Dr. Pusztaszeri thanks the Nuovo-Soldati Foundation for supporting his visiting fellowship to the Massachusetts General Hospital.
In contrast to other primary thyroid neoplasms and benign thyroid tissue, it has been demonstrated histologically that dendritic cells (DCs) are associated with papillary thyroid carcinoma (PTC). However, the presence and potential diagnostic value of DCs in thyroid fine-needle aspirations (FNAs) have not been previously described.
The authors quantitatively assessed for the presence of DCs that were positive for cluster of differentiation 1a (CD1a) (a 43-49 kD protein expressed on DCs and cortical thymocytes) in cytologic samples of histologically confirmed PTC (n = 31) and in a control group of benign thyroid nodules (BTNs) (n = 29) using immunocytochemical staining with antibodies against CD1a. A subset of the corresponding PTCs (n = 11) and BTNs (n = 10) from surgical resection specimens also were assessed immunohistochemically for both CD1a and Langerin (a type II transmembrane cell surface receptor produced by Langerhans cells).
CD1a-positive DCs were identified in 97% PTCs (n = 30 of 31 PTCs) in thyroid FNA specimens. DCs were largely present in 2 distinct patterns: either as isolated DCs in the background (n = 29 of 31) and/or associated with tumor cells (n = 30 of 31). Tumor-associated DCs (mean ± standard deviation: 6.44 ± 6.13 DCs per tumor cluster) exhibited multiple dendritic cytoplasmic processes extending over and between malignant cells within groups. The 3 PTC cases with the least DCs corresponded to the follicular variant at excision. In contrast, only 31% of BTNs (n = 9 of 29 BTNs; P = .0048) contained CD1a-positive DCs. When DCs were present in BTN, they were isolated primarily in the background (27%; n = 8 of 29), although 17% of BTNs (n = 5 of 29) contained rare DCs among thyrocytes, revealing both patterns in 4 cases. Both thyrocyte-associated DCs and background DCs were more numerous in PTC FNAs than in BTN FNAs, but only the thyrocyte-associated group of DCs was statistically significant (P < .0001 and P = .1173, respectively). Similar findings were reported in histologic samples in which all PTCs examined (n = 11 of 11) contained both CD1a-positive and Langerin-positive DCs; only 20% of BTNs (n = 2 of 10) contained rare DCs.
Dendritic cells (DCs) comprise a group of large, nonlymphocytic, class II major histocompatibility complex-positive mononuclear cells that are crucial in the presentation of antigen during the elicitation of primary immune responses against pathogens and tumors.1-5 They originate from hematopoietic precursors in the bone marrow and subsequently migrate to peripheral tissues, where they have the ability to capture and process antigens.1, 6 Langerhans cells (LCs) are a specific subset of DCs typically located in the skin and mucosal surfaces7 and are characterized ultrastructurally by the presence of Birbeck granules8 and immunohistochemically by the surface antigen expression of cluster of differentiation 1a (CD1a) (a 43-49 kD protein expressed on DCs and cortical thymocytes), Langerin/CD207 (a type II transmembrane cell surface receptor produced by Langerhans cells), and E-cadherin.1, 2, 9-12 After receiving an appropriate signal (eg, cytokine exposure), immature DCs, including LCs, migrate from peripheral tissues to lymphoid organs, where they participate in antigen-specific immune responses.7, 13, 14 In the course of migration, they lose their ability to capture and process antigens (including tumor antigens) and acquire the ability to stimulate resting T cells. Mature DCs are characterized by cell surface expression of human leukocyte antigen class I and II, adhesion, and costimulatory molecules as well as the lysosome-associated membrane glycoprotein DC-lysosome-associated membrane protein (LAMP)/CD208.15-18 Tumor-associated DCs have been recognized for many years in several tumors arising from various locations, including the thyroid gland, lung, breast, colon, and skin.2, 20-25 Moreover, immature CD1a-positive DCs tend to reside within the tumor, whereas mature DCs tend to be located in peritumoral areas (DC compartmentalization).3, 22
It has been observed that, in the thyroid gland, papillary thyroid carcinomas (PTCs) contain significantly higher numbers of DCs in surgical specimens than other follicular-derived neoplasms and normal thyroid tissues.13, 26-31 These cells usually are identified by immunohistochemistry or electron microscopy, and it has been proposed that the presence of DCs may serve as an ancillary diagnostic criterion for PTC, similar to psammoma bodies or multinucleated giant cells. Related to their important role in the immunologic defense mechanisms of the host against tumor, the quantity of DCs in tumors also has been correlated with clinical outcome and survival in several malignancies, including PTC.2, 12, 22, 25, 28, 32-34 In the study by Schroder et al,28 patients with a dense infiltrate of S-100-positive DCs in PTCs had a more favorable prognosis (less tumor recurrence and death resulting from cancer), independent of other morphologic and clinical features. However, a relationship between the extent of DC infiltrates and the prognosis of PTC was not confirmed in other studies.27 Still, thyroid cancers with a poor prognosis (eg, poorly differentiated thyroid carcinoma and undifferentiated carcinoma) were characterized by markedly reduced DC tumor infiltrates.27
Because fine-needle aspiration (FNA) is the leading method for the initial evaluation of a thyroid nodule, the presence of DCs in thyroid FNAs may be significant. To the best of our knowledge, the presence, amount, and diagnostic utility of DCs in thyroid FNAs have not been previously studied. The challenge of using DCs as a cytologic marker of PTC is that DCs may be difficult to recognize based on cytologic features alone without the aid of immunocytochemistry. In the current study, we used a series of 31 thyroid FNA PTC specimens to describe the presence and potential diagnostic utility of CD1a-positive DCs in thyroid FNA specimens.
MATERIALS AND METHODS
Thirty-one FNA specimens from sequential patients with PTC and 30 FNA specimens from sequential patients with benign thyroid nodules (BTNs), all with corresponding surgical resection specimens obtained between January 1, 2011 and December 31, 2011, were retrieved from the archival files of the Department of Pathology at the Massachusetts General Hospital (Boston, Mass). The FNA slides and the corresponding histologic tumor slides were reviewed to confirm the diagnosis. One BTN was excluded upon review of its histology (“atypical adenoma”) for a total of 29 cases of BTN. For each case, 1 cytologic slide (56 smears and 4 ThinPrep slides; Quest Diagnostics, Madison, NJ) that had adequate cellularity (≥10 clusters of ≥50 cells per slide) was selected for immunocytochemical evaluation. For comparison, a random subset of paraffin blocks from the corresponding resection specimens (11 cases of PTC and 10 cases of BTN) were selected for immunohistochemistry using CD1a and Langerin. The study was conducted under approval of the Massachusetts General Hospital Institutional Review Board (IRB protocol 2012-P-000324/1; MGH).
For immunocytochemical evaluation, selected cytologic slides were incubated at room temperature in xylene for 2 or 3 days to remove the coverslips. The slides were then transferred into 95% ethanol; a separate destaining of the Papanicolaou-stained smears was not performed. Slides were then routinely processed for immunocytochemistry with the primary antibody CD1a (prediluted antibody; Leica Microsystems Inc., Buffalo Grove, Ill). Immunocytochemical and immunohistochemical studies were performed on the Leica Microsystems Bond III automated tissue staining system (Leica Microsystems, Inc.) using the Polymer Refine Detection Kit. Antigen retrieval was marker dependent and consisted of heat-induced epitope retrieval using citrate, pH 6. The process was carried out for 30 minutes before primary antibody incubation. The tissues were washed and incubated with the primary antibodies indicated above. Endogenous peroxidase activity was blocked by H2O2, followed by incubation with postprimary and polymer. Antigen detection was performed using the diaminobenzidine chromogen step. Histologic tissues were counterstained with hematoxylin.
The conditions for optimal immunocytochemical staining of FNA smears were determined initially for both CD1a and Langerin (1:100 dilution; Leica Microsystems Inc., Buffalo Grove, Ill), because both are considered specific markers of LCs.2, 4, 5, 9-11 To do this, alcohol-fixed smears were prepared postoperatively from resections of papillary carcinoma. Brief fixation using 10% formaldehyde did not result in any improvement in immunocytochemical staining; and, under the conditions selected, nonspecific background staining was negligible. Both primary antibodies, CD1a and Langerin, produced identical immunocytochemical staining patterns, although CD1a immunostaining produced a more intense signal while maintaining a clean background. Therefore, CD1a was selected to evaluate the experimental group of FNA cases in this study.
For the immunohistologic assessment, 5-μ-thick sections of formalin-fixed, paraffin-embedded blocks corresponding to a randomly selected subset of the FNA study group and an FNA control group were prepared. Slides were deparaffinized and dehydrated before immunohistochemical evaluation with the primary antibodies CD1a and Langerin using the Leica Microsystems Bond III automated tissue staining system as described above. Positive controls (postoperative smears of PTC for CD1a immunocytochemistry and thymus and skin for CD1a and Langerin immunohistochemistry, respectively) were available for each round of immunochemistry.
Quantification of Dendritic Cells in Cytologic Specimens and on Histology
The number of DCs (defined as CD1a-positive cells with morphology consistent with DCs) was assessed quantitatively on the cytologic preparations and corresponding histologic slides. For cytologic preparations, 2 separate DC counts were made for each slide: 1 for isolated DCs in the background and 1 for DCs associated with tumor cell clusters or thyrocyte clusters for PTC and BTN, respectively. For isolated DCs, the total number of DCs was counted in 5 separate low-power fields at ×100 magnification from the areas in which they were the most numerous, and a mean value from these 5 fields was calculated. For tumor cell clusters or thyrocyte clusters, the number of DCs per cell cluster (approximately 50-100 cells per group) was counted, and the mean value of the 5 highest counts was calculated. For histologic evaluation of DCs, the total number of DCs in the lesion per slide was counted and was expressed as the total number of DCs per millimeter of lesion.
The significance of differences between groups was evaluated by using unpaired Student t tests (GraphPad Software, La Jolla, Calif). P values < 0.05 were considered statistically significant.
The clinical and pathologic features of the study cases are summarized in Table 1. We observed no statistically significant correlations between clinicopathologic features and the presence of CD1a-positive DCs in cytologic or histologic samples. Tables 2 and 3 list the distribution of CD1a-positive DCs in PTCs and BTNs according to cytology and histology, respectively. CD1a-positive DCs were identified in 97% of PTC FNA specimens (n = 30 of 31). There were isolated DCs in the background in the majority of cases (n = 29 of 31), but DCs were associated more typically with tumor cell clusters (n = 30 of 31). The 3 PTCs with the least DCs, including 1 without DCs (in the background or among tumor cells), corresponded histologically to the follicular variant of PTC. In contrast, only 31% of BTNs (n = 9 of 29) contained CD1a-positive DCs (P = .0048) (Fig. 1A). When DCs were present, they were primarily isolated cells in the background (n = 8 of 29), and 17% of BTNs (n = 5 of 29) contained rare DCs among thyrocytes. The mean (±standard deviation) number of thyrocyte-associated, CD1a-positive DCs in PTCs was 6.44 ± 6.13 DCs per cell cluster compared with 0.345 ± 1.02 DCs per cell cluster in BTNs (P < .0001) (Table 2). The average number (±standard deviation) of isolated background, CD1a-positive DCs also was greater in PTCs (7.6 ± 14.39 DCs) compared with BTNs (2.7 ± 8.71 DCs), but the difference did not reach statistical significance (P = .1173).
Table 1. Clinicopathologic Features of Papillary Thyroid Carcinomas and Benign Thyroid Nodules
Cluster of differentiation 1a (CD1a) is a 43-kD to 49-kD protein expressed on DCs and cortical thymocytes.
Background and tumor-associated DCs were not microscopically apparent in Papanicolaou-stained FNA PTC specimens. By using anti-CD1a immunocytochemical staining to observe DCs, tumor-associated DCs were characterized by multiple dendritic cytoplasmic processes radiating out from a central, polygonal-shaped cell body (Fig. 1B-E) imparting an arachnidian appearance. The DCs in tumor cell clusters exhibited multiple delicate, intercalating, cytoplasmic dendritic processes that extended over the length of 3 to 5 tumor cells (Fig. 1B-E). The DC cytoplasmic processes were approximately 5 μm thick and contained a small cytoplasmic expansion at the ends of the cell processes (Fig. 1E). In contrast, isolated background DCs had rounded morphology, either with or without abortive cytoplasmic extensions, imparting a more classic histiocytoid appearance (Fig. 1F).
Immunohistochemical evaluation of a subset of the corresponding histologic sections for both CD1a and Langerin were similar. One hundred percent of PTC specimens (n = 11 of 11) contained CD1a-positive and Langerin-positive DCs, whereas only 20% of BTN specimens (n = 2 of 10) contained rare DCs (P = .003) (Fig. 2). In PTC specimens, the majority of DCs were located at the periphery of the tumor, often within the fibrovascular core, interdigitating with tumor cells that lined the papillae (Fig. 2B,C). DCs were not identified in the normal thyroid tissue except in 1 BTN specimen that had rare intranodular DCs, in which scattered DCs also were present among thyrocytes that lined the normal thyroid follicles. Qualitatively, immunostaining for CD1a was similar on both histology and cytology.
To the best of our knowledge, this is the first study to specifically assess the cytologic appearance of CD1a-positive DCs in thyroid FNA specimens from PTC. We observed that CD1a-positive DCs were present in the majority of PTCs (97%) compared with FNA specimens from BTN, in which only 17% had thyrocyte-associated, CD1a-positive DCs (P < .0001). Our findings are compatible with previous histologic studies on thyroidectomy specimens, in which increased numbers of DCs also were reported in resected PTC specimens compared with other thyroid tumors, in which DCs were mostly absent or rare.13, 26-31 Our study further extends this peculiar feature of PTC to cytologic preparations.
Tumor-associated DCs in PTC have been described histologically as scattered cells within solid tumor islands extending characteristic dendritic processes.28, 29 The majority of CD1a-positive DCs have been described in the stroma, with a small number infiltrating malignant thyrocytes.29 Rarely, DCs were present in the luminal space.29 When viewed under the electron microscope, DC processes are long (>10 μm) and thin and are either spiny or sheet-like.4 When viewed under phase-contrast microscopy, DCs extend branching, delicate processes or veils in many directions from the cell body.4 In the current study, we used CD1a immunocytochemistry to highlight the in situ cytologic appearance of DCs and observed that DCs were intimately associated with multiple tumor cells within a group. Although intratumoral DCs had multiple cell processes, imparting a spider-like appearance, isolated DCs in the background had a histiocytoid morphology. It is noteworthy that 1 cell type was previously described in the literature as an unusual cell of uncertain lineage referred to as “histiocytoid” in cytologic smears of PTC FNA specimens.35 This finding raises the possibility that some of these cells may represent DCs; although, in those studies, the histiocytoid cells were not evaluated immunocytologically for CD1a or Langerin.35 In the only case that, to our knowledge, was tested immunohistochemically, the histiocytoid cells were positive for cytokeratin, suggesting that they were not of DC origin and more likely represented an unusual form of tumor cell.36
The pathophysiology of the immune system's response to tumors like PTC is complex. The association of DCs with PTC has been linked to the production of chemokines, which attract peripheral DCs.13, 14, 29 Yamakawa et al29 demonstrated that neoplastic PTC cells more frequently were reactive with antibodies against interleukin-1 alpha (IL-1α) and tumor necrosis factor alpha (TNF-α) than neoplastic follicular carcinoma cells, a finding that correlates with the increased frequency of tumor-associated DCs observed in our cohort and in other histologic studies.13, 26-31 Those data suggest that cytokines like IL-1α and TNF-α released from PTC tumor cells and from associated stromal cells may regulate tumor infiltration and the immunologic response of DCs.29 Another contributing chemokine may include PTC-expressed macrophage inflammatory protein 3α (MIP-3α), which have demonstrated the ability to specifically attract immunologically naive DCs.13, 14 It also has been suggested that qualitative and quantitative differences in the blood vessel supply to the tumor may contribute to an increased number of DCs in PTCs relative to follicular adenomas (FAs) and carcinomas. In support of this suggestion, intercellular adhesion molecule 2 (ICAM-2), which promotes the transendothelial migration of DCs, reportedly was expressed sparsely on endothelial cells in follicular carcinoma in contrast to its expression in PTC.13
It is noteworthy that the 3 PTCs in our series that had the fewest CD1a-positive DCs were follicular variants of papillary carcinoma (FVPTCs). This also is consistent with a histologic study by Proietti et al,37 who assessed the presence of inflammatory cells, including CD1a-positive DCs, in 91 FVPTCs and in 44 FAs. Although those investigators observed significantly greater numbers of intralesional and perilesional CD1a-positive DCs in FVPTCs compared with FAs (P = .0001), intralesional and perilesional CD1a-positive DCs were present in only 46% and 21% of FVPTCs, respectively, and in 3% of FAs. This contrasts with the higher reported rates of DCs in classic PTC.13 The authors also assessed the presence of immunologically mature (DC-LAMP-positive) DCs, but no differences were observed between FVPTCs and FAs.37
We observed no statistical correlation between several clinicopathologic features, including age, sex, tumor size, multifocality, lymph node metastases, or extrathyroidal extension and the quantity of PTC-associated, CD1a-positive DCs. However, in our study, we did not examine the relationship between the quantity of CD1a-positive DCs and clinical outcome. Several studies have suggested an association between the amount of DCs observed on histology in various cancers and prolonged survival, including in PTC,2, 22, 25, 28, 32, 33 but this remains controversial.25 Although a high number of tumor-infiltrating DCs may suggest a strong immune response against the tumor, the relationship is more complex. In patients with cancer and in animal tumor models, the DCs present in blood, tumor tissues, and draining lymph nodes are often functionally defective.38-40 Some studies have demonstrated that tumor-infiltrating DCs lack costimulatory molecules and possess poor T-cell–stimulatory capacity or induce anergy.41, 42 Moreover, tumor-derived factors, such as vascular endothelial growth factor, transforming growth factor-β and IL-10, can induce apoptosis and inhibit functional immunologic maturation of DCs.39, 40, 43, 44 Immature DCs may even mediate tolerance instead of immune activation.45
In conclusion, we have demonstrated that CD1a+ DCs were present in FNA specimens from PTC, typically as arachnidian cells in close association with tumor cells, whereas they were more rare in FNA specimens from BTN. Although these findings may be incompletely understood, they reflect the involvement of the immune response to PTC and suggest a possible ancillary diagnostic role for DCs in the evaluation of thyroid FNA specimens.
This study was supported by the Mini-Vickery Award grant from the Department of Pathology, Massachusetts General Hospital. M.P.P. received support from the Nuovo-Soldati Foundation for his visiting fellowship to Massachusetts General Hospital.