steve E. Kalloger, MSc; Hector H. Li-Chang, MD; Emina E. Torlakovic, MD; C. Blake Gilks, MD; David F. Schaeffer, MD
Context.—Immunohistochemical (IHC) testing for HER2/neu is becoming the standard of care for guiding adjuvant treatment of gastric carcinoma with trastuzumab. Objective.—To assess interlaboratory variation in IHC staining and interpretation across multiple laboratories. Design.—A tissue microarray consisting of 45 cores from 28 gastric cancers was distributed to 37 laboratories for HER2/neu assessment. The IHC results were compared against expert scores at an academic institution and correlated with in situ hybridization results from the originating specimen. Interlaboratory agreement was calculated using Cohen j statistic. Results.—The survey demonstrated several variations in IHC methods, including the primary antibodies in use. There was excellent agreement among laboratories in HER2/neuþ (IHC 3þ) cases (j ¼ 0.80 6 0.01) and very good agreement among laboratories in HER2/neu (IHC 0 or 1þ) cases (j ¼ 0.58 6 0.01). Less agreement was observed
among laboratories when scoring equivocal (IHC 2þ) cases (j ¼ 0.22 6 0.01). Sensitivity and specificity of HER2/neu IHC were 99% and 100%, respectively, when measured against expert review and consensus score as a reference standard. Conclusions.—There is substantial interlaboratory agreement in the interpretation of HER2/neu IHC despite variability in protocols. Although HER2/neu IHC is a highly sensitive and specific test, primary antibody selection may significantly affect IHC results. Furthermore, gastric tumors require a unique scoring system and expertise in interpretation. Intratumoral heterogeneity has a significant effect on HER2/neu scoring by IHC. Ongoing quality assurance exercises among laboratories will help ensure optimized HER2/neu testing. (Arch Pathol Lab Med. 2014;138:1495–1502; doi: 10.5858/arpa.2013-0604-OA)
G
Human epidermal growth factor receptor 2 (coded by the ERBB2 gene) gain of function is associated with increased cell motility, invasiveness, angiogenesis, resistance to apoptosis, and metastatic potential.4 Overexpression of HER2/neu in gastric cancers occurs most commonly as an amplification of the 17q12–q21 region and is detectable by in situ hybridization (ISH) techniques. This event occurs in 15% to 25% of gastric carcinomas; a meta-analysis showed an overall prevalence of 19%.5 Although HER2/neu amplification is associated with decreased overall survival in gastric cancer,6 the multinational Trastuzumab for Gastric Cancer (TOGA) clinical trial demonstrated increased progression-free and overall survival in HER2/neu-overexpressing gastric cancers when trastuzumab was added to conventional treatment.7 Trastuzumab is a humanized recombinant monoclonal antibody targeting the HER2/neu molecule. This drug is routinely used in the treatment of HER2/neu-amplified breast cancers,5 with HER2/neu testing and trastuzumab treatment also now becoming the standard of care for metastatic gastric cancer. Because only a subset of patients will benefit from this targeted therapy, there is a demand on medical laboratories to offer accurate and robust testing of HER2/neu status. Slide-based immunohistochemical (IHC) techniques have been validated and show high concordance with ISH studies
astric carcinoma contributes to 10% of cancer deaths worldwide with a case to fatality ratio of 70%.1 Despite recent advances in cancer treatment, the overall 5-year survival for patients with newly diagnosed gastric cancer is 28%.2 The treatment of gastric cancer is primarily surgical because the addition of perioperative chemotherapy has been associated with only marginal clinical benefits in overall and progression-free survival.3 Accepted for publication January 8, 2014. From the Division of Anatomic Pathology, Department of Pathology and Laboratory Medicine, Vancouver General Hospital and the University of British Columbia, Vancouver, Canada (Drs Sheffield, Gilks, and Schaeffer and Mr Garratt); the Canadian Immunohistochemistry Quality Control Program, Vancouver (Mr Garratt and Drs Torlakovic and Gilks); the Department of Pathology and Laboratory Medicine, British Columbia Cancer Agency and the University of British Columbia, Vancouver (Dr Li-Chang and Mr Kalloger); and the Department of Laboratory Hematology, University Health Network, University of Toronto, Toronto, Ontario, Canada (Dr Torlakovic). The authors have no relevant financial interest in the products or companies described in this article. Reprints: Brandon S. Sheffield, MD, Division of Anatomic Pathology, Department of Pathology and Laboratory Medicine, Vancouver General Hospital, 855 12 Ave W, Vancouver, BC V5Z 1M9, Canada (e-mail: [email protected]). Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1495
Table 1. Demographic Data of the Cases (n ¼ 28) Selected for Tissue Microarray Construction Characteristics
Cases, No. (%)
Sex Male Female
21 (79) 6 (21)
Age (at diagnosis) Range, y Mean, y Median, y
42–89 64.3 65
Gastric carcinoma subtype Intestinal Diffuse Mixed
20 (71) 3 (11) 5 (18)
Histologic grade Well differentiated Moderately differentiated Poorly differentiated
1 (4) 9 (32) 18 (64)
Site Proximal Distal Diffuse
17 (61) 10 (36) 1 (4)
Lymph node metastases Yes No
18 (64) 10 (36)
Stage IA IB IIA IIB IIIA IIIB IV
MATERIALS AND METHODS Appropriate research and ethics permissions were obtained from the University of British Columbia (Vancouver, Canada) and the participating health care centers.
Case Selection Twenty-eight gastric carcinoma surgical resection specimens from 2004 to 2011 were identified retrospectively from the archives of Vancouver General Hospital (Vancouver, British Columbia, Canada). These cases were randomly selected from an archival database and represented a range of gastric carcinoma subtypes (Table 1). Inclusion criteria were gastric adenocarcinoma undergoing surgical resection and available ISH data (available from previous research and validation studies), as well as availability of formalin-fixed, paraffin-embedded (FFPE) blocks for tissue microarray (TMA) construction. No exclusion criteria were used. Cases included HER2/neu-amplified and -nonamplified tumors as well as tumors showing a range of HER2/neu IHC staining intensity from 0 to 3þ.
TMA Construction and Interpretation
2 6 8 3 4 2 3
(7) (21) (29) (11) (14) (7) (11)
in the assessment of HER2/neu amplification and overexpression.8 Potential challenges in HER2/neu assessment of gastric tumors by IHC include novel criteria for assessment, which are different from those used in breast cancer.9 Intratumoral heterogeneity of HER2/neu amplification affects a significant proportion of gastric tumors10 and may affect IHC evaluation. In addition, many different antiHER2/neu primary antibodies are commercially available for conventional IHC testing, and two widely used antibodies (4B5 and SP3) have been shown to produce different staining patterns.11 Considering these variables to ensure accurate HER2/neu IHC testing is a vital process because test results have a direct influence on treatment choice and patient outcome. As HER2/neu IHC becomes widely implemented in the management of gastric cancer, so too, must appropriate quality control measures be implemented. Previous studies have shown significant interlaboratory variation in HER2/ neu IHC test results.12,13 One group has shown that a combination of IHC and ISH testing is reproducible among a few laboratories working collaboratively with a central reference laboratory.14 To date in North America, to our knowledge, no measures of interlaboratory variation in gastric cancer HER2/neu IHC test results have been reported in the literature. This study aimed to measure interlaboratory variation in HER2/neu testing by IHC. The methods were designed to capture significant differences in analytic and postanalytic aspects of IHC determination of HER2/neu status. 1496 Arch Pathol Lab Med—Vol 138, November 2014
A TMA was constructed using duplicate 0.6-mm cores, when available, derived from FFPE blocks corresponding to each of the study cases. All tissue was handled and processed at the anatomic pathology laboratory at Vancouver General Hospital according to the standard of care at our institution (cold ischemic time, approximately 0–4 hours; formalin fixation time in 10% neutralbuffered formalin, 24–48 hours). Resection specimens were used rather than biopsies to yield optimal TMA results. Tissue cores for the TMA were obtained from areas of tumor determined by routine microscopy on hematoxylin-eosin–stained sections. Unstained TMA slides were distributed to laboratories participating in a quality control exercise on a voluntary basis within the well-established Canadian Immunohistochemistry Quality Control Program (Vancouver, British Columbia, Canada).15 Laboratories were asked to provide information about their staining protocols, including primary antibody clone, dilution, and pretreatment, in addition to scoring each core on the TMA. No instruction or guideline was given about how to score the cases. All data were gathered electronically via online submission.
Expert Review Stained slides were returned to Vancouver General Hospital for review and analysis by 2 pathologists with expertise in IHC quality assurance and gastrointestinal pathology (C.B.G. and D.F.S., respectively), who achieved a consensus interpretation of these cases. Expert analysis of HER2/neu IHC staining intensity was made according to the system used in the TOGA trial.9 Briefly, 0 was defined as no reactivity or membranous reactivity in ,5 cells, 1þ staining was defined as faint/barely perceptible membranous reactivity in 5 cells, 2þ staining as weak to moderate complete or basolateral membranous reactivity in 5 cells, and 3þ staining as moderate to strong complete or basolateral membranous reactivity in 5 cells (TMA cores were scored as biopsies rather than resection specimens, see examples in Figure 1, A through D). A single set of observers, who were well versed in the abovementioned scoring system, analyzed stained slides returned from each of the 37 laboratories, establishing a consensus IHC score based on individual scores for each tissue core. False-positive and false-negative cases were identified as those where individual IHC scores differed from consensus score.
Statistical Analysis Individual, laboratory-scored TMAs were pooled, and a Cohen j statistic was calculated as a measure of interlaboratory agreement. Individual cores were considered to be negative (0 or 1þ IHC score), equivocal (2þ IHC score), or positive (3þ IHC score). Test characteristics were calculated using expert review and consensus interpretation of IHC staining as a gold standard. Scores Interlaboratory Variation in HER2/neu Testing—Sheffield et al
Figure 1. Example of 0 to 3þ scoring. a, 0, no immunoreactivity shown by laboratory 145 in core 25. b, 1þ, faint, barely perceptible membrane staining by laboratory 110 in core 14. c, 2þ, weak to moderate membranous staining by laboratory 152 in core 2. d, 3þ, strong membranous staining by laboratory 103 in core 8 (original magnification, 310 [a through d]).
of 0 and 1þ were considered negative results, whereas 3þ corresponded to a positive IHC result. When a participant core was assessed as 2þ (equivocal), it was, by default, determined to be concordant with the reference value, regardless of the actual reference value, because those patients would be referred for HER2/neu in situ hybridization and would be appropriately stratified for therapy based on those results.
IHC Staining Cases with discordance between consensus IHC staining and ISH data were identified. The source FFPE blocks were used to perform HER2/neu IHC testing on whole slides. The HER2/neu IHC tests were performed using the SP3 primary antibody (Thermo Scientific, Marietta, Ohio) at 1:100 dilution with the Optiview (Ventana Medical Systems, Oro Valley, Arizona) visualization system.
In Situ Hybridization All ISH data were collected from a previously established research database. The original assays were performed on whole Arch Pathol Lab Med—Vol 138, November 2014
slides derived from FFPE blocks using either fluorescence in situ hybridization (PathVysion HER2/neu kit, Abbott molecular, Des Plaines, Illinois) or silver in situ hybridization (HER2/neu and CEP17 probes, Ventana). In determining the HER2/neu status by ISH, the following cutoffs were used: (1) amplified, if the HER2/ neu:CEP17 ratio was greater than 2.2 or the HER2/neu was greater than 6 copies/nucleus; (2) equivocal, if the HER2/neu:CEP17 ratio was between 1.8 and 2.2; and (3) negative, if the HER2/neu:CEP17 ratio was less than 1.8. All cases were assessed by expert pathologists or cytogeneticists. Fluorescence in situ hybridization and silver in situ hybridization results were considered equivalents.
RESULTS Thirty-seven laboratories participated in the quality assurance exercise. Each of the 10 Canadian provinces was represented as well as 4 laboratories in the United States and 2 laboratories in Europe. Community, academic, and private laboratories participated in the exercise. Participating laboratories used a range of pretreatments, dilutions, and Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1497
Figure 2. Immunohistochemical (IHC) scores assigned to tissue microarray by participating laboratories. Individual cores are labeled 1 to 45. Individual laboratories are deidentified with a 3 digit code (101–199 nonconsecutively). Tissue microarray scoring was recorded as U, core unsatisfactory for scoring (white); 0, an IHC score of 0 (white); 1, an IHC score of 1þ (gray); 2, an IHC score of 2þ (amber); and 3, an IHC score of 3þ (red).
primary antibodies including: 4B5, SP3, A0485, and the HercepTest (Dako, Carpinteria, California). Details are provided in Table 2. The IHC scores, as reported by individual laboratories, are shown in Figure 2. Results of expert review and consensus interpretation are presented together with ISH data in Figure 3. Analysis of interlaboratory reproducibility showed excellent agreement among positive (IHC 3þ) cases (j ¼ 0.80 6 0.01), and very good agreement among negative (IHC 0 or 1þ) cases (j ¼ 0.58 6 0.01). Less agreement was observed among laboratories scoring equivocal (IHC 2þ) cases (j ¼ 0.22 6 0.01). Six of 45 cores represented on the TMA (cores 7, 8, 19, 20, 31, and 32) were identified on expert review as being positive by IHC (Figure 3). Those cores were identified by all but one of the participating laboratories (36 of 37) as being IHC 3þ or 2þ. A single laboratory (no. 161; 1 of 37) identified only 4 of those 6 cores (Figure 4, A and B). Upon expert review, that error was confirmed as a false-negative IHC 1498 Arch Pathol Lab Med—Vol 138, November 2014
test. That laboratory was using the HercepTest primary antibody. A single laboratory (no. 164; 1 of 37) showed falsepositive staining in 2 cores (cores 5 and 15), and upon expert review, it was confirmed that laboratory had incorrectly interpreted 2þ and 1þ staining patterns as 3þ in both instances (Figure 5, A and B). Overall the sensitivity and specificity of HER2/neu IHC were calculated as 99.1% (95% confidence interval, 98.1%–99.6%) and 99.8% (95% CI, 99.6%–100%), respectively. Four of 45 cores (cores 6, 12, 38, 39) showed HER2/neu amplification on ISH, but were, however, scored negative on IHC (0 or 1þ). Expert review confirmed that none of the TMA sections showed positive immunoreactivity for HER2/ neu in those 4 cores (Figures 2 and 3). Whole-slide sections stained for HER2/neu showed marked intratumoral heterogeneity in each of those 4 cases. Each case demonstrated regions of strong HER2/neu positivity alternating with areas of weak or no staining. In each of the 4 cases, the area cored for TMA construction (identifiable by the defect in the tissue Interlaboratory Variation in HER2/neu Testing—Sheffield et al
Table 2. Primary Antibodies and Dilutions Used by Participating Laboratories (n ¼ 37) Primary Antibody Dilution
Laboratories Reporting Use, No. (%)
4B5 Prediluted
18 (49)
SP3 1:50–1:100 1:100–1:200 ,1:200 Total SP3
5 3 2 10
(14) (8) (5) (27)
A0485 1:500–1:600 1:1000 1:1700 Total A0485
3 1 1 5
(8) (3) (3) (14)
HercepTest Prediluted
2 (5)
Unspecified polyclonal antibody 1:175–1:200
2 (5)
where the core had been removed) showed 0 or 1þ staining, whereas elsewhere in the same block, were areas of tumor with stronger (2þ or 3þ) staining (Figure 6, A and B). COMMENT The goals of this study were to assess interlaboratory variability in the assessment of HER2/neu status in gastric cancer and to survey current IHC methodologies used in that assessment. Secondarily, we were able to compare IHC
scores to expert and consensus IHC scores as well as ISH results for those cases. The results obtained in this study offer much encouragement for the role of IHC testing in determining HER2/neu status in gastric cancer. Although ISH is likely to remain the gold standard for HER2/neu status determination, IHC represents an effective and robust test that can be used for most specimens, with ISH being reserved for equivocal cases only. Immunohistochemistry offers many advantages over ISH, notably being available in a wider range of hospital laboratories, without the necessity for additional equipment and staff beyond the requirements of other commonly available IHC assays. Although these data are encouraging, agreement among laboratories was not 100%. Several factors contributed to the variability observed in this study, including differences in IHC methodologies, interpretive errors, and intratumoral heterogeneity. Preanalytic variables, such as cold-ischemic and fixation times, may also have a role in practice, although those elements were controlled in the current study. IHC Methodology A false-negative HER2/neu result was obtained using the HercepTest primary antibody. False-negative results with that antibody are documented in the literature, and improved diagnostic accuracy is achieved when using rabbit monoclonal antibodies, such as SP3 and 4B5.16 Although no gold standard method for IHC exists, this result reinforces that laboratories offering HER2/neu IHC should be aware of the test characteristics and staining patterns of the antibodies they employ and should always attempt to use the most appropriate primary antibodies. Additional pa-
Figure 3. HER2/neu status by expert review and in situ hybridization results for cores included in the tissue microarray (TMA). Each circle represents an individual core on the TMA. The top number indicates the core identification number (1–45); the central oval represents the in situ hybridization result for that tumor: amplified (Am) in red, equivocal (equ) in amber, and negative (neg) in white. The bottom number and the color of the entire circle represent the consensus immunohistochemical score at expert review: 0, white; 1þ, gray; 2þ, amber; and 3þ, red. Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1499
Figure 4. Core 32 stained by laboratory 161 (HercepTest primary antibody, ready to use dilution, with HercepTest antigen retrieval solution and visualization reagent) (a); and laboratory 189 (4B5 primary antibody, prediluted, mild CC1 antigen retrieval, UltraView DAB visualization) (b). Note false-negative staining in (a). Immunohistochemical stains directed toward HER2/neu (original magnification, 310 [a and b]). Figure 5. Core 15 stained by laboratory 164 (4B5 primary antibody, ultra CC1 antigen retrieval, BenchmarkUltra detection) (a); and laboratory 189 (4B5 primary antibody, prediluted, mild CC1 antigen retrieval, UltraView DAB visualization) (b). Note similar staining patterns in both cores, 1500 Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al
rameters, such as antibody dilution and antigen retrieval, should be internally validated and optimized. Table 2 shows the wide diversity of protocols used by laboratories assessing HER2/neu by IHC; however, collectively, this potpourri of methodologies was able to achieve excellent interlaboratory agreement, sensitivity, and specificity in selecting HER2/neuþ cases. Because false-negative results are typically generated by technical aspects of the test, the high sensitivity demonstrates robustness in the various materials and methods used.
siveness of focally amplified gastric carcinomas to targeted treatment is unknown. Thus, it is not possible to offer evidence-based guidelines on HER2/neu assessment of heterogeneously amplified tumors at present. Furthermore, ISH performed on equivocal IHC cases should be guided by an IHC-stained slide, rather than a routine hematoxylineosin–stained section, being certain to examine areas with IHC positive staining. In a heterogeneously amplified tumor, this targeted approach will ensure the most comprehensive cytogenetic assessment.
IHC Interpretation
Test Characteristics
The interpretive error identified emphasizes the necessity of expertise in applying the criteria for assessing HER2/neu IHC in gastric cancer. The main differences from assessing HER2/neu in breast cancer are that gastric cancer often does not show complete membranous staining8 (ie, staining is often basolateral) and that only 5 cells in a biopsy specimen are required to show immunoreactivity to be considered a positive result9 (whereas in breast cancer .10% immunoreactive cells are required). Caution should be used when attempting to score the edge of a tissue specimen as well as crushed tissue because both can lead to false-positive scores. Only membranous staining should be scored, and tissue showing nuclear or cytoplasmic staining should raise concerns about the validity of the test. Alternative scoring systems have been reported in the literature; caution should be exercised when using them, and internal or external validation is recommended. Gown et al17 recommended the use of a normalized IHC scoring system for breast cancer, wherein the HER2/neu score of adjacent normal tissue is subtracted from the HER2/ neu score of the tumor. That system has been shown to reduce the number of false-positive results and, although not validated for gastric carcinoma, should be assessed in the future. This modification to the traditional HER2/neu scoring system could not be evaluated in the TMA setting as ‘‘adjacent normal tissue’’ was unavailable for assessment. In general, interpretive errors lead to false-positive results. The high specificity observed in this study indicates consistent and accurate interpretation of IHC staining by pathologists.
Given the sensitivity and specificity calculated above, positive and negative likelihood ratios were established as 681 and 0.009, respectively. Given a pretest probability of 19% (the prevalence of HER2/neu amplification among gastric carcinomas), the posttest probability of amplification would be 99.9% given a positive test or 0.2% given a negative test. These characteristics offer much support to the use of IHC with reflex ISH for the detection of HER2/neu amplification, showing the extreme reliability of both positive and negative results.
Intratumoral Heterogeneity Intratumoral heterogeneity of HER2/neu overexpression is well documented in gastric cancer10,18,19 and was documented in the current study. Asioli et al10 have shown that assessing multiple FFPE blocks, as opposed to a single block, increases the number of HER2/neuþ gastric carcinomas detected by IHC. Literature reports, together with the current data, suggest that more-accurate HER2/neu status can be obtained by staining more tissue. Hence, resection specimens are preferred to biopsy specimens, and whole slides should be stained; although a TMA was used in this quality assurance exercise, TMAs should not be used to assess HER2/neu status in clinical practice. Data on optimal amount of tissue to evaluate are lacking, and the respon-
Preanalytic Variables Although preanalytic variables were not assessed in this study, they may additionally contribute to variability in HER2/neu results. Previous studies have shown that cold ischemic time, fixation time, and even fixative type have much less effect than originally thought on HER2/neu IHC in breast cancer.20,21 In unpublished data from our group, we have confirmed that variable formalin fixation times, ranging from 1 to 48 hours, do not affect HER2/neu IHC staining when conventional IHC techniques are used. Nonetheless, standardizing preanalytic treatment across laboratories would lead to better pathology practices, if not more reproducible immunohistochemistry. CONCLUSION Identification and targeted treatment of HER2/neuþ gastric cancer has become the standard of care in many centers. Rapid, robust, and reliable detection of HER2/neu gene amplification and overexpression will lead to the targeted treatment with effective systemic therapy for specific patients. Here, we have shown that HER2/neu assessment by IHC is reproducible among several laboratories with a high sensitivity and specificity. Despite a strong degree of interobserver agreement, improvements to the standardization of HER2/neu testing would be beneficial. Possibilities for future recommendations include standardization to rabbit monoclonal antibodies, incorporation of a normalized HER2/neu score, guidelines for testing multiple blocks or surgical specimens rather than biopsy specimens, and standardized preanalytic tissue handling regulations. As the field of cancer treatment progresses into the era of targeted therapy, medical laboratories will be faced with
incorrectly interpreted as 3þ, strong staining, by laboratory 164. Expert review agreed that either 1þ or 2þ interpretation of those cores was acceptable. Immunohistochemical stains directed toward HER2/neu (original magnification, 310 [a and b]). Figure 6. Example of intratumoral heterogeneity in a case that showed HER2/neu gene amplification by in situ hybridization. There is clear variability in HER2/neu staining, with areas of the tumor showing significant immunoreactivity (a), whereas the area surrounding the tissue core taken for tissue microarray construction shows no staining (b) (SP3 primary antibody, original magnification, 310 [a and b]). Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1501
increasing demands for biomarker testing. Trastuzumab is the first targeted treatment available for solid tumors, and accordingly, HER2/neu testing is paving the way for this cohort of laboratory tests. Continual refinement and analysis of the testing strategies used to identify HER2/neuþ breast and gastric carcinomas are crucial, not only for these disease entities but also for an identification paradigm of treatable molecular targets. Dr Li-Chang receives fellowship funding from the Terry Fox Foundation, Strategic Training Initiatives in Health Research in Cancer Research at the Canadian Institutes for Health Research. References 1. Guggenheim DE, Shah MA. Gastric cancer epidemiology and risk factors. J Surg Oncol. 2013;107(3):230–236. 2. American Cancer society. Stomach cancer detailed guide: stomach cancer survival rates. http://www.cancer.org/cancer/stomachcancer/. Accessed November 11, 2013. 3. Cunningham D, Allum WH, Stenning SP, et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med. 2006; 355(1):11–20. 4. Ross JS, Slodkowska EA, Symmans WF, Pusztai L, Ravdin PM, Hortobagyi GN. The HER-2 receptor and breast cancer: ten years of targeted anti–HER-2 therapy and personalized medicine. Oncologist. 2009;14:320–368. 5. Ross JS, McKenna BJ. The HER-2/neu oncogene in tumors of the gastrointestinal tract. Cancer Invest. 2001;19(1):554–568. 6. Garcia I, Vizoso F, Martin A, et al. Clinical significance of the epidermal growth factor receptor and HER2 receptor in resectable gastric cancer. Ann Surg Oncol. 2002;10(3):234–241. 7. Bang YJ, Van Cutsem E, Feyereislova A, et al; ToGA Trial Investigators. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet. 2010; 376(9742):687–697. 8. Hofmann M, Stoss O, Shi D, et al. Assessment of a HER2 scoring system for gastric cancer: results from a validation study. Histopathology. 2008;52(7):797–805. 9. Ruschoff J, Hanna W, Bilous M, et al. HER2 testing in gastric cancer: a practical approach. Mod Pathol. 2012;25(5):637–650.
1502 Arch Pathol Lab Med—Vol 138, November 2014
10. Asioli S, Maletta F, Verdun di Cantogno L, et al. Approaching heterogeneity of human epidermal growth factor receptor 2 in surgical specimens of gastric cancer. Hum Pathol. 2012;43(11):2070–2079. 11. Boers JE, Meeuwissen H, Methorst N. HER2 status in gastro-oesophageal adenocarcinomas assessed by two rabbit monoclonal antibodies (SP3 and 4B5) and two in situ hybridization methods (FISH and SISH) Histopathology. 2011; 58(3):383–394. 12. Perez EA, Suman VJ, Davidson NE, et al. HER2 testing by local, central, and reference laboratories in specimens from the North Central Cancer Treatment Group N9831 intergroup adjustment trial. J Clin Oncol. 2006;24(19):3032–3038. 13. Choritz H, Busche G, Kreipe H; Study Group HER2 Monitor. Quality assessment of HER2 testing by monitoring of positivity rates. Virchows Arch. 2011;459(3):283–289. 14. Fox SB, Kumarasinghe MP, Armes JE, et al. Gastric HER2 testing study (GaTHER): An evaluation of gastric/gastroesophageal junction cancer testing accuracy in Australia. Am J Surg Pathol. 2012;36(4):577–582. 15. Terry J, Torlakovic EE, Garratt J, et al. Implementation of a Canadian external quality assurance program for breast cancer biomarkers: an initiative of Canadian quality control immunohistochemistry (cIQc) and Canadian association of pathologists (CAP) national standards committee/immunohistochemistry. Appl Immunohistochem Mol Morphol. 2009;17(5):375–382. 16. Dekker TJ, Borg S, Hooijer GK, et al. Determining sensitivity and specificity of HER2 testing in breast cancer using a tissue micro-array approach. Breast Cancer Res. 2012;14(3):R93. doi:10.1186/bcr3208. 17. Gown AM, Goldstein LC, Barry TS, et al. High concordance between immunohistochemistry and fluorescence in situ hybridization testing for HER2 status in breast cancer requires a normalized IHC scoring system. Mod Pathol. 2008;21(10):1271–1277. 18. Fassan M, Ludwig K, Castoro C, et al. Human epithelial growth factor receptor 2 (HER2) status in primary and metastatic esophagogastric junction adenocarcinomas. Hum Pathol. 2012;43(8):1206–1212. 19. Kim MA, Lee HJ, Yang HK, Bang YJ, Kim WH. Heterogeneous amplification of HER2/neu in primary lesions is responsible for the discordant HER2/neu status of primary and metastatic lesions in gastric carcinoma. Histopathology. 2011;59(5):822–831. 20. Wolff AC, Hammond ME, Schwartz JN, et al; American Society of Clinical Oncology/College of American Pathologists. American Society of Clinical Oncology/College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. Arch Pathol Lab Med. 2007;131(1):18–43. 21. Moatamed NA, Nanjangud G, Pucci R, et al. Effect of ischemic time, fixation time, and fixative type on HER2/neuimmunohistochemical and fluorescence in situ hybridization results in breast cancer. Am J Clin Pathol. 2011;136(5):754–761.
Interlaboratory Variation in HER2/neu Testing—Sheffield et al
Context.—Immunohistochemical (IHC) testing for HER2/neu is becoming the standard of care for guiding adjuvant treatment of gastric carcinoma with trastuzumab. Objective.—To assess interlaboratory variation in IHC staining and interpretation across multiple laboratories. Design.—A tissue microarray consisting of 45 cores from 28 gastric cancers was distributed to 37 laboratories for HER2/neu assessment. The IHC results were compared against expert scores at an academic institution and correlated with in situ hybridization results from the originating specimen. Interlaboratory agreement was calculated using Cohen j statistic. Results.—The survey demonstrated several variations in IHC methods, including the primary antibodies in use. There was excellent agreement among laboratories in HER2/neuþ (IHC 3þ) cases (j ¼ 0.80 6 0.01) and very good agreement among laboratories in HER2/neu (IHC 0 or 1þ) cases (j ¼ 0.58 6 0.01). Less agreement was observed
among laboratories when scoring equivocal (IHC 2þ) cases (j ¼ 0.22 6 0.01). Sensitivity and specificity of HER2/neu IHC were 99% and 100%, respectively, when measured against expert review and consensus score as a reference standard. Conclusions.—There is substantial interlaboratory agreement in the interpretation of HER2/neu IHC despite variability in protocols. Although HER2/neu IHC is a highly sensitive and specific test, primary antibody selection may significantly affect IHC results. Furthermore, gastric tumors require a unique scoring system and expertise in interpretation. Intratumoral heterogeneity has a significant effect on HER2/neu scoring by IHC. Ongoing quality assurance exercises among laboratories will help ensure optimized HER2/neu testing. (Arch Pathol Lab Med. 2014;138:1495–1502; doi: 10.5858/arpa.2013-0604-OA)
G
Human epidermal growth factor receptor 2 (coded by the ERBB2 gene) gain of function is associated with increased cell motility, invasiveness, angiogenesis, resistance to apoptosis, and metastatic potential.4 Overexpression of HER2/neu in gastric cancers occurs most commonly as an amplification of the 17q12–q21 region and is detectable by in situ hybridization (ISH) techniques. This event occurs in 15% to 25% of gastric carcinomas; a meta-analysis showed an overall prevalence of 19%.5 Although HER2/neu amplification is associated with decreased overall survival in gastric cancer,6 the multinational Trastuzumab for Gastric Cancer (TOGA) clinical trial demonstrated increased progression-free and overall survival in HER2/neu-overexpressing gastric cancers when trastuzumab was added to conventional treatment.7 Trastuzumab is a humanized recombinant monoclonal antibody targeting the HER2/neu molecule. This drug is routinely used in the treatment of HER2/neu-amplified breast cancers,5 with HER2/neu testing and trastuzumab treatment also now becoming the standard of care for metastatic gastric cancer. Because only a subset of patients will benefit from this targeted therapy, there is a demand on medical laboratories to offer accurate and robust testing of HER2/neu status. Slide-based immunohistochemical (IHC) techniques have been validated and show high concordance with ISH studies
astric carcinoma contributes to 10% of cancer deaths worldwide with a case to fatality ratio of 70%.1 Despite recent advances in cancer treatment, the overall 5-year survival for patients with newly diagnosed gastric cancer is 28%.2 The treatment of gastric cancer is primarily surgical because the addition of perioperative chemotherapy has been associated with only marginal clinical benefits in overall and progression-free survival.3 Accepted for publication January 8, 2014. From the Division of Anatomic Pathology, Department of Pathology and Laboratory Medicine, Vancouver General Hospital and the University of British Columbia, Vancouver, Canada (Drs Sheffield, Gilks, and Schaeffer and Mr Garratt); the Canadian Immunohistochemistry Quality Control Program, Vancouver (Mr Garratt and Drs Torlakovic and Gilks); the Department of Pathology and Laboratory Medicine, British Columbia Cancer Agency and the University of British Columbia, Vancouver (Dr Li-Chang and Mr Kalloger); and the Department of Laboratory Hematology, University Health Network, University of Toronto, Toronto, Ontario, Canada (Dr Torlakovic). The authors have no relevant financial interest in the products or companies described in this article. Reprints: Brandon S. Sheffield, MD, Division of Anatomic Pathology, Department of Pathology and Laboratory Medicine, Vancouver General Hospital, 855 12 Ave W, Vancouver, BC V5Z 1M9, Canada (e-mail: [email protected]). Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1495
Table 1. Demographic Data of the Cases (n ¼ 28) Selected for Tissue Microarray Construction Characteristics
Cases, No. (%)
Sex Male Female
21 (79) 6 (21)
Age (at diagnosis) Range, y Mean, y Median, y
42–89 64.3 65
Gastric carcinoma subtype Intestinal Diffuse Mixed
20 (71) 3 (11) 5 (18)
Histologic grade Well differentiated Moderately differentiated Poorly differentiated
1 (4) 9 (32) 18 (64)
Site Proximal Distal Diffuse
17 (61) 10 (36) 1 (4)
Lymph node metastases Yes No
18 (64) 10 (36)
Stage IA IB IIA IIB IIIA IIIB IV
MATERIALS AND METHODS Appropriate research and ethics permissions were obtained from the University of British Columbia (Vancouver, Canada) and the participating health care centers.
Case Selection Twenty-eight gastric carcinoma surgical resection specimens from 2004 to 2011 were identified retrospectively from the archives of Vancouver General Hospital (Vancouver, British Columbia, Canada). These cases were randomly selected from an archival database and represented a range of gastric carcinoma subtypes (Table 1). Inclusion criteria were gastric adenocarcinoma undergoing surgical resection and available ISH data (available from previous research and validation studies), as well as availability of formalin-fixed, paraffin-embedded (FFPE) blocks for tissue microarray (TMA) construction. No exclusion criteria were used. Cases included HER2/neu-amplified and -nonamplified tumors as well as tumors showing a range of HER2/neu IHC staining intensity from 0 to 3þ.
TMA Construction and Interpretation
2 6 8 3 4 2 3
(7) (21) (29) (11) (14) (7) (11)
in the assessment of HER2/neu amplification and overexpression.8 Potential challenges in HER2/neu assessment of gastric tumors by IHC include novel criteria for assessment, which are different from those used in breast cancer.9 Intratumoral heterogeneity of HER2/neu amplification affects a significant proportion of gastric tumors10 and may affect IHC evaluation. In addition, many different antiHER2/neu primary antibodies are commercially available for conventional IHC testing, and two widely used antibodies (4B5 and SP3) have been shown to produce different staining patterns.11 Considering these variables to ensure accurate HER2/neu IHC testing is a vital process because test results have a direct influence on treatment choice and patient outcome. As HER2/neu IHC becomes widely implemented in the management of gastric cancer, so too, must appropriate quality control measures be implemented. Previous studies have shown significant interlaboratory variation in HER2/ neu IHC test results.12,13 One group has shown that a combination of IHC and ISH testing is reproducible among a few laboratories working collaboratively with a central reference laboratory.14 To date in North America, to our knowledge, no measures of interlaboratory variation in gastric cancer HER2/neu IHC test results have been reported in the literature. This study aimed to measure interlaboratory variation in HER2/neu testing by IHC. The methods were designed to capture significant differences in analytic and postanalytic aspects of IHC determination of HER2/neu status. 1496 Arch Pathol Lab Med—Vol 138, November 2014
A TMA was constructed using duplicate 0.6-mm cores, when available, derived from FFPE blocks corresponding to each of the study cases. All tissue was handled and processed at the anatomic pathology laboratory at Vancouver General Hospital according to the standard of care at our institution (cold ischemic time, approximately 0–4 hours; formalin fixation time in 10% neutralbuffered formalin, 24–48 hours). Resection specimens were used rather than biopsies to yield optimal TMA results. Tissue cores for the TMA were obtained from areas of tumor determined by routine microscopy on hematoxylin-eosin–stained sections. Unstained TMA slides were distributed to laboratories participating in a quality control exercise on a voluntary basis within the well-established Canadian Immunohistochemistry Quality Control Program (Vancouver, British Columbia, Canada).15 Laboratories were asked to provide information about their staining protocols, including primary antibody clone, dilution, and pretreatment, in addition to scoring each core on the TMA. No instruction or guideline was given about how to score the cases. All data were gathered electronically via online submission.
Expert Review Stained slides were returned to Vancouver General Hospital for review and analysis by 2 pathologists with expertise in IHC quality assurance and gastrointestinal pathology (C.B.G. and D.F.S., respectively), who achieved a consensus interpretation of these cases. Expert analysis of HER2/neu IHC staining intensity was made according to the system used in the TOGA trial.9 Briefly, 0 was defined as no reactivity or membranous reactivity in ,5 cells, 1þ staining was defined as faint/barely perceptible membranous reactivity in 5 cells, 2þ staining as weak to moderate complete or basolateral membranous reactivity in 5 cells, and 3þ staining as moderate to strong complete or basolateral membranous reactivity in 5 cells (TMA cores were scored as biopsies rather than resection specimens, see examples in Figure 1, A through D). A single set of observers, who were well versed in the abovementioned scoring system, analyzed stained slides returned from each of the 37 laboratories, establishing a consensus IHC score based on individual scores for each tissue core. False-positive and false-negative cases were identified as those where individual IHC scores differed from consensus score.
Statistical Analysis Individual, laboratory-scored TMAs were pooled, and a Cohen j statistic was calculated as a measure of interlaboratory agreement. Individual cores were considered to be negative (0 or 1þ IHC score), equivocal (2þ IHC score), or positive (3þ IHC score). Test characteristics were calculated using expert review and consensus interpretation of IHC staining as a gold standard. Scores Interlaboratory Variation in HER2/neu Testing—Sheffield et al
Figure 1. Example of 0 to 3þ scoring. a, 0, no immunoreactivity shown by laboratory 145 in core 25. b, 1þ, faint, barely perceptible membrane staining by laboratory 110 in core 14. c, 2þ, weak to moderate membranous staining by laboratory 152 in core 2. d, 3þ, strong membranous staining by laboratory 103 in core 8 (original magnification, 310 [a through d]).
of 0 and 1þ were considered negative results, whereas 3þ corresponded to a positive IHC result. When a participant core was assessed as 2þ (equivocal), it was, by default, determined to be concordant with the reference value, regardless of the actual reference value, because those patients would be referred for HER2/neu in situ hybridization and would be appropriately stratified for therapy based on those results.
IHC Staining Cases with discordance between consensus IHC staining and ISH data were identified. The source FFPE blocks were used to perform HER2/neu IHC testing on whole slides. The HER2/neu IHC tests were performed using the SP3 primary antibody (Thermo Scientific, Marietta, Ohio) at 1:100 dilution with the Optiview (Ventana Medical Systems, Oro Valley, Arizona) visualization system.
In Situ Hybridization All ISH data were collected from a previously established research database. The original assays were performed on whole Arch Pathol Lab Med—Vol 138, November 2014
slides derived from FFPE blocks using either fluorescence in situ hybridization (PathVysion HER2/neu kit, Abbott molecular, Des Plaines, Illinois) or silver in situ hybridization (HER2/neu and CEP17 probes, Ventana). In determining the HER2/neu status by ISH, the following cutoffs were used: (1) amplified, if the HER2/ neu:CEP17 ratio was greater than 2.2 or the HER2/neu was greater than 6 copies/nucleus; (2) equivocal, if the HER2/neu:CEP17 ratio was between 1.8 and 2.2; and (3) negative, if the HER2/neu:CEP17 ratio was less than 1.8. All cases were assessed by expert pathologists or cytogeneticists. Fluorescence in situ hybridization and silver in situ hybridization results were considered equivalents.
RESULTS Thirty-seven laboratories participated in the quality assurance exercise. Each of the 10 Canadian provinces was represented as well as 4 laboratories in the United States and 2 laboratories in Europe. Community, academic, and private laboratories participated in the exercise. Participating laboratories used a range of pretreatments, dilutions, and Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1497
Figure 2. Immunohistochemical (IHC) scores assigned to tissue microarray by participating laboratories. Individual cores are labeled 1 to 45. Individual laboratories are deidentified with a 3 digit code (101–199 nonconsecutively). Tissue microarray scoring was recorded as U, core unsatisfactory for scoring (white); 0, an IHC score of 0 (white); 1, an IHC score of 1þ (gray); 2, an IHC score of 2þ (amber); and 3, an IHC score of 3þ (red).
primary antibodies including: 4B5, SP3, A0485, and the HercepTest (Dako, Carpinteria, California). Details are provided in Table 2. The IHC scores, as reported by individual laboratories, are shown in Figure 2. Results of expert review and consensus interpretation are presented together with ISH data in Figure 3. Analysis of interlaboratory reproducibility showed excellent agreement among positive (IHC 3þ) cases (j ¼ 0.80 6 0.01), and very good agreement among negative (IHC 0 or 1þ) cases (j ¼ 0.58 6 0.01). Less agreement was observed among laboratories scoring equivocal (IHC 2þ) cases (j ¼ 0.22 6 0.01). Six of 45 cores represented on the TMA (cores 7, 8, 19, 20, 31, and 32) were identified on expert review as being positive by IHC (Figure 3). Those cores were identified by all but one of the participating laboratories (36 of 37) as being IHC 3þ or 2þ. A single laboratory (no. 161; 1 of 37) identified only 4 of those 6 cores (Figure 4, A and B). Upon expert review, that error was confirmed as a false-negative IHC 1498 Arch Pathol Lab Med—Vol 138, November 2014
test. That laboratory was using the HercepTest primary antibody. A single laboratory (no. 164; 1 of 37) showed falsepositive staining in 2 cores (cores 5 and 15), and upon expert review, it was confirmed that laboratory had incorrectly interpreted 2þ and 1þ staining patterns as 3þ in both instances (Figure 5, A and B). Overall the sensitivity and specificity of HER2/neu IHC were calculated as 99.1% (95% confidence interval, 98.1%–99.6%) and 99.8% (95% CI, 99.6%–100%), respectively. Four of 45 cores (cores 6, 12, 38, 39) showed HER2/neu amplification on ISH, but were, however, scored negative on IHC (0 or 1þ). Expert review confirmed that none of the TMA sections showed positive immunoreactivity for HER2/ neu in those 4 cores (Figures 2 and 3). Whole-slide sections stained for HER2/neu showed marked intratumoral heterogeneity in each of those 4 cases. Each case demonstrated regions of strong HER2/neu positivity alternating with areas of weak or no staining. In each of the 4 cases, the area cored for TMA construction (identifiable by the defect in the tissue Interlaboratory Variation in HER2/neu Testing—Sheffield et al
Table 2. Primary Antibodies and Dilutions Used by Participating Laboratories (n ¼ 37) Primary Antibody Dilution
Laboratories Reporting Use, No. (%)
4B5 Prediluted
18 (49)
SP3 1:50–1:100 1:100–1:200 ,1:200 Total SP3
5 3 2 10
(14) (8) (5) (27)
A0485 1:500–1:600 1:1000 1:1700 Total A0485
3 1 1 5
(8) (3) (3) (14)
HercepTest Prediluted
2 (5)
Unspecified polyclonal antibody 1:175–1:200
2 (5)
where the core had been removed) showed 0 or 1þ staining, whereas elsewhere in the same block, were areas of tumor with stronger (2þ or 3þ) staining (Figure 6, A and B). COMMENT The goals of this study were to assess interlaboratory variability in the assessment of HER2/neu status in gastric cancer and to survey current IHC methodologies used in that assessment. Secondarily, we were able to compare IHC
scores to expert and consensus IHC scores as well as ISH results for those cases. The results obtained in this study offer much encouragement for the role of IHC testing in determining HER2/neu status in gastric cancer. Although ISH is likely to remain the gold standard for HER2/neu status determination, IHC represents an effective and robust test that can be used for most specimens, with ISH being reserved for equivocal cases only. Immunohistochemistry offers many advantages over ISH, notably being available in a wider range of hospital laboratories, without the necessity for additional equipment and staff beyond the requirements of other commonly available IHC assays. Although these data are encouraging, agreement among laboratories was not 100%. Several factors contributed to the variability observed in this study, including differences in IHC methodologies, interpretive errors, and intratumoral heterogeneity. Preanalytic variables, such as cold-ischemic and fixation times, may also have a role in practice, although those elements were controlled in the current study. IHC Methodology A false-negative HER2/neu result was obtained using the HercepTest primary antibody. False-negative results with that antibody are documented in the literature, and improved diagnostic accuracy is achieved when using rabbit monoclonal antibodies, such as SP3 and 4B5.16 Although no gold standard method for IHC exists, this result reinforces that laboratories offering HER2/neu IHC should be aware of the test characteristics and staining patterns of the antibodies they employ and should always attempt to use the most appropriate primary antibodies. Additional pa-
Figure 3. HER2/neu status by expert review and in situ hybridization results for cores included in the tissue microarray (TMA). Each circle represents an individual core on the TMA. The top number indicates the core identification number (1–45); the central oval represents the in situ hybridization result for that tumor: amplified (Am) in red, equivocal (equ) in amber, and negative (neg) in white. The bottom number and the color of the entire circle represent the consensus immunohistochemical score at expert review: 0, white; 1þ, gray; 2þ, amber; and 3þ, red. Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1499
Figure 4. Core 32 stained by laboratory 161 (HercepTest primary antibody, ready to use dilution, with HercepTest antigen retrieval solution and visualization reagent) (a); and laboratory 189 (4B5 primary antibody, prediluted, mild CC1 antigen retrieval, UltraView DAB visualization) (b). Note false-negative staining in (a). Immunohistochemical stains directed toward HER2/neu (original magnification, 310 [a and b]). Figure 5. Core 15 stained by laboratory 164 (4B5 primary antibody, ultra CC1 antigen retrieval, BenchmarkUltra detection) (a); and laboratory 189 (4B5 primary antibody, prediluted, mild CC1 antigen retrieval, UltraView DAB visualization) (b). Note similar staining patterns in both cores, 1500 Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al
rameters, such as antibody dilution and antigen retrieval, should be internally validated and optimized. Table 2 shows the wide diversity of protocols used by laboratories assessing HER2/neu by IHC; however, collectively, this potpourri of methodologies was able to achieve excellent interlaboratory agreement, sensitivity, and specificity in selecting HER2/neuþ cases. Because false-negative results are typically generated by technical aspects of the test, the high sensitivity demonstrates robustness in the various materials and methods used.
siveness of focally amplified gastric carcinomas to targeted treatment is unknown. Thus, it is not possible to offer evidence-based guidelines on HER2/neu assessment of heterogeneously amplified tumors at present. Furthermore, ISH performed on equivocal IHC cases should be guided by an IHC-stained slide, rather than a routine hematoxylineosin–stained section, being certain to examine areas with IHC positive staining. In a heterogeneously amplified tumor, this targeted approach will ensure the most comprehensive cytogenetic assessment.
IHC Interpretation
Test Characteristics
The interpretive error identified emphasizes the necessity of expertise in applying the criteria for assessing HER2/neu IHC in gastric cancer. The main differences from assessing HER2/neu in breast cancer are that gastric cancer often does not show complete membranous staining8 (ie, staining is often basolateral) and that only 5 cells in a biopsy specimen are required to show immunoreactivity to be considered a positive result9 (whereas in breast cancer .10% immunoreactive cells are required). Caution should be used when attempting to score the edge of a tissue specimen as well as crushed tissue because both can lead to false-positive scores. Only membranous staining should be scored, and tissue showing nuclear or cytoplasmic staining should raise concerns about the validity of the test. Alternative scoring systems have been reported in the literature; caution should be exercised when using them, and internal or external validation is recommended. Gown et al17 recommended the use of a normalized IHC scoring system for breast cancer, wherein the HER2/neu score of adjacent normal tissue is subtracted from the HER2/ neu score of the tumor. That system has been shown to reduce the number of false-positive results and, although not validated for gastric carcinoma, should be assessed in the future. This modification to the traditional HER2/neu scoring system could not be evaluated in the TMA setting as ‘‘adjacent normal tissue’’ was unavailable for assessment. In general, interpretive errors lead to false-positive results. The high specificity observed in this study indicates consistent and accurate interpretation of IHC staining by pathologists.
Given the sensitivity and specificity calculated above, positive and negative likelihood ratios were established as 681 and 0.009, respectively. Given a pretest probability of 19% (the prevalence of HER2/neu amplification among gastric carcinomas), the posttest probability of amplification would be 99.9% given a positive test or 0.2% given a negative test. These characteristics offer much support to the use of IHC with reflex ISH for the detection of HER2/neu amplification, showing the extreme reliability of both positive and negative results.
Intratumoral Heterogeneity Intratumoral heterogeneity of HER2/neu overexpression is well documented in gastric cancer10,18,19 and was documented in the current study. Asioli et al10 have shown that assessing multiple FFPE blocks, as opposed to a single block, increases the number of HER2/neuþ gastric carcinomas detected by IHC. Literature reports, together with the current data, suggest that more-accurate HER2/neu status can be obtained by staining more tissue. Hence, resection specimens are preferred to biopsy specimens, and whole slides should be stained; although a TMA was used in this quality assurance exercise, TMAs should not be used to assess HER2/neu status in clinical practice. Data on optimal amount of tissue to evaluate are lacking, and the respon-
Preanalytic Variables Although preanalytic variables were not assessed in this study, they may additionally contribute to variability in HER2/neu results. Previous studies have shown that cold ischemic time, fixation time, and even fixative type have much less effect than originally thought on HER2/neu IHC in breast cancer.20,21 In unpublished data from our group, we have confirmed that variable formalin fixation times, ranging from 1 to 48 hours, do not affect HER2/neu IHC staining when conventional IHC techniques are used. Nonetheless, standardizing preanalytic treatment across laboratories would lead to better pathology practices, if not more reproducible immunohistochemistry. CONCLUSION Identification and targeted treatment of HER2/neuþ gastric cancer has become the standard of care in many centers. Rapid, robust, and reliable detection of HER2/neu gene amplification and overexpression will lead to the targeted treatment with effective systemic therapy for specific patients. Here, we have shown that HER2/neu assessment by IHC is reproducible among several laboratories with a high sensitivity and specificity. Despite a strong degree of interobserver agreement, improvements to the standardization of HER2/neu testing would be beneficial. Possibilities for future recommendations include standardization to rabbit monoclonal antibodies, incorporation of a normalized HER2/neu score, guidelines for testing multiple blocks or surgical specimens rather than biopsy specimens, and standardized preanalytic tissue handling regulations. As the field of cancer treatment progresses into the era of targeted therapy, medical laboratories will be faced with
incorrectly interpreted as 3þ, strong staining, by laboratory 164. Expert review agreed that either 1þ or 2þ interpretation of those cores was acceptable. Immunohistochemical stains directed toward HER2/neu (original magnification, 310 [a and b]). Figure 6. Example of intratumoral heterogeneity in a case that showed HER2/neu gene amplification by in situ hybridization. There is clear variability in HER2/neu staining, with areas of the tumor showing significant immunoreactivity (a), whereas the area surrounding the tissue core taken for tissue microarray construction shows no staining (b) (SP3 primary antibody, original magnification, 310 [a and b]). Arch Pathol Lab Med—Vol 138, November 2014
Interlaboratory Variation in HER2/neu Testing—Sheffield et al 1501
increasing demands for biomarker testing. Trastuzumab is the first targeted treatment available for solid tumors, and accordingly, HER2/neu testing is paving the way for this cohort of laboratory tests. Continual refinement and analysis of the testing strategies used to identify HER2/neuþ breast and gastric carcinomas are crucial, not only for these disease entities but also for an identification paradigm of treatable molecular targets. Dr Li-Chang receives fellowship funding from the Terry Fox Foundation, Strategic Training Initiatives in Health Research in Cancer Research at the Canadian Institutes for Health Research. References 1. Guggenheim DE, Shah MA. Gastric cancer epidemiology and risk factors. J Surg Oncol. 2013;107(3):230–236. 2. American Cancer society. Stomach cancer detailed guide: stomach cancer survival rates. http://www.cancer.org/cancer/stomachcancer/. Accessed November 11, 2013. 3. Cunningham D, Allum WH, Stenning SP, et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med. 2006; 355(1):11–20. 4. Ross JS, Slodkowska EA, Symmans WF, Pusztai L, Ravdin PM, Hortobagyi GN. The HER-2 receptor and breast cancer: ten years of targeted anti–HER-2 therapy and personalized medicine. Oncologist. 2009;14:320–368. 5. Ross JS, McKenna BJ. The HER-2/neu oncogene in tumors of the gastrointestinal tract. Cancer Invest. 2001;19(1):554–568. 6. Garcia I, Vizoso F, Martin A, et al. Clinical significance of the epidermal growth factor receptor and HER2 receptor in resectable gastric cancer. Ann Surg Oncol. 2002;10(3):234–241. 7. Bang YJ, Van Cutsem E, Feyereislova A, et al; ToGA Trial Investigators. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet. 2010; 376(9742):687–697. 8. Hofmann M, Stoss O, Shi D, et al. Assessment of a HER2 scoring system for gastric cancer: results from a validation study. Histopathology. 2008;52(7):797–805. 9. Ruschoff J, Hanna W, Bilous M, et al. HER2 testing in gastric cancer: a practical approach. Mod Pathol. 2012;25(5):637–650.
1502 Arch Pathol Lab Med—Vol 138, November 2014
10. Asioli S, Maletta F, Verdun di Cantogno L, et al. Approaching heterogeneity of human epidermal growth factor receptor 2 in surgical specimens of gastric cancer. Hum Pathol. 2012;43(11):2070–2079. 11. Boers JE, Meeuwissen H, Methorst N. HER2 status in gastro-oesophageal adenocarcinomas assessed by two rabbit monoclonal antibodies (SP3 and 4B5) and two in situ hybridization methods (FISH and SISH) Histopathology. 2011; 58(3):383–394. 12. Perez EA, Suman VJ, Davidson NE, et al. HER2 testing by local, central, and reference laboratories in specimens from the North Central Cancer Treatment Group N9831 intergroup adjustment trial. J Clin Oncol. 2006;24(19):3032–3038. 13. Choritz H, Busche G, Kreipe H; Study Group HER2 Monitor. Quality assessment of HER2 testing by monitoring of positivity rates. Virchows Arch. 2011;459(3):283–289. 14. Fox SB, Kumarasinghe MP, Armes JE, et al. Gastric HER2 testing study (GaTHER): An evaluation of gastric/gastroesophageal junction cancer testing accuracy in Australia. Am J Surg Pathol. 2012;36(4):577–582. 15. Terry J, Torlakovic EE, Garratt J, et al. Implementation of a Canadian external quality assurance program for breast cancer biomarkers: an initiative of Canadian quality control immunohistochemistry (cIQc) and Canadian association of pathologists (CAP) national standards committee/immunohistochemistry. Appl Immunohistochem Mol Morphol. 2009;17(5):375–382. 16. Dekker TJ, Borg S, Hooijer GK, et al. Determining sensitivity and specificity of HER2 testing in breast cancer using a tissue micro-array approach. Breast Cancer Res. 2012;14(3):R93. doi:10.1186/bcr3208. 17. Gown AM, Goldstein LC, Barry TS, et al. High concordance between immunohistochemistry and fluorescence in situ hybridization testing for HER2 status in breast cancer requires a normalized IHC scoring system. Mod Pathol. 2008;21(10):1271–1277. 18. Fassan M, Ludwig K, Castoro C, et al. Human epithelial growth factor receptor 2 (HER2) status in primary and metastatic esophagogastric junction adenocarcinomas. Hum Pathol. 2012;43(8):1206–1212. 19. Kim MA, Lee HJ, Yang HK, Bang YJ, Kim WH. Heterogeneous amplification of HER2/neu in primary lesions is responsible for the discordant HER2/neu status of primary and metastatic lesions in gastric carcinoma. Histopathology. 2011;59(5):822–831. 20. Wolff AC, Hammond ME, Schwartz JN, et al; American Society of Clinical Oncology/College of American Pathologists. American Society of Clinical Oncology/College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. Arch Pathol Lab Med. 2007;131(1):18–43. 21. Moatamed NA, Nanjangud G, Pucci R, et al. Effect of ischemic time, fixation time, and fixative type on HER2/neuimmunohistochemical and fluorescence in situ hybridization results in breast cancer. Am J Clin Pathol. 2011;136(5):754–761.
Interlaboratory Variation in HER2/neu Testing—Sheffield et al
