Expression of galectin-3 in fine-needle aspirates as a diagnostic marker differentiating benign from malignant thyroid neoplasms

Authors


Abstract

BACKGROUND

Galectin-3 is a β-galactoside-binding protein that has been reported to be expressed preferentially in thyroid malignancies. The current study was designed to substantiate this finding further and to establish a presurgical diagnostic modality of differentiating between benign and malignant thyroid neoplasms by analyzing galectin-3 expression in fine-needle aspirates.

METHODS

The expression of galectin-3 was examined immunohistochemically in total of 172 specimens: 45 primary and 20 metastatic papillary carcinomas, 8 primary and 2 metastatic follicular carcinomas, 5 primary and 3 metastatic anaplastic carcinomas, 3 primary medullary carcinomas, 25 follicular adenomas, 3 goiters, and 58 adjacent normal thyroid tissue. Alternatively, epithelial cells were isolated from the fine- needle aspirates of 14 thyroid nodules and subjected to immunoblotting analysis of galectin-3.

RESULTS

Immunohistochemical analysis revealed that all thyroid malignancies of follicular cell origin (including papillary, follicular, and anaplastic carcinomas) showed high and diffuse expression of galectin-3, whereas one of the three medullary carcinomas of parafollicular cell origin displayed weaker and focal expression of galectin-3. In contrast, neither benign thyroid adenomas, goiters, nor normal thyroid tissues expressed galectin-3. Immunoblot analysis of the isolated epithelial cells detected galectin-3 in nine thyroid nodules that were proven histologically to be malignant ( eight papillary carcinomas and one follicular carcinoma) after surgical intervention, whereas galectin-3 was not detected in five nodules proven to be benign follicular adenomas.

CONCLUSIONS

Galectin-3 serves as a marker of thyroid malignancy of follicular cell origin. Analysis of galectin-3 expression in fine-needle aspirates enhances the differential diagnostic accuracy between benign and malignant thyroid neoplasms. Cancer 1999;85:2475–84. © 1999 American Cancer Society.

The most frequently occurring endocrine malignancy is thyroid carcinoma, which is classified histologically into four principal subtypes: papillary, follicular, medullary, and anaplastic or undifferentiated.1 The incidence of each type has been reported to be 70–80%, 5–10%, ≤10%, and ≤10%, respectively.2, 3 Although papillary, follicular, and anaplastic carcinomas originate from the follicular epithelium, they show strikingly different clinical features. Papillary and follicular carcinomas are differentiated, producing thyroglobulin and usually pursuing an indolent course. In contrast, anaplastic carcinoma is a lethal entity and generally presents at an advanced stage. Medullary carcinoma arises from the parafollicular C cells of neural crest origin and produces calcitonin. Approximately 80–90% of this carcinoma occurs sporadically, whereas another 10–15% is associated with the inherited multiple endocrine neoplastic syndromes. Its general prognosis ranges between that of differentiated and anaplastic carcinomas.4, 5

A series of diagnostic procedures including ultrasonography, scintigraphy, and fine-needle aspiration biopsy (FNAB) currently are employed before treatment of thyroid neoplasms; FNAB is the most important modality among them. FNAB boasts excellent diagnostic accuracy for anaplastic and medullary carcinomas, but false-negative results are not unusual in papillary carcinoma.6, 7 Moreover, because the diagnosis of follicular carcinoma is defined by the presence of capsular or vascular invasion,3 FNAB is unable to differentiate benign follicular adenoma from its malignant counterpart. These shortcomings of FNAB become more manifest when a physician encounters a thyroid neoplasm in the absence of regional or distant metastases. Accordingly, it is evident that FNAB is limited in the differential diagnosis between benign and malignant thyroid neoplasms.

An increasing number of vertebrate lectins have been identified and characterized in a diversity of tissues and cells and assumed to play an important role in a variety of biologic processes through interaction with glycoconjugates.8, 9 Galectin is a growing family of β-galactoside-binding proteins defined by two properties: 1) consecutive amino acid sequence in the carbohydrate-binding site and 2) affinity for β-galactosides.10 Galectin-3 is a member of the family that includes Mr 30-kilodaltons (kD), which is comprised of two distinct structural domains: 1) an amino-terminal half containing proline-rich and glycine-rich tandem repeats and 2) a carboxy-terminal half representing the carbohydrate-binding site.11–14 Although galectin-3 is localized predominantly in the cytoplasm, it also is detected in the nucleus, on the cell surface, or in the extracellular environment,10 suggesting multifunctionality of the molecule. A variety of physiologic and pathologic functions have been demonstrated for galectin-3 including cell growth,15 cell adhesion,16, 17 inflammation,18, 19 neoplastic transformation,20 and apoptosis.21, 22

It has been shown in vitro that malignant progression may be associated with increased expression of galectin-3.23 Furthermore, a series of studies employing human surgical materials have suggested the possible involvement of galectin-3 in neoplastic progression in a variety of organs, including the stomach, colon, central nervous system, and thyroid.24–28 Xu et al. recently reported that galectin-3 is expressed preferentially in papillary and follicular thyroid carcinomas at a high level and in medullary carcinoma at a variable level.28 The objective of the current study was to confirm their results by analysis on a larger population and to address the issue of whether preferential overexpression of galectin-3 in thyroid malignancies can serve to help make a presurgical differential diagnosis between benign and malignant thyroid neoplasms by means of FNAB.

MATERIALS AND METHODS

Tissue Specimens

In the current study specimens used for immunohistochemical analysis were obtained from 87 patients who had undergone surgery without any prior treatment for thyroid nodules, including 3 patients with goiters, 25 with adenomas, 45 with papillary carcinomas, 8 with follicular carcinomas, 5 with anaplastic carcinomas, and 3 with medullary carcinomas at Osaka University Medical School Hospital and Osaka Teishin Hospital (Osaka, Japan) between 1986 and 1997. Twenty-five metastatic cervical lymph nodes from 20 papillary carcinomas, 2 follicular carcinomas, and 3 anaplastic carcinomas as well as 58 adjacent normal thyroid tissues from 2 goiters, 20 adenomas, 30 papillary carcinomas, 4 follicular carcinomas, 1 anaplastic carcinoma, and 1 medullary carcinoma also were available for immunohistochemistry. The specimens for histopathologic and immunohistochemical analyses were fixed in 10% neutral formaldehyde solution and embedded in paraffin. Histologic classification was performed according to the World Health Organization.1 For flow cytometric analysis surgically excised fresh materials were obtained from three additional patients with papillary carcinoma at Osaka Teishin Hospital in 1997. The specimens were divided into two groups: adjacent normal tissues and tumor tissues. Alternatively, FNAB specimens obtained from another 14 patients with thyroid nodules were subjected to immunoblot analysis. Histopathologic examination of these nodules was made after surgical intervention at Osaka University Medical School Hospital and Osaka Teishin Hospital between 1997 and 1998, and revealed that these nodules were comprised of eight papillary carcinomas, one follicular carcinoma, and five follicular adenomas. Informed consent was obtained from patients whose surgical materials and FNAB specimens were used for flow cytometric and immunoblot analyses, respectively.

Immunohistochemistry

Immunohistochemistry was performed using a modification of the avidin-biotin peroxidase complex technique. Briefly, 4-μm tissue sections were deparaffinized in xylene, rehydrated in ethanol, incubated with freshly prepared 3% hydrogen peroxide in methanol for 10 minutes to inhibit endogenous peroxidase activity, and washed with phosphate-buffered saline (pH 7.4) (PBS). Normal rabbit serum (5%) was applied to block nonspecific binding sites and removed by blotting. The sections were incubated at room temperature for 2 hours with 2-fold diluted hybridoma supernatant fluid containing rat antigalectin-3 antibody M3/38 produced by TIB166 (American Type Culture Collection, Rockville, MD), washed with PBS, and then incubated at room temperature for 30 minutes with biotinylated secondary antibody from the Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA). After washing with PBS, the sections were incubated with avidin-biotin complex reagent from the kit (Vector Laboratories) for 30 minutes followed by washing with PBS. Subsequently, the sections were incubated with diaminobenzidine (Abbott Laboratories, Abbott Park, IL) for 1 minute to visualize the bound antibody by colored peroxidase reaction product, washed with tap water, counterstained with hematoxylin, washed with tap water and PBS, dehydrated, and finally mounted. Controls receiving either no primary antibody or a nonspecific rat immunoglobulin (Ig) G exhibited no background staining.

Alternatively, immunohistochemistry for Ber-EP4 antigen was performed according to a procedure essentially identical to that mentioned earlier with the following exceptions. Tissue sections were digested with 0.1% trypsin (Sigma Chemical Company, St. Louis, MO) at 37 °C for 60 minutes prior to blocking of nonspecific binding sites, in which normal rabbit serum was replaced with normal horse serum. The primary antibody was mouse antihuman epithelial antigen Ber-EP4 (Dakopatts, Glostrup, Denmark), which was used at a dilution of 1:50 for incubation at 4 °C overnight. Controls receiving either no primary antibody or a nonspecific mouse IgG exhibited no background staining.

Flow Cytometric Analysis

From freshly excised surgical specimens single cell suspensions were made in PBS containing 0.1% bovine serum albumin, followed by fixation and permeabilization with 70% ethanol at -20 °C for > 24 hours. The cells were washed with PBS containing 10% normal rabbit serum (FACS medium), and aliquots of 2 × 105 cells were prepared. The aliquots were incubated overnight at 4 °C with a 1:10 dilution of hybridoma supernatant fluid containing rat antigalectin-3 and a 1:100 dilution of mouse antithyroglobulin (Neomarkers, Fremont, CA). After 3 washes with FACS medium, the cells were labeled with a 1:100 dilution of fluorescein-conjugated goat antirat IgG (Organon Teknika, Durham, NC) and a 1:100 dilution of rhodamine-conjugated goat antimouse IgG (Chemicon International, Temecula, CA) at 4 °C for 30 minutes to localize galectin-3 and thyroglobulin, respectively, and washed 3 times with FACS medium. Subsequently, the cells were analyzed by a FACScan flow cytometer (Becton Dickinson, San Jose, CA). No cross-reactivity between the rat antigalectin-3 primary antibody and the goat antimouse IgG secondary antibody or between the mouse antithyroglobulin primary antibody and the goat antirat IgG secondary antibody was observed. Controls receiving either no primary antibody, a nonspecific rat IgG, or a nonspecific mouse IgG exhibited no detectable labeling.

Isolation of Epithelial Cells from FNAB Specimens

The procedure of FNAB has been described in detail elsewhere.29 Epithelial cells were purified from FNAB specimens by the immunomagnetic isolation technique. Cells in fine needles obtained by aspiration from thyroid nodules were rinsed off with ice-cold RPMI-1640 and collected by centrifugation. All the following procedures were performed at 4 °C. For lysis of red blood cells the cells were resuspended with 400 μL of 0.4% NaCl and immediately with an additional 700 μL of 1.4% NaCl. Subsequently the cells were centrifuged, washed twice, and resuspended with 500 μl of RPMI-1640 containing 1% fetal bovine serum (FBS). The cell suspension was rotated gently for 30 minutes after the addition of 1.6 μl of Dynabeads Anti-Epithelial Cell (Dynal, Oslo, Norway), which had been prewashed according to the supplier protocol. Rosetted cells were isolated by placing the cell suspension in a magnetic field for 5 minutes by means of Dynal MPC-E (Dynal). The isolated cells were washed 3 times by centrifugation and finally resuspended with 200 μL of RPMI-1640 containing 1% FBS. The number of cells was counted with a hemocytometer.

Immunoblotting

The isolated cell-immunobead complexes were dissolved and boiled in reducing sodium dodecyl sulfate (SDS) sample buffer (2% SDS, 62.5 mM Tris-HCl [pH 6.8] 10% glycerol, and 5% β-mercaptoethanol) for 5 minutes and aliquots of the supernatant fluids obtained after centrifugation were subjected to SDS-polyacrylamide gel electrophoresis on 12.5% gels. After electrophoresis, proteins were transferred to Hybond ECL membranes (Amersham, Buckinghamshire, U K) in the presence of 12.5 mM Tris and 120 mM glycine and the membranes were quenched overnight at 4 °C with 5% skim milk (Difco, Detroit, MI) in PBS containing 0.1% Tween 20 and 0.1% NaN3 (PBS-T) to block remaining free protein-binding sites. Blots were probed with 200-fold diluted hybridoma supernatant fluid containing rat antigalectin-3 antibody M3/38 for 1 hour at room temperature, washed with PBS-T, and treated with horse radish peroxidase-conjugated rabbit antirat IgG (Zymed, San Francisco, CA) at 1:4000 dilution for 1 hour at room temperature. After further washing with PBS-T, the blots were processed for revelation with enhanced chemiluminescence (ECL Western blotting detection reagents; Amersham).

RESULTS

Immunohistochemical Analysis of Galectin-3

To address the possible involvement of galectin-3 in neoplastic transformation of the thyroid gland, retrospective analysis of galectin-3 expression was performed. The rat antigalectin-3 antibody recognized a single band migrating at 30 kD when protein extracts from thyroid carcinomas were analyzed by immunoblotting, verifying the specificity of the antibody and its suitability for immunohistochemical analysis (data not shown). As shown in Figures 1A and 1B , no epithelial cells of either normal thyroid or benign adenoma tissues expressed galectin-3. In sharp contrast, all papillary, follicular, and anaplastic carcinomas of follicular cell origin showed diffuse expression of galectin-3 (Figs. 1C, 1D, and 1E). Staining intensity in papillary and follicular carcinomas was higher than that in anaplastic carcinoma. It should be emphasized that in one case in which benign adenoma and papillary carcinoma were present simultaneously in the same individual only papillary carcinoma was positive for galectin-3 and that in another case of papillary and follicular carcinomas combined both carcinomas were positive for galectin-3. Galectin-3 was expressed predominantly in the cytoplasm with localization at the luminal cell surface. Nuclear localization also was observed in some portion of carcinoma cells. Moreover, galectin-3 expression was maintained consistently in carcinoma cells metastasized to regional lymph nodes. With regard to medullary carcinomas of parafollicular C cell origin one case showed a less intense and focal expression of galectin-3 (Fig. 1F), and the other two cases were negative for galectin-3. Collectively, as summarized in Table 1 , neither normal thyroid tissues, benign follicular adenomas, nor nodular goiters expressed galectin-3, whereas all thyroid carcinomas consistently expressed galectin-3 at both primary and metastatic sites with the exception that medullary carcinomas expressed galectin-3 at a variable and inconsistent level. In the stroma fibroblasts, polymorphonuclear inflammatory cells and endothelial and smooth muscle cells of vessels displayed galectin-3. It should be noted that galectin-3 mRNA was detected in all thyroid lesions, including normal and neoplastic tissues (data not shown), when analyzed by reverse transcriptase-polymerase chain reaction (RT-PCR). Most likely, RT-PCR was too sensitive to show the difference in mRNA. Quantitative RT-PCR is required for comparison in mRNA.

Figure 1.

Immunohistochemical analysis of galectin-3 in thyroid lesions. (A) Normal thyroid tissue; (B) follicular adenoma; (C) papillary carcinoma; (D) follicular carcinoma; (E), anaplastic carcinoma; and (F) medullary carcinoma. Note that the papillary, follicular, and anaplastic carcinomas showed diffuse expression of galectin-3, whereas medullary carcinoma displayed focal expression of galectin-3. Neither normal thyroid tissue nor follicular adenoma were found to express detectable amount of galectin-3 (original magnification, ×150).

Table 1. Expression of Galectin-3 in Thyroid Lesions
 MalignantNonmalignant
PapillaryFollicularAnaplasticMedullaryNormalAdenomaGoiter
  • NT: not tested.

  • a

    Number of positive specimens/total number of specimens.

Primary45/45 (100%)a8/8 (100%)5/5 (100%)1/3 (33.3%)0/58 (0%)0/25 (0%)0/3 (0%)
Metastatic20/20 (100%)2/2 (100%)3/3 (100%)NT

Flow Cytometric Analysis of Galectin-3

To substantiate the aforementioned immunohistochemical results further, we performed flow cytometry, which allowed us to analyze expression of galectin-3 more quantitatively. Permeabilized cells were used to increase detection sensitivity inasmuch as the subcellular localization of galectin-3 prominently was cytoplasmic, as shown in Figure 1. In addition, we performed double labeling of galectin-3 and thyroglobulin, which would eliminate possible false-positive results caused by stroma cells positive for galectin-3. Figure 2 depicts representative results of flow cytometric analysis of papillary carcinoma and its adjacent normal thyroid tissue. In normal thyroid tissues all the cells exhibited a thyroglobulin positive (+) galectin-3 negative phenotype. On the contrary, > 90% of the cells from papillary carcinoma tissues showed a thyroglobulin+ galectin-3+ phenotype. The remaining ≤10% of the cells exhibiting the thyroglobulin+ galectin-3 phenotype most likely were contaminating normal follicular epithelial cells because they constituted a population of small-sized cells on forward-angle light scatter (data not shown). These results indicate that galectin-3 is expressed in follicular epithelial cells only after malignant transformation although no histologic subtypes other than papillary carcinoma were examined by flow cytometry.

Figure 2.

Flow cytometric analysis of galectin-3. Single cell suspensions were labeled for two-color flow cytometric analysis using rat antigalectin-3 and mouse antithyroglobulin antibodies. (A) Normal thyroid tissue and (B) papillary thyroid carcinoma.

Immunohistochemical Analysis of Ber-EP4 Antigen

The immunohistochemical and flow cytometric results described earlier indicate the possibility that analysis of galectin-3 expression in FNAB specimens of thyroid neoplasms may enable clinicians to make a better differential diagnosis between benign and malignant phenotypes. If crude materials of FNAB are subjected to analysis of galectin-3 expression directly, false-positive findings may result from stroma cells that are positive for galectin-3. Accordingly, we decided to isolate the epithelial cell population from crude materials of FNAB by means of an immunomagnetic isolation technique, in which magnetic bead-conjugated Ber-EP4 was used. Ber-EP4 is a monoclonal antibody directed against the protein moiety of two 34-kD and 39-kD glycopolypeptide chains, which are not bound covalently.30 Ber-EP4 antigen is highly specific for epithelial cells and negative for nonepithelial cells.30 As an initial step we examined expression of Ber-EP4 antigen in thyroid lesions immunohistochemically to confirm whether magnetic bead-conjugated Ber-EP4 is effective for isolating thyroid epithelial cells. As representatively depicted in Figure 3 , all thyroid neoplastic cells of follicular cell origin, with the exception of anaplastic carcinoma, consistently displayed intense and predominantly membranous positivity for Ber-EP4 antigen with weaker cytoplasmic antigenicity. No Ber-EP4 antigenicity was detected in fibroblasts, polymorphonuclear inflammatory cells, or endothelial and smooth muscle cells of vessels that were positive for galectin-3. Normal thyroid follicular cells showed the same staining pattern whereas medullary carcinoma cells of parafollicular cell origin were negative for Ber-EP4 antigen (data not shown). These results indicate that the immunomagnetic isolation technique using magnetic bead-conjugated Ber-EP4 is useful to purify neoplastic cells from heterogeneous materials obtained by FNAB of thyroid neoplasms except anaplastic and medullary carcinomas.

Figure 3.

Immunohistochemical analysis of Ber-EP4 antigen in thyroid lesions. (A) Follicular adenoma; (B) papillary carcinoma; (C) follicular carcinoma; and (D) anaplastic carcinoma. Note that endothelial and smooth muscle cells of vessels and fibroblasts were negative for Ber-EP4 antigen, whereas all neoplastic cells except anaplastic carcinoma showed intense antigenicity (original magnification, A, B, and C: ×150 and D: ×125).

Immunoblotting Analysis of Galectin-3 in FNAB

We routinely performed FNAB twice on each patient with thyroid nodules and succeeded by the immunomagnetic isolation technique to purify a sufficient amount of epithelial cells to allow an immunoblot analysis (cell number ×10−5: 1.47 ± 0.31, mean± the standard error). When papillary carcinoma was examined by immunoblotting, galectin-3 was detected with a cell number as small as 1 × 104 (Fig. 4A). Conversely, in the case of follicular adenoma, galectin-3 was not detected even with a cell number as large as 5 × 104 (Fig. 4A). Accordingly, we used 2.5 × 104 of isolated epithelial cells for further immunoblot analyses. As depicted in Figure 4B, a total of 14 samples were examined, with 9 samples found to be positive for galectin-3. Histopathologic examination after surgical intervention revealed that all nine samples that were found to be positive for galectin-3 were derived from malignancies (eight papillary carcinomas and one follicular carcinoma), whereas all five samples found to be negative for galectin-3 were from benign follicular adenomas. Samples of papillary carcinomas (Lanes 2, 3, 7, 9, 10, 11, 13, and 14) showed stronger expression of galectin-3 than those of follicular carcinoma (Lane 12). It should be noted that the immunoblot analysis of 5 × 104 cells yielded similar results (data not shown). These results suggest that analysis of galectin-3 expression in FNAB is useful in differentiating between benign and malignant thyroid neoplasms without surgical intervention.

Figure 4.

Immunoblotting analysis of galectin-3 in fine-needle aspirates. (A) Varying number of epithelial cells isolated from fine-needle aspirates of follicular adenoma and papillary carcinoma were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by immunoblotting with anti-galectin-3 antibody. (B) 2.5 × 104 of epithelial cells isolated from fine-needle aspirates of thyroid nodules were resolved by SDS-PAGE followed by immunoblotting with anti-galectin-3 antibody. Histopathologic examination after surgical intervention revealed that Lanes 1, 4, 5, 6, and 8, Lane 12, and Lanes 2, 3, 7, 9, 10, 11, 13, and 14 were follicular adenomas, follicular carcinoma, and papillary carcinomas, respectively. Migration positions of molecular mass markers are shown in the left margin.

DISCUSSION

A series of in vitro evidence has been accumulated to support the contention that galectin-3 plays an important role in malignant transformation and metastasis. Highly metastatic melanoma and fibrosarcoma cells express a higher level of galectin-3 than their nonmetastatic or weakly metastatic counterparts.23 T-cells infected with human T-cell leukemia virus type 1 exhibit a high level of galectin-3 in contrast to uninfected cells, which do not express a detectable amount of the protein.31 Overexpression of recombinant galectin-3 in immortalized mouse fibroblasts results in the acquisition of increased tumorigenic and metastatic potential.20 Moreover, we recently have shown that cell surface galectin-3 molecules mediate homotypic cell aggregation through interaction with soluble complementary glycoconjugates, culminating in the enhancement of metastatic formation.16, 17 We also have shown that modified citrus pectin, capable of binding to galectin-3, inhibits experimental lung colonization of murine melanoma cells when injected intravenously32 as well as spontaneous lung metastases of rat prostate carcinoma cells when administered orally.33 This inhibition most likely is due to the modified citrus pectin preventing the tumor cells from undergoing complementary glycoconjugate-induced homotypic aggregation and adhering to laminin.34

Additional evidence further supporting the involvement of galectin-3 in malignant transformation and metastasis comes from several experiments using surgically excised specimens. It has been shown that galectin-3 expression is increased along with neoplastic progression in the colon and is in proportion to advanced Dukes stages as well as decreased long term patient survival.24, 25 Likewise, galectin-3 expression has been demonstrated to be elevated in certain primary gastric carcinomas and their metastases compared with adjacent normal mucosa.26 In the central nervous system galectin-3 has been suggested to correlate with the malignant potential of tumors.27 Alternatively, galectin-3 has been shown to be a useful tumor marker for anaplastic large cell lymphoma.35 It is increasingly evident that an enhanced expression of galectin-3 correlates with malignant transformation and progression toward metastasis in some type of tissues and cells.

There has been accumulating evidence that galectin-3 is overexpressed in thyroid malignancies. A series of recent reports have demonstrated that papillary carcinomas display an intense and consistent expression of galectin-3 without exception.28, 36–38 This is substantiated further by the current study showing that all cases of papillary carcinoma express galectin-3 at a high level. Overexpression of galectin-3 in papillary carcinoma is beyond question. Alternatively, expression of galectin-3 in follicular carcinomas is inconsistent between the reports. Xu et al. and Orlandi et al. have shown galectin-3 positivity in all of 7 and all of 17 cases, respectively,28, 38 whereas Fernandez et al. and Cvejic et al. detected positivity in 4 of 8 and 11 of 15 cases, respectively.36, 37 In the current study we have demonstrated that all eight follicular carcinomas expressed a high level of galectin-3 in a diffuse pattern. Overall 47 of 55 follicular carcinomas (85.5%) were found to be positive for galectin-3. Although it appears likely that a small population of follicular carcinomas are negative for galectin-3, it remains elusive whether the negativity is true or false. Expression of galectin-3 in follicular carcinomas predominantly is diffuse, but occasionally focal.28, 36–38 Orlandi et al. failed to detect galectin-3 positivity in some follicular carcinomas, whereas they succeeded in showing focal positivity in the same cases when analyzed with differently prepared sections.38 Immunohistochemical analysis of follicular carcinomas with focal galectin-3 expression possibly may yield a false-negative result when the areas negative for galectin-3 are sampled. Analyses with immunohistochemistry of serial sections and/or immunoblotting on a large population of follicular carcinomas will determine whether they all are positive for galectin-3. It is evident that at least the majority of follicular carcinomas express galectin-3. With regard to anaplastic carcinomas, we found galectin-3 positivity in all five examined cases and the staining intensity was weaker than that of differentiated carcinomas (i.e., papillary and follicular carcinomas), whereas medullary carcinomas expressed a variable and inconsistent amount of galectin-3, which may reflect the difference in origin in that this carcinoma arises from parafollicular C cells of neural crest origin. These results are in agreement with the recent reports by other investigators.36–37

Collectively, thyroid carcinomas of follicular epithelial cell origin (i.e., papillary, follicular, and anaplastic carcinomas) express a high level of galectin-3 in contrast to benign thyroid adenomas, goiters, and normal thyroid tissues, which do not show a detectable amount of the protein. Galectin-3 can serve as a marker to differentiate between benign and malignant thyroid neoplasms. It is interesting to note that increased expression of galectin-3 is conserved in metastatic regional lymph nodes. It is possible that the acquisition of the phenotype expressing a high level of galectin-3 occurs as an early event during malignant transformation in the follicular epithelium of the thyroid gland and is maintained throughout metastatic cascade. However, it remains unknown whether the increased expression of galectin-3 is a consequence or a cause of the acquisition of the malignant phenotype. A clue toward solving this problem will be obtained from experiments such as transfection of cDNA-encoding galectin-3 into normal follicular cells of the thyroid.

The immunohistochemical analysis described in the current study revealed that galectin-3 is distributed prominently in the cytoplasm with occasional nuclear localization. A possible explanation for this heterogeneous subcellular distribution pattern can be drawn from the previous findings. It has been shown that in quiescent mouse 3T3 cells galectin-3 is phosphorylated and localized mainly in the cytoplasm, whereas in proliferating cells the lectin is concentrated in the nucleus in both phosphorylated and unphosphorylated forms.15, 39 Thus, the cells exhibiting nuclear galectin-3 may represent a population in proliferating state. With regard to a possible involvement of galectin-3 in cell growth regulation, of special interest is a recent finding that galectin-3 is a cell death suppressor interfering with a common pathway of apoptosis that involves bcl-2.21, 22

Although we believe FNAB is the most useful diagnostic modality for thyroid nodules, false-negative results are not unusual for papillary carcinoma, which is the most common thyroid malignancy.6, 7 The accuracy of FNAB in the preoperative diagnosis of nodular thyroid diseases primarily depends on the skill and experience of cytopathologists. Moreover, because the diagnosis of follicular carcinoma is defined by the presence of capsular or vascular invasion,3 FNAB is significantly unable to differentiate benign follicular adenoma from its malignant counterpart. Alternatively, FNAB generally allows a definitive diagnosis of anaplastic and medullary carcinomas.6, 7 Aggressive tumor growth and spread and an increased serum level of calcitonin represent clinical features characteristic of anaplastic and medullary carcinomas, respectively, which facilitate their diagnosis. Collectively, improvement of FNAB is required, especially for the diagnosis of papillary and follicular carcinomas.

In the current study we developed the immunomagnetic isolation technique of purifying epithelial cells from FNAB specimens by use of Ber-EP4 and analyzed expression of galectin-3 in the isolated cells by immunoblot analysis. Because anaplastic carcinoma is negative for Ber-EP4 antigen (most likely due to dedifferentiation) and medullary carcinoma is not of epithelial origin, the immunomagnetic isolation technique described in the current study is incapable of purifying these carcinoma cells from FNAB. This appears less significant because FNAB and clinical symptoms are good diagnostic indicators for anaplastic and medullary carcinomas as mentioned earlier. In turn, when the immunomagnetic isolation technique fails to capture epithelial cells from FNAB despite the presence of a significant number of cells after red blood cell lysis, it is indicative of either anaplastic or medullary carcinomas. Accordingly, the immunoblot analysis of galectin-3 in epithelial cells isolated from FNAB should serve to make a differential diagnosis between benign follicular adenomas and differentiated carcinomas (i.e., papillary and follicular carcinomas). As shown in Figure 4, immunoblot analysis of samples obtained from thyroid nodules that later were proven to be either papillary or follicular carcinomas by histopathologic examination after surgical intervention never failed to detect galectin-3. In sharp contrast, histopathologically proven benign follicular adenomas showed no detectable amount of galectin-3. Although only one case of follicular carcinoma was included in the current study, it is interesting to note that the immunoblot analysis detected galectin-3 in a sample prepared from the case because FNAB failed to make a definitive diagnosis of follicular carcinoma. Because follicular carcinomas occasionally demonstrate focal expression of galectin-3 as discussed earlier, it is necessary to sample the area positive for galectin-3. Plural analyses of galectin-3 expression in FNAB specimens independently sampled from a thyroid nodule appear important in reducing the possibility of false-negative results, especially when conventional cytologic examination reveals features typical of malignancy, such as an increasing nuclear:cytoplasmic ratio. It should be noted that a challenge to our results was published independently while our study results were being written. Orlandi et al. prepared serial paraffin embedded sections from FNAB samples, examined the expression of galectin-3 by immunohistochemistry, and showed the same results as the current study.38

Galectin-3 is a valuable tumor maker for thyroid malignancies of follicular epithelial cell origin. Immunoblotting analysis of galectin-3 in FNAB serves as a useful presurgical diagnostic modality with which to differentiate benign follicular adenoma from its malignant counterpart as well as papillary carcinoma. Thus, the analysis appears able to overcome the intrinsic limitations of FNAB and eliminate potentially unnecessary surgical intervention.

Acknowledgements

The authors thank Masahide Sakai for technical assistance in flow cytometry.

Ancillary