MicroRNA-222 and MicroRNA-146b are tissue and circulating biomarkers of recurrent papillary thyroid cancer

Authors

  • James C. Lee FRACS,

    Corresponding author
    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Endocrine Surgical Unit, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    • Corresponding author: James C. Lee, FRACS, Cancer Genetics Laboratory, Level 9, Kolling Institute of Medical Research, Corner Reserve Road and First Avenue, Royal North Shore Hospital, St. Leonards, NSW, 2065, Australia; Fax: (011) 61 294371732; lee.jamesc@gmail.com

    Search for more papers by this author
  • Jing Ting Zhao PhD,

    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Roderick J. Clifton-Bligh FRACP, PhD,

    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Department of Endocrinology, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Anthony Gill FRCPA,

    1. Department of Pathology, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Northern Translational Cancer Research Unit, Sydney, New South Wales, Australia
    Search for more papers by this author
  • Justin S. Gundara MBBS(Hons),

    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Endocrine Surgical Unit, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Julian C. Ip BMedSc(Hons), MBBS,

    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Endocrine Surgical Unit, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Anthony Glover FRACS,

    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Endocrine Surgical Unit, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Mark S. Sywak FRACS, MMedSci(Epi),

    1. Endocrine Surgical Unit, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Leigh W. Delbridge FRACS, MD,

    1. Endocrine Surgical Unit, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Bruce G. Robinson FRACP, MD,

    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Department of Endocrinology, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author
  • Stanley B. Sidhu FRACS, PhD

    1. Kolling Institute of Medical Research, Cancer Genetics Laboratory, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    2. Endocrine Surgical Unit, Royal North Shore Hospital and University of Sydney, St. Leonards, New South Wales, Australia
    Search for more papers by this author

  • We acknowledge the assistance of Dr. Aleksandra Popadich and Dr. Mark Versnick for the collection of some preoperative blood samples, and we thank statistician Miss Jillian Patterson for statistical support.

Abstract

BACKGROUND

Papillary thyroid cancer (PTC) persistence or recurrence and the need for long-term surveillance can cause significant inconvenience and morbidity in patients. Currently, recurrence risk stratification is accomplished by using clinicopathologic factors, and serum thyroglobulin is the only commercially available marker for persistent or recurrent disease. The objective of this study was to determine microRNA (miRNA) expression in PTC and determine whether 1 or more miRNAs could be measured in plasma as a biomarker for recurrence.

METHODS

Patients with recurrent PTC (Rc-PTC) and those without recurrence (NR-PTC) were retrospectively recruited for a comparison of their tumor miRNA profiles. Patients with either newly diagnosed PTC or multinodular goiter who were undergoing total thyroidectomy were prospectively recruited for an analysis of preoperative and postoperative circulating miRNA levels. Healthy volunteers were recruited as the control group.

RESULTS

MicroRNA-222 and miR-146b were over-expressed 10.8-fold and 8.9-fold, respectively, in Rc-PTC tumors compared with NR-PTC tumors (P = .014 and P = .038, respectively). In plasma from preoperative PTC patients, levels of miR-222 and miR-146b were higher compared with the levels in plasma from healthy volunteers (P < .01 for both). Reductions of 2.7-fold and 5.1-fold were observed in the plasma levels of miR-222 and miR-146b, respectively, after total thyroidectomy (P = .03 for both).

CONCLUSIONS

This study demonstrated that tumor levels of miR-222 and miR-146b are associated with PTC recurrence and that miR-222 and miR-146b levels in the circulation correspond to the presence of PTC. The potential of these miRNAs as tumor biomarkers to improve patient stratification according to the risk of recurrence and as circulating biomarkers for PTC surveillance warrants further study. Cancer 2013;119:4358–4365. © 2013 American Cancer Society.

INTRODUCTION

Papillary thyroid cancer (PTC) is the most common endocrine malignancy, and its incidence has been rising over the last few decades.[1, 2] For the majority of patients with PTC, the prognosis is excellent, with overall 10-year survival rates of >90%. However, up to 30% of patients suffer locoregional or distant metastatic recurrence at 10 years.[3-5] Therefore, one of the key issues in the management of patients with PTC is minimizing the morbidity and mortality associated with recurrent disease. This is echoed by the Revised American Thyroid Association (ATA) guidelines, which state that areas for ongoing research include improving risk stratification and dealing with the problem of antithyroglobulin (anti-Tg) antibodies (ATA guidelines location keys D10 and D12).[6] MicroRNAs (miRNAs) as biomarkers are potential candidates to help achieve these goals.

Risk stratification to identify the subset of patients at higher risk of recurrence currently is achieved by using various clinical and pathologic factors. As many as 17 risk stratification tools have been suggested based on those factors, indicating the lack of accuracy or consensus in the matter.[7] Various molecular markers, the most studied of which is the proto-oncogene for the B-isoform of the Raf kinase protein (BRAF), have been investigated in the hope of more accurate risk stratification.[8-11] However, no conclusive evidence has been published. The objectives of this study were to identify miRNA biomarkers of recurrence in PTC tissue and to determine whether they can be measured in the circulation, thereby providing a noninvasive method of detecting disease recurrence.

MATERIALS AND METHODS

This study was granted ethics approval by the local Human Research Ethics Committee before commencement.

Patient Groups

Four main groups of patients were recruited for this study, and written informed consent was obtained from all participants. Two groups were recruited retrospectively for the comparison of tissue miRNA expression in PTCs with and without recurrence: patients with PTC who were diagnosed with recurrence after initial curative treatment (Rc-PTC) and patients with PTC confined to the thyroid gland (no lymph node metastasis) who had not been diagnosed with recurrence at their latest follow-up (NR-PTC). Patients who had adequate archival tissue samples were selected randomly from the surgical database. Curative treatment was defined as undetectable stimulated thyroglobulin (Tg) and no uptake of radioactive iodine (RAI) at 1 year after initial surgery and RAI ablation. Therefore, recurrence was defined as the subsequent appearance of histologically confirmed PTC or RAI-avid lesions either in the neck or at a distant site.

Two groups were recruited prospectively for measurements of plasma miRNA levels: patients who had PTC without lateral lymph node or distant metastasis and were undergoing total thyroidectomy (Pl-PTC) and patients who had multinodular goiter (MNG) and were undergoing total thyroidectomy (Pl-MNG). Patients who had a history of other malignancies or who had deranged thyroid function at presentation were excluded from this part of the study. In addition, normal thyroid tissues from laryngectomy patients were used as normal controls (NC) in the microarray, whereas plasma samples from healthy volunteers with normal thyroids were used as normal controls (Pl-NC) in the plasma miRNA measurements.

Clinical Management of Papillary Thyroid Cancer

All patients with PTC were treated in accordance with ATA guidelines, including prophylactic central lymph node dissection if the PTC was diagnosed before surgery.[6] Patients also underwent RAI ablation in accordance with ATA recommendations.

Plasma Samples

Venous blood was drawn from each patient both on the day of surgery before commencement of the procedure and between 2 and 6 weeks after surgery. Blood was drawn with the Vaccutainer venepuncture system into ethylene diamine tetraacetic acid (EDTA) tubes (Becton, Dickinson and Company, Melbourne, Australia). The first 3 mL were discarded to avoid possible contamination with miRNAs of the skin plug from the puncture. The blood was then centrifuged at ×800g for 15 minutes. A maximum of 4 hours between the blood draw and separation into plasma was the standardized protocol in this study, although it has been reported that miRNAs are stable for up to 24 hours of incubation at room temperature.[12] The supernatant (plasma) was separated from the cellular layer by pipetting, and the 3 to 5 mm of plasma just above the interphase was sacrificed to prevent disturbance of the cellular layer. The plasma was stored at −80°C until RNA extraction.

RNA Extraction

Total RNA was extracted from approximately 5 to 10 mm3 of formalin-fixed, paraffin-embedded (FFPE) samples using the FFPE RNeasy Kit (Qiagen, Hilden, Germany) and from 200 μL of thawed plasma with the miRNeasy Mini Kit (Qiagen), in accordance with the manufacturer's instructions. The PTC tissue samples for RNA extraction were macrodissected in archival FFPE blocks from areas that contained >90% malignant tissue. The plasma RNA extraction yielded 34 μL of RNA from each 200-μL sample of plasma. The RNA quality and concentration were assessed using a NanoDrop ND1000 Spectrophotometer (ThermoFisher Scientific, Waltham, Mass).

Tissue Microarray

RNA samples were labeled with fluorescence Hy3 using the miRCURY Hy3/Hy5 power labeling kit (Exiqon A/S, Vedbaek, Denmark), and miRNA microarray profiling was performed by Exiqon A/S on the miRCURY LNA microRNA Array (fifth generation; Exion A/S), which targets a total of 1261 human, mouse, and rat miRNA sequences registered in the miRBASE database (version 14.0). The average log median ratios (LMRs) of samples, the difference in LMRs (ΔLMR), and the corresponding P values between sample groups were calculated.

Selection of MicroRNAs

In building the panel of miRNAs with the potential to identify PTC associated with recurrence, the results of the tissue microarray study and the published literature were taken into account. A fold-change (2−ΔLMR) over 1.4-fold and P values < .01 favored the selection of the miRNAs as well as frequent reports of being a differentiator of PTC from benign thyroid pathology. We selected a panel of 8 miRNAs.

The selection of miRNAs for the plasma investigation also was based on several factors. The miRNAs that were identified as differentiators of recurrent and nonrecurrent PTC were included as a priority. In addition, miRNAs that were frequently reported as being over-expressed in PTC compared with benign thyroid disease also were included, whereas miRNAs that were highly expressed in healthy individuals were ruled out.[13-16] Because a reduction in plasma miRNA levels was hypothesized, only miRNAs that were over-expressed in PTC were considered. A panel of 6 miRNAs was selected for the plasma miRNA investigation.

Quantitative Reverse Transcriptase-Polymerase Chain Reaction

The expression levels of individual miRNAs were measured with quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) using TaqMan miRNA assays (Applied Biosystems, Foster City, Calif). The relative expression (RQ) was obtained using the ΔΔCt method. Briefly, total RNA was reverse transcribed to complementary DNA (cDNA) using TaqMan miRNA primers according to the manufacturer's instructions. Twenty nanograms of tumor total RNA and 2 μL of plasma total RNA were used as the starting material. The PCR products were then amplified from cDNA using the TaqMan miRNA probes with the ABI 7900HT Real-time PCR System according to the manufacturer's instructions. Standard TaqMan miRNA cycling conditions were used, and all samples were run in triplicate.

RNU48 was used as an endogenous control of the tissue samples, and miR-16 was used as an endogenous control for the plasma samples. Although there is no consensus on the optimal gene for endogenous control,[12, 17] many other investigators also have reported the use of miR-16 in their studies of serum or plasma samples.[18, 19] All samples across all PCR plates were calibrated against commercially available human thyroid total RNA (Ambion; Life Technologies Corporation, Carlsbad, Calif).

BRAF V600E Immunochemistry

BRAF immunohistochemistry was performed on FFPE tissue using a mouse monoclonal antibody (clone VE1). This antibody was raised to react with a substitution of glutamic acid (E) for valine (V) at residue 600 of BRAF (BRAFV600E), but not with wild-type BRAF or non-V600E mutations.[20, 21] The immunohistochemical staining was read as either positive or negative. This method of detecting BRAF mutations was recently validated against Sanger sequencing reportedly and was highly sensitive and specific.[21]

Statistics

For the analysis of clinical data, Fisher exact tests and chi-square tests were used, as appropriate, to test for the significance of categorical data, and unpaired t tests were used for continuous data. Statistical significance was set at P ≤ .05. For comparisons of miRNA expression, Mann-Whitney U tests or Student t tests were used to analyze unpaired data, and paired t tests were used to analyze paired data. All statistical analyses were performed using Stata/SE version 12.1 (StataCorp, College Station, Tex).

RESULTS

Clinicopathologic Characteristics

Recurrent versus nonrecurrent papillary thyroid cancer

Patient demographics and pathologic tumor features are listed in Table 1. Other than age, no other significant differences were observed between the Rc-PTC group (N = 9) and the NR-PTC group (N = 17). In particular, BRAF mutations and other high-risk pathologic features were represented similarly in both groups. Significant differences were observed in the rates of lymph node metastasis and disease-free survival because of the selection criteria. The median follow-up for the Rc-PTC group was 72 months, which was significantly longer than that for the NR-PTC group (24 months; P = .03).

Table 1. Characteristics of Patients
 Tissue Profiling GroupsPlasma Profiling Groups
CharacteristicRc-PTC, N = 9NR-PTC, N = 17Pl-PTC, N = 42Pl-MNG, N = 36
  1. Abbreviations: BRAF, proto-oncogene for the B-isoform of the Raf kinase protein; DFS, disease-free survival; ETE, extrathyroidal extension; FV, follicular variant; LN, lymph nodes; LVI, lymphovascular invasion; NA, not applicable; NR-PTC, nonrecurrent papillary thyroid cancer group; Pl-MNG, plasma multinodular goitre group; Pl-PTC, plasma papillary thyroid cancer group; R1, microscopic positive; Rc-PTC, recurrent papillary thyroid cancer group.

  2. a

    P = .03.

  3. b

    P < .001.

  4. c

    P = .02.

  5. d

    P < .001(because of selection).

  6. e

    P = .02 (because of selection).

  7. f

    P = .03.

Mean age (range), y57 (42-85)a44 (24-62)a49 (24-77)b60 (28-81)b
Men:women5:43:1413:29c3:33c
Histologic variants, %    
Classic:FV100:070:3083:17NA
BRAF mutation10069NANA
Primary tumor size: Mean ± SE, mm29 ± 4.922 ± 3.316 ± 1.55NA
Multifocal PTC, %404145NA
LVI, %601833NA
ETE, %44621NA
R1 margin, %251212NA
Positive LN, %100d0d57NA
DFS, %43e100eNANA
Follow-up: Median (range), mo72 (15-216)f24 (12-60)fNANA
Gland weight: Mean (range), gNANANA64 (20-230)

Of the 9 patients who had recurrent disease, 1 developed a recurrence in the central neck lymph nodes, 2 developed recurrences in the lung and lateral neck lymph nodes, and the remaining 6 patients developed recurrences in the lateral neck lymph nodes. The median time to recurrence was 24 months (range, 14-192 months).

Plasma microRNA profiling groups

For plasma miRNA expression measurement, there were 42 patients in the Pl-PTC group and 36 patients in the Pl-MNG group. The mean age was 49 years (range, 24-77 years) in the Pl-PTC group and 60 years (range, 28-81 years) in the Pl-MNG group (P < .001), and the ratios of men to women were 13:29 and 3:33, respectively (P = .02). The pathologic tumor features of patients in the Pl-PTC group are listed in Table 1. Postoperative plasma was available for analysis from 32 patients in the Pl-PTC group and 24 patients in the Pl-MNG group.

In the Pl-PTC group, there were 24 pathologic T1 (pT1) tumors, including 6 pT1a tumors, 9 pT2 tumors, 6 pT3 tumors, and 3 pT4 tumors. Although most tumors measured <40 mm, 9 tumors were classified as pT3 or pT4 by virtue of extrathyroidal extension (ETE). Of the 17 patients who had lymph nodes in their surgical specimen for analysis, 57% had lymph node metastases to the central lymph nodes. Some patients who were diagnosed with PTC incidentally or on diagnostic hemithyroidectomy specimens had no lymph nodes available for assessment. By exclusion, no patient had lateral lymph node involvement or distant metastasis. In the Pl-MNG group, the mean gland weight was 64 g (range, 20-230 g).

Tissue Microarray Analysis

MicroRNA microarray profiling was performed in a subset of 5 Rc-PTC tumors, in a subset of 10 NR-PTC tumors, and in 5 NC samples. All group-wise comparisons identified a total of 40 miRNAs as potential differentiators in PTC samples of recurrent and nonrecurrent disease. Seven of those 40 miRNAs were selected for further investigation with qRT-PCR from the total cohort of 9 Rc-PTC tumors and 17 NR-PTC tumors. In addition, miR-155 also was selected for this analysis on the basis of previous studies indicating that it is over-expressed in PTC compared with normal thyroid tissue.

Tissue Quantitative Reverse Transcriptase-Polymerase Chain Reaction

Recurrence versus nonrecurrence

The 8 miRNAs that were selected for qRT-PCR in Rc-PTC tumors versus NR-PTC tumors were miR-221, miR-222, miR-146b, miR-155, miR-1299, miR-193b, miR-620, and miR-890. The tissue expression of miR-222 was 10.8-fold higher (P = .014) and the tissue expression of miR-146b was 8.9-fold higher (P = .038) in tumors that were associated with recurrence (Table 2). The remaining 6 miRNAs in the panel either did not differ significantly in expression between the groups or were not expressed by >20% of the tumors and, thus, were excluded from analysis.

Table 2. Tissue MicroRNA Expression: Recurrent Papillary Thyroid Cancer Versus Nonrecurrent Papillary Thyroid Cancer
 RQ: Median (IQR) 
MicroRNARc-PTC, N = 9NR-PTC, N = 17P
  1. Abbreviations: IQR, interquartile range; NE, not expressed in >20% of samples; NR-PTC, nonrecurrent papillary thyroid cancer group; Rc-PTC, recurrent papillary thyroid cancer group; RQ, relative expression.

miR-221119.2 (98.0-214.8)70.2 (43.2-108.6).15
miR-222113 (39.8-206.4)38.5 (19.8-71.3).014
miR-146b522.2 (381.7-893.4)261.5 (114.8-607.3).038
miR-155179.7 (50.4-280.3)67.7 (28.6-146.6).28
miR-129931.2 (3.2-108.5)5.8 (2.7-7.7).23
miR-193b0.42 (0.32-0.73)0.27 (0.17-0.30).38
miR-620NENE 
miR-890NENE 

BRAF V600E versus BRAF wild type

BRAF staining was available for all Rc-PTC tumors and for 16 of 17 NR-PTC tumors. One hundred percent of Rc-PTC tumors had the BRAF mutation, whereas 69% of NR-PTC tumors had the BRAF mutation (P = .12).

Three miRNAs were significantly over-expressed in tumors that harbored the BRAF mutation compared with tumors that had the wild-type BRAF gene (Table 3). MicroRNA-146b exhibited the largest differential expression, with 35-fold over-expression in BRAF-mutated tumors (P = .0025), whereas miR-221 and miR-222 exhibited 6.5-fold and 13-fold over-expression, respectively (P = .005 and P = .012, respectively). No difference in miRNA expression was observed when we compared other high-risk features, such as ETE, lymphovascular invasion (LVI), and microscopic positive resection (R1) margins.

Table 3. Tissue MicroRNA Expression: BRAF Mutated Versus BRAF Wild Type
 RQ: Median (IQR) 
MicroRNABRAF Mutated, N = 23BRAF Wild Type, N = 7P
  1. Abbreviations: BRAF, proto-oncogene for the B-isoform of the Raf kinase protein; IQR Interquartile range; RQ, relative expression.

miR-221106.5 (50.3-214.8)11.4 (4.5-70.2).005
miR-22271.3 (30.2-119.4)6.0 (1.7-60.1).012
miR-146b396.6 (261.5-992.2)25.1 (5.8-272.1).0025

Plasma Quantitative Reverse Transcriptase-Polymerase Chain Reaction

The expression levels of 6 miRNAs (miR-221, miR-222, miR-146b, miR-21, miR-181b, and miR-31) were measured in 42 Pl-PTC patients before thyroidectomy (preoperative) and again after thyroidectomy (postoperative) in a subgroup of 32 patients. Preoperative and postoperative plasma miRNA expression levels were compared with baseline levels established in plasma from the Pl-NC group. MicroRNA-222 and miR-146b were included in the panel because they were over-expressed in Rc-PTC tumors; and miR-221, miR-21, miR-181b, and miR-31 were included because they all reportedly were over-expressed in PTC compared with MNG or normal thyroid tissue.[13-16]

Compared with baseline levels in the Pl-NC group, miR-221, miR-222, and miR-146b levels were significantly higher in the preoperative Pl-PTC group (P = .011, P = .004, and P = .001, respectively) (Table 4). Conversely, plasma miR-21 levels were not different in the preoperative Pl-PTC and Pl-NC groups. MicroRNA-181b and miR-31 did not exhibit sufficient levels in >30% of the plasma samples and, thus, were excluded from analysis.

Table 4. Preoperative Plasma MicroRNA Expression Among Patients in the Plasma Papillary Thyroid Cancer Group Versus the Plasma Normal Control Group
 RQ: Mean ± SE 
MicroRNAPre-op Pl-PTC, N = 42Pl-NC, N = 5P
  1. Abbreviations: Pl-NC, plasma normal control group; Pl-PTC, plasma papillary thyroid cancer group; Pre-op, preoperative; RQ, relative expression; SE, standard error.

miR-2210.65 ± 0.150.23 ± 0.06.011
miR-2220.060 ± 0.0120.022 ± 0.004.004
miR-146b0.56 ± 0.130.083 ± 0.019.001
miR-210.048 ± 0.0110.037 ± 0.006.40

A significant reduction in miR-221, miR-222, and miR-146b expression was observed in postoperative Pl-PTC patients. MicroRNA-221 decreased by 10.8-fold (P = .017), miR-222 decreased by 2.7-fold (P = .034), and miR-146b decreased by 5.1-fold (P = .032) in postoperative samples compared with preoperative samples in the Pl-PTC group (Table 5, Fig. 1).

Table 5. Plasma MicroRNA Expression of Preoperative Versus Postoperative Patients in the Plasma Papillary Thyroid Cancer Group
 RQ: Mean ± SE 
MicroRNAPre-op Pl-PTC, N = 32Post-op Pl-PTC, N = 32FC: Mean ± SEP
  1. Abbreviations: FC, fold change; Pre-op, preoperative; Pl-PTC, plasma papillary thyroid cancer group; Post-op, postoperative; RQ, relative expression; SE, standard error.

miR-2210.51 ± 0.150.12 ± 0.0310.8 ± 3.67.017
miR-2220.057 ± 0.0140.026 ± 0.0042.7 ± 0.62.034
miR-146b0.48 ± 0.150.15 ± 0.0285.1 ± 1.70.032
miR-210.031 ± 0.0110.016 ± 0.0052.5 ± 0.54.066
Figure 1.

Expression levels of (a) miRNA-222 (miR-222) and (b) miRNA-146b (miR-146b) are illustrated in patients who had papillary thyroid cancer without lateral lymph node or distant metastasis and were undergoing total thyroidectomy (preoperative [Pre-op] vs postoperative [Post-op]).

A similar pattern was observed with miRNA expression in the Pl-MNG group. Plasma levels of miR-221, miR-222, and miR-146b in the preoperative Pl-MNG group also were over-expressed significantly compared with baseline Pl-NC levels (P = .019, P = .004, and P = .003, respectively). No significant difference was observed between the preoperative Pl-PTC group and the preoperative Pl-MNG group with regard to the expression of miR-221, miR-222, or miR-146b. In addition, miR-221, miR-222, and miR-146b also had significantly reduced expression after total thyroidectomy in the postoperative Pl-MNG group (P = .05, P = .016, and P = .011, respectively).

In summary, we observed that miR-222 and miR-146b levels were significantly over-expressed in tumors that were associated with recurrence; whereas the levels of miR-221, miR-222, and miR-146b in plasma were significantly higher in patients who had either PTC or MNG before thyroidectomy than in healthy volunteers. Furthermore, a significant reduction in plasma levels of miR-221, miR-222, and miR-146b were observed after total thyroidectomy in both the PTC group and the MNG group.

DISCUSSION

Molecular predictors of PTC recurrence are currently lacking. Although the BRAF mutation has been suggested as a potential marker of aggressive PTC behavior, including recurrence,[7, 8, 22] its utility seems to reside more in aiding fine-needle aspiration diagnosis.[8-10, 23, 24] Other molecular markers, such as p27, p21, cyclin D1, CEACAM-1 (carcinoembryonic antigen-related cell adhesion molecule 1), osteopontin, and E-cadherin, also have been studied; however, to date, none have reached the stage of clinical application.[10, 11]

MicroRNAs are small molecules approximately 22 nucleotides in length and have been well conserved evolutionally. Although they are noncoding genes, they are involved in the regulation of a wide range of biologic processes through translational modulation.[25, 26] MiRNA profiles have been used to accurately identify the tissues of origin of poorly differentiated cancer tissues.[27] In thyroid pathology, they have been used to classify thyroid tumor types and to differentiate malignant tumors from benign counterparts.[28]

In this study, we initially identified a panel of miRNAs using a microarray that distinguished Rc-PTC from NR-PTC. By using qRT-PCR, we confirmed that miR-222 and miR-146b levels were significantly over-expressed in Rc-PTC compared with NR-PTC. Although the rate of BRAF mutation did not differ significantly between our Rc-PTC and NR-PTC groups, we did observe that tissue levels of miR-146b, miR-221, and miR-222 all were strongly associated with the presence of this oncogene. This observation suggests that high levels of miR-146b, miR-221, and miR-222 are characteristic of an oncogenic signature that may be associated with more aggressive disease. Moreover, our data suggest that these miRNAs indeed are more strongly associated with PTC recurrence than BRAF, likely because miRNA levels can be quantified as a continuous variable in contrast to the dichotomous nature of BRAF expression.

This study is the first to specifically investigate the miRNA expression profile in recurrent PTC. In a recent study that included tumors associated with both recurrence and distant metastasis, Yip et al reported that miR-146b and miR-222 were over-expressed in aggressive PTC.[29] Hopefully, refining the study group to be more homogeneous will allow investigators to tease out the specific roles played by miRNAs in different tumor behaviors that, collectively, are considered aggressive. For instance, the different roles that miRNAs play in RAI resistance and LVI may result in either locoregional recurrence or distant metastasis. Although, to our knowledge, there is no published study demonstrating a correlation between miRNA dysregulation and RAI resistance in PTC, it has been reported that miR-221 and miR-222 regulate radiosensitivity in nonthyroid cell lines through the PTEN/AKT pathway.[30]

The significant differences in miR-146b, miR-221, and miR-222 expression observed between tumors with and without the BRAF mutation is in partial agreement with the results reported by Chou et al,[31] who also observed that miR-146b was highly expressed in BRAF-mutated PTC tumors. Their follow-up in vitro study demonstrated that miR-146b over-expression increased cell migration and invasion and also increased resistance to chemotherapy-induced apoptosis.[32] Our results are in line with the conclusion of Chou et al that miR-146b over-expression contributes to the increased aggressiveness of PTC and, thus, can be used as a prognostic biomarker. Because we observed no differential expression of miR-146b in tumors with high-risk pathologic features (ETE, LVI, and R1 resection margin), a potential mechanism to be considered is resistance to RAI ablation. To our knowledge, no study has yet investigated the role of miR-146b in RAI resistance.

The ability to accurately identify patients who have a high risk of recurrence using tumor miRNAs, such as miR-222 and miR-146b, would allow clinicians to recommend more aggressive adjuvant therapy and/or more rigorous surveillance for this subgroup. At the same time, unwarranted intensive treatment and surveillance can be avoided for those with a low risk of recurrence. Several groups have published reports on the ability to measure miRNA levels in fine-needle aspiration samples.[33-35] If differential levels of recurrence-associated miRNAs can be demonstrated in fine-needle aspiration samples, then even surgical decisions, such as the extent of lymph node dissection, may be guided by miRNA levels.

In recent years, there has been intense interest in the use of miRNAs as circulating biomarkers for the presence of malignant and nonmalignant diseases. More than just a diagnostic tool, some miRNAs have been reported as prognostic markers, staging tools, extracellular communicators, and markers of disease progression.[36-38]

The current results have confirmed that miRNAs associated with PTC can be measured in plasma. More specifically, the 2 miRNAs (miR-222 and miR-146b) that were over-expressed in tumors from patients with recurrent PTC, along with miR-221, were over-expressed in plasma from patients with PTC compared with plasma from healthy individuals. Furthermore, levels of these miRNAs were decreased significantly in plasma from patients after total thyroidectomy and central lymph node dissection. This observation strongly suggests that plasma levels of these miRNAs correlate closely to the presence of PTC. These observations, however, were tempered by the finding that these circulating miRNAs also were elevated in plasma from patients with MNG and were reduced postoperatively. Therefore, these miRNAs in the circulation would not be useful in the diagnosis of de novo PTC, because they do not seem to be able to distinguish between malignant and benign follicular growth. However, this is entirely in keeping with the utility of the only existing biomarker—serum Tg—for follow-up, but not initial diagnosis, of PTC.

Yu et al reported that plasma expression levels of miR-222, miR-151-5p, and let-7e were significantly higher in patients with PTC compared with either patients who had benign thyroid nodules or healthy volunteers, and the levels were decreased in a subset of 9 post-thyroidectomy PTC patients.[39] However, it was not clear whether the patients with benign nodules in that study had solitary nodules, or MNG, or both.

Therefore, we speculate that circulating miR-221, miR-222, and miR-146b levels are a potential tool for long-term surveillance and may be especially useful in patients for whom measurement of serum Tg is not suitable: namely, those with anti-Tg antibodies or whose tumors have lost Tg expression. To our knowledge, this is the largest study comparing preoperative and postoperative plasma miRNA levels in patients with PTC, and it is the first to compare them with both preoperative and postoperative plasma miRNA expression in patients with MNG. However, larger and longitudinal studies will be needed to confirm this potential.

Finally, although miR-222 and miR-146b have been identified as molecular markers of recurrence in both tumor and blood, their functions within those environments almost certainly are different. Further in vitro studies will be required to clarify the mechanism of action of these miRNAs within PTC tissue in the context of recurrence as well as their mechanism of secretion into, and functions within, the circulation.

CONCLUSIONS

In this study, we identified tumor miR-222 and miR-146b as strong predictors of PTC recurrence, and they may be useful in guiding adjuvant therapy and surveillance intensity. Circulating levels of the same miRNAs also were correlated with the presence of PTC and MNG. Therefore, these are potential, noninvasive markers of PTC recurrence to be used in the context of thyroid cancer surveillance.

FUNDING SUPPORT

This project was partly funded by an institutional cancer grant to the University of Sydney. Dr Lee was the recipient of the Neville Brown Scholarship and Australian Postgraduate Award in 2012.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary