Effect of formalin tissue fixation and processing on immunohistochemistry

Although immunohistochemistry is routinely performed by many pathology laboratories, its standardization still lags behind. A major cause of variation in the reproducibility of immunohistochemical staining is induced by tissue fixation and, to a lesser degree, tissue processing. This report, stemming from the first meeting of the International Consensus Group on Standardization and Quality Control (ICGSQC) in Nice, France, summarizes the problem and suggests solutions to begin to achieve standardization of fixation and processing. Most laboratories use neutral-buffered formalin (10%) for tissue fixation which introduces cross-links, whereas coagulative fixatives are less popular. Problems with formalin fixation comprise delay of fixation and variations in the duration of the fixation mainly. Solutions to these problems could be to start fixation soon (<30 min) after surgical removal of the tissue and to avoid overfixation (>24-48 hrs). For tissue processing, the most important problem is inadequate tissue dehydration prior to paraffin embedding. This can be prevented by preparing all solutions freshly every week, depending on the volume of tissue processed. If consistently applied, these procedures could eliminate some of the sources of variation in immunohistochemical stains.

Rearrangement of the p53 gene in human osteogenic sarcomas

p53 is a 53-kDa nuclear protein that is associated with malignant transformation in several tumor model systems. In a survey of 134 human carcinomas, sarcomas, leukemias, and lymphomas obtained at surgery or from peripheral blood, we found rearrangements of the p53 gene only in osteogenic sarcomas (3 of 6 osteogenic sarcomas examined). Normal tissue from one of these patients had an unrearranged gene, indicating that the genetic abnormality in the tumor was acquired. Two of the sarcomas with rearranged genes expressed levels of p53 protein that were elevated relative to other tumors. Rearranged p53 genes were also found in human osteogenic sarcoma cell lines.

Alterations of myc, myb, and rasHa proto-oncogenes in cancers are frequent and show clinical correlation

Alterations of c-myc, c-rasHa, or c-myb oncogenes were found in more than one-third of human solid tumors. Amplification of c-myc occurred in advanced, widespread tumors or in aggressive primary tumors. Apparent allelic deletions of c-rasHa and c-myb can be correlated with progression and metastasis of carcinomas and sarcomas.

Amplification of c-erbB-2 oncogene in human adenocarcinomas in vivo

There are two genes related to the viral erbB gene in the human genome. c-erbB-1 is the same as the gene for the epithelial-growth-factor (EGF) receptor, and c-erbB-2 encodes a receptor-like protein very similar to, but distinct from, the EGF receptor. Hybridisation analysis of DNA from 101 fresh human malignant tumours showed that the c-erbB-2 gene was amplified in 5 of 63 adenocarcinomas and none of 38 other types of tumours, whereas the c-erbB-1/EGF-receptor gene was amplified only in 1 of 8 squamous-cell carcinomas. Thus, the protein products of the amplified c-erbB-2 gene may have a role in the evolution of adenocarcinomas, as does the EGF receptor in some squamous-cell carcinomas.

The human hematopoietic progenitor cell antigen (CD34) in vascular neoplasia

The human hematopoietic progenitor cell antigen CD34 is synthesized and expressed by early normal hematopoietic progenitor cells and by many acute leukemias. Anti-CD34 antibodies also have been reported to stain blood vessels in tissue sections, and, more recently, CD34 mRNA has been detected in vascular endothelial cells. Therefore, the authors studied the diagnostic utility of immunohistochemical CD34 antigen detection in tumors of endothelial cell derivation and compared the results with stains for von Willebrand (vW) factor. A wide variety of epithelial and mesenchymal neoplasms also were examined to assess the specificity of CD34 for vascular neoplasia. Seven cases of angiosarcoma (seven of seven), five cases of Kaposi's sarcoma (five of five), and eight cases of epithelioid hemangioendothelioma (eight of eight) were moderately to strongly positive for CD34. This reactivity was equally intense in frozen sections, alcohol-fixed tissue, and formalin-fixed specimens. In many cases, the malignant endothelial cells stained more strongly than adjacent benign endothelium. Moreover, in most cases CD34 positivity was quantitatively and qualitatively stronger than staining for vW factor. Two cases of hemangiopericytoma (two of two) were CD34 positive but stained less intensely than the angiosarcomas, Kaposi's sarcomas, or hemangioendotheliomas. Five of six cases of hemangioma also stained positively for CD34; the nonreactive tumor in this group was the only one among 28 vascular neoplasms studied that was not reactive for CD34. In comparison, 9 of the 28 vascular tumors did not stain for vW factor. Three hundred fifty-seven tumors of nonvascular derivation also were examined for CD34 antigen expression. Focal light staining was seen in one pulmonary squamous cell carcinoma; moderate to intense staining was observed in half of the epithelioid sarcomas studied (8 of 16) and in a minority of leiomyosarcomas (3 of 22). These findings indicate that CD34 is a sensitive and relatively specific marker for neoplasms of vascular origin.

The distribution of vimentin and keratin in epithelial and nonepithelial neoplasms. A comprehensive immunohistochemical study on formalin- and alcohol-fixed tumors

Vimentin has been regarded as the intermediate filament characteristic of normal and neoplastic mesenchymal cells and keratins as typical of epithelial cells and the neoplasms derived from them. However, many epithelial cells in tissue culture or in effusion, as well as cells in some solid epithelial neoplasms such as renal, endometrial, ovarian, pulmonary, and thyroid adenocarcinomas, have been shown to coexpress vimentin and keratin. The recent availability of monoclonal antibodies that work reasonably well in paraffin-embedded tissue led us to carry out a comprehensive immunohistochemical study on formalin- and alcohol-fixed specimens of neoplasms in which we used monoclonal antibodies against vimentin. These results were compared with our previous study in which the same tumor tissues were investigated using antibodies against keratin. The antigenicity of vimentin was found to be preserved in all alcohol-fixed specimens and in 63% of formalin-fixed tissues. Vimentin was the sole intermediate filament present in virtually all sarcomas, meningiomas, schwannomas, and melanomas. In addition, variable percentages (10-57%) of carcinomas, neuroendocrine carcinomas, neuroblastomas, thymomas, and mesotheliomas were positive for vimentin, which, except in the neuroblastomas, was coexpressed with keratins. Among the adenocarcinomas, more than 50% of papillary carcinomas of the thyroid, as well as renal, endometrial, ovarian, and lung carcinomas, coexpressed keratins and vimentin. A distinctive paranuclear and basal localization of vimentin was observed in the cells of many of these tumors, in contrast to the predominantly apical distribution of the keratins. The authors conclude that coexpression of vimentin and keratin is more widespread than previously reported and that antibodies to vimentin, by themselves, are of limited value for the differentiation of epithelial from mesenchymal neoplasms.

Spindle cell carcinoma: ultrastructural evidence of squamous origin and collagen production by the tumor cells

Two cases of spindle cell carcinoma (SCC) of the esophagus and skin, respectively, were studied by transmission electron microscopy. The tumor cells were closely associated with collagen fibrils and had abundant rough endoplasmic reticulum with dilated cisternae. Except for their irregular shape and atypical nuclei they resembled actively synthesizing fibroblasts. In addition to these features, some cells contained numerous tonofibrils and occasional well-developed desmosomes. A gradual transition to typical squamous cells was noted in the skin tumor. These findings suggest that the pseudosarcomatous component of SCC originates from mesenchymal metaplasia of squamous cells and that collagen is produced by these metaplastic cells.

Further evidence that "malignant angioendotheliomatosis" is an angiotropic large-cell lymphoma

Malignant angioendotheliomatosis is a rare, generally fatal disease characterized by a multifocal proliferation of neoplastic mononuclear cells within the lumens of small blood vessels. Although the disease primarily involves the vasculature of the skin and central nervous system, vascular involvement of other organs may occur and may produce a variety of clinical findings. Some early investigators concluded that malignant angioendotheliomatosis was a neoplasm of endothelial cells, but recently others have suggested that it is of hematopoietic origin. We have studied three patients with the disease and have characterized the immunophenotype of the neoplasm on cryostat-cut fresh-frozen tissues. A detailed antigenic phenotyping of neoplastic lymphoid cells showed that one patient had the immunophenotype T11+, Leu-1+, Leu-3+, Leu-2+, B1-, B2-, SIg-, LN1-, LN2-, the predominant phenotype for peripheral T-cell lymphoma; the others had T11-, Leu-1-, Leu-3-, Leu-2-, B1+, B2+, SIg+, LN1+, LN2+, consistent with a B-cell-derived lymphoma. On the basis of our results, we suggest that angiotropic (intravascular) large-cell lymphoma would be more appropriate than malignant angioendotheliomatosis as a name for this disease.

Malignant fibrous histiocytoma. An electron microscopic study

Sixteen cases of malignant fibrous histiocytoma are presented. Electron microscopy of 15 cases demonstrated fibroblast-like and mononuclear and multinucleated histiocyte-like cells. A small capillary was at the center of all storiform areas examined. Ultrastructural examination can be diagnostically useful within the context of a narrow differential diagnosis by conventional microscopy and the ability, by electron microscopy, to eliminate other mesenchymal cell types. In 13 cases, follow-up information was available from 18 months to 9 years following histological diagnosis. Five patients are alive and 8 patients have died, including two non-tumor related deaths. In 3 cases follow-up was less than 4 months. The biologic behavior of the tumor in this series was generally not related to histopathological parameters. The issue of histogenesis is largely unresolvable. Ultrastructural studies of various types of fibrous histiocytomas, suggesting cells of origin other than histiocytes, give credence to the concept that the histiocyte may represent a morphologic state of a given mesenchymal cell rather than a particular cell type.

The influence of protease digestion and duration of fixation on the immunostaining of keratins. A comparison of formalin and ethanol fixation

The effect of three proteases--trypsin, pepsin, and pronase--on the immunohistochemical staining of keratins with a broad-spectrum monoclonal antibody was investigated in paraffin sections of formalin and ethanol-fixed tissues by means of the peroxidase-antiperoxidase method. Both the length of exposure to the fixative and the duration of proteolysis were varied over a wide range. Ethanol-fixed tissues showed excellent preservation of the antigenicity of keratins, and no appreciable differences in immunostaining related to the length of fixation were found. The use of proteolytic enzymes did not improve these results; on the contrary, it caused rapid tissue disintegration. Formalin-fixed epithelial tissues stained weakly or failed to stain unless they were treated with a proteolytic enzyme. The optimal length of proteolysis varied with the degree of fixation; tissues that were fixed for long periods of time in formalin required longer exposure to a proteolytic enzyme and were more resistant to digestion than were tissues that were fixed briefly. No significant advantage of one protease over another was found in this study. We conclude that a proteolytic step must precede immunostaining for keratins if the tissue is fixed in formalin, but that the digestion period must be adjusted according to the length of exposure to the fixative. The superiority of alcohol over formalin fixation for the preservation of the antigenicity of keratins is confirmed by this study.

Interpretation and quantification of immunostains

Despite the fact that immunohistochemistry is widely used in routine diagnostic work and is a very common part of scientific reports in pathology and cytology, its standardization still lags behind. Interpretation of immunostains should be based on microanatomic distribution of the staining, proportion of positively stained cells, staining intensity, if relevant, and cutoff levels. These parameters should be shown to be reasonably reproducible and should be clearly defined in publications. Uniformity in the setting of thresholds could probably benefit from interlaboratory control materials containing defined amounts of the target antigen. Reliable and precise quantitative immunohistochemistry requires the use of control materials containing defined amounts of the target antigen and processed alongside the specimen combined with automated computer-assisted microspectrophotometry. Application of these suggestions is hoped to improve standardization and to facilitate communication in the field of immunohistochemistry.

The immunohistochemical diagnostic panel for epithelial mesothelioma: a reevaluation after heat-induced epitope retrieval

The immunohistochemical diagnosis between epithelial mesothelioma and adenocarcinoma is currently based on the use of a panel of antibodies to adenocarcinoma-associated antigens and a few antibodies to mesothelial-associated antigens. Since the introduction of epitope retrieval methods, the sensitivity of many antibodies has been enhanced. Thus, a reevaluation of the mesothelioma/adenocarcinoma diagnostic panel becomes necessary. We studied 268 paraffin-embedded formalin-fixed tumor samples that included 57 epithelial mesotheliomas and 211 adenocarcinomas of various origins, comparing an extensive antibody panel with and without heat-induced epitope retrieval (HIER). Marked increase in the sensitivity of several antibodies, with no loss of specificity, was found when HIER was used. After statistical analysis, the antibodies to the epithelial glycoproteins carcinoembryonic antigen, BerEp4, and Bg8 emerged as the best discriminators between adenocarcinoma and epithelial mesothelioma within the entire panel. The mesothelium-associated antibodies, HBME-1, calretinin, and thrombomodulin were less sensitive and less specific than the former, although they were found to be useful on certain cases. Antibodies to cytokeratins and vimentin, although of minor diagnostic value in this context, may be helpful to evaluate the quality of antigen preservation. This study confirms the value of immunohistochemistry to accurately distinguish mesothelioma from adenocarcinoma when an antibody panel approach is used. The addition of heat-induced epitope retrieval methods increases the effectiveness of the procedure and is recommended for most of the antibody panel members.

Leu-M1 antigen in human neoplasms. An immunohistologic study of 400 cases

Several studies have shown that the Leu-M1 antigen, a monocyte/granulocyte-related marker, is consistently expressed by the neoplastic cells of patients with Hodgkin's disease (HD). It has been suggested that reactivity of Reed-Sternberg cells with Leu-M1 can be used in support of a morphologic interpretation of HD, and that it is helpful in the differential diagnosis of HD from morphologically similar lesions. To evaluate the significance of the Leu-M1 positivity of Reed-Sternberg cells in the diagnosis of HD, we investigated the distribution of Leu-M1 antigen in a series of patients with HD, non-Hodgkin's lymphomas, and nonhematopoietic neoplasms. We were able to demonstrate the presence of Leu-M1 antigen not only in the majority of patients with HD, but also in 12 of 18 (67%) peripheral T-cell lymphomas, as well as in a variety of nonhematopoietic neoplasms, which included 113 of 199 carcinomas, most of them (58%) adenocarcinomas. Only one of 34 sarcomas showed a focal positive reaction. Leu-M1-related antigen was not detected in any of 18 mesotheliomas, 15 germ cell tumors, 13 melanomas, three schwannomas, or three astrocytomas. Our study indicates that Leu-M1 positivity has no value in supporting the diagnosis of HD in situations where the histologic diagnosis of HD is doubtful. However, since anti-Leu-M1 reacted positively in the majority of adenocarcinomas but was absent in mesotheliomas, melanomas, and most sarcomas, this antigen could serve as a new marker that may be helpful in situations in which carcinoma is a part of the differential diagnosis.

Angiotropic (intravascular) large cell lymphoma. A clinicopathologic study of seven cases with unique clinical presentations

The authors recently reported the antigenic phenotypes of three cases of so-called "malignant angioendotheliomatosis" and suggested that angiotropic large cell lymphoma (ALCL) is a more appropriate designation for this disease. The authors now report an additional seven cases of ALCL with unique clinical presentations. One patient presented with prostate enlargement, the second with lytic bone lesions and thickened nasal sinus mucosa, the third had diffuse myalgia, the fourth had dyspnea and pulmonary infiltrates, the fifth had gangrene of the lower extremities, total-body skin involvement, and pancytopenia, the sixth had a lesion of the foreskin mimicking squamous cell carcinoma, and the seventh had a mediastinal mass. In all cases histologic features were characteristic of ALCL with, in two cases, extravascular spread into soft tissue. Immunohistologic studies showed a B-cell phenotype in five cases and a T-cell phenotype in one case. Two patients received combination chemotherapy using established treatment protocol for large cell lymphoma, and remain in complete clinical remission and two patients are responding clinically to combination chemotherapy. Two patients died shortly after receiving combination chemotherapy. One patient has only recently been diagnosed as having ALCL and no long-term follow-up is available. These data indicate that, although ALCL affects predominantly the central nervous system and skin, unusual clinical presentations may occur, and patients with ALCL may respond to combination chemotherapy for large cell lymphoma.

The use of antikeratin antiserum as a diagnostic tool: thymoma versus lymphoma

Indirect immunofluorescence staining in two thymomas, one case of thymic hyperplasia, 10 malignant lymphomas and three seminomas was done with an antibody prepared against keratins from human epidermis. Staining was observed only in the epithelial cells of the thymomas and thymic hyperplasia and correlated well with electron microscopic studies. Immunofluorescence staining of thymic tumors with antikeratin antibody provides a simple, specific, and sensitive method for distinguishing thymoma from lymphoma and seminoma. The method may also prove to be useful in other instances in the distinction between epithelial and nonepithelial tumors.

Proto-oncogene abnormalities in human breast cancer: correlations with anatomic features and clinical course of disease

DNAs from fifty-three primary breast cancers were hybridized with 16 different proto-oncogene or oncogene probes. Abnormalities of one or more of five proto-oncogenes were found in fifty-eight percent of tumors at the time of mastectomy. Amplification of c-myc and c-erbB-2, and allelic deletions of c-ras-Ha and c-myb were the most common abnormalities. The presence of altered proto-oncogenes correlated with clinical stage of the cancers. Fifteen of 43 evaluable tumors of stages I to III recurred, and four of five evaluable stage IV tumors progressed within 16 to 24 months of surgery. All but one of the cancers that recurred or progressed had detectably altered proto-oncogenes (P less than .001). Analysis of proto-oncogenes may have prognostic value in breast cancer.

Distinction of mesothelioma from adenocarcinoma. An immunohistochemical approach

The authors investigated the expression of keratin, carcinoembryonic antigen (CEA), and an epithelial marker derived from milk fat globule membranes in 12 mesotheliomas and 100 diverse adenocarcinomas with immunohistochemical methods. The authors employed a monoclonal antibody to keratin designated as AE1, as well as the following commercially available antisera: rabbit anti-whole human keratin, rabbit anti-CEA, and a monoclonal antibody to an epithelial factor designated as MFG-2. Expression of keratin was found in all the mesotheliomas and adenocarcinomas with antibody AE1 as well as with the rabbit antiserum; CEA was detectable in 65% of the adenocarcinomas but two mesotheliomas also reacted weakly. With antibody MFG-2, positive results were obtained in 85% of the adenocarcinomas and in none of the mesotheliomas. All of 64 (100%) breast-, lung- and ovary-derived adenocarcinomas immunostained positively with antibody MFG-2. This is of particular significance because pulmonary and ovarian adenocarcinoma frequently may be indistinguishable clinically and histologically from epithelial mesothelioma. The authors conclude that antikeratin antibodies are not useful in the distinction of adenocarcinoma from mesothelioma. Because of its greater sensitivity and specificity, MFG-2 is superior to CEA in this differential diagnosis.

The use of antikeratin antibodies in the immunohistochemical distinction between neuroendocrine (Merkel cell) carcinoma of the skin, lymphoma, and oat cell carcinoma

Paraffin sections of formalin-fixed tumor samples from 26 patients with neuroendocrine (Merkel cell) carcinoma of the skin (NECS) were studied immunohistochemically with three monoclonal antibodies to low molecular weight keratin (MAB-K) and with antibodies to leukocyte common antigen (LCA), neurofilament (NF), neuron-specific enolase (NSE), S100 protein (S100), and chromogranin (CGN), to investigate the relative diagnostic value of these antibodies. Samples from 20 lymphomas, 10 non-oat cell undifferentiated carcinomas, 10 oat cell carcinomas, and 10 melanomas served as controls. Keratin was found in 25 of the 26 NECS and in all undifferentiated and oat cell carcinomas. A ball-like immunostaining for keratins, resembling an inclusion body was seen only in cases of NECS and some carcinoids. Neurofilament, NSE, and CGN were expressed by fewer NECS than was keratin and all NECS were negative for LCA and S100. None of the lymphomas and melanomas contained detectable keratin, NF, NSE, or CGN. Only the lymphomas stained with LCA. Only the melanomas were S100-positive. It is concluded that keratin is the most useful single discriminating marker in the separation of neuroendocrine (Merkel cell) carcinoma of the skin from lymphoma, melanoma and, when the characteristic inclusion-like pattern is seen, from metastatic oat cell carcinoma.

Malignant vascular tumors of the serous membranes mimicking mesothelioma. A report of 14 cases

Malignant endothelial neoplasms involving the serous membranes are rare, and only a few cases have been documented. We report 14 patients with epithelioid hemangioendothelioma (EHE) or epithelioid angiosarcoma (EA) diffusely involving the pleural, peritoneal, or pericardial cavities, resulting in a picture closely resembling mesothelioma. The mean age at diagnosis was 52 (range, 34-85). The patients included two women and one man with peritoneal tumors, eight men with pleural tumors, and three men with pericardial tumors. A shared histological appearance was a diffuse sheet-like and clustered pattern of tumor growth with variable degrees of vascular differentiation. A tubulopapillary growth pattern, often seen in mesothelioma, was prominent in four cases. Nine cases showed a variable number of spindle cells, some neoplastic, others reactive, focally producing a biphasic growth pattern, further suggesting mesothelioma. Initial interpretations included mesothelioma, adenocarcinoma, and, in one case with prominent spindle-cell components, leiomyosarcoma. Immunohistochemically, strong vimentin staining and negative or weak to moderate cytokeratin staining were observed in all 14 cases. The tumor cells coexpressed at least two of the four endothelial markers used in the study (CD31, CD34, von Willebrand factor, and Ulex europaeus agglutinin-I [UEA-I)]. Detection of abortive vessel formation was facilitated by staining for collagen type IV. Markers of mesothelial, epithelial, muscular, and neuronal differentiation were all negative in the subset of cases studied. As a control group, 39 mesotheliomas and more than 60 adenocarcinomas of various origins were studied using the same antibody panel. This group revealed strong keratin staining, moderate or negative vimentin staining, and no expression of any of the endothelial-lineage markers, with the exception of positive staining for UEA-I in occasional adenocarcinomas. Clinically, these endothelial tumors were highly aggressive; 12 patients presented with disseminated disease, and most died within months of the initial presentation. These findings indicate that, although uncommon, EHE/EA should be included in the differential diagnosis of serous membrane neoplasms with histological and clinical features of malignant mesothelioma. The diagnosis of an endothelial neoplasm can be suspected by the presence of abortive vessel formation and by the strong expression of vimentin, with absent or low-level expression of cytokeratin. The demonstration of immunoreactivity for two or more endothelial-associated markers is essential in confirming the diagnosis.

Sclerosing cholangitis after continuous hepatic artery infusion of FUDR

Eight of 46 (17.4%) patients treated in our trial of continuous hepatic artery infusion (CHAI) of fluorodeoxyuridine (FUDR) by Infusaid pump developed biliary strictures. The lesions were clinically, radiographically, and pathologically identical to the idiopathic sclerosing cholangitis frequently seen in association with inflammatory bowel disease. Treatment included immediate cessation of intraarterial FUDR, and surgical or percutaneous drainage of the biliary tree if it was dilated. Two of the eight patients died of the complication. Three patients stabilized after biliary system drainage, and two patients improved on observation only. The pathogenesis of this complication is not understood. This report details the clinical and pathological features of this entity.