Tissue microarray cores of differentiated thyroid cancer (DTC) (A, C, E, G, I, K, M, and O) and associated anaplastic thyroid carcinoma (ATC) (B, D, F, H, J, L, N, and P) exhibiting decreased expression of thyroglobulin (A and B), Bcl-2 (C and D), vascular endothelial growth factor (E and F), E-cadherin (G and H), and β-catenin (I and J) and increased expression of p53 (K and L), MIB-1 (M and N), and topoisomerase II-α (O and P) (original magnification ×200).
Hierarchical clustering of all 62 markers. A indicates anaplastic carcinoma; F, follicular carcinoma; and P, papillary carcinoma. Each number identifies 1 of 12 thyroid tumors studied. Tumors with the same number came from the same patient. See Table 3 for expansions of the marker abbreviations.
Hierarchical clustering of 8 significant markers. CTNNB1 indicates β-catenin; E-CAD, E-cadherin; TG, thyroglobulin; TOPO-II, topoisomerase II-α; and VEGF, vascular endothelial growth factor. See Table 3 for expansions of the marker abbreviations.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Wiseman SM, Griffith OL, Deen S, et al. Identification of Molecular Markers Altered During Transformation of Differentiated Into Anaplastic Thyroid Carcinoma. Arch Surg. 2007;142(8):717–729. doi:10.1001/archsurg.142.8.717
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
A change in tumor expression profile will be observed during the transformation of differentiated into anaplastic thyroid carcinoma.
Population-based sample (British Columbia).
Sequential archival cases of anaplastic thyroid cancer with an adjacent associated differentiated thyroid cancer focus, and with available paraffin blocks, that had been diagnosed and treated in British Columbia during a 20-year period (12 cases; January 1, 1984, through December 31, 2004) were identified through the provincial tumor registry for tissue microarray construction.
Main Outcome Measure
Significant associations between marker staining and tumor pathologic diagnosis (differentiated vs anaplastic) were determined with contingency table and marginal homogeneity tests. A classifier algorithm was also used to identify useful and important molecular classifiers.
Overall, there were 3 up-regulated and 5 down-regulated markers when comparing the anaplastic carcinoma with associated differentiated thyroid cancers. Contingency table statistics identified 5 markers (thyroglobulin, Bcl-2, MIB-1, E-cadherin, and p53) to be significantly differentially expressed by the anaplastic and differentiated tumor foci. These 5 markers and 3 others (β-catenin, topoisomerase II-α, and vascular endothelial growth factor) were significant when evaluated using the marginal homogeneity test. Clustering and classification analysis based on these same 8 markers readily separated differentiated and anaplastic thyroid tumors with a high degree of accuracy.
The markers we observed to change during thyroid tumor progression may not only show promise as molecular diagnostic or prognostic tools but also warrant further study as potential targets for treatment of disease.
Transformation is a term commonly used to describe the biological process in which normal or premalignant cells undergo a change and become cancerous.1 For thyroid cancer, transformation or anaplastic transformation describes an intratumoral evolution, or progression, from differentiated thyroid cancer (DTC) into anaplastic thyroid carcinoma (ATC).2 Differentiated thyroid cancer, which includes papillary thyroid carcinoma (PTC) and follicular thyroid carcinoma (FTC), accounts for more than 90% of newly diagnosed thyroid malignancies, and with current treatment few individuals die of their disease.3 Even individuals who present with evidence of metastatic DTC can experience long-term survival.4 In contrast, ATC is considered one of the most lethal of all human cancers but accounts for less than 2% of all newly diagnosed thyroid malignancies.5 Despite advances in the understanding and treatment of human cancer gleaned during the past several decades, little improvement has occurred in the outcomes of individuals diagnosed as having ATC.6-8 Individuals diagnosed as having ATC tend to present with a rapidly enlarging symptomatic neck mass, and many have evidence of metastatic disease.6-8 A report that evaluated the outcome of 516 patients with ATC from the National Cancer Institute's Surveillance, Epidemiology, and End Results database found that at presentation 38% of newly diagnosed ATC cases had evidence of either extrathyroidal tumor extension or lymph node invasion and 43% of cases had distant metastasis.9 The observed overall cause-specific mortality of this ATC cohort was 68.4% at 6 months and 80.7% at 12 months.9 Death from ATC is caused by either local and/or regional disease extension, which leads to suffocation or multiorgan failure as a consequence of overwhelming distant metastatic disease.6-8 In selected patients with ATC who are able to tolerate aggressive multimodal treatment protocols, local disease may be controlled and death occurs as a consequence of distant metastases.10,11
The rarity and rapidly fatal disease course of ATC have made the study of this cancer difficult in both the clinic and laboratory. Most clinical literature that has investigated ATC is composed of small, retrospective institutional case series.6,7 Recently, clinical trials that evaluated the efficacy of several treatment protocols have been reported for a few small cohorts of patients with ATC.10-12 Much of the laboratory investigation of ATC has consisted of ATC cell line–based studies. The ATC cell lines grown in culture or as xenografts in mice have served as important models for the study of tumor biological characteristics and novel new cancer treatments.13
Accumulated clinical, pathologic, and experimental evidence has led to acceptance of the hypothesis that ATC transforms or evolves from preexisting DTC.2 However, little understanding exists of the specific molecular alterations and mechanisms that underlie this intrathyroidal tumor evolution. Expression profiling of human tumors for a panel of molecular markers, which represent a variety of cellular processes and signaling pathways, represents a powerful technique for gaining insight into tumor biological characteristics. Thus, to develop a better understanding of thyroid tumor progression, the overall objective of this study was to evaluate the change in the tumor expression profile that occurs during the transformation of DTC into ATC.
Sequential archival cases of ATC with an adjacent associated DTC focus, and with available paraffin blocks, that had been diagnosed and treated in British Columbia during a 20-year period (January 1, 1984, through December 31, 2004) were identified through the provincial tumor registry for tissue microarray (TMA) construction. The study was approved by the research ethics boards of the University of British Columbia, British Columbia Cancer Agency, Vancouver Coastal Health, and Providence Health Care (Vancouver, British Columbia). All patients had newly diagnosed ATC, and all clinical data were retrospectively collected from hospital medical records. Clinicopathologic data collected included patient age, patient sex, type of therapy, patient follow-up, and survival. The ATC and DTC pathologic diagnoses were confirmed by review of all cases by 3 pathologists (A.R., H.M., and B.G.). Hematoxylin-eosin–stained sections of each tumor were examined, and areas of ATC or DTC (PTC or FTC) were marked on both the slide and the corresponding paraffin block for TMA construction. The TMA construction was performed in a manner that has been previously described.14 Adequate tissue was present for immunohistochemical staining of 12 ATC and associated adjacent DTC foci for the 63 markers summarized in Table 1. The 63 molecular markers were chosen because they represent a wide variety of gene products important in the maintenance of normal cellular function. Some of these proteins are thyroid specific (eg, thyroglobulin, thyrotropin, and thyroid transcription factor 1), others have previously been identified as altered by anaplastic transformation (eg, p53, E-cadherin, and β-catenin), and others have never been well characterized in a thyroid cancer cohort (eg, Wilms tumor gene product, mismatch repair enzyme hMLH1, and autocrine motility factor receptor).
The antibodies used and the antigen retrieval methods performed are also summarized in Table 1, and sample cores stained for thyroglobulin, Bcl-2, MIB-1, E-cadherin, p53, β-catenin, topoisomerase II-α, and vascular endothelial growth factor (VEGF) are presented in Figure 1. The scoring of the TMA sections stained for all markers evaluated was performed in a semiquantitative manner; the scoring system types are summarized in Table 2, and the scoring systems for all markers evaluated are summarized in Table 3. All samples were evaluated and scored simultaneously by 2 of 5 pathologists (A.R., H.M., B.G., L.G., and A.G.) who were blinded to all patient clinical data, and any interpathologist discrepancy in the scoring of a specific tissue core was immediately resolved. If a discrepancy in the scores assigned to the 2 cores from the same tumor was found, the tumor was assigned the higher score. All marker data were logged into a standardized score sheet that matched each TMA section, and a master study database was subsequently created by incorporation of all clinical, pathologic, and marker data (Microsoft Excel Microsoft, Redmond, Washington).
Significant associations between marker staining and pathologic status (ATC vs DTC) were determined using contingency table statistics (Pearson χ2 or Fisher exact test where appropriate) using the SPSS statistical software package (version 13.0; SPSS Inc, Chicago, Illinois). Two marker score groupings were analyzed. In the first grouping, marker scores were grouped as either negative (score of 0) or positive (score of ≥ 1). In the second grouping, marker scores were grouped as either negative/low (score of 0 or 1) or medium/high (score of ≥ 2). A marginal homogeneity (MH) test for 2 related samples was also used to test for a significant trend toward increasing or decreasing score for a marker between the pairs of ATC and DTC samples. This test does not require the marker scores to be grouped but instead uses the actual semiquantitative scores. Both the contingency table and MH statistics were corrected for multiple testing using the Benjamini and Hochberg (BH) correction.15 The BH correction is a simple step-up, false-discovery rate–controlling procedure that is much less stringent than the more commonly used Bonferroni correction. All statistical tests were 2 sided and considered statistically significant at P < .05 (after correction).
The samples and markers were clustered using a simple hierarchical clustering algorithm and heat maps generated to visualize the data using the gplots library (version 2.3.0) for the R programming language (version 2.3.1; R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria). The markers were also evaluated for their utility in classification (ATC vs DTC) using the Random Forests classifier algorithm (Random Forests, version 1.0; Salford Systems, San Diego, California). Although distinguishing ATC from DTC tumor tissue is not a challenge for the pathologist, the classification approach represents a useful method for identifying the discriminating, biologically interesting molecular markers.
Of 94 cases of ATC diagnosed and treated in British Columbia during a 20-year period (January 1, 1984, through December 31, 2004), 32 cases (34%) had adequate tissue available for evaluation and 12 (38%) of these cases had associated foci of DTC. Of the DTC foci, 9 (75%) were FTC and 3 (25%) were PTC. The histologic subtypes of the anaplastic tumors in the study cohort were 9 epithelioid tumors, 2 spindle cell tumors, and 1 squamoid tumor. The median study patient age was 67 years (range, 54-81 years), and the study cohort was composed of 7 women and 5 men. Two patients in the study cohort had a history of remote (>15 years before ATC presentation) treatment of hyperthyroidism with radioactive iodine. Both these patients had FTC associated with their ATC. None of the study patients had a history of head and neck irradiation or a personal or family history of thyroid carcinoma. Three study patients (25%) presented with evidence of distant metastatic disease.
The most common treatment given to the study cohort, which was received by 8 patients, was surgery (lobectomy or total thyroidectomy) and external beam radiotherapy (67%). One patient was treated with surgery and chemotherapy, 1 patient received external beam radiotherapy and chemotherapy, 1 patient received external beam radiotherapy alone, and 1 patient underwent surgery alone. The 2 patients who did not undergo surgery as part of their disease treatment had undergone an open tissue biopsy for diagnosis. The patients who were treated with external beam radiotherapy received a median of 35 Gy (range, 20-60 Gy) of radiation, and of the 2 patients who received chemotherapy, one received doxorubicin and the other doxorubicin and cisplatin. The median survival of patients in the ATC cohort from their date of ATC diagnosis was 22.5 weeks (range, 2-302 weeks) or less than 6 months. A single patient in our cohort who was treated with surgery and external beam radiotherapy did not die of ATC and has remained disease free when evaluated at 302 weeks of clinical follow-up. None of the study tumors showed evidence of calcitonin expression, ensuring that a medullary thyroid tumor was not erroneously included in the study cohort.
The contingency table statistics identified 5 markers (thyroglobulin, Bcl-2, MIB-1, E-cadherin, and p53) with significant associations between marker staining and disease status (Table 4; after BH multiple testing correction). Because the contingency table statistics depend on how the marker scores are grouped, we tried an alternate grouping. When marker scores were grouped as negative/low vs medium/high instead of negative vs positive, 5 markers (MIB-1, β-catenin, thyroglobulin, topoisomerase II-α, and VEGF) were significant according to contingency table statistics (Table 5; after BH multiple testing correction). Both score groupings produced 5 significant markers, and although they do not perfectly overlap, all significant markers from both groupings were found to be significant using the MH test, which does not depend on arbitrary score groupings (Table 4). This finding suggests that all 8 significant markers from the MH test may be of biological relevance. In total, it appears that there are 3 up-regulated (overexpressed) and 5 down-regulated (underexpressed) markers in the ATC samples compared with the DTC samples. Three patients had alterations in their marker expression between DTC and ATC (in the expected direction) for all 8 markers, with an overall median of 7 alterations (range, 3-8) for the entire cohort of 12 patients.
The clustering analysis and heat maps (Figure 2 and Figure 3) show that simple hierarchical clustering can separate ATC (light green sidebar) from DTC (dark green sidebar) with high accuracy (especially when nondiscriminate markers are excluded). The heat maps also provide a useful visualization of the marker staining patterns and show the 3 up-regulated (yellow sidebar) and 5 down-regulated (orange sidebar) markers clustered together, as expected.
The classification analysis identified the same 8 markers (thyroglobulin, MIB-1, β-catenin, Bcl-2, E-cadherin, p53, VEGF, and topoisomerase II-α) as the most important variables for discriminating patient-matched ATC from DTC samples (Table 4). A classifier based on all markers was able to correctly classify ATC vs DTC with high accuracy, sensitivity, and specificity. Specifically, cross-validation analysis reported a receiver operating characteristic curve integral of 0.98, overall accuracy of 95.8%, sensitivity of 100%, and specificity of 91.7%. Only a single DTC sample was misclassified as ATC.
Anaplastic transformation represents a postmalignant tumor progression. Specifically, transformation is a terminal event, with ATC representing the end point of thyroid tumor evolution.2 In the current study, by comparing the ATCs with the DTCs from which they arose, we have identified 8 significantly altered markers (3 up-regulated and 5 down-regulated) involved in the transformation process.
The 3 markers identified as being significantly up-regulated, or overexpressed, in the ATCs when compared with the DTCs from which they presumably evolved were p53, MIB-1, and topoisomerase II-α. p53 is a tumor suppressor gene often referred to as the “guardian of the genome,”16(p1219) and approximately half of human tumors exhibit a p53 alteration.16,17 Its gene product, the p53 protein, acts as a transcription factor that regulates downstream genes that are involved in DNA repair, cell cycle arrest, and apoptosis.16,17 The status of p53 has also been linked to tumor chemosensitivity, radiosensitivity, and prognosis of many human cancer types.16-19 p53 is also believed to play an important role in transformation because it is rarely altered in DTC but commonly mutated in ATC. In a multistudy review20 of 265 ATC cases, 52% of tumors exhibited either p53 gene or p53 protein alteration. The important role of p53 in anaplastic transformation has led several investigators to report the development of DTC characteristics with reintroduction of wild-type p53 into ATC.2 Thus, the alteration of the p53 protein we have identified in the study cohort is consistent with the current literature and further highlights the importance of this tumor suppressor gene in thyroid tumor progression.
The rapid rate of cellular proliferation that characterizes ATC has been demonstrated in both ATC cell lines and by the clinical observation that ATC tumor volume may double when observed during short periods.8,21 MIB-1 is an antibody that binds to the Ki-67 nuclear antigen, and its level of expression correlates with measurements of cellular proliferation, which include bromodeoxyuridine uptake and S-phase fraction.22 Kjellman et al23 evaluated MIB-1 expression with immunohistochemical analysis in a cohort of 144 thyroid tumors (including 40 DTCs and 8 ATCs) and found that expression was higher in ATCs (median, 16.2%) when compared with DTCs (median, 1.9% in PTCs and 2.7% in FTC). These authors also reported higher MIB-1 expression (1.85% or greater) to represent a significant independent risk factor for a less favorable disease course in individuals diagnosed as having PTC.23 The increased expression of MIB-1 we observed in the ATCs, relative to the DTCs from which they evolved, is also consistent with current literature reports.
The topoisomerase II enzymes, which in mammalian cells include the α and β isoforms, are important for many aspects of DNA metabolism, which include replication, transcription, chromosome segregation, and cellular proliferation.24,25 Topoisomerase II-α is a nuclear enzyme that is required for chromatin condensation and segregation during mitosis and is expressed in the S, G2, and M phases of the cell cycle.24,25 As a consequence of the expression of topoisomerase II-α being coupled with the cell cycle, like MIB-1, it is also considered a marker of cellular proliferation.24,25 It has been most extensively studied in breast cancer because the topoisomerase II-α gene is located adjacent to the HER2 oncogene and is either amplified or deleted in approximately 90% of HER2-amplified primary breast tumors.26 Notably, topoisomerase II-α may serve as a target for many anticancer drugs, including anthracyclines, and the presence of gene amplification, deletion, or protein expression may predict response to treatment.26,27 Currently, the topoisomerase II inhibitor doxorubicin, which has a reported response rate of 5% to 22%, is considered to be the most effective drug for treatment of ATC.6-8 Few reports have evaluated topoisomerase II-α expression by thyroid tumors. Using immunohistochemical analysis, Lee et al24 observed a higher topoisomerase II-α expression in ATC compared with DTC (labeling index, 7.8, 4.8, and 2.6 for ATC, PTC, and FTC, respectively).24 Poorly differentiated DTC is considered an intermediate form in the progression of DTC to ATC.28 Fluge et al29 compared the gene expression profiles of DTC and clinically aggressive poorly differentiated DTC with a complementary DNA microarray approach. These investigators identified topoisomerase II-α to be markedly up-regulated in the poorly differentiated DTCs when compared with the DTCs.29 Several mutations have been identified in the human topoisomerase II-α gene in cell lines that resist topoisomerase II inhibitors, and Satake et al30 evaluated 10 ATC cell lines and 3 human ATCs for the presence of these mutations (altered amino acids 439, 450, and 803). Interestingly, none of these mutations were identified in either ATCs or ATC cell lines.30 Thus, the increased expression of topoisomerase II-α we identified in the ATCs, compared with the DTCs from which they transformed, is consistent with the current literature.
The 5 markers identified as being significantly down-regulated, or showing decreased expression in the ATCs when compared with the DTCs from which they transformed, were thyroglobulin, E-cadherin, β-catenin, Bcl-2, and VEGF. The production and expression of the thyroglobulin protein are suggestive of a differentiated thyroid tumor phenotype, and thyroglobulin protein expression is commonly absent in ATC.31 Cellular discohesiveness and detachment are necessary for the development and dissemination of cancer.32 E-cadherin is a transmembrane glycoprotein that is important for cell-to-cell adhesion and complexes with catenin proteins via an intracellular domain for function. Either β-catenin or γ-catenin binds to a common E-cadherin domain and then binds to α-catenin, which anchors the actin cytoskeleton to the cadherin-mediated adhesion complex.32 β-catenin also functions as a regulator of cell growth and survival as a downstream effector of the Wnt signaling pathway.33 We have previously evaluated and reviewed evidence that suggested that derangement of the E-cadherin–catenin complex is involved in anaplastic transformation of thyroid cancer.34 The observed down-regulation, or decreased expression, of E-cadherin and β-catenin observed in the progression of DTC into ATC is also consistent with current reports.
The protein encoded by the Bcl-2 proto-oncogene is responsible for prolongation of cell survival by blocking apoptosis.35-37 In a study that evaluated 134 thyroid tumors for Bcl-2 protein expression with immunohistochemical analysis, Pollina et al35 reported down-regulation of Bcl-2 in ATC. In this study, Bcl-2 immunoreactivity was identified in 60 of 70 DTCs (85.7%) and only 8 of 24 ATCs (33.3%).35 Similar observations have been reported by other investigators.36 In an ATC cell line model, Kim et al37 reported success in using a Bcl-2 antisense oligonucleotide to enhance apoptosis and increase ATC drug sensitivity. Thus, the down-regulation of Bcl-2 we observed in the progression of DTC into ATC is consistent with current literature reports.
For cancer cells to survive, they must be able to receive nutrients and oxygen and dispose of the wastes and byproducts of cellular metabolism.38 A solid tumor cannot grow larger than 1 mm3 without developing its own network of microvessels through a process of neovascularization or angiogenesis.38 Angiogenesis is important for both local tumor growth and distant cancer spread.38 Vascular endothelial growth factor is a multifunctional cytokine with secretion that is regulated by a variety of cytokines and growth factors, that plays an important role in angiogenesis, and that is overexpressed in many human malignancies.39 Huang et al40 evaluated VEGF expression in a cohort of 117 thyroid tumors, which included 88 DTCs (76 PTCs and 12 FTCs) and 8 ATCs. All the PTCs exhibited strong diffuse staining, whereas the ATCs showed weak and infrequent immunoreactivity. In a cohort that included 52 DTCs (34 PTCs and 18 FTCs) and 8 poorly differentiated DTCs, Vieira at al41 reported VEGF expression to be significantly more prevalent in PTC (79%) than either FTC (50%) or poorly differentiated DTC (37%). Thus, the decrease in VEGF expression observed during transformation of DTC to ATC is consistent with prior reports. Although the underlying mechanisms are unknown, these observations suggest a potentially important role for alternate pathways of angiogenesis in these tumors. Currently, antiangiogenesis agents, including drugs that target VEGF and other mediators of angiogenesis, are being studied for the treatment of ATC.42-44
By using a TMA-based approach and evaluating 12 coexisting adjacent DTC and ATC tumors, we have evaluated the change in expression profile for a panel of 63 molecular markers, with the aim of identifying the molecular alterations that occur during the transformation of DTC into ATC. Not only did analysis of the tumor expression profiles reveal 8 markers (thyroglobulin, MIB-1, β-catenin, Bcl-2, E-cadherin, p53, VEGF, and topoisomerase II-α) as being significantly altered when comparing patient-matched ATC and DTC samples, but independent of tumor histologic features, a classifier based on all markers was able to correctly differentiate ATC and DTC with high accuracy, sensitivity, and specificity. It is clear that the specific intratumoral molecular alterations that occur during transformation warrant further evaluation as molecular prognosticators for DTC. Risk stratification systems, which are used to guide the treatment of individuals diagnosed as having DTC, are based on both patient and tumor characteristics that have been identified through retrospective evaluation of outcomes in large cohorts of patients diagnosed as having thyroid cancer.45 The DTC patient risk stratification systems currently guide the extent of surgery and the use of adjuvant radioactive iodine therapy, although no single system has been universally accepted or applied.46 Further evaluation of the prognostic utility of 1 or more of the 8 markers we have identified as being significantly altered during transformation, in large DTC patient cohorts, could potentially lead to improved treatment selection and patient outcomes. The molecular markers we and others have identified as being significantly altered in transformation may also represent important targets for the treatment of ATC. Molecular targets that are important for the transformation of DTC could potentially prevent ATC development or, as has been demonstrated in the laboratory with reintroduction of p53 into ATC cell lines, lead to the development of more differentiated tumor characteristics.2 We have recently evaluated and reviewed many promising targeted therapeutic approaches for the treatment of ATC.13
In conclusion, by evaluating the change in immunohistochemical expression of coexisting adjacent ATC and DTC tumors for a panel of 63 molecular markers, we have identified alterations in a broad range of important underlying cellular processes. These changes, which occur in cells that exhibit a DTC phenotype, include decreased control of cell cycle regulation and apoptosis, leading to uncontrolled cellular proliferation (p53, Bcl-2, topoisomerase II-α, and MIB-1), derangement of normal cellular adhesion and signaling molecules (E-cadherin and β-catenin), loss of normal cellular biosynthetic functions (thyroglobulin), and progressive development of neovascularization (VEGF). Further study of the underlying biological characteristics of the anaplastic transformation of thyroid cancer is warranted and may provide important insights for the development of effective treatments for individuals diagnosed as having this fatal thyroid malignancy.
Correspondence: Sam M. Wiseman, MD, FRCSC, Department of Surgery, St Paul's Hospital, University of British Columbia, C303-1081 Burrard St, Vancouver, BC, Canada V6Z-1Y6 (email@example.com).
Accepted for Publication: March 12, 2007.
Author Contributions: Dr Wiseman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Wiseman. Acquisition of data: Wiseman, Deen, Rajput, Masoudi, Gilks, Goldstein, and Gown. Analysis and interpretation of data: Wiseman, Griffith, Deen, Rajput, Masoudi, Gilks, Goldstein, Gown, and Jones. Drafting of the manuscript: Wiseman and Griffith. Critical revision of the manuscript for important intellectual content: Wiseman, Griffith, Deen, Rajput, Masoudi, Gilks, Goldstein, Gown, and Jones. Statistical analysis: Wiseman, Griffith, and Jones. Obtained funding: Wiseman and Jones. Administrative, technical, and material support: Wiseman, Deen, Rajput, Masoudi, Gilks, Goldstein, and Gown. Study supervision: Wiseman.
Financial Disclosure: None reported.
Funding/Support: Dr Wiseman is a Michael Smith Scholar, and this work was supported by the Michael Smith Foundation for Health Research (MSFHR) and the Department of Surgery, University of British Columbia. Mr Griffith was supported by the Canadian Institutes of Health Research and MSFHR. Dr Jones was supported by MSFHR and the British Columbia Cancer Foundation.
Previous Presentation: This paper was presented at the 78th Annual Meeting of the Pacific Coast Surgical Association; February 18, 2007; Kohala Coast, Hawaii; and is published after peer review and revision. The discussions that follow this article are based on the originally submitted manuscript and not the revised manuscript.
David Byrd, MD, Seattle, Washington: This presentation by Dr Wiseman and colleagues is a multiparameter analysis of protein expression in a fairly small sample of 12 patients for whom archival tissue was available to compare anaplastic thyroid carcinoma (ATC) with adjacent differentiated thyroid carcinoma (DTC) in specimens from the same patients. Microarray analyses are being extensively used to classify tumors for diagnosis, identify prognostic factors, and identify predictive factors to assess response to treatment. Tissue microarray data rely heavily on the interpretation of staining intensity and specificity of the antibodies. The antibodies used in this immunohistochemical (IHC) analysis are generally well characterized and widely accepted for use both as research tools and in pathology departments in the clinical care of patients. The use of 2 pathologists to independently evaluate and score the specimens was appropriate. I am familiar with the careful and highly reproducible IHC work performed in the laboratory of Dr Gown, one of the authors, with consistent attention to the use of positive and negative controls in all analyses. The authors demonstrate their understanding of the limitations of interpreting tissue marker staining by their use of both contingency table statistics using score groupings and marginal homogeneity tests for actual semiquantitative scores.
Dr Wiseman's group identified significant differential protein expression in 8 of 63 antibodies tested in adjacent specimens containing ATC against DTC from the same patients. The results showed up-regulation in 2 proteins associated with increased cell proliferation and overexpression of p53 protein, a manifestation of mutated p53. These findings are not surprising in view of the rapid growth seen in ATC and have been previously reported. The 5 proteins found to show down-regulation have been individually studied in patients with ATC or cell lines with similar results. The unique features of this study are the comparisons of ATC and DTC protein expression from the same patient specimens and the microarray technique that allows clustering and classification analysis in thyroid cancer specimens. The authors comment on the possible use of ATC-specific protein changes to identify targets of therapy. Since protein changes are downstream events from genomic changes that may be causally related or just associated consequences, this may not identify the upstream gene targets that start the cascade of transformation. Since ATC has been resistant to nearly all pharmacological treatment, it would seem clinically most relevant to identify protein changes in DTC compared with normal thyroid or other DTC not associated with ATC to look for the earliest triggers to transformation.
My questions for Dr Wiseman are (1) Have you looked at normal adjacent thyroid in these same specimens using your Vogelstein colorectal analogy of malignant transformation? (2) Similarly, have you compared normal adjacent thyroid with DTC not associated with ATC to look at alterations in the DTC that may predict progression to ATC before transformation? (3) Have you identified and compared ATC found years later after treatment for DTC and seen the same changes? Does late development of ATC involve the same transformation expression profile? (4) How do the authors plan to use clustering profiles to identify early transformation to ATC before there are obvious clinical and histological findings? The finding of down-regulation of VEGF is a potentially disappointing observation in ATC, since there is considerable interest in antiangiogenesis-targeted therapy of patients with refractory recurrent DTC, where we occasionally have seen a sudden change in phenotype toward ATC. (5) Do the authors predict that these targeted therapies are too late to affect survival in patients with ATC?
Dr Wiseman: First, did we evaluate normal adjacent thyroid tissue with the DTC and ATC components? The answer is, no, we did not evaluate normal adjacent tissue. Many of these specimens were from excisional biopsies done at the time of tracheostomy or as part of various palliative surgical procedures. For most cases there was either little or no normal thyroid tissue present. It certainly would be interesting to look at the alteration of these molecular markers earlier on in thyroid tumor progression.
I think that this relates to your next question, which was did we evaluate the marker panel in normal thyroid tissue not associated with ATC but with DTC? I think I hinted at this question in my future directions slide, which boils down to the question, what is the utility of this marker panel as potential molecular prognosticators or predictors of transformation for DTC patients? We hope to answer this question soon and are currently carrying out similar expression profiling experiments in a large DTC tissue microarray-based study.
In terms of your next question, have we evaluated the expression profile of ATC not associated with a DTC component? This will also be addressed by current experiments in which we are evaluating the expression profile of a cohort of anaplastic tumors, which do not have an associated DTC component. I agree with your suggestion that it would be interesting to evaluate a cohort of anaplastic tumors that developed years after treatment of DTC and see if the same changes are present that we identified in the coexisting ATCs and DTCs. Currently, I don't have access to this type of an ATC cohort.
In terms of this question regarding the vascular endothelial growth factor (VEGF), certainly I don't want to pass the message on here that antiangiogenesis strategies for ATC don't work. I think that I would say quite the opposite. I believe that preclinical data evaluating such agents actually do show great promise. I also believe what we have seen in our study ATC cohort with respect to decreased expression of VEGF, when compared with DTC, is consistent with what has been described by other groups. You have to remember a few things when interpreting these observations in the context of antiangiogenesis strategies. Remember, VEGF is a cytokine; it's not the receptor on the cell. There is a group of VEGF receptors that are directly involved in cellular signal transduction and warrant further study in ATC. There have also been several recent interesting preclinical studies evaluating some of the newer drugs that target angiogenesis and other cellular signaling molecules, including the epidermal growth factor receptor, that I think also show exciting potential. Actually, the very first ATC patient ever to receive targeted therapy had a complete response in an early phase clinical trial utilizing an antiangiogenesis drug. This has stimulated the development of many clinical trials utilizing antiangiogenesis agents for treatment of ATC, which are currently ongoing.
Quan-Yang Duh, MD, San Francisco, California: I have 2 questions for you. First, have you corroborated what you have with the data from New York and Baltimore with expression microarray? I know that you did not have frozen tissues, so studying messenger RNA (mRNA) microarray would not be possible, but did you compare what you found in the protein expression level with the studies of mRNA expression?
Second, what is your future plan of research regarding poorly differentiated cancer? It will be nice to be able to tell the difference between a poorly differentiated cancer and an anaplastic cancer on a fine needle aspiration (FNA). Is your technique applicable to a fine needle biopsy type specimen?
Dr Wiseman: The first question about looking at the gene level and are our observations consistent with what has been described by groups in New York and Baltimore: interestingly, for ATC there is really very little published in terms of gene expression profiling. There has been a gene expression study from Japan, which reported on a small cohort of ATC and ATC cell lines. They reported genetic alterations in many of the same processes that we are seeing altered in our cohort at the protein level. My group recently published a meta-review evaluating all of the reported gene expression profiling studies for thyroid cancer in the Journal of Clinical Oncology several months ago, which also highlights how few ATCs have been profiled. I think that this really underscores the importance of prospective collection of ATC tissue. We have developed tissue banking for uncommon thyroid tumors at our center.
The other question about can an FNA differentiate poorly differentiated thyroid cancer (PDTC) and ATC: the answer is potentially yes. Like I mentioned in my future directions slide, I would like to evaluate the marker panel in a cohort of poorly differentiated thyroid tumors. Unfortunately, PDTC is uncommon and its specific histologic definition is currently controversial. If you ask investigators in New York or in Italy you will get different answers in terms of what defines a PDTC. This controversy has made studying these uncommon thyroid tumors quite frustrating. Limited reports of some of the markers we have evaluated suggests their expression is intermediate to DTC and ATC. It is possible to evaluate FNAs for the molecular markers we have described in this study. Several molecular markers have already been reported in clinical trials from FNA samples of thyroid cancer. So I think that the answer to your question is potentially yes, but further study of our marker panel in a PDTC cohort would need to first be carried out.
Orlo H. Clark, MD, San Francisco: We know that about 80% of patients with ATC have a coexistent DTC. Some of these differentiated tumors are follicular and some are papillary. Did you look at the gene profiles of your control, that is, your DTCs, and did they differ?
Dr Wiseman: Certainly you have made an excellent point here, which is the harder you look in an anaplastic tumor, the greater the proportion that you will find harbor a differentiated component. This is the key clinical observation, which led to emergence of the concept that DTC transforms into ATC. Regardless of the type of DTC (papillary thyroid carcinoma [PTC] or follicular thyroid carcinoma [FTC]), and as was also illustrated in the heat map, a classifier based on the 8-marker panel was able to separate the ATC and DTC (PTC or FTC) with an overall accuracy of 95.8%, sensitivity of 100%, and specificity of 91.7%. Though these tumors are discrete at the genetic level, I believe our observations suggest that there may be some similarities in the alterations that occur during transformation of PTC and FTC into ATC. We are currently evaluating the larger DTC cohort that I mentioned earlier for differences in the expression profile of PTC and FTC for these molecular markers.
Dr Clark: I am also sure that you are aware that a BRAF mutation is a very common mutation in DTC, and RET/PTC mutations are also common in patients with PTC who have a previous history of radiation exposure. BRAF mutations occur in well-differentiated, poorly differentiated, and in anaplastic thyroid cancers. Did you identify either of these mutations in the 2 patients you said had previously received radioiodine therapy for Graves disease?
Dr Wiseman: We did not carry out any gene mutation analysis in this study; specifically, we did not sequence the genes of any tumors to look for either BRAF or RET mutations. We did, however, evaluate RET protein expression in the study cohort, because there is literature that suggests that thyroid tumors that harbor RET alteration or mutation do not progress into anaplastic cancer. We saw negligible RET protein alteration in any of the study tumors, including the 2 patients who had received prior radioactive iodine therapy, which is consistent with these prior reports. I believe that there are many other factors, or underlying molecular mechanisms, that come into play during thyroid tumor progression. The overall objective of this study was to try and identify underlying processes involved in anaplastic transformation so that we could get an inkling of some potentially important targets for treatment of this fatal thyroid malignancy.