Sensitivity and Specificity of Antibodies on Necrotic Tumor Tissue

Transcription

1 Anatomic Pathology / ORIGINAL ARTICLE Sensitivity and Specificity of Antibodies on Necrotic Tumor Tissue Alexander R. Judkins, MD, Kathleen and Matt van de Rijn, MD, PhD T. Montone, MD, Virginia A. LiVolsi, M D, Key Words: Necrosis; Immunohistochemistry; Cytokeratins; S100; LCA Immunohistochemistry occasionally is used to determine the lineage of entirely necrotic tumors. However, the sensitivity and specificity of antibodies on necrotic tissue are unknown. To determine the usefulness of immunohistochemistry with necrotic lesions, a series of 24 known tumors consisting of 14 carcinomas, 2 lymphomas, 2 melanomas, and 6 sarcomas (all with extensive necrosis) was examined for reactivity with 6 cytokeratin antibodies, SI00, and LCA. Carcinomas stained positively with at least 1 cytokeratin antibody in 78% of the cases. The cytokeratin antibodies with the highest sensitivity were AE1, AE1/3, S903, and PANCK. These antibodies also retained specificity for epithelial differentiation; no reactivity was observed in the 10 necrotic nonepithelial tumors. LCA retained its reactivity with necrotic lymphoma, but SI 00 reacted with only one third of the necrotic lesions. Unexpectedly, reactivity for LCA and S100 occurred in some necrotic carcinomas. Keratin markers can be used on necrotic tissue to determine epithelial differentiation, but the results obtained with S100 and LCA on necrotic tissue should be interpreted with caution. Immunohistochemistry occasionally is used in an attempt to determine the lineage of entirely necrotic tumor masses or in necrotic lymph nodes suspected of containing metastatic tumor. It is well recognized that infarcted lymph nodes can be the first indication of malignant lymphoma.13 They may also be the first diagnostic tissue seen in cases of metastatic carcinoma. When pathologists are confronted with a necrotic tumor mass, immunohistochemical studies may be useful to determine the lineage of the necrotic lesion. Several studies of antigen preservation in necrotic tissue have been undertaken and support the use of immunohistochemical analysis on necrotic tissue.46 However, 2 of the studies focus on postmortem autolysis or its experimental equivalent.4,6 Only 1 report directly addressed the issue of in situ tumor necrosis; however, that study was limited to a consideration of lymphoid neoplasms.5 Here we investigate the sensitivity and specificity of immunohistochemical stains on necrotic tumor tissue, with an emphasis on the antibodies most frequently used to distinguish carcinomas, lymphomas, and melanomas. Materials and Methods The surgical pathology files of the Department of Pathology and Laboratory Medicine at the University of Pennsylvania were searched for cases in which a significant portion of tumor had undergone necrosis. A total of 24 cases were identified: 14 carcinomas, 2 lymphomas, 2 melanomas, and 6 sarcomas. Cases were included if they contained both viable tumor sufficient for diagnosis and large areas of tumor necrosis. Alternatively, cases of entirely necrotic metastatic tumor masses were included if the patient had a prior, confirmed diagnosis of malignancy of known histogenesis. The use of archival tissue blocks in this study was approved by the University of Pennsylvania's Institutional Review Board. Downloaded from by guest on 09 October 2018 Abstract Am J Clin Pathol 1998; 110:

2 Judkins et al / IMMUNOHISTOCHEMISTRY ON NECROTIC TUMORS Table II Panel of Antibodies and Sources Antibody Specificity AE1 AE1/3 34pEl2(S903) DEK13 PANCK Ks208 (CK20) S100 PD7/2B11 (LCA) Dilution Keratins 10, 14, 15, Keratins 18, 10, Keratins 1, 5, 10, 14 Keratins 10, 13 Keratins 5, 6, 8, 18 Keratin 20 S100 protein (polyclc)na ) Panlymphocyte marker, LCA Source , Boehringer Mannheim (Indianapolis, Ind) Boehringer Mannheim (Indianapolis, Ind) Enzo Diagnostics, Inc (Farmingdale, NY) Novocastra Laboratories (Burlingame, Calif) Novocastra Laboratories (Burlingame, Calif) DAKO (Carpinteria, Calif) DAKO (Carpinteria, Calif) DAKO (Carpinteria, Calif) Cytokeratin Antibodies Case Carcinoma b G Lymphoma Melanoma Sarcoma Site Diagnosis AE1 AE1/3 S903 Neck nodule Breast GE junction Met. Met. adenocarcinoma Met. NPCA Ductal CA Met. Adenocarcinoma Adenocarcinoma Met. adenocarcinoma PDCA PDCA Colon PANCK DEK13 Large Bcell lymphoma Tcell lymphoma Liver Soft tissue Metastatic melanoma Metastatic melanoma Esophagus Uterus Jejunum Stomach Soft tissue Gl stromal tumor Leiomyosarcoma Gl stromal tumor Gl stromal tumor MPNST Leiomyosarcoma S100 LCA CK20 Met. = metastatic; = squamous cell carcinoma; NPCA = nasopharyngeal carcinoma;cancer; CA =GE = gastroesophageal; PDCA = poorly differentiated carcinoma; Gl = gastrointestinal; MPNST = malignant peripheral nerve sheath tumor. *A11 specimens had appropriate reactivity with the antibodies in viable areas of tumor. Immunohistochemical analysis was performed on deparaffinized sections by using the Ventana/BIOTEK automated staining system (Ventana, Tucson, Ariz). The antibodies used in this study are listed and summarized in Table II. The antigen retrieval techniques used included the following: 0.01% trypsin digestion (AE1, AE 1/3, and PANCK); trypsin digestion followed by heatinduced epitope retrieval (CK20); and pronase digestion followed by heatinduced epitope retrieval (S903). After their endogenous peroxidase activity had been inhibited, the sections were incubated with the primary antibodies, washed, and incubated with a biotinylated antimouse secondary antibody. 642 AmJCIinPathol 1998;110: After incubation with avidinbiotin complex, the slides were developed with 3'diaminobenzidine serving as the chromogen. The slides were then counterstained with Gill's hematoxylin, dehydrated, and coverslipped. All stains were performed with appropriate positive and negative controls. Results The 24 cases selected for review were stained with the panel of immunohistochemical markers listed in Table 1. The results are summarized in liable 21. Fourteen carcinomas Downloaded from by guest on 09 October 2018 liable 21 Immunohistochemical Staining of Necrotic Tumor Tissue*

3 Anatomic Pathology / ORIGINAL ARTICLE The antigen SI00, which is associated with a variety of tumors,78 demonstrated poor preservation in necrotic tumor tissue. Only 1 of 2 melanomas demonstrated positive staining, in a cytoplasmic pattern, in necrotic tumor cells. In the other melanoma (case 17), viable tumor showed nuclear S100 reactivity, but adjacent necrotic tumor was negative. Similarly, a malignant peripheral nerve sheath tumor of the arm demonstrated nuclear SI00 staining in viable tumor but no staining in areas of tumor necrosis (case 23). A further problem with SI00 antibody in this series of cases was positive cytoplasmic staining of necrotic areas of a moderately differentiated squamous cell carcinoma in the lung that had metastasized from the larynx (case 1) llmage 21. Both lymphoma cases demonstrated strong LCA reactivity with a membranous pattern of staining in areas of necrotic tumor. However, LCA also demonstrated a similar staining pattern in 3 necrotic carcinomas (cases 6, 7, and 14). Illustrative is case 6, a nasopharyngeal carcinoma metastatic to a left neck lymph node. Necrotic areas of this tumor demonstrated both cytokeratin and LCA immunoreactivity Image 31. Cases 7 (infiltrating ductal carcinoma) and 14 (poorly differentiated nonsmall cell carcinoma of lung) showed LCA immunoreactivity of necrotic but not viable tumor. Although rare scattered intratumor lymphocytes were present and highlighted by LCA in the viable tumor, neither of these cases was intranodal or had a significant accompanying inflammatory reaction. Discussion Completely necrotic tissue masses can pose a diagnostic dilemma for the surgical pathologist. This may be encountered as a lymph node biopsied in the setting of high clinical suspicion for lymphoma or metastatic carcinoma, or as a diminutive biopsy of suspected tumor to establish diagnosis. The increasing use of minimally invasive diagnostic biopsy techniques suggests this will become a more common problem. Only a few studies have addressed the reactivities of antibodies on necrotic tissue. In a study of autopsy cases, Pallesen and Knudsen demonstrated that specific immunoreactivity is preserved for a wide range of leukocyte antigens up to 72 hours postmortem in snapfrozen material.4 Norton and associates evaluated antigen preservation in infarcted lymph nodes in 11 cases of malignant lymphoma and 1 case of testicular infarction associated with lymphoma. A panel of leukocyte and epithelial antibodies was tested on formalinfixed, paraffinembedded tissue. Good preservation of several of the leukocyte markers, including CD45, was observed. In addition, there were no falsepositives reported for the epithelial markers that included EMA and CAM Finally, Pelstring and colleagues evaluated antigen preservation in 2 lymph nodes 1 with metastatic breast cancer and the other with reactive hyperplasia after allowing the tissue to autolyze for periods of 0 to 72 hours.6 The tissue was then snapfrozen or fixed with formalin, and a panel of immunohistochemical stains were performed. They reported preservation of the leukocyte markers LCA (CD45) and UCHL1 (CD45RO) in both frozen and formalinfixed tissue that had undergone autolysis for 72 hours. Other leukocyte markers were less well preserved; when autolysis was allowed to proceed for over 24 hours, staining results were less reliable in the formalinfixed tissue. Epithelial markers, including EMA, AE1, and AE1/3, were also examined. These markers showed variable preservation in formalinfixed tissues; EMA was the most durable, with positive staining after autolysis for 72 hours. AE1 was positive in tissues that had undergone up to 48 hours of autolysis; however, the extent to which this study of in vitro autolysis reflects the changes in antigenicity that occur during in vivo necrosis is uncertain. Our study was undertaken to establish the reactivity of necrotic tumor tissue against a panel of immunohistochemical stains. To accomplish that objective, we selected a series of cases with extensive necrosis in which adjacent viable tumor, or previously diagnosed viable tumor, was available for comparison and definitive tumor identification. The panel of immunohistochemical stains was selected to include a Am J Clin Pathol 1998; 110: Downloaded from by guest on 09 October 2018 were examined. In 11 cases, adjacent viable tumor was present in the stained sections. The remaining 3 cases were entirely necrotic but were assumed to be cases of carcinoma inasmuch as each of the patients in question had a recent history of primary carcinoma. Eleven necrotic carcinomas stained positively with at least 1 cytokeratin antibody. AE1 and AE1/3 were the most frequently reactive antibodies, with positive staining in 10 of 14 cases. S903 and PANCK were also frequently positive (each was positive in 9 of 14 cases). This pattern of strong cytokeratin reactivity is illustrated in llmage II (case 1). In 2 cases (cases 4 and 8), cytokeratin immunoreactivity was observed within macrophages surrounding areas of tumor necrosis. The histiocytic nature of these cells was demonstrated by their reactivity with anticd68 antibody, and this cytokeratin immunoreactivity could represent positive staining of phagocytized keratin debris within the cells (data not shown). Necrotic portions of 3 tumors (cases 11, 12, and 13) failed to show immunoreactivity for any of the cytokeratin stains, whereas viable tumor from the same cases was positive for AE1, AE1/3, and PANCK. Keratins 10, 13, and 20, as detected by DEK13 and CK20, were less commonly expressed in the carcinomas. Viable areas of 4 carcinomas reacted with these antibodies, but only 2 showed reactivity in necrotic areas (cases 1 and 10). In no instance was there staining for cytokeratins in necrotic tumor that was not also expressed in viable tumor. Significantly, no falsepositive staining was observed for any of the antikeratin antibodies in the necrotic nonepithelial tumors in this series.

4 Judkins ot al / IMMUNOHISTOCHEMISTRY ON NECROTIC TUMORS broad range of cytokeratin markers, as well as LCA and S100. These are among the most commonly available stains, and if found to be sufficiently sensitive and specific, they would be most useful in establishing the lineage of an entirely necrotic tumor. Two specific questions were addressed. First, are the antigenic targets of these immunohistochemical stains well preserved in necrotic tumor tissue? For cytokeratins, the answer is yes. Cytokeratin markers are relatively well preserved, showing positive staining in 78% (11 of 14) of necrotic carcinomas. No cases of positive cytokeratin staining in necrotic tumor tissue alone were encountered; positive staining in necrotic carcinomas always correlated with the staining pattern of the viable tumor tissue. The most widely expressed cytokeratins were those recognized by antibodies AE1 and AE1/3. This is not surprising inasmuch as these antibodies are specific for the largest number of different cytokeratins, both acidic and basic, and they have a welldemonstrated utility in identifying carcinomas. Although epithelial markers have previously been reported to be preserved in necrotic tissue,5 our results are in contrast to the findings of Pelstring and associates,6 who reported preservation of AE1 after only 48 hours of autolysis and no preservation of AE3 in formalinfixed material (although staining of both antibodies was slightly better maintained in frozen tissue). As in previous reports, LCA proved to be well preserved in the 2 lymphoma cases in our study.56 Both cases demonstrated strong positive membranous staining in viable and necrotic tumor tissue. The results of S100 staining were mixed, however. In only 1 of 2 SlOOpositive melanomas did necrotic tissue demonstrate preservation of S100 staining, and a malignant peripheral 604 Am J Clin Pathol 1998;110: nerve sheath tumor with positive staining of viable tumor yielded no staining in adjacent areas of necrosis. The second question addressed in this study is the reliability of the staining observed in necrotic tumor tissue. No evidence of aberrant cytokeratin staining was observed among the nonepithelial tumors in our series, indicating that these antibodies retain their specificity for epithelial differentiation. The results were less reassuring for the other 2 immunohistochemical markers examined in this study. One necrotic squamous cell carcinoma showed cytoplasmic immunoreactivity to SI00, and LCA stained necrotic tissue in 3 carcinomas. Nuclear rather than cytoplasmic staining for SI00 is generally seen as a more significant reaction. In the presence of widespread tumor necrosis, however, nuclear SI00 staining may not be achievable. These results suggest caution when relying on an exclusively cytoplasmic staining pattern in necrotic tissue with this stain. LCA is a widely used marker of leukocyte differentiation. The DAKOLCA stain is a mixture of 2 antibodies (PD7 and 2B11) that, when combined, are reactive with all members of the LCA family.910 The antibody is well characterized and has been shown to be nonreactive with nonhemopoietic tissues, as well as with a wide range of nonhemopoietic neoplasms including many carcinomas, melanomas, and sarcomas."14 Previous studies had shown that LCA continued to be expressed on necrotic lymphoid tissue.46 Indeed, Norton and coworkers reported that when the surface membrane pattern of staining is used to assess the specificity of the reaction, the presence of LCA reactivity could be used to confirm the lymphoid nature of the infarcted tissue.5 Our results would suggest a more cautious Downloaded from by guest on 09 October 2018 Image I I Photomicrographs of necrotic squamous cell carcinoma metastatic from larynx to lung (case 1). A, Squamous cell carcinoma with extensive necrosis and focally residual viable tumor (H&E, x60). B, AE1/3 stain of same area showing strong positive staining of both viable and necrotic tumor (x60).

5 Anatomic Pathology / ORIGINAL ARTICLE r^~. c am % ' m f "i~ y *V> ~*,Jr* f*^ 2 Image 21 Photomicrographs of a necrotic squamous cell carcinoma that has metastasized from larynx to lung (case 1). A, S100 stain shows cytoplasmic staining of necrotic tumor cells (x40). B, S100 stain showing no immunoreactivity on a viable tumor; positive staining is present among entrapped dendritic cells (x40). s r X Ml i ^ " m ft* T " /*:4R '.*' * ;*, * * ^s* d "*>**<. «tf ;? /.. Image 31 Photomicrographs of a necrotic nasopharyngeal carcinoma, metastatic to a level III neck lymph node (case 6). A, Lymph node with metastatic nasopharyngeal carcinoma and extensive necrosis, stained with AE1/3, showing a strong positive reaction on necrotic tumor cells (x40). B, Same field stained with LCA, showing decoration of membranes of necrotic tumor cells (x40). endorsement of the use of LCA on necrotic tissue suspected to harbor tumor. In addition to being well preserved in both necrotic lymphoma cases examined, LCA also demonstrated strong surface membrane staining in necrotic areas of 3 carcinomas. One of the cases was a carcinoma metastatic to a lymph node, and the reactivity noted may reflect the presence of contaminating lymphocytes. However, the remaining 2 cases were extranodal and without significant inflammatory infiltration. The staining for LCA that is seen on necrotic epithelial cells may be the result of diffusion of LCA antigen from necrotic intratumor lymphocytes. The falsepositive reaction for SI00 is more difficult to explain but may be related to the fact that SI00 is an antiserum rather than a monoclonal antibody. Our results demonstrate good preservation of cytokeratin markers and the absence of falsepositive cytokeratin staining in the 10 necrotic nonepithelial tumors in this series. Antibody markers such as AE1/3 should be considered reliable in demonstrating the epithelial nature of suspected necrotic carcinomas. LCA also demonstrated good preservation but showed a pattern Am J Clin Pathol 1998;110: Downloaded from by guest on 09 October jq*>

6 Judkins et al / IMMUNOHISTOCHEMISTRY ON NECROTIC TUMORS 5. Norton AJ, Ramsay AD, Isaacson PG. Antigen preservation in infarcted lymphoid tissue. A novel approach to the infarcted lymph node using monoclonal antibodies effective in routinely processed tissues. Am] Surg Pathol. 1988;12: Pelstring RJ, Allred DC, Esther RJ, et al. Differential antigen preservation during tissue autolysis. Hum Pathol. 1991;22: Perentes E, Rubinstein U. Recent applications of immunoperoxidase histochemistry in human neurooncology. From the Department of Pathology and Laboratory Medicine, An update. Arch Pathol Lab Med. 1987;111: University of Pennsylvania Medical Center, Philadelphia, 8. Nakajima T, Watanabe S, Sato Y, et al. An Pennsylvania. immunoperoxidase study of S100 protein distribution in normal and neoplastic tissues. Am J Surg Pathol. Presented in abstract form at USCAP meeting, Orlando, 1982;6: Florida, March Streuli M, Morimoto C, Schrieber M, et al. Characterization This study was conducted under the auspices of the of CD45 and CD45R monoclonal antibodies using transfected University of Pennsylvania Institutional Review Board Approval mouse cell lines that express individual human leukocyte for use of human tissues (expedited review). common antigens. J Immunol. 1988;141: Address reprint requests to Dr Judkins: Division of Anatomic 10. Dalchau R, Kirkley J, Fabre JW. Monoclonal antibody to a Pathology, University of Pennsylvania Medical Center, 3400 human leukocytespecific membrane glycoprotein probably Spruce St, 6 Founders, Philadelphia, PA homologous to the leukocytecommon (LC) antigen of the rat. Eur J Immunol. 1980;10: Acknowledgments: The authors would like to thank Shelly11. Michie SA, Spagnolo DV, Dunn KA, et al. A panel approach Roberts, MS, Terry Pasha, and Bob Caron, MS, for their outstanding technical work in the immunohistochemical staining of to the evaluation of die sensitivity and specificity of antibodies for the diagnosis of routinely processed these cases. histologically undifferentiated human neoplasms. Am] Clin Pathol. 1987;88: Pulido R, Cerebrian M, Acevedo A, et al. Comparative biochemical and tissue distribution study of four distinct References CD45 antigen specificities. J Immunol. 1988;140: Li CY, LazcanoVillareal O, Pierre RV, et al. 1. Davies JD, Stansfeld AG. Spontaneous infarction of Immunocytochemical identification of cells in serous superficial lymph nodes. ] Clin Pathol. 1972;25: effusions. Technical considerations. Am] Clin Pathol. 2. Cleary KR, Osborne BM, Butler JJ. infarction 1987;88: foreshadowing malignant lymphoma. Am J Surg Pathol. 1982;6: Kurtin PJ, Pincus GS. Leukocyte common antigen a diagnostic discriminant between hematopoietic and 3. Maurer R, Schmid U, Davies JD, et al. Lymphnode infarction nonhematopoietic neoplasms in paraffin sections using and malignant lymphoma: a multicentre survey of European, monoclonal antibodies. Hum Pathol. 1985;16: English and American cases. Histopathobgy. 1986;10: of staining in 3 of 14 necrotic carcinomas, which could be interpreted as positive (especially in limited tissue samples). Lastly, SI00 showed poor preservation in known SlOOpositive neoplasms, and like LCA, showed positive cytoplasmic staining in 1 of 14 necrotic carcinomas. In at least 4 cases in this series, overreliance on a single positive LCA or SI00 stain could have resulted in misclassification of the primary tumor type. 646 AmJCIinPathol 1998;110: Downloaded from by guest on 09 October Pallesen G, Knudsen LM. Leucocyte antigens in post mortem tissues: their preservation and loss as demonstrated by monoclonal antibody immunohistological staining. Histopathology. 1985;119: