A novel assessment of the quality of immunohistostaining overcomes the limitations of current methods

Transcription

1 ARTICLE IN PRESS Pathology Research and Practice 204 (2008) ORIGINAL ARTICLE A novel assessment of the quality of immunohistostaining overcomes the limitations of current methods Avi Eisenthal, Leonor Trejo, Alexander Shtabsky, Faina Bedny, Eli Brazowski Pathology Institute, Tel-Aviv Sourasky Medical Center, 6 Weizmann Street, Tel-Aviv 64239, Israel Received 16 August 2007; accepted 18 December 2007 Abstract Quality assurance has become an integral part of surgical pathology. Despite the development of interdisciplinary quality systems, however, the means for objective analysis in surgical pathology are limited. Immunohistostaining is a multi-factorial procedure that depends on the quality of reagents and antibodies employed in the process and on technical methodology. In the present study, we aim to establish a straightforward procedure for objective quality evaluation of the components involved in immunohistostaining. The quality of two of these components, the primary antibody and the automated staining device, was assessed by employing each component from two different sources, one serving as the test substance and the second as the reference. Assessment was performed by at least two pathologists in a blinded fashion using pre-established quality criteria and scores. The quality analysis of two automated devices revealed a significant difference between the reference and tested devices ( and , respectively, P40.05), while the analysis of two selected antibodies did not reveal any statistical difference. The described method provided objective quality assessment of selected components affecting immunohistostaining by elaborating numeric values that enabled statistical analysis. This approach is applicable to any given component in various surgical pathology procedures. r 2007 Elsevier GmbH. All rights reserved. Keywords: Quality control; Immunohistostaining; Automated device Introduction Quality assurance has become an integral part of surgical pathology as witnessed by the introduction of laboratory systems, the reporting of quality control, and the development of interdisciplinary quality systems. One of the important tools of surgical pathology is the process of immunohistostaining, which consists of the interaction between antibodies and antigens expressed in specific cells and tissues [6]. This multi-factorial procedure depends on a variety of independent components, each affecting the intensity and specificity of staining. Corresponding author. Tel.: ; fax: address: eisenthal@tasmc.health.gov.il (A. Eisenthal). These components include the technical methodology, technical skills of the person performing the staining, quality of reagents and antibodies, as well as the reliability of the automated devices employed in the process. In addition, determination of the quality of immunohistostaining carried out by expert pathologists suffers from obvious subjective limitations which stem from the absence of numeric values produced during the immunostaining procedures which could facilitate the establishment of accepted quality standards. In this report, we describe a straightforward method that essentially overcomes such limitations. The approach consists of comparisons between two different sources for each of the component tested, and it enabled us to achieve a relatively objective determination of quality /$ - see front matter r 2007 Elsevier GmbH. All rights reserved. doi: /j.prp

2 324 ARTICLE IN PRESS A. Eisenthal et al. / Pathology Research and Practice 204 (2008) for each tested component. This approach may be applied to any component affecting the quality of immunohistostaining, as well as to other procedures employed in surgical pathology. Materials and methods Two different components, each affecting the quality of the immunohistostaining procedure, were analyzed separately. The first was the automated immunostaining device, i.e., the ES was compared to the Benchmark XT, both produced by Ventana (Arizona, USA). The second analysis was that of antibodies with similar specificities from two different sources, with each of the antibodies being employed at pre-established optimal dilutions and pretreatment regimens. Control tissues Formalin-fixed, paraffin-embedded tissues positively stained by each of the antibodies were pre-selected by the pathologist. Two identical 3 5 mm sections were cut from each block by the same technician (designated Z1) using an automated microtome (RM 2055, Leica, Germany). Antibodies Antibodies and antigen retrieval conditions are described in Table 1. Automated immunostaining procedure This procedure was performed according to the manufacturer s protocol using the Ventana-automated ES and Benchmark XT devices [1]. The XT instrument processes the slides from baking through counterstain. In addition, all slides are held on an individual thermopad. The thermopad temperature is controlled individually, so that each slide can be processed at a different temperature and time of incubation. The XT is controlled by the NexES software; thus, all pieces of information regarding reagents, protocols, and runs are saved on the computer and can be used for reporting and tracking. The ES runs Immunostaining on slides that have been deparaffinazed and heat-treated outside the instrument. The ES runs all the slides at 37 1C. The data obtained are stored on a floppy disk. The protocol used in both devices included the following steps: antigen retrieval by short exposure to heat in the presence of buffers, reduction of nonspecific immunoglobulin binding to tissues by pre-incubation with human serum albumin, and neutralization of endogenic peroxidase activity following the exposure to peroxide (H 2 O 2 ). After incubation with the primary antibodies (listed above), a second biotinylated-labeled antibody directed against the first antibody was applied, followed by exposure to avidin peroxidase complex and thereafter diaminobenzidine as the substrate to the peroxidase. Finally, the tissues were incubated with hematoxylin, which served as a counterstain. The entire process described above was completed within 4.5 h. Parameters selected for evaluation Two components of the immunostaining procedure were selected and assessed in two separate tests. In the first test, the quality of anti-vimentin antibodies produced by two different manufactures (Dako and Lab Vision) was compared. In the second test, the quality of two automated device (ES and Benchmark XT both from Venatna) was compared. All components of the immunostaining procedure, excluding the analyzed component, were identical in each of the tests. Preparation and evaluation of immunostained tissue sections Selected paraffin-embedded tissues that were known as being positively stained for the applied antibodies Table 1. List of antibodies and antigen retrieval reagents Antibody Manufacture (Catalog no.) Dilution Antigen retrieval procedure Synaptophysin [1] Zymed, USA (180130) 1:50 Citrate buffer ph 6.0, Zymed, USA (Catalog no ) microwave: 97 1C 12 min Vimentin [2] Dako, Denmark (M0725) 1:400 Citrate buffer ph 6.0 microwave: 97 1C 12 min VEGF [3] Zymed, USA (180254) 1:50 Citrate buffer ph 6.0 microwave: 97 1C 12 min CD117/c-kit [4] Dako, Denmark (M7140) 1:50 Without pretreatment Synaptophysin Lab Vision (9111-S0) 1:100 CC1 buffer, Ventana, USA (Catalog no. V ) microwave: 97 1C 8 min Vimentin Lab Vision (9120-S0) 1:100 CC1 buffer microwave: 97 1C 8 min VEGF Lab Vision (9128-R1) 1:100 CC1 microwave: 97 1C 8 min CD117/c-kit Lab Vision (9038-S0) 1:100 Without pretreatment

3 ARTICLE IN PRESS A. Eisenthal et al. / Pathology Research and Practice 204 (2008) were cut by a lab technician (designated as Z1) in four consecutive sections, and two of each were placed on one slide. All slides were then numbered arbitrarily by the senior lab technician (designated Z2), and these numbers were documented (Tables 2 and 5). The slides were then analyzed by two or three pathologists (designated as X1, X2, and X3), and the obtained data were recorded (Tables 3 and 6). The data obtained from the pathologists were then transferred to the senior lab technician (designated as Z2) for summarizing and documentation (Tables 4 and 7). Quality assessment and scoring of stained tissue The quality of each immunostained tissue section was evaluated by pre-defined criteria, which included the specificity, homogeneity and the intensity of staining compared to an unstained background. Quality was scored from 1 to 5, where 1 represents the lowest and 5 represents the highest quality level. Statistical analysis Statistical analysis was performed by a 2-tailed Student s t-test. Results Comparison between the quality of two anti-vimentin antibodies from different companies In the first test, we compared the quality of antivimentin antibody from Lab Vision (USA) with a routinely used reference antibody from Dako (Denmark). Paraffin-embedded tissues were cut by the same technician (designated as Z1) and placed on duplicate slides as described in the Materials and methods section. The slides were then stained by the Benchmark XT-automated device under previously established, optimal conditions for the pretreatment step and antibody dilution. The slides were then numbered arbitrarily as shown in Table 2. Pathologist quality scoring of anti-vimentin staining Slides with immunostained tissues were sent separately to three different pathologists who also received the data sheet illustrated in Table 3. Each pathologist evaluated the staining independently and recorded its quality. Table 3 shows the results of one of three pathologists (designated as X1). Summary of quality evaluation of three pathologists Data sheets with the quality scores of anti-vimentin immunostaining as determined by three pathologists were sent to a senior technician. The data were summarized on the data sheet after the codes for each slide were opened (Table 4). Although there were some differences in the quality scoring between pathologists X1, X2 ( ), and X3 ( ) regarding the evaluation of the tested antibody (4.8 and 3.8, respectively), analysis of all the scores for the two antibodies revealed that the differences in the mean scoring of the three pathologists ( and for the reference and tested antibodies, respectively) were not significant (P ¼.27). Comparison between the quality of two automated immunostaining devices In the second evaluation, we compared the quality of two automated immunostaining devices, the ES and the Benchmark XT, both manufactured by Ventana. Paraffin-embedded tissues were cut by the same technician (Z1) from two different tissue-embedded blocks (Table 5) for each of the four antibodies employed (synaptophysin, vimentin, VEGF and c-kit) and placed on duplicate Table 2. Data sheet describing the preparation of slides employed in the blinded quality assessment Prepared by: Z1 Role: Lab technician Date: 25/1/07 Signature: Antibody Block no. a Slide no. Reference antibody Dako (M0725) Tested antibody Lab Vision (9120-S0) Vimentin C 2A Vimentin E 1C Vimentin B 2F Vimentin K 3D Vimentin F 4A a Matched arbitrary number to the original block number to avoid patient identification.

4 326 ARTICLE IN PRESS A. Eisenthal et al. / Pathology Research and Practice 204 (2008) slides as described in the Materials and methods section. One set of slides was then stained using either the ES or the Benchmark XT-automated device under previously determined optimal conditions. The slides were then numbered arbitrarily (Table 5). Pathologists quality scoring of staining performed on the ES and Benchmark XT-automated devices Arbitrarily numbered slides with immunostained tissues were sent to two different pathologists. Each pathologist evaluated and documented the staining Table 3. staining Data sheet with quality scoring of anti-vimentin Parameter assessed: antibody employed in immunostaining Quality assessed by: X1 MD License no: XXXXX Role: Senior pathologist Signature: Date: 28/1/07 Slide no. Antibody Quality scoring C Vimentin + 2A Vimentin + 7E Vimentin + 1C Vimentin + 5B Vimentin + 2F Vimentin + 4K Vimentin + 3D Vimentin + 3F Vimentin + 4A Vimentin + independently and was unaware of the other pathologist s scores. Table 6 illustrates the scoring obtained from one of two pathologists and includes both the score for each of three quality criteria and the final score. Summary of quality evaluation of two pathologists The data sheets of two pathologists that contained the final quality scores for the staining of the four antibodies carried out by the ES and the Benchmark XT-automated devices were sent to a senior technician. The data were summarized using the data sheet shown in Table 6 after opening the codes for each slide (Table 7). The scores of two pathologists did not differ significantly: they were and for the reference device (ES) and for the tested device (XT), respectively, as evaluated by the first pathologist (X1), and and for the ES and XT devices, respectively, as evaluated by the second pathologist (X2). The mean scores of the two pathologists for the XT device were significantly higher than those for the ES device ( and , respectively, P ¼.047), making it superior for routine application. Discussion The absence of built-in numerical values in current methods of evaluation hampers efforts to establish objective standard values for quality in surgical pathology. In the present study, we describe an approach for minimizing such limitations. This approach consists of several guidelines: (1) the quality of a given component of the immunostaining procedure is established by Table 4. Summary of three pathologists final immunostaining quality scores Slide no. Reference Tested Pathologist X1 X2 X3 X1 X2 X Mean of each pathologist s score Mean7s of 3 pathologists scores P-value (tested vs. reference) P ¼.27 (not significant) Based on pathologists evaluation tested anti-vimentin antibody approved for routine application: Yes: V No: Performed and approved by: Z2-PhD Date: 30/1/07 Signature:

5 ARTICLE IN PRESS A. Eisenthal et al. / Pathology Research and Practice 204 (2008) Table 5. Data sheet describing the preparation of slides employed in the blinded quality assessment 1. Reference device Manufacture: Ventana Model: ES SN: Tested device Manufacture: Ventana Model: Benchmark XT SN: Prepared by: Z1 Role: Lab technician Date: 13/11/06 Signature: Antibody Block no. Slide no. Reference device Tested device Synaptophysin 3232; A 2B Vimentin 1414; D 1K VEGF 5252; F 5E CD117/c-kit 1717; C 3D Table 6. Data sheet with quality scoring of anti-vimentin staining including quality criteria Parameter assessed: Antibody employed in immunostaining Quality assessed by: X1 MD License no: Y1 Role: Senior pathologist Signature: Date: 15/11/06 Score value (1 5) Final score (1 5) Slide no. Antibody Intensity Specificity Homogeneity Synaptophysin 2B Vimentin 4D Vimentin 1K Synaptophysin 2A VEGF 3F CD117/c-kit 3D VEGF 5E CD117/c-kit 5C Table 7. Summary of two pathologists final immunostaining quality scores Device Reference (ES) Tested (Benchmark XT) Pathologist X1 X2 X1 X2 Antibody Synaptophysin Vimentin VEGF CD117/c-kit Mean of each pathologist s score Mean7s of 2 pathologists scores P-Value (tested vs. reference).047 (n ¼ 8) Based on pathologists evaluation of Benchmark XT device approved for routine application: Yes: V No: Performed and approved by: Z2-PhD Date: 16/11/06 Signature:

6 328 ARTICLE IN PRESS A. Eisenthal et al. / Pathology Research and Practice 204 (2008) comparison between two different sources of the same component (e.g., antibodies sharing the same antigenic specificity from two different manufacturers); (2) analysis is performed for a single component, while other components of the procedure remain identical; (3) each stained slide is marked arbitrarily to enable anonymous evaluation; (4) criteria for quality and score levels are pre-established [1,3 6]; (5) each source of a given component is tested for at least four or five different cases (i.e., several different antibodies are used to evaluate two automated staining devices), thus enabling analysis of the difference between the sources with statistical significance; (6) the evaluation is performed independently by at least two different pathologists. Using these guidelines, we were able to produce numeric values in each of the two quality evaluation tests and reached levels of statistical significance for the differences between them. In the first example, we demonstrated the quality evaluation of anti-vimentin antibody from a new manufacture (Lab Vision, 9120-S0) compared to a reference antibody (Dako, M-0725). All other components remained unchanged for both of the tested antibodies: specifically, the automated device (Benchmark XT), the reagents (Ventana pretreatment reagents and detection kit), the tissue paraffin-embedded block which stained positive with the antibody, and the lab technician who performed the staining. There were minimal differences in the scoring of the three pathologists, with the exception of the mean score of the third pathologist for the tested antibody (Table 4). There was no significant difference between the tested and the reference antibody (P ¼.27) when the mean scores of the three pathologists were calculated together, thus demonstrating the suitability of the tested antibody for routine work. In the second example, we compared the staining quality of two automated devices, the ES and the Benchmark XT, and showed similar mean score values for them both. The mean scores for pathologists X1 and X2 were and , respectively, for the ES device, and and , respectively, for the Benchmark XT device. The mean score for the ES device was 19% higher for pathologist X1 compared to pathologist X2, whereas the mean score for the Benchmark device was 13% higher for pathologist X1 compared to pathologist X2. The mean combined scores of both pathologists (Table 7) were and for the ES and the Benchmark XT devices, respectively (P ¼.047). These results establish the basis for accepting the Benchmark XT device as more suitable for immunhistostaining. In the present study, we described a straightforward approach for establishing a relatively objective quality evaluation of immunohistostaining. This approach is based on the creation of numeric values that facilitate analysis of statistical significance of the differences between the tested parameters. It differs from a method described by others [2] who employed tissue arrays as a quality assurance tool for diagnostic immunohistochemistry: in ours, only one among several variables are tested separately, while other variables remain identical. Based on the guidelines described for quality evaluation, this approach could be applied for the evaluation of any given component in various surgical pathology procedures. Acknowledgment Esther Eshkol is thanked for editorial assistance. References [1] M. Bar-Shira Maymon, H. Yavetz, L. Yogev, et al., Detection of calretinin expression in abnormal immature Sertoli cells in non-obstructive azoospermia, Acta Histochem. 107 (2005) [2] F.D. Hsu, T.O. Nielsen, A. Alkushi, et al., Tissue microarrays are an effective quality assurance tool for diagnostic immunohistochemistry, Mod. Pathol. 15 (2002) [3] D.W. Leung, Vascular endothelial growth factor is a secreted angiogenic mitogen, Science 246 (1989) [4] Y. Tsutsumi, H. Kubo, Immunohistochemistry of desmin and vimentin in smooth muscle tumors of the digestive tract, Acta Pathol. Jpn. 38 (1988) [5] Y. Tsuura, H. Hiraki, K. Watanabe, et al., Preferential localization of c-kit product in tissue mast cells, basal cells of skin, epithelial cells of breast, small cell lung carcinoma and seminoma/dysgerminoma in human: immunohistochemical study of formalin-fixed, paraffin-embedded tissues, Virch. Arch. 424 (1994) [6] B. Wiedenmann, K. Schwechheimer, R. Waldherr, et al., Synaptophysin identified in metastases of neuroendocrine tumors by immunocytochemistry and immunoblotting, Am. J. Clin. Pathol. 87 (1987)