USES AND LIMITATIONS OF ANCILLARY TECHNIQUES
APPLIED TO CYTOPATHOLOGY

Jeffrey S. Ross, M.D.

INTERPRETATION FACTORS INFLUENCING IMMUNOCYTOCHEMISTRY The three main uses of IHC in cytologic specimens has been the diagnosis of malignant versus benign or reactive cells; the differential diagnosis of malignant cells; and prognosis assessment of malignant tumors. Significant interpretation factors based on the evaluation of the staining reactions are important considerations for the avoiding of errors and false conclusions from the studies. Diagnosis of Malignant versus Benign, Reactive or Normal Cells Epithelial membrane antigen has been utilized to resolve the differential diagnosis in problematic serous effusions and found to have high specificity and sensitivity for the diagnosis of carcinoma versus reactive mesothelial cells. Featuring only a 3% false positive rate and, although unable to discriminate malignant mesothelioma from metastatic carcinoma, the EMA reaction was judged as highly useful to confirm the presence of malignancy in this type of specimen.34  A panel of markers may be of even great use for the detection of malignant cells in an effusion cytology sample.35  Another example of the use of IHC to differentiate malignant from benign cells is the identification of myoepithelial cells by staining for muscle specific actin in the differential diagnosis of difficult breast fine needle aspirates.36  In both of these IHC applications improper technique resulting in a false negative or false positive staining could lead to serious errors in diagnostic interpretation. The impact of false positive IHC due to technical errors or excess antigen retrieval may result in the false immunolocalization of tumor specific antigens in benign or reactive cells. Over-reliance on IHC in cases with non-diagnostic routine cytologic features can produce a disastrous final cytology report in which a technical error has led to a misdiagnosis of cancer. Similarly, technical errors, particularly in fixation, processing and the nature of the specimen available for IHC may lead to false negative staining of certain tumor specific antigens (CEA, PSA, CA125, B72.3, etc.) and a similar misdiagnosis in which the presence of malignant cells are not specifically reported. Additional pitfalls which may impact on interpretation of IHC in the diagnosis of malignancy include the endogenous biotin activity in normal hepatocytes leading to false positive immunoreactivity for alpha fetoprotein and the misdiagnosis of well differentiated hepatocellular carcinoma and the non-specific binding of glycoprotein antibodies to mesothelial cells leading to a misdiagnosis of metastatic carcinoma in an effusion specimen. Differential Diagnosis of Cytologic Specimens Diagnostic for Malignancy on Routine Cytology Preparations In certain settings of differential diagnosis of malignancy, the use of IHC to render a specific diagnosis may feature technical or biologic errors that lead to an erroneous interpretation. Carcinoma Versus Lymphoma The main pitfall in IHC in this differential diagnosis is that, even in well preserved specimens, certain poorly differentiated large cell malignant lymphomas may stain negatively for common leukocyte antigen. This fact coupled with appropriate negative epithelial marker staining may lead to a non-specific final diagnosis of "undifferentiated malignant tumor". Moreover, on occasion, a CLA negative large cell lymphoma may be diagnosed as a carcinoma based on solely on the CLA negative IHC.37  The use of additional lymphoid markers is recommended to resolve the differential diagnosis of CLA negative, epithelial marker negative large cell lymphomas. Carcinoma Versus Sarcoma, Mesothelioma, Melanoma In this setting, false negative epithelial marker staining due to technical factors may confuse the differential diagnosis. Exfoliation into fluid specimens may result in significant loss of epithelial marker immunoreactivity. Interestingly, HMB45 immunoreactivity, originally considered diagnostic of malignant melanoma has now been identified in non-melanoma cells including angiomyolipoma.36  New antibodies have been developed to resolve the important differential diagnosis of reactive mesothelial cells versus malignant mesothelioma.38-41  The AMAD-2 antibody shows particular promise in this regard.42  In the differential diagnosis of epithelail malignant mesothelioma versus metastatic adenocarcinoma, a variety of immunocytochemcal strategies have been employed.43-44  The markers MOC-31, CEA and BG-8 have been recommended as the best discriminators.43  A variety of epithelial markers have also been used to detect scarce n umbers of malignant cells in effusion cytology specimens.45-46  Lymphoma Phenotyping Errors and pitfalls in lymphoma phenotyping are considered in the molecular cytopathology section focused on lymph node fine needle aspiration diagnosis. Of importance is the staining heterogeneity of these neoplasms and the aberrant marker expression seen in lymphoid cells when present in effusions versus the immunophenotype of the same tumor obtained by fine needle aspiration.37  Interestingly, although the ThinPrepR processing technique has worked quite well for general immunocytochemistry applications,47  a recent study indicated that the results of lymphoma immunophenotyping studies using this method were inconclusive.48  Immunocytochemistry performed on smears and FNA specimens has compared with flow cytometric results of immunophenotyping of the same lymph node specimen.49  The combination of FNA and lymphoma immunophenotyping by flow cytometry can provide sufficient detailed information required for treatment and reduce the need for open lymph node surgical biopsy.50  Neuroendocrine Versus Non-Neuroendocrine Neoplasm This differential diagnosis is highly impacted by technical factors and the biologic nature of the exfoliated cells. Neuroendocrine tumors with minimal neurosecretory cytoplasmic granules may stain negatively for neuroendocrine markers due to limited antigen available. Technical problems resulting in antigen loss or use of relatively insensitive antibodies may produce false negative staining reactions. False positive staining reactions for cytologic specimens is seen particularly with neuron specific enolase which may non-specifically bind to cells devoid of neurosecretory granules when evaluated by electron microscopy. Recently, an antibody to the MIC-2 gene protein product has shown high sensitivity and specificty for the detection of primitive neuroectodermal tumors on fine needle aspiration biopsy51  Sarcoma Subtype As previously mentioned exfoliation may causes down-regulation of certain intermediate filament expression and up-regulation of vimentin. Exfoliated tumor cells may feature low cellular intermediate filament antigen concentrations and fail to immunostain especially when insensitive antibodies are used. Interestingly, antigen retrieval procedures may cause false positive reactions when antibodies specific for intermediate filaments, muscle proteins, or other target antigens non-specifically bind to the reagents after overheating. Glioma Versus Lymphoma or Carcinoma in Spinal or Ventricular Fluid The glial fibrillary acidic protein antibodies are relative specific for glial cells. However, non-neoplastic glial cells will also stain for GFAP and may be present in CSF or, particularly, ventricular fluid samples. False negative staining for GFAP in these specimens will occur when improper antigen preservation techniques are utilized. However, in general immunocytochemistry has worked quite well in resolving this differential diagnosis when applied to small samples such as aspirates and stereotactic needle biopsies.52  Germ Cell Versus Non-Germ Cell Malignancy The AFP stain is a generally unreliable marker for exfoliated cells with an unacceptable high false negative staining rate even for ideally preserved specimens. The serum AFP level may be markedly elevated and cytologic specimens of germ cell tumors will still stain negatively. The AFP antigen may not be available for antibody complex formation during the IHC procedure resulting from non-specific bonding of the antigen binding sites to other proteins within the cytoplasm. Alpha-1-antitrypsin and placental lactogen are better IHC markers for germ cell differentiation. Human chorionic gonadotropin antibodies are non-specific and do not prove the presence of germ cell differentiation in a malignant neoplasm. Adenocarcinomas, particularly poorly differentiated specimens, may stain for HCG and thus produce a false interpretation of germ cell tumor when this marker is used alone. Cytokeratin panels have been used to indentify the primary site of malignant neoplasms first encountered as metastases on cytologic specimens.53  A summary table (Table 4) is shown below as a suggestion for the work-up of anaplastic tumors. Table 4. Immunomarkers of Poorly Differentiated Tumors Markers Probable Tumor Keratin + Vimentin + LCA + S100 + Probable artifact. Check controls and repeat. Keratin - Vimentin + LCA + S100 - Lymphoma. May rarely be LCA - and even more rarely keratin +. Keratin + Vimentin -/+ LCA - S100 -/+ Carcinoma. Possible germ cell tumor. Others rare. Keratin - Vimentin + LCA - S100 + Melanoma. May rarely be S100 negative. HMB45 and other markers (melastatin) useful. Melanotic sarcomas also. Keratin - Vimentin + LCA - S100 - Sarcomas. Rarely seminoma. Keratin - Vimentin - LCA - S100 - Artifact. Check internal vimentin control. Repeat with antigen retrieval.