ANNALS O F CLINICAL AND LABORATORY SCIENCE, Vol. 28, No. 4 Copyright 1998, Institute for Clinical Science, Inc. Reproducibility of LSI HER-2/new SpectrumOrange and CEP 17 SpectrumGreen Dual Color Deoxyribonucleic Acid Probe Kit For Enum eration o f G ene Amplification in P araffin-em bedded Specim ens: A M ulticenter C linical Validation Study* SHAHLA MASOOD, M.D.,t MARILYN M. BUI, M.D., P h.d, JAR-FEE YUNG, Ph.D.,* HON FONG L. MARK, Ph.D., EDITH Y. WONG, M.S. J JILL M. BIRKMEIER, M.S.,! SU-JAN YANG, M.S.,f and PING HSU, Ph.D.! f University o f F brida Health Science Center, Jacksonville, FL 32206 and fm ercy Hospital and Medical Center Chicago, IL 60616. and Rhode Island Hospital and Brown University School o f Medicine, Providence, R I 02903 and fvysis, Downers Grove, IL 60515 ABSTRACT Overexpression and/or amplification of HER-2/new gene have been found to be prognostic and predictive in breast and other cancers. Fluorescence in situ hybridization (FISH) assay, with its sensitivity and specificity, can be a superior method of detection when its performance characteristics are demonstrated. A multicenter study was initiated to evaluate the reproducibility of the LSI HER-2/neu Spectrum- Orange and CEP 17 SpectrumGreen Dual Color DNA Probe for enumeration of both the HER-2/neu gene and chromosome 17 (signals) in interphase cells. Section slides were prepared from four cell lines (H, E, R, and N) with known ratios of the HER-2/neii to CEP 17 copy numbers (approximately H = 1.10, E = 1.70, R = 4.50, N = 9.0). The study variable was the ratios of the HER-2/neu to chromosome 17 copy numbers. * Send reprint requests to: Shahla Masood, M.D., Professor and Associate Chair of Department of Pathology, University o f Florida Health Science Center/Jacksonville, 655 W est Eighth Street, Jacksonville, FL 32206. 2 15 0091-7370/98/0700-0215 $02.25 Institute for Clinical Science, Inc.
2 1 6 MASOOD, BUI, YUNG, MARK, WONG, BIRKMEIER, YANG, AND HSU Reproducibility with respect to assay, site, lot, day and reader was evaluated at 3 centers. Out of 120 specimen slides, 100 percent were successfully assayed. There were no significant differences among: (1) four repeated assays of the same specimen (p = 0.99), (2) the four probe lots (p = 0.33), or (3) the four study days (p = 0.54). There were statistically significant, but not important differences among centers and between readers. The ratios of the HER-2/neu to chromosome 17 copy num bers were estimated with accuracy and precision; the mean ratios (and sd) for specimens, H, E, R, and N were 1.05 (0.06), 1.81 (0.12), 4.48 (0.28), and 8.60 (1.23), respectively. In summary, assays with the LSI HER-2/new and CEP 17 Dual Color DNA Probe Kit, conducted at three sites by 6 different technicians, over 8 assay days, using kits from four lots, were performed with a high success rate in paraffin-embedded specimens. The signal enumeration was also accurate and precise. This study demonstrated that the results obtained by using the LSI HER-2/neu SpectrumOrange and CEP 17 SpectrumGreen Dual Color DNA Probe Kit are reliable and reproducible. Introduction The HER-2 /neu or C-erb-B2 oncogene codes for a 185-Kd transmembrane oncoprotein (pl85). This HER-2/new m em brane receptor protein is partially homologous to the epidermal growth factor receptor.1 Approximately 25 to 30 percent of breast carcinomas have been shown to have amplification of this gene.2 Amplification and overexpression of this gene were also found to be associated with rapid proliferation,3 low estrogen receptor content,1 and high tumor grade,4 which indicates its potential role in the progression of breast cancer. Therefore, it is important to investigate the prognostic value of the HER- 2/neu gene in order to improve the care of breast cancer patients.5,6,7 Immunohistochemical staining has been a popular analytical technique for assessing the alteration of the HER-2/new oncogene, by detecting the increase of HER-2/neu protein content (over-expression). Compared to this indirect method of studying gene amplification, fluorescence In situ hybridization (FISH) technique is a more specific approach. By FISH, a fluorescence-labeled HER-2 /neu probe will hybridize to its target sequence in the nuclei of breast cancer cells. The increase of HER-2/Wm gene copy numbers can be detected by an increase in the number of fluorescence signals. In addition, FISH can quantitatively assess the gene amplification in tumor cells while the morphology of the tissue is maintained. Owing to the obvious advantages of FISH, it is rational to adopt this method to study the prognostic markers of breast cancer. There are only a few commercial kits available for the detection of the HER-2/neM gene using FISH. As a pilot study of the amplification and overexpression of a panel of biomarkers in atypical hyperplasia, ductal carcinoma in situ, and invasive breast carcinomas, this study was designed to evaluate the reproducibility performance of the Vysis LSI HER-2/ neu and CEP 17 Dual Color Probe Kit*, which quantifies the HER-2/new gene amplification using paraffin-embedded breast tissue. Performance characteristics such as reproducibility with respect to different days, sites, lots, assays, and observers were the study variables. Materials and Methods V ysis L S I H E R -2/,\' (/ S p e c t r u m O r a n g e a n d C E P 17 S p e c t r u m G r e e n D u a l C o lo r DNA P r o b e K it The probes are fluorescent-labeled nucleic acid fragments for use in the FISH assay of tissue sections fixed on slides. The LSI HER- 2/neu probe contains DNA sequences specific for the HER-2/ne«human gene locus and hybridizes to 1 7 q ll.2-ql2 region of human * Vysis, Inc., Downers Grove, IL 60515.
REPRODUCIBILITY OF LSI HER-2/NEU SPECTRUMORANGE AND CEP17 SPECTRUMGREEN 2 1 7 chromosome 17. The CEP 17 probe contains alpha-satellite DNA that hybridizes to the D17Z1 locus. These two probes were premixed and pre-denatured in hybridization buffer for ease of use. The CEP 17 probe was used as a control to determine copy number of chromosome 17 in order to adjust for the effects of aneuploid chromosome 17 when the HER-2/neu copy numbers were counted. The ratios of average copy numbers per cell were calculated to establish the presence of amplified HER-2/neu. The LSI HER-2/neu and CEP 17 probe kit from four manufactured lots were supplied to each site for this study. Each kit included the following reagents necessary to perform 20 assays. (1) LSI HER-2/neu SpectrumOrange and CEP 17 SpectrumGreen probe mix, (2) DAPI counterstain, (3) NP-40, (4) 20 x SSC. St u d y S p e c im e n s Human breast tumor cell lines with normal and amplified HER-2/new gene copy numbers were embedded in paraffin block. Sections (4 to 6 microns) from the specimens with four different ratios of HER-2/ne«to chromosome 17 were prepared and placed on microscope slides. The four ratios were 1.0 to 1.2, 1.6 to 2.0, 3-5, and 7 to 11 as determined by FISH (table I). Study Sample3 TABLE i Four Designated Ratios of HER-2/neuto CEP 17 Amplification Ratio H 1.0-1.2 E 1.6-2.0 R 3.0-5.0 N 7.0-11.0 U Unknown asampies were from breast cancer cell lines. Sample U was for randomization control. St u d y O b jec tiv es a n d E x p e r im e n t a l D e s ig n In order to investigate the reliability and reproducibility, the following performance characteristics of the LSI HER-2/new and C EP 17 probe kit w ere tested: (1) the b etw een-assay rep ro d u cib ility, (2) the between-site reproducibility, (3) the betweenlot reproducibility, (4) the between-day reproducibility, and (5) the betw een-observer reproducibility. The experimental design for week 1 was a simple repetition design (table II), and for week 2 was a 4 x 4 Latin Square Design (table III). E x p e r im e n t a l M e t h o d s (1) Paraffin pretreatment: Slides were deparaffinized in Xylene, followed by dehydration with absolute ethanol. HCl treatment and protease treatment were applied to the slides before they were fixed in neutral buffered formalin. The necessary reagents were supplied separately in the Vysis Paraffin P retreatment Kit. (2) LSI HER-2/neu and CEP 17 assay. After pretreatment, slides were treated in denaturation solution (70 percent formamide/2x SSC) at 72 C. Slides were dehydrated in serial ethanol solutions (70 percent, 85 percent, and 100 percent, consecutively), then dried on a 45-50 C slide warmer. Ten microliters of HER- 2/neu probe mix was applied to the prewarmed slides and hybridized at 37 C for 14 to 18 hrs. Post-hybridization wash was then applied using 2x SSC/0.3 percent NP-40 at 73 C for 2 TABLE II The Simple Repetition Design for Week 1 Study Replication Day 1 Day 2 Day 3 Day 4 1 N1 R1 E1 H1 2 N2 R2 E2 H2 3 N3 R3 E3 H3 4 N4 R4 E4 H4 # U U U U
2 1 8 MASOOD, BUI, YUNG, MARK, WONG, BIRKMEIER, YANG, AND HSU TABLE III The Latin Square Design for Week 2 Study Day 1 Day 2 Day 3 Day 4 1 N R E H 2 R E H N 3 E H N R 4 H N R E # U U U U Four lots were used; one lot was used on one specimen each day. minutes. Ten microliters of DAPI I counterstain (4,6 diamidino-2-phenylinodole) was applied to the target area of the slide. T e st A nalysis (1) Fluorescence microscopy: Enumeration of HER-2/new and chromosome 17 was viewed under fluorescence microscopy with a 100 watt mercury lamp at 630x or looox magnification. Dual and triple band pass filters were utilized for the detection and enumeration of the HER-2/new probe signal, CEP 17 probe signal, and the DAPI counterstain. The two colors of fluorescent signals (orange for the HER-2/new gene and green for chromosome 17) stood out brightly against the general blue fluorescence of the nuclear DNA provided by DAPI counterstain, making the counting of probe signals unambiguous. (2) Rules of signal scoring: Only the nuclei that had discrete signals were enumerated according to Vysis guidelines (figure 1). One hundred nuclei were counted per slide. (3) Computer images: FISH images were processed at looox magnification utilizing an Olympus MX60 fluorescence microscope with a 100 watt mercury lamp. Separate narrow band pass filters were used for the detection of the HER-2/neu probe signal (SpectrumOrange), CEP 17 probe signal (SpectrumGreen), and DAPI counterstain. Fluorochrome signals were captured individually and images were generated via com puter digitization and pseudo-colorization processing with a CytoVison System and CytoProbe Software.* Completed original images were exported to a graphic.tif format. The images were printed through a laser printer. (4) Statistical methods: The ratio of HER- 2/neu to chromosome 17 copy number was the primary variable of analysis. A ratio greater than 1.5 was considered as positive for HER- 2/neu gene amplification. The intra-assay variation (precision) for each specimen was determined by pooling the individual variance across all six observers at the three sites. For assessment of inter-assay variation the mean, standard deviation, and the coefficient of variation for the ratio across all assay days, each kit lot, for each site and for each observer were calculated for each specimen. Correlation and regression analysis techniques w e re u se d to an aly ze th e b e tw e e n - observer reproducibility. The ratio of HER-2/new gene to chromosome 17 copy number was also analyzed by the Analysis of Variance technique using models appropriate for a multi-center simple repetition design and for a multi-center 4 x 4 Latin Square Design. In these analyses, the technologist who had the smaller overall mean ratio was designated as observer #1 and the other technologist, observer #2, for all sites. In this manner, the worst case observerobserver variation was analyzed. Results Info rm ativeness This study was conducted at three different sites in two consecutive weeks. It was initiated at the first site on August 11, 1997 and closed at the last site on November 7, 1997. At each site, two independent technologists performed the assays following Vysis Protocol 300. Five slides were processed per day. A total of 40 slides were processed at each institution. There were only two assay failures in the 120 * Applied Imaging, Inc.
REPRODUCIBILITY OF LSI HER-2/NEU SPECTRUMORANGE AND CEP17 SPECTRUMGREEN 2 1 9 1 ~ s. Nuclei are overlapping and all areas of both of the j f ^ N \ nuclei are not visible but signals are not in j \0 @ J o J overlapping area. Count as two orange and two j ------- ^ --------' green in each nucleus. 1! 2 ^ -------- n. Count as two orange signals and two green signals. / \ One orange signal is diffuse.! ( ^ J j ) 3 ^ ----- ys' X Don t count. Nuclei are overlapping, all areas of i /^~~j O / j nuclei are not visible and some signals are in / o \ \ overlapping area. ( O QJOoZUt 0 4 s 0 -------- s. Count as two orange signals and two green signals. f \ One orange signal is split.. y 5 / ---- O o " ^ \ Count as one orange signal and two green signals. f A One green signal is split and the orange signal is y s p l i I ' 6 O s' ~ s. Count as two orange signals and one green signal. 7 \ Count as three orange signals and one sreen signal. ( 0 )!! 8 ^ Count as four orange signals and two green signals, j n 0 ^! V 6 * V! F ig u r e 1. Signal counting guide. Key: O = green probe; = orange probe. slides assayed. In one case, tissue section fell off the slide and in the other case, the signals faded. Hybridization of both replacement slides was successful. A c c u r a c y Based on a cutoff ratio of 1.5 for defining HER-2/new amplification, none of the slides in this study were misclassified, ie, the results of all slides within a range of 1.0-1.2 were <1.5, while the rest of the slides were >1.5. This was true for the results from each individual observer for all study specimens. Four classes of ratios were observed as shown in figure 2. R e p r o d u c ib il it y The results of statistical analysis for week 1 (intra-assay reproducibility) and week 2 (inter-
220 MASOOD, BUI, YUNG, MARK, WONG, BIRKMEIER, YANG, AND HSU 2A 2B F ig u r e s 2A and 2B. Observed ratios o f HER-2/neu to CEP 17. (A) 1.00 to 1.2; (B) 1.6 to 2.0.
REPRODUCIBILITY OF LSI HER-2/NEU SPECTRUMORANGE AND CEP17 SPECTRUMGREEN I 4T 2C F ig ur es 2C a n d 2D. Observed ratios o f H ER-2/neu to CEP 17. (C) 3.0 to 5.0; (D) 7.0 to 11.0.
2 2 2 MASOOD, BUI, YUNG, MARK, WONG, BIRKMEIER, YANG, AND HSU assay reproducibility) studies are summarized in table IV. There was no significant difference among the four probe lots (p = 0.3490) or among the four assay days (p = 0.5868). There was a significant difference between observers (p = 0.0072), reflecting the subjectivity involved in signal interpretation and enumeration (table IV). However, the 100 percent correct classification of slides as normal or amplified for HER-2 showed that this difference was not clinically important. There was variation between study sites (table V). Again, the 100 percent correct classification of slides as normal or amplified for HER-2/new showed that this difference was not clinically important. The results showed high inter-observer correlation. The correlation coefficient was 0.98, and the regression coefficient was 1.12, which is close to 1.0, indicating high reproducibility of signal enumeration between observers (figure 3). This high between-observer correlation indicated that determination of HER-2/ngw gene amplification with the HER-2/neu and CEP 17 assay requires only one trained, experienced technologist. Conclusions The purpose of this study was to establish the intra-assay reproducibility of the LST TABLE V Between-site Reproducibility of FISH Assay (Week 2) Ratios of HER-2/neu to CEP 17 Mean Site 1 Site 2 Site 3 1.0-1.2 1.08 1.01 1.07 1.6-2.0 1.81 1.71 1.78 3.0-5.0 4.39 3.65 4.49 7.0-11.0 7.21 8.26 8.23 HER-2/neu and CEP 17 probe kit (assay-toassay repeatability), and to establish the interassay reproducibility with respect to probe lot, day of assay, institution, and observer. Excellent reproducibility was observed, indicating that the assay method and enumeration guide were clear and easy to follow. A close to 100 percent overall success rate of hybridization at first try dem onstrated the consistency in manufacture quality, and the reliability of the kit and procedure. The results of this study demonstrated that enumerating a maximum of 100 nuclei per slide is adequate for the quantitation of different levels of HER-2/neM gene amplification in paraffin-embedded human breast tissue speci- Study Objective TABLE IV Summary of Results from Week 1 and Week 2 p-vaiue (From Analyses of Variances on HER-2/neu to CEP 17 Ratios) Week 1 Week 2 Reproducibility Between-day (4 days) Not applicable 0.5868 Reproducibility: Between-site (3 sites) < 0.0001 0.0630 Reproducibility: Between-lot (4 probe lots) Not applicable 0.3490 Reproducibility: Between-assay (4 assays) 0.9832 Not applicable Reproducibility: Between-observer (2 observers at each site) 0.0148 0.0072
REPRODUCIBILITY OF LSI HER-2/NEU SPECTRUMORANGE AND CEP17 SPECTRUMGREEN 2 2 3 mens. One trained, experienced technologist is needed to perform the assay and enumeration. This is a feasible procedure that can be followed w ith quality assurance in a p ra c tical setting. A cknow ledgm ent R atio o fh E R -2/neu copy to CEP 17 by O bserver #1 F i g u r e 3. Correlation o f ratio of HER-2/neu copy number to CEP 17 between two observers (week 1). The authors appreciate the generosity o f Dr. Robert Zori and Mr. Brain Gray at the Cytogenetics Lab of the Division o f Genetics, Department of Pediatrics, College of Medicine, University of Florida for their help in capturing the images in figure 2. The technical support of the staff of the University Medical Center in Jacksonville, the Mercy Hospital and Medical Center in Chicago, and the Lifespan A cadem ic M edical C enter C ytogenetics laboratory at R hode Island H ospital in Providence is gratefully acknowledged. mens. One additional advantage of this assay is that the slides can be easily recounted later (up to three months), if they are kept in a freezer. The high between-observer correlation indicated that the determination of HER-2/neu gene amplification with the LST HER-2/neu and CEP 17 assay requires only one trained, experienced technologist. The analysis of recount data also indicated that a discrepant result between observers was an indication for repeat analysis and that recount was an effective quality control measure. In conclusion, the LST HER-2/new and CEP 17 Dual Color DNA Probe Kit is of high quality. It provides reliable quantitative information on HER-2/neu gene amplification in paraffin-embedded human breast tissue speci- TABLE VI The Precision of the Observed Quantitation of HER-2/neu Amplification (Week 1) Ratios of HER-2/neu to CEP 17 Mean Standard Deviation Coefficient of Variation (%) 1.0-1.2 1.05 0.03 3.04 1.6-2.0 1.81 0.08 4.29 3.0-5.0 4.47 0.25 5.53 7.0-11.0 8.59 0.32 3.68 References 1. Craft PS, Harris AL. Clinical prognostic significance of tumor angiogenesis and breast cancer. Breast Cancer 1994;8:51. 2. Ciocca DR, Fujimura FK, Tandon AK, Clark GM, Mark C, Lee-Chen G-J, Pounds GW, Vendely P, Owens MA, Pandian MR, McGuire WL. Correlation o f HER-2/neu amplification with expression and with other prognostic factors in 1103 breast cancers. J of the National Cancer Institute 1992;84:1279-82. 3. Kerns B-JM, Jordan PA, Huper G, Marks JR, Iglehart JD, Layfield LJ. Assessment o f c-erbb-2 amplification by im m unohistochem istry in paraffin-em bedded breast cancer. Modem Pathology 1993;6:673-8. 4. Slamon DJ, Godolphin W, Jones LA, Holt JA, Wong SG, Keith D E, Levin WJ, Stuart SG, Udove J, Ullrich A, Press MF. Studies o f the H ER -2/neu protooncogene in human breast and ovarian cancer. Science 1989;244:707-12. 5. Allred DC, Clark GM, Molina R, Tandon AK, Schnitt SJ, Gilchrist KW, Osborne CK, Tormey DC, McGuire WL. Overexpression of HER-2/neu and its relationship with other prognostic factors change during the progression o f In Situ to invasive breast cancer. Human Pathology 1992;23:974-9. 6. Allred DC, Clark GM, Tandon AK, Molina R, Tormey DC, Osborne CK, Gilchrist KW, Mansour EG, Abeloff M, Eudey L, McGuire WL. HER-2/neu in nodenegative breast cancer: prognostic significance of overexpression influenced by the presence o f in situ carcinoma. J Clin Oncology 1992;10:599-605. 7. Muss HB, Thor AD, Berry DA, Kute T, Liu ET, Koemer F, Cirrincione CT, Budman DR, W ood WC, Barcos M, Henderson IC. C-erbB-2 expression and response to adjuvant therapy in women with nodepositive early breast cancer. N Eng J Med 1994;330: 1260-6.