Presented in part at the 95th Annual Meeting of the United States and Canadian Academy of Pathology (USCAP), Atlanta, Georgia, February 11–17, 2006.
CD57 (Leu7), a marker for natural killer lymphocytes, and glucose transporter-1 (GLUT-1), a facilitative cell surface glucose transport protein, are expressed in a wide spectrum of epithelial malignancies. The usefulness of CD57 and GLUT-1 immunostaining was evaluated as an adjunct to assist in the diagnosis of papillary carcinoma in fine-needle aspirations (FNAs) of the thyroid.
Immunohistochemical staining for CD57 and GLUT-1 was performed on paraffin-embedded cell blocks of 50 thyroid FNA cases with the following cytologic diagnoses: 1) papillary carcinoma (15 cases); 2) atypical cytology, cannot exclude papillary carcinoma (14 cases); and 3) benign thyroid (21 cases).
Fourteen of 15 cases with an unequivocal diagnosis of papillary carcinoma were positive for CD57 and 1 case was negative for CD57. Tissue follow-up confirmed papillary carcinoma in all 15 cases. Of the 14 cases with a diagnosis of atypical cytology, cannot exclude papillary carcinoma, 6 cases were positive for CD57 and subsequent excision confirmed papillary carcinoma in all 6 cases. The remaining 8 cases negative for CD57 included surgically confirmed goiter (5 cases), adenoma (2 cases), and papillary carcinoma (1 case). The follicular cells in all 21 cases with a cytologic diagnosis of benign thyroid were negative for CD57. Histologic follow-up of these cases confirmed the benign cytologic diagnoses. GLUT-1 was negative in all cases. The sensitivity and specificity of CD57 was 91% and 100%, respectively.
Fine needle aspiration (FNA) biopsy is an accepted diagnostic modality for evaluating thyroid nodules, and often is the initial step in diagnostic assessment of thyroid lesions. FNA is now considered to be the most accurate, cost-effective, and simplest screening test for rapid diagnosis of thyroid nodules, with accuracy approaching 95%.1–3 Papillary thyroid carcinoma is the most frequently detected malignant thyroid neoplasm.4, 5 A majority of cases of papillary carcinoma are identified by FNA based on the identification of classic nuclear, cytoplasmic, and architectural features. However, in a subset of cases that lack ≥1 of these classic features, the accurate diagnosis of papillary carcinoma can be diagnostically challenging.
CD57 (Leu7) was initially identified in a subset of normal lymphocytes with natural killer cell activity.6 Subsequent studies found CD57 expression in a wide variety of epithelial and nonepithelial tumors.7–13 Studies in the surgical pathology literature have demonstrated the usefulness of CD57 immunohistochemical staining in the diagnosis of papillary carcinoma of the thyroid.14–18 Other authors have found partially conflicting results, reporting CD57 expression in thyroid adenomas and benign thyroid tissue.19, 20 Glucose transporter-1 (GLUT-1), a facilitative cell surface glucose transport protein, is normally expressed in erythrocytes, perineurium, blood-brain barrier, and placenta.21, 22 Recent studies have reported immunohistochemical expression of GLUT-1 in thyroid carcinomas, but not in adenomas or goiters.23, 24
The purpose of the current study was to examine the usefulness of CD57 and GLUT-1 immunolocalization in FNA-derived cell block material in the diagnosis of papillary carcinoma of the thyroid. To our knowledge, no published study to date has specifically assessed the use of these 2 immunohistochemical markers in FNA-derived cell block material for diagnosing papillary thyroid carcinoma.
MATERIALS AND METHODS
Consecutive samples were obtained from the archives of the Department of Pathology, Upstate Medical University (Syracuse, NY) from January 1998 to December 2004. Only cases with adequate cell blocks and available histologic follow-up were included. The cytologic diagnoses rendered on the cell blocks in conjunction with Papanicolaou and Diff-Quik-stained cytology smears were grouped into 3 categories: 1) papillary carcinoma (15 cases); 2) atypical cytology, cannot exclude papillary carcinoma (14 cases); and 3) benign thyroid (21 cases).
For each case, immunohistochemical staining was performed on 5-μm sections of cellblocks using a monoclonal antibody against CD57 (Dako Corporation, Carpinteria, CA) and a polyclonal antibody against GLUT-1 (NeoMarkers, Fremont, CA) on an automatic immunostainer, which incorporates standard capillary gap and streptavidin-biotin-peroxidase techniques. Positive control for CD57 and GLUT-1 consisted of a histologic block of papillary carcinoma that had been previously demonstrated to be CD57-positive and placenta tissue, respectively. Strong staining of erythrocyte membranes was observed with GLUT-1 and used as an internal positive control. Negative controls included tissue sections from each case with replacement of the primary antibody by nonimmune mouse serum. For analysis of immunoreactivity the membranous staining of the tumor cells was assessed. Immunoreactivity was considered positive if >10% of the follicular cells demonstrated distinct membranous staining. The staining reaction in each case was graded on the basis of percentage of follicular cells stained (0, <10%; 1+, >10%–25%; 2+, >25%–50%; 3+, >50%–75%; and 4+, >75%).
Smears from the 15 cases with a cytologic diagnosis of papillary thyroid carcinoma demonstrated characteristic cytomorphologic features of papillary carcinoma. These features included nuclei with irregular membranes, grooving, pale chromatin, and intranuclear cytoplasmic invaginations. Cell blocks in most of these cases revealed cells in sheets, clusters, or papillae with similar cytologic nuclear features.
Aspirate smears of 14 cases with a cytologic diagnosis of “atypical cytology, cannot exclude papillary carcinoma” lacked ≥1 features of classic papillary carcinoma, especially intranuclear cytoplasmic invaginations. This category also included 4 FNAs with features suspicious for the follicular variant of papillary carcinoma (cellular smears with a predominantly follicular cytoarchitectural pattern, occasional nuclear grooves, pale chromatin, and scant colloid). Cell blocks revealed groups and sheets of follicular cells with occasional nuclear grooves and nuclear clearing. Prominent follicular arrangement was also noted; however, intranuclear inclusions were absent and colloid was scant.
Smears from 21 cases of benign lesions revealed cytologic features of goiter (12 cases; moderately cellular smears with bland follicular cells in macrofollicular groups and histiocytes admixed with abundant colloid), lymphocytic thyroiditis (5 cases; follicular cells and Hurthle cells admixed with a polymorphous population of lymphocytes and scattered fragments of thick colloid), or colloid nodules (4 cases; abundant watery or inspissated colloid admixed with a few groups of bland follicular cells in flat sheets). Cell blocks revealed sheets or clusters of follicular cells and variably sized follicles filled with colloid. A prominent lymphoid component and histiocytes were identified in cases with cytologic diagnoses of lymphocytic thyroiditis and goiter, respectively.
A summary of results for the cytologic diagnosis, immunohistochemistry, and final histologic diagnosis are presented in Table 1. Both positive and negative controls for CD57 and GLUT-1 demonstrated appropriate results. Among the 15 cases with an unequivocal cytologic diagnosis of papillary carcinoma, 14 cases (93%) demonstrated strong and diffuse 4+ staining for CD57 (Fig. 1) and 1 case (7%) was negative for CD57. GLUT-1 was negative in all 15 cases (100%). Total thyroidectomy performed in all 15 cases confirmed the diagnosis of papillary thyroid carcinoma in each case.
Table 1. CD57 and GLUT-1 Expression in Aspirates of 50 Thyroid Cases with Follow-Up Histologic (Final) Diagnosis
Of the 14 cases with a diagnosis of “atypical cytology, cannot exclude papillary carcinoma,” 6 cases (43%) exhibited strong and diffuse 4+ immunostaining for CD57 (Fig. 2) and subsequent excision confirmed papillary carcinoma in all 6 cases (4 of these 6 cases were the follicular variant of papillary carcinoma). The remaining 8 atypical cases (57%) were negative for CD57 and included surgically confirmed goiter (5 cases), follicular adenoma (2 cases), and papillary carcinoma (1 case). GLUT-1 was negative in all 14 cases (100%).
All 21 cases (100%) with the cytologic diagnosis of benign thyroid were negative for CD57 (Fig. 3) and GLUT-1. Four cases did demonstrate weak 1+ luminal staining of the follicular cells for CD57 but there was an absence of complete membranous staining and therefore these cases were considered negative. Total thyroidectomy performed in all cases confirmed benign lesions in each cases (adenomatous goiter [16 cases] and lymphocytic thyroiditis [5 cases]).
Based on the overall immunohistochemical findings using all 50 thyroid FNA cases in our study, the sensitivity, specificity, and positive predictive value for the detection of papillary carcinoma using CD57 were determined to be 91%, 100% and 100%, respectively.
FNA of the thyroid is often the initial step in the evaluation of a thyroid lesion and hence cytologic interpretation can play a very important role in guiding the clinical management of the patient. The diagnosis of papillary carcinoma in FNA specimens is usually straightforward when classic cytologic features are present (including nuclei with irregular membranes, longitudinal grooves, pale chromatin, and intranuclear cytoplasmic invaginations), dense squamoid cytoplasm, and evidence of papillary architecture.25–28 In a subset of cases, however, it is difficult to make an unequivocal cytologic diagnosis of papillary thyroid carcinoma. The diagnostic dilemma is related to the observation that some of the typical cytologic features of papillary carcinoma (nuclear grooves, giant cells, psammoma bodies, papillary fragments) can also be observed in nonneoplastic lesions and follicular neoplasms of the thyroid.29–33 Follicular adenomas, adenomatous nodules, and hot autonomous nodules can show papillary fragments, nuclear enlargement, and nuclear grooves; optical clearing and nuclear atypia and enlargement can also be observed in lymphocytic thyroiditis, and dystrophic calcifications in goiter may be overinterpreted as psammoma bodies. Thus, benign lesions can occasionally be misdiagnosed as papillary carcinoma. A marker that would readily differentiate papillary carcinoma from nonneoplastic thyroid lesions and follicular neoplasms would be useful in resolving these common diagnostic dilemmas.
A variety of immunohistochemical stains have been evaluated in thyroid FNA cytology to aid in the diagnosis of papillary thyroid carcinoma.34–40 Ghali et al.14 initially reported diffuse membranous staining of CD57 in all of their surgically resected papillary thyroid carcinomas and concluded that CD57 may be a valuable additional tool in routine surgical pathology for the diagnosis of thyroid carcinoma. A number of other studies in the surgical pathology literature also found CD57 to be a useful adjunct in differentiating papillary thyroid carcinoma from benign thyroid lesions.15–18 However, Loy et al.19 reported that CD57 was not specific for thyroid carcinoma in tissue sections because 33% of benign thyroid lesions in their study were positive for CD57. Another study performed on thyroid aspirates also reported CD57 staining in benign thyroid conditions, including 6 of 14 follicular adenomas.20 GLUT-1 expression in tissue sections has also been shown in some thyroid carcinomas, although it was reportedly absent in follicular adenomas and other benign conditions of the thyroid.23, 24
In the current study, we evaluated the utility of CD57 and GLUT-1 as an ancillary tool for the diagnosis of papillary thyroid carcinoma in FNA-derived cell block material. In our experience, cell blocks are the preferred method for application of immunohistochemical stains because appropriate controls are available. In some cases, we noticed false-positive and false-negative immunostaining when using thin-layer preparations and destained smears in our laboratories. We found CD57 immunolocalization to be of considerable help in the diagnosis of papillary carcinoma and its distinction from benign thyroid lesions. In our study, strong and diffuse CD57 immunostaining was detected not only in 14 of 15 cases (93%) for which the cytologic diagnosis was papillary carcinoma, but also in 6 of 7 cases (86%) with an equivocal cytologic diagnosis (“atypical cytology, cannot exclude papillary carcinoma”) that were confirmed to be papillary carcinoma by histologic follow-up of resection specimens. In contrast, a lack of CD57 staining did not exclude papillary carcinoma, because 2 cases (9%) of papillary thyroid carcinoma (1 case with obvious cytologic features of papillary carcinoma and another case with equivocal cytology) were negative for CD57 in our study. The reason for this false-negative staining of papillary thyroid carcinoma was unclear, but may reflect sampling limitations; with cell block material, only a small fraction of the tumor is available for staining.
None of the benign thyroid lesions (confirmed by tissue follow-up) stained positively for CD57 in the current study. Although 4 cases in this group did demonstrate weak 1+ luminal staining of the follicular cells, they were considered negative because the immunoreactivity was only luminal, with the absence of a complete membranous staining pattern. Our results differ from those published by Loy et al.19 and Ostrowski et al.,20 in which significant expression of CD57 was reported in benign thyroid lesions. In those 2 studies, however, the staining pattern for CD57 was either cytoplasmic and/or membranous, whereas using the criteria for our study only cases with distinct membranous staining were considered positive. The reason for these differences is not known but may be related to technique, including antigen retrieval, types of antibodies used, or both. The study by Ostrowski et al.20 specifically mentions that all of their papillary thyroid carcinomas revealed distinct membranous and cytoplasmic staining for CD57, whereas the benign thyroid lesions demonstrated only cytoplasmic staining. The study by Loy et al.19 also evaluated both membranous and cytoplasmic staining for CD57, but provided no details regarding distinction between cytoplasmic and membranous staining in their cases.
All the cases in the current study were negative for GLUT-1 immunoreactivity, although the positive control and red blood cells in the cell block serving as an internal positive control stained appropriately. Our results differ from that of Haber et al.23 and Yasuda et al.,24 which reported that GLUT-1 was a highly specific marker for the diagnosis of thyroid carcinoma in tissue sections. The differences between their results and ours for GLUT-1 immunostaining are not known but might be attributable to the differences in criteria for evaluating the immunostain results, primary antibody, or limitations related to the small amount of cellular material present in some cell blocks. Similar discrepancies have been observed in other studies between cell block and tissue sections that were immunostained using different antibodies.41, 42
The findings of our retrospective study indicate that distinct membranous CD57 immunostaining of cell blocks from thyroid aspirates with features suspicious but not diagnostic for papillary carcinoma serves as a useful adjunct. Positive CD57 staining in this context has both a high sensitivity and specificity for the diagnosis of papillary thyroid carcinoma. In the current study, GLUT-1 was found to be of no diagnostic utility.