Objective To investigate the expression of connexin 43 (Cx43) in benign vs malignant thyroid tissue for potential use as a diagnostic marker.
Design Retrospective study.
Subjects Thyroid specimens were obtained from 50 patients who underwent partial or total thyroidectomy at the New York Eye and Ear Infirmary, New York, New York, between 1999 and 2007. They included goiter (n = 5), follicular adenoma (n = 15), follicular carcinoma (n = 17), papillary thyroid carcinoma (PTC) (n = 7) and the follicular variant of PTC (n = 6).
Interventions Tissue sections (5 μm) were immunohistochemically stained for Cx43 with the avidin-biotin-peroxidase method using an automated stainer. The Cx43 membrane staining pattern was evaluated.
Results Twenty-three of 30 cancer specimens (77%) revealed a loss of Cx43 plaque staining at the cellular membrane compared with only 3 of 20 benign specimens (15%). Among the malignant specimens, loss of Cx43 plaque staining was observed in 11 of 17 follicular carcinomas (65%), 5 of 6 follicular variants of PTC (83%), and 7 of 7 PTCs (100%). In contrast, only 3 of 15 adenomas (20%) and 0 of 5 goiter samples demonstrated loss of Cx43 plaque staining at the cell membrane.
Conclusion Our data provide evidence that the absence of Cx43 plaque staining is associated with thyroid cancer and thus holds potential clinical utility as a marker for malignant disease.
Fine-needle aspiration (FNA) is recommended as a diagnostic tool for the initial screening of patients with thyroid nodules. Approximately 10% of thyroid FNAs are considered “suspicious for malignancy,” yet only a small number of these aspirates are confirmed to be malignant on subsequent thyroid excision. A common feature in nondiagnostic aspirates deemed to be suspicious is the presence of a hypercellular, microfollicular architecture with scant colloid in the absence of signature cytologic characteristics of malignancy. Well-differentiated follicular carcinoma cannot be reliably distinguished from a benign adenoma on the basis of an aspirate alone, thus posing a particular challenge.
Studies on thyroid folliculogenesis have shown that the formation of fully functional thyroid follicles is highly dependent on connexin (Cx) gap junction (GJ) proteins.1-3 Gap junctions are transmembrane channels that are involved in intercellular communication and coordinated secretion between cells. Each GJ is composed of 2 connexons, or hemichannels, which in turn are composed of 6 Cxs. Twenty-one different types of Cxs have been identified to date, and different Cx isotypes may be found within the same GJ.4 Connexins are synthesized in the endoplasmic reticulum, assembled into connexons in the Golgi apparatus, and transported to the cell membrane, where they aggregate into GJs, clump together into plaques, and interact with GJs of neighboring cells. The formation and degradation of GJs are dynamic processes, with a half-life of 1.5 to 5 hours in most tissue types.4 Loss of GJ intercellular communication via Cx downregulation has been implicated in carcinogenesis in a variety of neoplasms; therefore, Cxs have been thought to be a type of tumor suppressor gene.5 Decreased expression of Cx43 in particular has been reported in human prostate, lung, brain, and breast cancers.5 Cx43, Cx32, and Cx26 have been identified in thyrocytes, and animal studies have shown a link between Cx32 expression and thyroid folliculogenesis.3 To our knowledge, no studies to date have examined the expression of Cx43 in thyroid carcinoma.
As Cx family GJ protein expression is associated with thyroid follicular differentiation, its loss may therefore be a biomarker of the malignant phenotype. The objectives of this study were to examine the immunohistochemical expression of Cx43 in normal and neoplastic human thyroid tissue and to determine its potential as a tumor marker for future aspirate studies.
After institutional review board approval was obtained, patients who had undergone partial or total thyroidectomy at the New York Eye and Ear Infirmary, New York, New York, between 1999 and 2007 for both benign and malignant disease were identified. A total of 50 thyroid patient specimens were collected. Tissues included 5 hyperplastic goiters, 15 follicular adenomas, 17 follicular carcinomas, 7 papillary thyroid carcinomas (PTCs), and 6 follicular variants of PTC (FVPTCs). Clinical staging and prognostic data were obtained from patient medical records. The parameters that were recorded include age at diagnosis, sex, primary tumor size, evidence of lymph node disease or distant metastasis, and evidence of extrathyroidal invasion. Various staging and prognostic indexes, including the TNM, AMES (age, distant metastasis, tumor extent, and size), and MACIS (distant metastasis, age, completeness of primary tumor resection, local invasion, and tumor size) systems, were calculated.
Immunohistochemical analysis was performed with the avidin-biotin-peroxidase method. All specimens were fixed in formalin for 24 hours before paraffin embedding and storage. Five-micron sections of each specimen were placed on glass slides, together with a section of normal thyroid tissue from the same patient (internal positive control) and a section of normal thyroid tissue from a different patient (external positive control). One additional section of normal thyroid tissue from the source patient, untreated with primary antibody, was used as a negative control. The specimens were deparaffinized, pretreated with Tris-based buffer (CC1, Ventana Medical Systems, Tucson, Arizona) (pH, 8) for 30 minutes, and incubated with primary rabbit, polyclonal, anti–Cx43 antibody (Cell Signaling, Danvers, Massachusetts) at a 1:50 dilution for 1 hour at 37°C. An automated stainer (Benchmark XL; Ventana Medical Systems) was used for color development with an indirect biotin-streptavidin system for detecting mouse IgG, mouse IgM, and rabbit primary antibodies (iVIEW DAB Detection Kit; Ventana Medical Systems), which includes a biotinylated immunoglobulin secondary antibody composed of goat anti–mouse IgG and IgM, goat anti–rabbit IgG, and a protein block. The secondary antibody incubation was performed for 8 minutes at 37°C. Specimens were then mounted and photographed with a digital microscope. Qualitative changes in staining of Cx43 between benign and malignant thyroid tissue were analyzed by the second author (A.D.P.) and by a senior member of the pathology department (S.M.). Specimens were evaluated based on the intensity of staining of particular cellular structures and compared with staining exhibited on normal tissue within the same slide. The presence or absence of discrete plaques at the cell membrane was recorded.
Functional Cx43 normally demonstrates staining at the cell membrane in close association with tight junctions. Distinct clumping of Cx43 staining at the cell membrane was defined as “plaque staining” and represents the configuration of individual Cx proteins into formed GJs (Figure). Membranous and cytoplasmic Cxs that do not aggregate into plaques are nonfunctional. Cytoplasmic Cx43 reflects molecules that have not been transported and inserted properly into the cell membrane because of changes in the transformed cell. Because GJ plaques represent the only functional unit, only plaque staining was scored.
For purposes of using Cx43 staining as a possible diagnostic parameter, the loss of Cx43 plaque staining is considered a positive test result and retention of Cx43 staining is considered a negative test result. Specificity was determined by dividing the number of samples that retained Cx43 plaque staining (negative test result) divided by the number of benign samples. Sensitivity was determined by dividing the number of samples that lost Cx43 plaque staining (positive test result) divided by the number of malignant samples. Positive predictive value was obtained by dividing the number of malignant samples that lost Cx43 plaque staining by the total number of samples that lost Cx43 plaque staining. Negative predictive value was obtained by dividing the number of benign samples that retained Cx43 plaque staining by the total number of samples that retained Cx43 plaque staining.
The demographic and staging data are shown in the Table. Patient age did not differ significantly between groups (P = .30), and, overall, approximately 80% of patients were female, which is consistent with the prevalence of thyroid disease. Approximately 75% of the malignant specimens were T1 or T2 (≤4 cm), while 25% were T3 (>4 cm). Only 2 specimens had evidence of nodal metastasis (1 PTC and 1 follicular carcinoma). None of the patients had distant metastasis. Approximately 85% of the benign lesions were 4 cm or less, while 15% were larger than 4 cm.
The immunohistochemical staining pattern for Cx43 at the cell membrane was evaluated qualitatively. Connexin 43 normally demonstrates staining at the cell membrane in close association with tight junctions.6 Distinct clumping of Cx43 staining at the cell membrane, which was defined as plaque staining, represents the configuration of individual Cx proteins into formed GJs (Figure, A and B). The Cx43 GJ staining pattern of all the specimens is shown below.
Twenty-three of 30 cancer specimens (77%) revealed a lack of Cx43 plaque staining at the cellular membrane, compared with only 3 of 20 benign specimens (15%).
Among the malignant specimens, loss of Cx43 plaque staining was observed in 11 of 17 follicular carcinomas (65%), 5 of 6 FVPTCs (83%), and 7 of 7 PTCs (100%) (Figure, C and D). In contrast, only 3 of 15 adenoma (20%) and 0 of 5 goiter samples demonstrated an absence of Cx43 plaque staining at the cell membrane. The utility of the loss of GJ plaque staining as a biomarker of malignant disease is summarized in the preceding tabulation.
The absence of staining as an indicator of malignant disease has a sensitivity of 0.76, a specificity of 0.85, a positive predictive value of 0.88, and a negative predictive value of 0.71. These data are highly significant (Fisher exact test, P < .001). Of particular interest is the loss of GJ plaques on follicular thyroid cancer cells compared with follicular adenomas.
The loss of staining as an indicator of follicular carcinoma vs follicular adenoma has a sensitivity of 0.65, a specificity of 0.80, a positive predictive value of 0.79, and a negative predictive value of 0.67 (Fisher exact test, P = .02).
A significant number of thyroid operations are performed for benign disease as a result of limitations in our current diagnostic techniques. While FNA is a highly useful tool in the analysis of thyroid nodules, its utility is restricted by the finding of suspicious, nondiagnostic pathology. Improved diagnostic markers are needed to help distinguish benign from malignant lesions before definitive surgical treatment.
Differentiating follicular carcinoma from benign follicular lesions presents a particular challenge that typically precludes FNA as a stand-alone diagnostic tool and requires macroscopic tissue architecture assessment to demonstrate invasive features. Attempts to predict the likelihood of malignancy on a nondiagnostic aspirate are based on degrees of cellularity, follicle formation, and colloid production. Benign-appearing lesions generally exhibit a macrofollicular architecture, whereas malignancy is associated with a microfollicular, trabecular, or solid appearance.7 Key findings suggestive of follicular carcinoma include a significant proportion of microfollicles (small groups of 6 to 10 follicular cells in a ring with or without a small amount of central colloid), trabeculae, cellular crowding, and overlapping clusters of follicular cells.7
Many studies have supported the idea that decreased expression or dysfunction of GJs may play an important role in carcinogenesis. Intercellular communication of GJs is usually decreased in cancer cells,8-10 and restoration of this communication in neoplastic cells by means of Cx transfection has led to normalization of growth and differentiation patterns both in vivo and in vitro.11-15 Mice knockout studies have also shown that Cx-depleted cells are more susceptible to tumor induction by chemicals and radiation.16-19 Connexin 43, which is the most widely expressed Cx in humans, is present in at least 46 cell and 34 tissue types.20 The expression of Cx43 has been shown to be decreased in several tumor types, including human prostate, lung, brain, and breast cancers.5
Connexin 32 and Cx43 are known to be present in thyrocytes, where they comprise homotypic GJs (consisting of only 1 Cx type) that reside in different regions of the cell membrane.6 Proper thyroid folliculogenesis appears to have an association with GJ intercellular communication and Cx expression. Studies have demonstrated thyroid-stimulating hormone–induced folliculogenesis in monolayered cells in association with an increase in Cx expression.1 Porcine thyrocyte studies have focused on Cx32 as a key player in follicle formation, as it is present in follicular structures but is lost in a cellular monolayer.3 The role of Cx43 in human thyroid folliculogenesis and thyroid function is unknown.
In this study, the formation of Cx43 plaques at the cell membrane indicates the gathering of individual Cx proteins into formed GJs. Also, 76% of the malignant specimens, compared with only 15% of the benign specimens, demonstrated a lack of plaque staining at the cell membrane, indicating a failure to form Cx43 GJs. In particular, plaque staining was absent in 83% of follicular carcinoma specimens, compared with only 20% of follicular adenoma specimens. This is of significant interest as there are currently no biomarkers to differentiate follicular carcinoma from adenoma on FNA biopsy. As a result, all thyroids that contain neoplasms with follicular architecture (carcinoma or adenoma) are surgically excised. Therefore, there is less concern regarding false-positives (all of which are currently removed) compared with false-negatives, which might argue against surgery, despite the true presence of cancer. This pilot study used a small number of samples, and we expect that sensitivity, specificity, and positive and negative predictive values will change as the study is expanded.
Loss of Cx43 plaque expression appears to be associated with thyroid cancer and loss of normal follicular architecture. Additional studies are required to investigate the qualitative and quantitative aspects of Cx43 staining of FNAs and to correlate the aspirates with pathology specimens to determine the clinical utility of Cx43 staining as a marker of malignant disease. Staining of serial FNA biopsy sections for other Cx molecules, such as Cx32 and Cx26, may further contribute to the diagnostic specificity and sensitivity of altered Cx expression in follicular carcinoma.
Correspondence: Jan Geliebter, PhD, Department of Microbiology and Immunology, New York Medical College, Basic Sciences Bldg 311, Valhalla, NY 10595 (jan_geliebter@nymc.edu).
Submitted for Publication: March 19, 2011; final revision received May 26, 2011; accepted September 18, 2011.
Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Darr, Patel, Tiwari, and Geliebter. Acquisition of data: Patel, Komorowski, McCormick, Schantz, and Geliebter. Analysis and interpretation of data: Patel, Yu, Schantz, and Geliebter. Drafting of the manuscript: Darr, Patel, Yu, Schantz, and Geliebter. Critical revision of the manuscript for important intellectual content: Patel, Komorowski, McCormick, Tiwari, and Geliebter. Statistical analysis: Yu, Schantz, and Geliebter. Obtained funding: Tiwari. Administrative, technical, and material support: Darr, Patel, Komorowski, McCormick, and Geliebter. Study supervision: McCormick, Schantz, and Geliebter.
Financial Disclosure: None reported.
Funding/Support: This study was supported in part by grant 1R01 CA131946-01A2 from the National Cancer Institute (Dr Tiwari).
Previous Presentation: This study was presented at the American Head and Neck Society 2011 Annual Meeting; April 27, 2011; Chicago, Illinois.
Additional Contributions: Fred Moy, PhD, New York Medical College, provided guidance in statistical analysis.
1.Munari-Silem Y, Guerrier A, Fromaget C, Rabilloud R, Gros D, Rousset B. Differential control of connexin-32 and connexin-43 expression in thyroid epithelial cells: evidence for a direct relationship between connexin-32 expression and histiotypic morphogenesis.
Endocrinology. 1994;135(2):724-7348033821
PubMedGoogle ScholarCrossref 2.Statuto M, Audebet C, Tonoli H, Selmi-Ruby S, Rousset B, Munari-Silem Y. Restoration of cell-to-cell communication in thyroid cell lines by transfection with and stable expression of the connexin-32 gene: impact on cell proliferation and tissue-specific gene expression.
J Biol Chem. 1997;272(39):24710-247169305943
PubMedGoogle ScholarCrossref 3.Tonoli H, Flachon V, Audebet C,
et al. Formation of three-dimensional thyroid follicle-like structures by polarized FRT cells made communication competent by transfection and stable expression of the connexin-32 gene.
Endocrinology. 2000;141(4):1403-141310746644
PubMedGoogle ScholarCrossref 4.Leithe E, Kjenseth A, Bruun J, Sirnes S, Rivedal E. Inhibition of connexin 43 gap junction channels by the endocrine disruptor ioxynil.
Toxicol Appl Pharmacol. 2010;247(1):10-1720510257
PubMedGoogle ScholarCrossref 5.Naoi Y, Miyoshi Y, Taguchi T,
et al. Connexin26 expression is associated with aggressive phenotype in human papillary and follicular thyroid cancers.
Cancer Lett. 2008;262(2):248-25618191019
PubMedGoogle ScholarCrossref 6.Guerrier A, Fonlupt P, Morand I,
et al. Gap junctions and cell polarity: connexin32 and connexin43 expressed in polarized thyroid epithelial cells assemble into separate gap junctions, which are located in distinct regions of the lateral plasma membrane domain.
J Cell Sci. 1995;108(pt 7):2609-26177593302
PubMedGoogle Scholar 7.Atkinson B, ed. Atlas of Diagnostic Cytopathology. 2nd ed. Philadelphia, PA: WB Saunders Co; 2003
8.Leithe E, Kjenseth A, Sirnes S, Stenmark H, Brech A, Rivedal E. Ubiquitylation of the gap junction protein connexin-43 signals its trafficking from early endosomes to lysosomes in a process mediated by Hrs and Tsg101.
J Cell Sci. 2009;122(pt 21):3883-389319808888
PubMedGoogle ScholarCrossref 11.Langlois S, Cowan KN, Shao Q, Cowan BJ, Laird DW. The tumor-suppressive function of connexin43 in keratinocytes is mediated in part via interaction with caveolin-1.
Cancer Res. 2010;70(10):4222-423220406988
PubMedGoogle ScholarCrossref 12.Qin H, Shao Q, Curtis H,
et al. Retroviral delivery of connexin genes to human breast tumor cells inhibits in vivo tumor growth by a mechanism that is independent of significant gap junctional intercellular communication.
J Biol Chem. 2002;277(32):29132-2913812042301
PubMedGoogle ScholarCrossref 13.Eghbali B, Kessler JA, Reid LM, Roy C, Spray DC. Involvement of gap junctions in tumorigenesis: transfection of tumor cells with connexin 32 cDNA retards growth in vivo.
Proc Natl Acad Sci U S A. 1991;88(23):10701-107051660148
PubMedGoogle ScholarCrossref 14.Goldberg GS, Bechberger JF, Tajima Y,
et al. Connexin43 suppresses MFG-E8 while inducing contact growth inhibition of glioma cells.
Cancer Res. 2000;60(21):6018-602611085522
PubMedGoogle Scholar 15.Omori Y, Yamasaki H. Gap junction proteins connexin32 and connexin43 partially acquire growth-suppressive function in HeLa cells by deletion of their C-terminal tails.
Carcinogenesis. 1999;20(10):1913-191810506104
PubMedGoogle ScholarCrossref 16.Avanzo JL, Mesnil M, Hernandez-Blazquez FJ,
et al. Increased susceptibility to urethane-induced lung tumors in mice with decreased expression of connexin43.
Carcinogenesis. 2004;25(10):1973-198215166089
PubMedGoogle ScholarCrossref 17.Dagli ML, Yamasaki H, Krutovskikh V, Omori Y. Delayed liver regeneration and increased susceptibility to chemical hepatocarcinogenesis in transgenic mice expressing a dominant-negative mutant of connexin32 only in the liver.
Carcinogenesis. 2004;25(4):483-49214688024
PubMedGoogle ScholarCrossref 18.King TJ, Lampe PD. The gap junction protein connexin32 is a mouse lung tumor suppressor.
Cancer Res. 2004;64(20):7191-719615492231
PubMedGoogle ScholarCrossref 19.Temme A, Buchmann A, Gabriel HD, Nelles E, Schwarz M, Willecke K. High incidence of spontaneous and chemically induced liver tumors in mice deficient for connexin32.
Curr Biol. 1997;7(9):713-7169285723
PubMedGoogle ScholarCrossref 20.Rackauskas M, Neverauskas V, Skeberdis VA. Diversity and properties of connexin gap junction channels.
Medicina (Kaunas). 2010;46(1):1-1220234156
PubMedGoogle Scholar