MicroRNA Expression Profiling of Thyroid Tumors: Biological ...


Published on

Published in: Health & Medicine
  • Be the first to comment

  • Be the first to like this

MicroRNA Expression Profiling of Thyroid Tumors: Biological ...

  1. 1. ORIGINAL ARTICLE E n d o c r i n e C a r e MicroRNA Expression Profiling of Thyroid Tumors: Biological Significance and Diagnostic Utility Marina N. Nikiforova, George C. Tseng, David Steward, Donna Diorio, and Yuri E. Nikiforov Department of Pathology (M.N.N., Y.E.N.), University of Pittsburgh Medical Center, and Department of Biostatistics (G.C.T.), University of Pittsburgh, Pittsburgh, Pennsylvania 15261; Department of Otolaryngology-Head and Neck Surgery (D.S.), University of Cincinnati College of Medicine, Cincinnati, Ohio 45267; and Department of Pathology (D.D.), Cincinnati Children’s Hospital, Cincinnati, Ohio 45229 Objective: MicroRNA (miRNA) expression is deregulated in many types of human cancers. We sought to investigate the expression patterns of miRNA in all major types of thyroid tumors, including tumors carrying distinct oncogenic mutations, and to explore the utility of miRNA pro- filing for the preoperative diagnosis of thyroid nodules. Design: miRNA expression levels were detected in 60 surgically removed thyroid neoplastic and nonneoplastic samples and in 62 fine-needle aspiration (FNA) samples by RT-PCR using TaqMan MicroRNA Panel or individual miRNA sequence-specific primers. miRNA expression levels were calculated relative to normal thyroid tissue. All tumors were genotyped for most common mutations. Results: Various histopathological types of thyroid tumors, including those deriving from the same cell type, showed significantly different profiles of miRNA expression. Oncocytic tumors, conven- tional follicular tumors, papillary carcinomas, and medullary carcinomas formed distinct clusters on the unsupervised hierarchical clustering analysis. Significant correlation between miRNA expres- sion patterns and somatic mutations was observed in papillary carcinomas. A set of seven miRNAs (miR-187, miR-221, miR-222, miR-146b, miR-155, miR-224, and miR-197) that were most differen- tially overexpressed in thyroid tumors vs. hyperplastic nodules in the surgical samples was validated in the FNA samples, showing high accuracy of thyroid cancer detection. Conclusions: In this study, we demonstrate that various histopathological types of thyroid tumors have distinct miRNA profiles, which further differ within the same tumor type, reflecting specific oncogenic mutations. A limited set of miRNAs can be used diagnostically with high accuracy to detect thyroid cancer in the surgical and preoperative FNA samples. (J Clin Endocrinol Metab 93: 1600 –1608, 2008) M icroRNAs (miRNAs) represent a recently identified class of endogenous noncoding RNAs that act as negative regulators of the protein-coding gene expression (1, 2). miRNA miRNA expression may be important in tumor development and progression (6 – 8). Many miRNAs are expressed in a tissue-specific manner and dysregulation is common in cancer cells, and a rapidly growing exhibit expression profiles that are different between normal and number of studies provide evidence for miRNA involvement in neoplastic tissues and between tumors with distinct biological carcinogenesis (3, 4). miRNAs regulate the expression of well- properties (6, 9, 10). Some data suggest that miRNA profiles known oncogenes and tumor suppressor genes (3, 5). Specific allow reliable identification of the cell origin of tumors (11, 12). subsets of overexpressed or down-regulated miRNAs have been However, it remains to be fully understood whether variable identified in various cancer types, suggesting that aberrations in tumor types originating from the same cell type have different 0021-972X/08/$15.00/0 Abbreviations: AC, Anaplastic carcinoma; FA, follicular adenoma; FC, follicular carcinoma; Printed in U.S.A. FNA, fine-needle aspiration; MC, medullary carcinoma; miRNA, microRNA; PC, papillary carcinoma; PCA, principal component analysis; PDC, poorly differentiated carcinoma. Copyright © 2008 by The Endocrine Society doi: 10.1210/jc.2007-2696 Received December 6, 2007. Accepted January 31, 2008. First Published Online February 12, 2008 1600 jcem.endojournals.org J Clin Endocrinol Metab. May 2008, 93(5):1600 –1608 Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  2. 2. J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 jcem.endojournals.org 1601 miRNA profiles and whether the profiles distinguish malignant Materials and Methods tumors from benign hyperplastic/reactive processes. In this regard, thyroid cancer represents an attractive model Thyroid tissue samples to study because it encompasses several histopathological tumor Snap-frozen tissue from surgically removed thyroid samples was col- lected at the Department of Pathology, University of Cincinnati follow- types originating from the same cell and tumors with distinct ing the University of Cincinnati Institutional Review Board approval or levels of differentiation. Most thyroid carcinomas originate from obtained through the Cooperative Human Tissue Network. In total, 60 thyroid follicular cells and are subdivided into well-differenti- thyroid neoplastic and nonneoplastic samples were analyzed including ated papillary carcinoma (PC) and follicular carcinoma (FC) (the 23 PCs [18 classical PCs and five follicular variant of PC (PC, FV)], nine FCs of conventional or oncocytic (Hurthle cell) type, eight FAs of con- ¨ latter further subclassified into conventional and oncocytic type) ventional or oncocytic (Hurthle cell) type, four ACs, four PDCs, two ¨ (13). Both PCs and FCs may progress to poorly differentiated MCs, five normal thyroid tissues, and five hyperplastic nodules. The age carcinoma (PDC) or may completely lose differentiation to give of patients ranged from 21–79 yr, and the female to male ratio was 3.3:1. rise to anaplastic carcinoma (AC). Follicular adenomas (FAs) are All tumors were classified according to the widely accepted diagnostic benign thyroid tumors and can be of either conventional type or histological criteria (13). oncocytic type. Less than 5% of cells within the thyroid gland are Thyroid FNA samples C-cells that give rise to medullary carcinoma (MC). Although Sixty-two thyroid FNA samples were collected as part of the pro- several recent studies have assessed the miRNA expression pro- spective molecular study on thyroid FNA samples at the University of files in specific types of thyroid cancer (14 –17), miRNA expres- Cincinnati, which was approved by the University of Cincinnati Insti- sion signatures of all major types of thyroid neoplasms have not tutional Review Board. During the FNA procedure, half of the first or been analyzed and compared in a single study to our knowledge. second pass of the aspirated material was directly collected into DNA/ RNA preservative solution (Roche Molecular Biochemicals, Manheim, Significant information has been accumulated on carcino- Germany) and frozen at 80 C. genic mutations in thyroid cancer (18). Development of PCs, the most common thyroid malignant tumor, is known to involve the RNA isolation activation of the MAPK signaling pathway either as a result of Total RNA was extracted from surgical specimens using Trizol re- BRAF or RAS point mutations or RET/PTC rearrangement. agent (Invitrogen, Carlsbad, CA) as previously described (25). RNA These mutually exclusive mutations are found in more than 70% quality was assessed by 1% agarose gel electrophoresis in the presence of of PCs and each of them is associated with the distinct pheno- ethidium bromide. RNA samples that did not show intact 18S and 28S ribosomal bands were excluded from the study. Total nucleic acids were typical and biological properties of these tumors (19). FCs are isolated from FNA samples using magnetic glass particles (Roche) ac- known to harbor either RAS mutations or PAX8/PPAR rear- cording to the manufacturer’s protocol. rangements, which are identified in 50 – 80% of conventional- type FCs and with lower prevalence in oncocytic tumors (20). It miRNA expression analysis is not known, however, whether miRNA expression profiles are Quantitation of mature miRNA expression levels in thyroid tumors different among tumors carrying specific oncogenic mutations. and normal thyroid tissue was performed by RT-PCR using TaqMan One of the main diagnostic problems in the thyroid field in- MicroRNA Assays Human Panel (Applied Biosystems Inc., Foster City, CA), which was designed to detect 158 human miRNAs. One tumor volves the preoperative assessment of thyroid nodules. Palpable sample (PTC30) was assayed twice to test the reproducibility of the nodules are common in the adult population, with an estimated detection. A good correlation (0.912) in miRNA expression levels was prevalence in the United States in the range of 4 –7% or 10 –18 found between the two runs. Expression of individual miRNAs was million affected individuals (21, 22). Thyroid fine-needle aspi- analyzed using miRNA sequence-specific primers (Applied Biosystems). Briefly, 10 ng total RNA was reverse transcribed using High-Capacity ration (FNA) is an important method for preoperative evaluation cDNA Archive kit (Applied Biosystems) followed by amplification on the of thyroid nodules, although in 10 –20% of samples, the precise ABI 7500 Real-Time PCR System (Applied Biosystems). All RT-PCR diagnosis cannot be reached, and they are reported as indeter- were performed in triplicate. Two endogenous controls were used for the minate or atypical (21). Most of these patients undergo surgery, normalization of RNA input: let7-a miRNA (according to the manufac- although only 8 –17% of surgically removed nodules are found turer’s protocol) and small nucleolar RNA RNU44 (Applied Biosys- tems). To evaluate the appropriateness of these endogenous controls for to be malignant (21). Some improvement in the diagnostic ac- use in thyroid tissue, their expression levels were determined in 16 ran- curacy can be achieved by additional testing of the FNA material dom thyroid tumor and normal samples. All samples demonstrated low for somatic mutations known to occur in thyroid tumors (23, variability in the expression levels of let7a and RNU44, validating their 24), although its sensitivity is limited because a significant pro- use as normalization controls. Two nonhuman miRNAs, ath-mir159a and cek-lin-4, were used as negative controls. portion of PCs and FCs do not have any known mutations. miRNA expression levels were calculated by relative quantitation Therefore, additional methods to improve the preoperative di- using the ABI 7500 Real-Time PCR SDS 1.2 software (Applied Biosys- agnosis are highly desirable and would result in a major impact tems) and the fold expression changes were determined by 2 CT on the clinical care. method (26). The data are presented as the fold change of miRNA ex- In this study, we 1) determined and compared miRNA ex- pression in tumors relative to normal thyroid tissues after normalization to an endogenous control (let7-a or RNU44). pression profiles of all major types of thyroid tumors, 2) explored the correlation between miRNA expression patterns and specific Detection of mutations oncogenic mutations, and 3) determined the diagnostic utility of Tumor DNA was tested for BRAF V600E, NRAS codon 61, HRAS the detection of specific miRNAs in the preoperative assessment codon 61, KRAS codons 12/13 mutations using real-time LightCycler of thyroid nodules. PCR as previously reported (19). RET/PTC1, RET/PTC3, and PAX8/ Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  3. 3. 1602 Nikiforova et al. miRNA Profiles in Thyroid Tumors J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 TABLE 1. Ten most up-regulated miRNAs in various thyroid tumors: PCs, conventional and oncocytic FCs, and PDCs PCs Conventional FCs Oncocytic FCs PDCs miRNA Fold miRNA Fold miRNA Fold miRNA Fold miR-187 73.7 miR-187 33.4 miR-187 53.7 miR-181b 24.7 miR-221 19.1 miR-181b 13.6 miR-221 46.9 miR-187 23.5 miR-222 17.2 miR-200a 6.7 miR-339 42.5 miR-221 20.6 miR-181b 14.4 miR-224 6.2 miR-183 26.6 miR-129 17.3 miR-146b 10.5 miR-182 5.7 miR-222 26.0 miR-222 13.5 miR-155 9.5 miR-155 5.5 miR-181b 19.7 miR-146b 12.4 miR-122a 8.9 miR-222 4.5 miR-182 15.6 miR-339 6.1 miR-31 7.5 miR-221 3.6 miR-213 14.3 miR-183 5.7 miR-205 6.8 miR-96 3.5 miR-96 10.3 miR-213 4.3 miR-224 6.2 miR-146b 2.8 miR-197 7.2 miR-181a 3.9 The fold change of miRNA expression was calculated relative to normal thyroid tissue. The bolded miRNAs signify those that were also significantly overexpressed in thyroid tumors as compared with hyperplastic nodules. PPAR rearrangements were detected from RNA by RT-PCR as previ- Results ously reported (20, 25). Overall miRNA expression in thyroid normal and tumor Statistical analysis tissues Agglomerative hierarchical clustering between thyroid specimens Initially, a set of five normal thyroid tissues, two hyperplastic was performed in R software. miRNA expression data underwent fil- nodules, and 24 thyroid tumors including nine PCs, five FCs, tering by computation of the number of missingness for each miRNA. four FAs, two ACs, two PDCs, and two MCs was used to eval- The purpose of the filtering was to remove miRNAs with no detectable expression across all thyroid specimens that may introduce noise to the uate the expression of 158 human mature miRNAs. Overall, 148 clustering. A subset of 59 miRNAs remained after filtering and was used (94%) of miRNAs were found to be expressed in normal and for hierarchical clustering. hyperplastic thyroid tissue. In the majority of thyroid tumors, 47 For class comparisons to identify differentially expressed miRNAs, a of 148 miRNAs (32%) were consistently up-regulated when t test was first performed for each miRNA for two class comparisons,the compared with normal tissue, and 57 of 148 (38%) were down- and Benjamini-Hochberg procedure (27) was applied to control the false discovery rate at 5% (i.e. among detected miRNAs, 5% of them are false regulated with more than a 2-fold change. positives on average). For class predictions in the TaqMan MicroRNA When the expression levels of miRNAs in all well-differenti- Human Panel data, a nearest-shrunken-centroid method by Prediction ated thyroid carcinomas (PCs, FCs, and MCs) were compared Analysis of Microarray (28) software was applied. The software inte- with those in less differentiated carcinomas (PDCs and ACs), 31 grates selection of predictive miRNAs while constructing the prediction miRNAs showed higher expression levels in the well-differenti- model and performs 10-fold cross-validation. Linear discriminant anal- ysis was used for the class prediction of the set of individual miRNAs ated carcinomas, whereas 27 miRNAs were expressed at higher without feature selection. levels in less differentiated carcinomas, showing no statistically For individually studied miRNAs, one-way ANOVA was used to significant difference between the numbers of up-regulated and detect the statistical significance of mutation effects for each miRNA. down-regulated genes in the two groups (P 0.69). Principal component analysis (PCA) was applied to provide an unsuper- vised visualization and investigation of the relationship between miRNA Expression of miRNAs in various types of thyroid expression and mutation type. For each pair of mutation-specific groups, PCA was performed to project the samples to the first principal compo- tumors nent, and a simple t test was applied to test for significance of separation To determine whether different histopathological types of between the two groups. thyroid tumors have distinct miRNA profiles, the unsupervised TABLE 2. Ten most up-regulated miRNAs in various thyroid tumors: ACs, conventional and oncoytic FAs, and MCs ACs Conventional FAs Oncocytic FAs MCs miRNA Fold miRNA Fold miRNA Fold miRNA Fold miR-302c 114.2 miR-200a 35.0 miR-31 35.2 miR-323 142.2 miR-205 75.6 miR-181b 23.1 miR-339 18.2 miR-370 136.4 miR-137 64.9 miR-200b 18.1 miR-183 17.5 miR-129 123.3 miR-187 19.1 miR-339 13.4 miR-182 15.6 miR-137 116.4 miR-214 16.3 miR-224 11.0 miR-181b 14.1 miR-10a 63.6 miR-181b 15.3 miR-205 10.1 miR-221 12.8 miR-124a 61.9 miR-155 13.2 miR-210 9.6 miR-96 10.8 miR-224 55.2 miR-224 12.0 miR-190 9.1 miR-182 8.4 miR-127 52.3 miR-222 9.5 miR-328 8.1 miR-224 6.6 miR-9 42.1 miR-221 9.4 miR-342 7.9 miR-203 6.2 miR-154 32.3 The fold change of miRNA expression was calculated relative to normal thyroid tissue. The bolded miRNAs signify those that were also significantly overexpressed in thyroid tumors as compared with hyperplastic nodules. Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  4. 4. J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 jcem.endojournals.org 1603 hierarchical clustering analysis of miRNA expression was per- tested for the presence of BRAF and RAS point mutations and formed. It revealed four major clusters: oncocytic follicular tu- RET/PTC and PAX8/PPAR rearrangements. Among PCs, two mors (adenomas and carcinomas), conventional follicular tu- tumors were positive for RET/PTC1, one for RET/PTC3, two mors (adenomas and carcinomas), PCs and MCs (Fig. 1). The for BRAF, and two for NRAS, and two were found negative for first three clusters were located closer to each other, whereas the any of these mutations. Both PCs positive for RAS mutation were MC cluster was at the greatest distance, consistent with their the follicular variants. Among FCs, two were positive for PAX8/ different cell type origins, i.e. thyroid follicular cells and C-cells. PPAR and one for NRAS, and two were negative for these The oncocytic tumor cluster was most segregated of the three alterations. The unsupervised hierarchical clustering analysis of follicular cell-derived tumor clusters. Less differentiated tumors miRNA expression in the tumors positive for mutations was (PDCs and ACs) did not form distinct clusters and were situated performed. It demonstrated that BRAF-, RET/PTC-, and PAX8/ either close or within the papillary or follicular clusters or sep- PPAR -positive tumors formed individual clusters, whereas tu- arately, supporting their origin from the well-differentiated PCs mors with RAS mutations did not form a separate cluster and and FCs and their propensity for profound dedifferentiation. were positioned between other clusters (Fig. 2). These pointed Next, we searched for individual miRNAs that had the toward the possible variation in miRNA signatures depending on highest levels of overexpression in specific tumor types. The the mutational status. top 10 up-regulated miRNAs in each type of malignant and To explore this possibility further, an additional 14 PCs with benign thyroid tumor are shown in Tables 1 and 2. There was BRAF, RAS, RET/PTC, or none of these mutations were ana- virtually no overlap in the highly expressed miRNAs between lyzed for the expression of six miRNAs with the highest overall MCs and the rest of the tumors, all of which derive from expression levels in PCs: miR-187, miR-221, miR-222, miR- thyroid follicular cells. Seven miRNAs, miR-187, miR-221, 181b, miR-146b, and miR-155 (Fig. 3A). The one-way ANOVA miR-222, miR-181b, miR-146b, miR-155, and miR-224, analysis of all 23 PCs revealed significant differences in the ex- were most consistently overexpressed in all follicular cell-de- pression levels of five of these miRNAs between the groups (P rived carcinomas, although their expression levels varied sig- 0.09 to P 0.0001), whereas the expression of miR-181b was nificantly between individual tumor types. In ACs, these not significantly different (P 0.95). PCA was used for an un- miRNAs were overexpressed at lower levels, whereas miR- supervised assessment of the relationship between mutation type 302c, miR-205, and miR-137 showed most dramatic up-reg- and expression levels of the six miRNAs. It revealed a strong ulation. Additionally, miRNA-339, miR-183, and miR-197 relationship between miRNA expression and mutational status were highly expressed in oncocytic follicular carcinomas. (Fig. 3B). Similar results were obtained using pairwise compar- Among benign adenomas, tumors of conventional type and ison between the individual groups, which revealed significant oncocytic type had quite distinct sets of mostly up-regulated differences in the expression profiles between all group pairs (P miRNAs, with miR-200a being at the top of the list in con- 0.003 or lower) except for the RET/PTC and RAS groups (P ventional follicular adenomas and miR-31 in oncocytic 0.46). adenomas. Diagnostic utility of miRNA detection in clinical thyroid Expression of miRNAs in tumors carrying different samples oncogenic mutations The major challenge in the preoperative FNA diagnosis of To explore whether miRNA expression profiles vary in tu- thyroid nodules is to differentiate between the follicular- mors with different initiating mutations, all PCs and FCs were patterned thyroid cancers and hyperplastic nodules, which are RAS RET/PTC RAS BRAF RAS OFC/OFA CFC/CFA PC MC RET/PTC PAX8/PPARγ FIG. 1. Cluster dendrogram of miRNA expression of thyroid tumors showing four major clusters: oncocytic FC (OFC) and FA (OFA), FIG. 2. Cluster dendrogram of miRNA expression in thyroid tumors with conventional FC (CFC) and FA (CFA), PC, and MC. different mutations. Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  5. 5. 1604 Nikiforova et al. miRNA Profiles in Thyroid Tumors J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 2.5 2 Log10 RQ 1.5 1 0.5 0 PC, RET/PTC PC, BRAF PC, RAS PC, NONE A miR-187 miR-221 miR-222 miR-146b miR-155 miR-181b B FIG. 3. A, Expression levels of selected miRNAs in PCs with various mutations (mean SE); B, PCA of log-transformed data for all mutational group probe sets. Blue, BRAF mutants; green, RAS mutants; red, RET/PTC mutants; yellow, tumors with no mutation. common in the general population (21, 22). Therefore, a di- setting, we selected a subset of miRNAs based on 1) at least 2-fold agnostically valid assay must distinguish thyroid tumors not overexpression in thyroid cancers compared with hyperplastic only from normal thyroid tissue but also from hyperplastic nodules (Tables 1 and 2, data in bold), and 2) their up-regulation nodules. in different types of thyroid cancer. A subset of seven miRNAs To explore the diagnostic utility of miRNA profiling in this (miR-187, miR-221, miR-222, miR-146b, miR-155, miR-224, Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  6. 6. J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 jcem.endojournals.org 1605 and miR-197) was selected. Expression levels of these miRNAs sion in all types of thyroid cancers compared with hyperplastic were evaluated in surgical samples from additional 14 malignant nodules (Fig. 4A). This set of miRNAs was highly effective in thyroid tumors (six PCs, four FCs, two ACs, and two PDCs) and separating thyroid cancer from hyperplastic nodules in all sur- three hyperplastic nodules, showing different patterns of expres- gical samples studied, because supervised prediction analysis by 2.5 2 1.5 1 Log10 RQ 0.5 0 -0.5 -1 HN PC CFC OFC PDC AC -1.5 miR-187 miR-221 miR-222 miR-146b miR-155 miR-197 miR-224 A 2.5 2 1.5 1 Log 10 RQ 0.5 0 -0.5 -1 PC PC OFC PC PC PC PC PC HN HN HN HN HN -1.5 miR-187 miR-221 miR-222 miR-146b miR-155 miR-197 miR-224 B FIG. 4. A, Expression levels of selected miRNAs in various types of thyroid cancer based on the analysis of surgically removed tumors (mean SE); B, expression of selected miRNAs in thyroid preoperative FNA samples from patients who subsequently underwent surgery. The surgical pathology diagnoses are shown at the bottom. CFC, Conventional FC; HN, hyperplastic nodules; OFC, oncocytic FC. Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  7. 7. 1606 Nikiforova et al. miRNA Profiles in Thyroid Tumors J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 linear discriminant analysis misdiagnosed only one (2.4%) of 41 been a subject of a longstanding debate in the field. Although the malignant tumors and hyperplastic nodules. most recent World Health Organization classification of thyroid Then, the diagnostic accuracy of this panel was validated in tumors designates oncocytic thyroid adenomas and carcinomas 62 consecutive thyroid FNA samples. Thirteen of these patients as a variant of follicular tumors (13), some histological features underwent surgery based on cytological diagnosis (four positive and mutational profiles (29) support their independent origin. and eight atypical cytology) or clinical suspicion (one case), The results of miRNA analysis provide more evidence to support yielding the diagnosis of malignant tumor in eight cases (seven the latter notion, because it showed that oncocytic tumors had a PCs and one FC of oncocytic type) and hyperplastic nodule in five distinct set of up-regulated miRNA and clustered separately cases. Expression of miR-187, miR-221, miR-222, miR-146b, from conventional follicular tumors. Second, the clusters of fol- miR-155, miR-224, and miR-197 was analyzed in these FNA licular tumor and oncocytic tumor included both adenomas and samples without knowing the surgical pathology diagnosis. All carcinomas, supporting the possibility of stepwise progression of the eight malignant cases revealed more than 2-fold overex- for each tumor type. pression of one to six miRNAs tested (Fig. 4B). Among them, Some reports have suggested that overall, most miRNAs are seven PCs showed at least three up-regulated miRNAs, whereas expressed at lower levels in tumors and tumor-derived cell lines one case of oncocytic FC demonstrated up-regulation of compared with normal tissues and in poorly differentiated tu- miR-221 only. Among the four hyperplastic nodules that were mors compared with their well-differentiated counterparts (11, surgically removed based on atypical cytology, only one case 30). Other studies have not confirmed such a tendency (12). We showed more than 2-fold overexpression of one of these did not observe this trend in thyroid tumors, where a significant miRNAs (miR-197). The other 49 samples from patients that did proportion of miRNAs was found to be up-regulated and over- not undergo surgery revealed no up-regulation of any of these expressed at high levels in tumors compared with normal thyroid miRNAs in 46 cases, up-regulation of one miRNA in two cases, tissue. Likewise, we did not observe significant differences in and strong up-regulation of three miRNAs (miR-221, miR-222, miRNA expression in tumors depending on their differentiation and miR-146b) in one case. The latter case had atypical cytology status. but no surgery during the follow-up and therefore was removed Most interestingly, we observed that miRNA expression pro- from further analysis. All other cases had negative cytology and files had substantial variability within specific tumor types. The no further evidence of thyroid disease on follow-up and were first study of miRNA expression in thyroid PCs has reported the assumed to be negative for malignancy. Based on these data, the overall highest up-regulation of miR-221, miR-222, and miR- panel of seven miRNAs allowed discrimination between malig- 146b (14). We confirmed these data but show that these miRNAs nant tumors and hyperplastic nodules with high accuracy. Spe- are not equally expressed in all tumors of this type and mostly cifically, when at least one miRNA was overexpressed more than up-regulated in tumors carrying BRAF mutation. Another study 2-fold, the sensitivity of tumor detection was 100%, specificity has shown significant up-regulation of miR-221, miR-222, and 94%, and accuracy 95%. When three or more miRNAs were miR-181b in PCs compared with normal thyroid tissue (15). Our up-regulated, the sensitivity of tumor detection was 88%, spec- data suggest that of these three markers, miR-181b is overex- ificity 100%, and accuracy 98%. pressed in virtually all types of follicular cell-derived thyroid tumors and also in thyroid hyperplastic nodules. We also iden- tified several additional miRNAs that are highly up-regulated in Discussion PCs. One of them, miR-187, appears to be the most up-regulated miRNA in tumors harboring RET/PTC rearrangement and RAS In this study, we demonstrate that various histopathological mutations but is expressed at significantly lower levels in tumors types of thyroid tumors derived from the same cell have distinct with BRAF mutation. miR-224, in contrast, is a pan-PC marker miRNA profiles, which further differ within the same tumor type because it is overexpressed, albeit at lower levels, in all PCs. reflecting specific oncogenic mutations found in these tumors. A Overall, we observed significant correlation between the mu- limited set of miRNAs can be used diagnostically with high ac- tational status of PCs and miRNA expression. The oncogenic curacy to detect thyroid cancer in the surgical and preoperative mutations in PCs, RET/PTC, BRAF, and RAS are all capable of FNA samples. activation of the MAPK pathway and rarely overlap in the same From the early stages of their discovery, it has been known tumor, suggesting that activation of a single effector of this path- that many miRNAs are expressed in a tissue-specific manner (1, way is sufficient for transformation (31). However, each of these 2). Therefore, it was not surprising to find markedly different mutations is associated with distinct coding gene expression pro- profiles of miRNA expression between C-cell-derived thyroid files, phenotypical characteristics, and biological properties of MCs and all other thyroid tumors that derives from follicular PCs (19, 32, 33). Of those, the association between BRAF mu- cells. However, the results of this study show that miRNA pro- tation and tumor recurrence and treatment failure has produced files have significant variability between different histopatho- the most substantial effect on clinical management of patients logical types of thyroid tumors that originate from the same cell. with thyroid cancer (34). Previous attempts have been made to The importance of this information for classification of thy- identify the dysregulated miRNAs in thyroid cell lines carrying roid tumors and refining their scheme of progression is at least mutant BRAF and RET/PTC1 in comparison with the normal 2-fold. First, it supports the independent notion that oncocytic thyroid cell line (35, 36). Herein, we show that miRNA expres- tumors represent a distinct class of thyroid neoplasms, which has sion profiles of human thyroid tumors carrying BRAF and other Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  8. 8. J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 jcem.endojournals.org 1607 mutations are substantially different. Although the nature of the and progression in chronic lymphocytic leukemia. N Engl J Med 353:1793– 1801 correlation between BRAF mutation and miRNA expression re- 10. Porkka KP, Pfeiffer MJ, Waltering KK, Vessella RL, Tammela TL, Visakorpi mains unclear, this finding suggests that miRNA profiling may T 2007 MicroRNA expression profiling in prostate cancer. Cancer Res 67: also be potentially used as a prognostic marker for thyroid 6130 – 6135 cancer. 11. Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, Peck D, Sweet-Cordero A, Ebert BL, Mak RH, Ferrando AA, Downing JR, Jacks T, Horvitz HR, Golub Weber et al. (17) have found significant overexpression of TR 2005 MicroRNA expression profiles classify human cancers. Nature 435: miR-197 and miR-346 in FCs. In our series, up-regulation of 834 – 838 miR-197 was predominantly seen in the oncocytic type of FC, 12. Volinia S, Calin GA, Liu CG, Ambs S, Cimmino A, Petrocca F, Visone R, Iorio M, Roldo C, Ferracin M, Prueitt RL, Yanaihara N, Lanza G, Scarpa A, Vec- whereas miR-346 was not studied. We identified several other chione A, Negrini M, Harris CC, Croce CM 2006 A microRNA expression miRNAs up-regulated in these tumors, with miR-187 showing signature of human solid tumors defines cancer gene targets. Proc Natl Acad the highest levels of overexpression in both conventional and Sci USA 103:2257–2261 13. DeLellis RA, Lloyd RV, Heitz PU, Eng C, eds 2004 World Health Organization oncocytic FCs. classification of tumours: pathology and genetics of tumours of endocrine The results obtained by the analysis of surgically removed organs. Lyon, France: IARC Press tumors allowed us to assemble a small panel of miRNAs for the 14. He H, Jazdzewski K, Li W, Liyanarachchi S, Nagy R, Volinia S, Calin GA, Liu preoperative diagnosis of thyroid nodules. Using a large series of CG, Franssila K, Suster S, Kloos RT, Croce CM, de la Chapelle A 2005 The role of microRNA genes in papillary thyroid carcinoma. Proc Natl Acad Sci well-characterized FNA samples, we demonstrated the feasibil- USA 102:19075–19080 ity of miRNA testing in FNA samples and provided evidence for 15. Pallante P, Visone R, Ferracin M, Ferraro A, Berlingieri MT, Troncone G, high diagnostic potential of a limited panel of miRNAs in cancer Chiappetta G, Liu CG, Santoro M, Negrini M, Croce CM, Fusco A 2006 MicroRNA deregulation in human thyroid papillary carcinomas. Endocr Relat diagnosis. Despite overall lower levels of miRNA overexpression Cancer 13:497–508 in the FNA samples compared with surgical material, the panel 16. Visone R, Pallante P, Vecchione A, Cirombella R, Ferracin M, Ferraro A, provided high diagnostic accuracy. Although our findings need Volinia S, Coluzzi S, Leone V, Borbone E, Liu CG, Petrocca F, Troncone G, Calin GA, Scarpa A, Colato C, Tallini G, Santoro M, Croce CM, Fusco A 2007 to be further confirmed in a larger series of samples, they lay the Specific microRNAs are downregulated in human thyroid anaplastic carcino- foundation for the use of miRNA profiling as an effective diag- mas. Oncogene 26:7590 –7595 nostic tool for the preoperative assessment of thyroid nodules. 17. Weber F, Teresi RE, Broelsch CE, Frilling A, Eng C 2006 A limited set of human MicroRNA is deregulated in follicular thyroid carcinoma. J Clin En- docrinol Metab 91:3584 –3591 18. Kondo T, Ezzat S, Asa SL 2006 Pathogenetic mechanisms in thyroid follicular- cell neoplasia. Nat Rev 6:292–306 Acknowledgments 19. Adeniran AJ, Zhu Z, Gandhi M, Steward DL, Fidler JP, Giordano TJ, Bid- dinger PW, Nikiforov YE 2006 Correlation between genetic alterations and Address all correspondence and requests for reprints to: Dr. Marina microscopic features, clinical manifestations, and prognostic characteristics of Nikiforova, Department of Pathology, University of Pittsburgh, 3550 thyroid papillary carcinomas. Am J Surg Pathol 30:216 –222 Terrace Street, Pittsburgh, Pennsylvania 15261. E-mail: nikiforovamn@ 20. Nikiforova MN, Lynch RA, Biddinger PW, Alexander EK, Dorn GW, 2nd, upmc.edu. Tallini G, Kroll TG, Nikiforov YE 2003 RAS point mutations and PAX8- This work was supported by National Institutes of Health Grant R01 PPAR rearrangement in thyroid tumors: evidence for distinct molecular path- CA88041 (to Y.E.N.). G.C.T. is partially supported by the National ways in thyroid follicular carcinoma. J Clin Endocrinol Metab 88:2318 –2326 Institutes of Health (1 KL2 RR024154-02). 21. Mazzaferri EL 1992 Thyroid cancer in thyroid nodules: finding a needle in the Disclosure Statement: The authors have nothing to disclose. haystack. Am J Med 93:359 –362 22. Mazzaferri EL 1993 Management of a solitary thyroid nodule. N Engl J Med 328:553–559 23. Salvatore G, Giannini R, Faviana P, Caleo A, Migliaccio I, Fagin JA, Nikiforov YE, Troncone G, Palombini L, Basolo F, Santoro M 2004 Analysis of BRAF References point mutation and RET/PTC rearrangement refines the fine-needle aspiration diagnosis of papillary thyroid carcinoma. J Clin Endocrinol Metab 89:5175– 1. Bartel DP 2004 MicroRNAs: genomics, biogenesis, mechanism, and function. 5180 Cell 116:281–297 24. Xing M, Tufano RP, Tufaro AP, Basaria S, Ewertz M, Rosenbaum E, Byrne 2. Ambros V 2004 The functions of animal microRNAs. Nature 431:350 –355 PJ, Wang J, Sidransky D, Ladenson PW 2004 Detection of BRAF mutation on 3. Esquela-Kerscher A, Slack FJ 2006 Oncomirs: microRNAs with a role in can- fine needle aspiration biopsy specimens: a new diagnostic tool for papillary cer. Nat Rev 6:259 –269 thyroid cancer. J Clin Endocrinol Metab 89:2867–2872 4. Negrini M, Ferracin M, Sabbioni S, Croce CM 2007 MicroRNAs in human 25. Nikiforova MN, Caudill CM, Biddinger P, Nikiforov YE 2002 Prevalence of cancer: from research to therapy. J Cell Sci 120:1833–1840 5. Calin GA, Croce CM 2006 MicroRNA signatures in human cancers. Nat Rev RET/PTC rearrangements in Hashimoto’s thyroiditis and papillary thyroid 6:857– 866 carcinomas. Int J Surg Pathol 10:15–22 6. Yanaihara N, Caplen N, Bowman E, Seike M, Kumamoto K, Yi M, Stephens 26. Livak KJ, Schmittgen TD 2001 Analysis of relative gene expression data using RM, Okamoto A, Yokota J, Tanaka T, Calin GA, Liu CG, Croce CM, Harris real-time quantitative PCR and the 2( CT) method. Methods 25:402– 408 CC 2006 Unique microRNA molecular profiles in lung cancer diagnosis and 27. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I 2001 Controlling the false prognosis. Cancer Cell 9:189 –198 discovery rate in behavior genetics research. Behav Brain Res 125:279 –284 7. Iorio MV, Ferracin M, Liu CG, Veronese A, Spizzo R, Sabbioni S, Magri E, 28. Tibshirani R, Hastie T, Narasimhan B, Chu G 2002 Diagnosis of multiple Pedriali M, Fabbri M, Campiglio M, Menard S, Palazzo JP, Rosenberg A, cancer types by shrunken centroids of gene expression. Proc Natl Acad Sci USA Musiani P, Volinia S, Nenci I, Calin GA, Querzoli P, Negrini M, Croce CM 99:6567– 6572 2005 MicroRNA gene expression deregulation in human breast cancer. Cancer 29. Nikiforova MN, Biddinger PW, Caudill CM, Kroll TG, Nikiforov YE 2002 Res 65:7065–7070 PAX8-PPAR rearrangement in thyroid tumors: RT-PCR and immunohisto- 8. Chan JA, Krichevsky AM, Kosik KS 2005 MicroRNA-21 is an antiapoptotic chemical analyses. Am J Surg Pathol 26:1016 –1023 factor in human glioblastoma cells. Cancer Res 65:6029 – 6033 30. Gaur A, Jewell DA, Liang Y, Ridzon D, Moore JH, Chen C, Ambros VR, Israel 9. Calin GA, Ferracin M, Cimmino A, Di Leva G, Shimizu M, Wojcik SE, Iorio MA 2007 Characterization of microRNA expression levels and their biological MV, Visone R, Sever NI, Fabbri M, Iuliano R, Palumbo T, Pichiorri F, Roldo correlates in human cancer cell lines. Cancer Res 67:2456 –2468 C, Garzon R, Sevignani C, Rassenti L, Alder H, Volinia S, Liu CG, Kipps TJ, 31. Kimura ET, Nikiforova MN, Zhu Z, Knauf JA, Nikiforov YE, Fagin JA 2003 Negrini M, Croce CM 2005 A microRNA signature associated with prognosis High prevalence of BRAF mutations in thyroid cancer: genetic evidence for Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009
  9. 9. 1608 Nikiforova et al. miRNA Profiles in Thyroid Tumors J Clin Endocrinol Metab, May 2008, 93(5):1600 –1608 constitutive activation of the RET/PTC-RAS-BRAF signaling pathway in pap- sion profiles discovered by DNA microarray analysis. Oncogene 24:6646 – illary thyroid carcinoma. Cancer Res 63:1454 –1457 6656 32. Xing M, Westra WH, Tufano RP, Cohen Y, Rosenbaum E, Rhoden KJ, 34. Xing M 2005 BRAF mutation in thyroid cancer. Endocr Relat Cancer 12: Carson KA, Vasko V, Larin A, Tallini G, Tolaney S, Holt EH, Hui P, 245–262 Umbricht CB, Basaria S, Ewertz M, Tufaro AP, Califano JA, Ringel MD, 35. Cahill S, Smyth P, Denning K, Flavin R, Li J, Potratz A, Guenther SM, Henfrey Zeiger MA, Sidransky D, Ladenson PW 2005 BRAF mutation predicts a R, O’Leary JJ, Sheils O 2007 Effect of BRAFV600E mutation on transcription poorer clinical prognosis for papillary thyroid cancer. J Clin Endocrinol and post-transcriptional regulation in a papillary thyroid carcinoma model. Metab 90:6373– 6379 Mol Cancer 6:21 33. Giordano TJ, Kuick R, Thomas DG, Misek DE, Vinco M, Sanders D, Zhu Z, 36. Cahill S, Smyth P, Finn SP, Denning K, Flavin R, O’Regan EM, Li J, Potratz Ciampi R, Roh M, Shedden K, Gauger P, Doherty G, Thompson NW, Hanash A, Guenther SM, Henfrey R, O’Leary JJ, Sheils O 2006 Effect of ret/PTC 1 S, Koenig RJ, Nikiforov YE 2005 Molecular classification of papillary thyroid rearrangement on transcription and post-transcriptional regulation in a pap- carcinoma: distinct BRAF, RAS, and RET/PTC mutation-specific gene expres- illary thyroid carcinoma model. Mol Cancer 5:70 Downloaded from jcem.endojournals.org at Univ Pittsburgh Falk Lib Hlth Sciences on June 17, 2009