Development of a MicroRNA-Based Molecular Assay for the Detection of Papillary Thyroid Carcinoma in Aspiration Biopsy Samples

Abstract

Background:

Although thyroid nodules are common and diagnosed in over 5% of the adult population, only 5% harbor malignancy. Patients with clinically suspicious thyroid nodules need to undergo fine-needle aspiration biopsy (FNAB). The main limitation of FNAB remains indeterminate cytopathology. Only 20%–30% of the indeterminate nodules harbor malignancy, and therefore up to 80% of patients undergo unnecessary thyroidectomy. The aim of this study was to identify and validate a panel of microRNAs (miRNAs) that could serve as a platform for an FNAB-based diagnostic for thyroid neoplasms.

Methods:

The study population included 27 consecutive patients undergoing total thyroidectomy for FNAB-based papillary thyroid cancer (n = 20) and benign disorders (n = 7). Aspiration biopsy was performed from the index lesion and from the opposite lobe normal tissue in all study patients at the time of operation. RNA was extracted from all aspiration biopsy samples. Quantitative polymerase chain reaction on a panel of previously selected miRNAs was performed. Polymerase chain reaction results were compared with final histopathology. miRNA from tumor tissues was amplified using the highest value of each miRNA expression in normal tissue as a threshold for malignancy detection.

Results:

Diagnostic characteristics were most favorable for mir-221 in differentiating benign from malignant thyroid pathology. mir-221 was overexpressed in 19 patients (p < 0.0001) with a sensitive yield of 95%. Specificity, negative and positive predictive value, and accuracy of the miRNA panel were 100%, 96%, 100%, and 98%, respectively.

Conclusions:

miRNA quantification for differential diagnosis of thyroid neoplasms within aspiration biopsy samples is feasible and may improve the accuracy of FNAB cytology.

Introduction

T hyroid cancer constitutes only 1% of all epithelial malignancies worldwide. However, it is the most common endocrine neoplasm, and it represents 95% of all endocrine malignancies. As many as 37,000 new cases of thyroid cancer are diagnosed annually in the United States with an estimated annual incidence of 5.1/100,000 (1). The biology of thyroid malignancy ranges from curable, incidental, well-differentiated micro-carcinomas to almost uniformly fatal poorly differentiated (anaplastic) carcinomas. Fortunately, over 90% of patients found to have thyroid malignancies will have well-differentiated carcinoma associated with very favorable prognosis (20-year survival >95%) (2).

Thyroid nodules are very common and are diagnosed in over 5% of the adult population. With the growing use of diagnostic imaging (e.g., neck ultrasonography) the number of thyroid nodules identified and undergoing further diagnostic evaluation (e.g., fine-needle aspiration biopsy [FNAB]) is steadily growing. Nevertheless, only 5% of all thyroid nodules harbor malignancy; therefore, preoperative differentiation of benign from malignant thyroid nodules is imperative. One such approach involves defining nodules suspicious for malignancy on the basis of sonographic criteria. Patients with suspicious thyroid nodules should undergo ultrasound-guided FNAB (3). This is a safe, straightforward, sensitive, office-based diagnostic procedure that represents an accepted standard of practice. Its main limitation, however, remains the FNAB cytopathological category of indeterminate nodule (e.g., atypia of undetermined significance, follicular neoplasm and suspicious for malignancy), as definitive diagnosis of malignancy requires a morphological finding of capsular and/or vascular invasion by the tumor that is only identifiable in a resected thyroid specimen. Although FNAB is the current gold-standard diagnostic test for thyroid nodules, it continues to be limited in the differential diagnosis of indeterminate lesions, which are found in up to 20% of aspirations. Only 20%–30% of the indeterminate lesions harbor malignancy and this means that up to 80% of patients with cytologically indeterminate thyroid nodules undergo thyroidectomy solely for the purpose of diagnosis, thereby creating an important gap in the clinical decision pathway for thyroid nodules (4,5). Obviously, unnecessary thyroidectomies harbor risks like recurrent laryngeal nerve injury, hypoparathyroidism, and the need for lifelong thyroid replacement.

Despite extensive research to date, the scientific community has struggled to translate biomarkers into useful clinical tools and thus attain the full clinical potential of biomarkers. In particular, the clinical application of diagnostic biomarkers remains to be realized for patients with indeterminate thyroid nodules. To become clinically useful, biomarkers must be validated prospectively, quantitatively assessed for impact on clinical decision making, and made available to the clinical community at large. The ability to perform high-throughput molecular screening of human tissues and body fluids led to accumulating molecular data that may translate into diagnostic tests, which improve the accuracy of FNAB. A molecular assay based on aberrantly expressed genes in thyroid cancer was recently described (6). The identification and validation of a predictive molecular biomarker panel useful in the distinction of thyroid tumor subtypes associated with indeterminate FNAB cytology, which can also be applied to FNAB diagnosis in the preoperative setting, is the current clinical challenge, one that would address an important unmet need in distinguishing benign from malignant tumors in patients with indeterminate thyroid nodules. The ability to more accurately diagnose indeterminate nodules preoperatively could potentially have a dramatic positive impact on clinical outcomes and health care costs by avoiding unnecessary total thyroidectomy and the need for lifelong thyroid hormone replacement in the case of benign lesions, as well as reduce the need for completion thyroidectomy and quality of life-altering surgical morbidity in the case of malignant lesions.

MicroRNAs (miRNAs) are a newly discovered class of endogenous short (18–24 nucleotides) noncoding RNAs. As many as 530 human miRNAs have been identified and this number is constantly increasing. miRNAs have a role in a wide variety of physiologic cellular processes, including differentiation, proliferation, and apoptosis. miRNA dysregulation is a common finding in malignancy, and there is mounting evidence supporting a role of miRNAs in carcinogenesis (7,8). Specific tumors exhibit unique expression profiles of miRNAs differentiating them from one another, more importantly from normal tissues (9,10). Several recent studies have utilized miRNA microarrays to demonstrate unique molecular expression signatures differentiating benign from malignant thyroid lesions, particularly papillary thyroid cancer (PTC), which is the most common well-differentiated thyroid malignancy encountered in clinical practice (11,12). Specifically, miRNAs 21, 31, 146b, 187, 221, and 222 are differentially expressed in malignant and normal thyroid tissue (13 –17). Most of the studies to date have extracted RNA from formalin-fixed, paraffin-embedded (FFPE) tissue or from fresh tissue samples taken directly from the thyroid gland (13,16,17). Extracting high-quality RNA from paucicellular FNAB samples is challenging and remains an investigational approach. The aim of the present study was to develop a reliable and reproducible method to extract RNA from aspiration biopsy samples and to identify and validate a panel of miRNAs that could serve as a platform for an FNAB-based diagnostic for thyroid neoplasms.

Materials and Methods

Patients

The study protocol was reviewed and approved by the Independent Ethics Committee, Hadassah-Hebrew University Medical Center (IEC, Helsinki committee, protocol # HMO-0265).

Adult male and female patients aged 18–75 years referred for elective total thyroidectomy were offered participation in the study. IEC-approved informed consent was obtained from 20 consecutive study subjects with preoperative diagnosis of PTC on FNAB and from 7 patients who were operated on for benign conditions, which served as negative controls. All patients underwent total thyroidectomy under general anesthesia. Immediately after resection, the thyroid gland was taken to a side table. Aspiration biopsies were performed by 2–3 passes using a 19-gauge needle. In patients with PTC, palpation was used to identify the malignant nodule (Index nodule) and aspiration biopsy was performed of that nodule. Another aspiration biopsy was performed from the contra-lateral thyroid lobe in an area free of grossly apparent, palpable abnormality. Ultrasonography results were carefully reviewed in the operating room to ensure avoidance of nodules in the contra-lateral lobe. In patients operated-on for benign disease, aspiration biopsy was performed from normal-appearing thyroid tissue. All patients with benign disease were confirmed to be free of malignancy by standard histopathology.

Demographic and clinical data, including age, gender, ethnicity, co-morbidities, and concomitant medications as well as preoperative diagnosis, indication for operation, thyroid nodule size, and glandular consistency by physical examination, nodule size, nature, vascularity, and extent by ultrasonography, and preoperative FNAB cytology were recorded on a Case Report Form designed for the trial. Histopathological examination including maximum index lesion diameter, histological diagnosis, tumor grade, capsular, and/or vascular invasion in the case of malignancy, and presence of additional benign or malignant nodules were also recorded.

RNA extraction

Cell samples were acquired from the aspiration biopsy needle by air injection into a collecting tube containing 1 mL of QIazol lysis reagent (Qiagen Inc., Valencia, CA). Thyroid follicular cells were also collected by repeated irrigation of the syringe and the needle using pipette and plastic tips. All the accumulated cells were homogenized in the lysis reagent until complete cell wall disruption.

Total RNA was extracted using mirVana^® miRNA isolation kit (Ambion, Austin, TX) according to the manufacturer's protocol. Elution was performed in 30 mL diethyl pyrocarbonate (DEPC) nuclease-free water. The concentration was quantified using NanoDrop^® Spectrophotometer (ND-1000; Nano-Drop Technologies, Wilmington, DE) and the cell sample stored at −80°C until further use. RNA quality was assessed by gel electrophoresis.

Reverse transcription

cDNA was synthesized from total RNA using gene-specific primers according to the TaqMan^® MicroRNA reverse transcription kit protocol (Applied Biosystems, Foster City, CA). Reverse transcriptase reactions contained 50 ng of RNA samples, 50 nM stem-loop RT primer, 1 × RT buffer, 1 mM each of dNTPs, 3.33 U/μL MultiScribe reverse transcriptase, and 0.25 U/μL RNase Inhibitor. The reactions were incubated in an Applied Biosystems 9600 Thermo-Cycler (Applied Biosystems) in a 96-well plate for 30 minutes at 16°C, 30 minutes at 42°C, 5 minutes at 85°C, and then maintained at 4°C. cDNA was synthesized using TaqMan MicroRNA assay reverse transcription loop-primers specific for each mature miRNA of interest only, obviating the need to apply DNAse treatment on RNA samples. cDNA was stored at −20°C until used.

Real-time polymerase chain reaction quantification

Real-time polymerase chain reaction (PCR) was performed using an Applied Biosystems 7500 Sequence Detection system (Applied Biosystems). The 20 μL PCR included 1.34 μL (4.33 ng) RT product, 1 × TaqMan Universal PCR master mix (no AmpErase UNG), and 1 μL of primers and probe mix of the TaqMan MicroRNA Assay protocol (Applied Biosystems). The reactions were incubated in a 96-well optical plate at 50°C for 2 minutes and 95° for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60° for 1 minutes.

Each sample was checked in duplicate with water being used as the negative control. miRNA expression levels were normalized to endogenous RNU43. Comparison between different cell samples was performed by relative quantification (RQ) of miRNA expression using sample 4NN as a calibrator. The results were analyzed by 7500 SDS version 1.2 software (Applied Biosystems 7500 Sequence Detection software). The data are presented as the RQ of miRNA expression in samples relative to normal thyroid sample (2NN) after normalization to endogenous control (RNU43). ([Ct_sample− Ct_endo]/[Ct_calibrator − Ct_endo]).

Statistical analysis

Summary statistics were calculated using established methods. Assuming normal distribution of measurements, continuous variables (miRNA expression in tumor samples and normal samples) were compared using Student's t-test. Statistical calculations were performed using statistical software SPSS version 13.0 (SPSS, Inc., Chicago, IL) and a p-value < 0.05 was considered to represent statistical significance for all comparisons.

To eliminate type I errors the threshold for specificity calculations was determined as the highest RQ value of the normal samples.

Definitions

True-positive (TP): A sample where miRNA expression was higher than the previously set threshold value in the RNA extracted from the FNAB and where histopathology examination of the same lesion showed malignancy.

True-negative (TN): A sample where miRNA expression was lower than the previously set threshold value in the RNA extracted from the FNAB and where histopathology examination of the same lesion was not malignant.

False-positive (FP): A sample where miRNA expression was higher than the previously set threshold value in the RNA extracted from the FNAB and where histopathology examination of the same lesion was not malignant.

False-negative (FN): A sample where miRNA expression was lower than the previously set threshold value in the RNA extracted from the FNA and where histopathology examination of the same lesion showed malignancy.

Sensitivity = [TP/(TP + FN)]; Specificity = [TN/(TN + FP)]; positive predictive value (PPV) = [TP/(TP + FP)]; negative predictive value (NPV) = [TN/(TN + FN)]; Accuracy = [(TP + TN)/(TP + TN + FP + FN)].

Results

Study subjects were recruited over a 6-month period (July 2009 through December 2009). Twenty consecutive patients with preoperative FNAB cytology diagnosis of PTC were included in the study. Seven patients operated for benign thyroid disease served as negative controls. The indications for operation for the control group were large multi-nodular goiter (n = 2), Grave's disease (n = 3), and toxic adenoma (n = 2).

There were 20 women and 7 men in the patient group, with a mean age of 44.7 ± 15.9 years (range 19–74). Demographic and clinical data as well as preoperative thyroid function and ultrasonography findings are presented in Table 1.

Table 1.

Demographics, Thyroid Function, and Ultrasonography Findings

Patient lab number	Age	Gender	Thyroid function	Ultrasonography finding
1	26	F	N	Nodule
2	34	M	N	Nodule
3	57	M	N	Mult nodules
4	49	F	N	Mult nodules
5	27	F	N	Nodule
6	39	M	Hypo	Mult nodules
7	74	F	N	Mult nodules
8	28	F	N	Nodule
9	61	M	N	Mult nodules
10	19	F	N	Nodule
11	56	M	N	Mult nodules
12	44	M	N	Mult nodules
13	36	F	N	Mult nodules
14	60	F	N	Mult nodules
15	26	F	N	Nodule
16	64	F	N	Mult nodules
17	69	F	N	Mult nodules
18	49	F	N	Mult nodules
19	42	F	N	Mult nodules
20	47	F	N	Nodule
Patients with benign conditions
NN-1	31	M	Hyper	MNG
NN-2	24	F	Hyper	Normal
NN-3	48	F	Hyper	MNG
NN-4	37	F	Hyper	MNG
NN-5	65	F	N	Mult nodules
NN-6	62	F	N	MNG
NN-7	33	F	Hyper	Normal

MNG, multinodular goiter; N, normal; Mult, multiple.

All patients with a preoperative FNAB cytology diagnosis of PTC were confirmed to have PTC by standard histopathology. Four patients (20%) had follicular variant of papillary carcinoma. The mean tumor size was 16 ± 9.7 mm (range 5–40) and in 17 patients (85%) the tumor was found to be multi-focal. In 5 patients (25%) cervical lymph node involvement was identified (Table 2).

Table 2.

Patients’ Tumor Characteristics

Patient lab number	Pathology	Tumor size (mm)	Multifocal	Lymph node involvement
1	Papillary	10	N	Y
2	Papillary	13	Y	N
3	Papillary	7	Y	N
4	Papillary	14	Y	N
5	Papillary, follicular variant	28	Y	Y
6	Papillary	18	Y	N
7	Papillary	29	N	N
8	Papillary	5	Y	N
9	Papillary, follicular variant	14	Y	N
10	Papillary	28	Y	Y
11	Papillary	14	N	N
12	Papillary	40	Y	N
13	Papillary	30	Y	Y
14	Papillary	10	Y	N
15	Papillary	14	Y	Y
16	Papillary, follicular variant	14	Y	N
17	Papillary	6	Y	N
18	Papillary, follicular variant	9	Y	N
19	Papillary	6	Y	N
20	Papillary	12	Y	N

In total, 47 aspiration biopsy samples were obtained from the surgical specimens of 27 patients. In patients with PTC, 20 aspiration biopsy samples were obtained from the palpated nodule (tumor—T) and another 20 from the contra-lateral lobe (normal—N). In patients operated for benign conditions, 7 aspiration biopsies were performed from benign thyroid tissue (normal-normal—NN). In all aspiration biopsy samples, RNA was extracted as defined by sufficient concentration quantified using NanoDrop Spectrophotometer and quality assessed by 1% agarose gel electrophoresis. miRNA expression level was determined for each 47 samples.

The log RQ values of each sample for every miRNA are presented in Figure 1.

FIG. 1.

MicroRNA expression (log RQ) in different samples.

The mean miRNA relative quantification (RQ) of the FNAB samples obtained from patients with benign thyroid tissue (NN); mean miRNA RQ of the FNAB samples obtained from the contra-lateral lobe (N) of subjects with PTC; threshold RQ considered positive for tumor; mean miRNA RQ for tumor samples; mean miRNA RQ difference between tumor and normal samples; and diagnostic sensitivity, specificity, NPV and PPV, and accuracy were determined for mir-21, mir-31, mir-146b, mir-187, mir-221, and mir-222 as shown in Table 3. Diagnostic characteristics were most favorable for mir-221 and mir-222 in differentiating benign from malignant thyroid pathology. miRNA-221 identified the tumor in 19 (95%) patients, for 100% specificity, 95% sensitivity, 96% NPV, 100% PPV, and 98% accuracy. miRNA-222 identified the tumor in 18 (90%) patients, for 100% specificity, 90% sensitivity, 93% NPV, 100% PPV, and 96% accuracy.

Table 3.

Different MicroRNA Relative Quantification and Diagnostic Values

Characteristic	miRNA-21	miRNA-31	miRNA-146b	miRNA-187	miRNA-221	miRNA-222
RQ NN	1.07 (0.2–2.10)	0.63 (0.03–1.06)	11.39 (1.0–30.44)	0.67 (0.26–1.32)	1.08 (0.11–3.07)	2.15 (0.08–4.24)
RQ N	1.94 (0.07–5.13)	0.65 (0.01–3.23)	19.64 (0.94–57.24)	0.51 (0.1–1.5)	2.91 (0.3–9.31)	5.98 (1.22–15.5)
Threshold RQ	≥5.13	≥3.23	≥57.24	≥1.5	≥9.31	≥15.5
Mean RQ T	7.09 (0.50–27.86)	6.23 (0.01–45.95)	582.31 (5.55–1,548)	7.57 (0.14–79.29)	71.2 (4.49–239.46)	127.9 (6.45–862.24)
Mean RQ (T-N)	7.09 vs.1.72 p < 0.001	6.23 vs. 0.65 p = 0.007	582.31 vs.17.5 p < 0.0001	7.57 vs. 0.55 p = 0.04	71.2 vs.2.43 p < 0.0001	127.9 vs.4.99 p = 0.001
Sensitivity	50%	45%	80%	50%	95%	90%
Specificity	100%	100%	100%	100%	100%	100%
NPV	73%	71%	87%	73%	96%	93%
PPV	100%	100%	100%	100%	100%	100%
Accuracy	79%	77%	91%	79%	98%	96%

miRNA, microRNA; NPV, negative predictive value; PPV, positive predictive value; RQ, relative quantification.

Discussion

Although thyroid nodules are very common, only 5% harbor malignancy, and hence the importance of correctly identifying the nature of such nodules. Fine-needle aspiration of thyroid nodules is considered the gold standard for differentiating between benign and malignant thyroid nodules. However, its predictive value is limited in indeterminate nodules (e.g., follicular neoplasm, suspicious for malignancy, and atypia of undetermined significance), leading to diagnostic thyroid lobectomies or thyroid resections in many patients. The aims of this study were, first, to extract high-quality RNA from aspiration biopsy samples containing a limited number of thyroid cells, and, second, to select a panel of miRNAs that may be utilized to accurately identify malignancy in aspiration biopsy samples. We extracted high-quality RNA from all aspiration biopsy samples. Moreover, in an ongoing different study, we were also able to extract in vivo at the radiology suit high-quality RNA from leftover cells after the majority of cells in the FNAB samples were sent to cytology. Apparently, some cells are not collected during regular FNAB and are left within the needle cup. These scarce cells are sufficient for RNA extraction sparing the inconvenience of additional needle stick for the molecular diagnosis. In most previous reports, investigators used FFPE tissue or used fresh tissue samples taken from the thyroid gland to extract RNA (13,17). Chen et al. demonstrated the possibility of RNA extraction from frozen FNA samples (16). The present study demonstrates that RNA can be extracted from fresh aspiration biopsy samples or even from few cells left at the needle cup after the FNAB was performed. This will enable future studies to compare miRNA analysis to FNAB cytohistological findings performed on the very same FNAB sample.

The miRNA panel we selected from reviewing the literature and public domains was proven to be highly specific with a variable sensitivity among the various selected miRNA fragments. We established that miRNA-21, 31, 146B, 187, 221, and 222 expression was significantly increased in aspiration biopsy tumor samples compared with aspiration biopsy samples obtained from nonmalignant thyroid tissues (p = 0.0002, 0.007, 0.000001, 0.04, 0.000004, and 0.001, respectively). The most accurate miRNAs were 146b, 221, and 222 with an accuracy rate of 96%, 98%, and 92%, respectively. All three miRNAs achieved specificity values of 100%, and miRNA-221 identified tumor correctly in 19 out of 20 patients. miRNA-221 was overexpressed in the lesion of patient no. 14 when analysis was done from the whole tissue, but not in the aspiration biopsy sample taken from it, and this suggests a sampling error, an inherent limitation of aspiration biopsy. Therefore, theoretically our assay may reach 100% accuracy if sampling accuracy is verified.

Altered miRNA expression in thyroid neoplasms has been shown by others. Tetzlaff et al. extracted RNA from FFPE and showed that expression of miRNAs 21, 31, 221, and 222 was increased in papillary tumors of the thyroid (13). Similarly, Chen et al. used FNAB and FFPE samples and singled out miRNAs 146b, 221, and 222 to have the highest expression rates (16). Nikiforova et al. extracted RNA from snap-frozen tissue and FNAB samples and identified 10 miRNA fragments differentially expressed in thyroid malignancies. The highest one was miRNA-187 with a 73-fold expression in PTC compared with normal thyroid tissue. Other miRNAs that this group identified as differentially expressed in thyroid tumors were miRNAs-31, 122a, 146b, 155, 181b, 205, 221, 222, and 224 (15).

Some of these miRNAs are not specific to thyroid tumors and are highly expressed in other tumor types. miRNA-21 expression is increased in adenocarcinomas of the pancreas, breast, uterus, liver, and in chronic lymphocytic leukemia. miRNA-31 is overexpressed in colorectal cancer, and miRNA-221 was shown to be overexpressed in pancreatic, gastric, brain, and liver malignancies. Lastly, miRNA-222 is overexpressed in adenocarcinoma of the pancreas and stomach (10). Therefore, the panel we studied may be used to diagnose many other tumor types using FNAB and sparing patients with pancreatic or liver lesions the need for core needle biopsies.

One hundred percent specificity and 95% sensitivity was achieved by testing mir-221, and this may reduce the value of the other miRNAs in the panel. However, one must realize that thyroid neoplasms are variable and may represent a common phenotype of many genetic abnormalities. In larger cohorts of patients, these underlying genetic abnormalities may have an impact on miRNA expression.

At present, the exact role of miRNAs in carcinogenesis of tumors is yet to be elucidated. MiRNAs regulate gene expression at the post-transcriptional level and thereby may control cellular processes such as developmental transitions, organ morphology, cell proliferation, and apoptosis (9). It is postulated that each miRNA regulates up to 100 different mRNAs and that >10,000 mRNAs appear to be directly regulated by miRNAs (18). In carcinogenesis, miRNAs can either regulate known oncogenes or tumor suppressor genes at the post-transcriptional level or act themselves as oncogenes or tumor suppressor genes (18 –23). Thyroid neoplasms attracted attention because of the variability of malignancies and the known genetic mutations related to each malignancy. Nikiforova et al. demonstrated that miRNA-187 is overexpressed in patients with RET/PTC gene mutation as opposed to patients with BRAF gene mutation (15). In contrast, expression of miRNAs 221 and 222 is increased in patients' BRAF of RAS gene mutation. RAS-positive PTCs expressed the highest amount of miR-146. Several target proteins have been identified for miRNAs-221 and 222, both located on chromosome X. cKIT (also known as CD117) is a cytokine receptor that is lost in most patients with PTC (24). Moreover, the p27^kip1 protein level is reduced by both miRNAs. The p27^kip1 protein plays a key role in the cell cycle particularly in the cell growth arrest at the G1/S transition. Thus, overexpression of miRNAs-221 and 222 stimulates thyroid carcinoma cells to overcome the G1/S block (14,25). Several target proteins were also identified to be affected by miRNA-146b. These proteins include c-Kit mRNA, nuclear factor kB, IL-1, IL-8, TNF receptor-associated factor 6, C-X-C chemokine receptor type 4, MMP9, and epidermal growth factor receptor (26 –29).

As mentioned before, only one tumor sample (patient no. 14) was not identified as malignant by any of the six miRNAs. This sample was aspirated from a 60-year-old woman with a 10-mm tumor in the left thyroid lobe. We used RNA extraction from FFPE of the tumor to confirm our suspicion that the aspiration was not performed from the tumor itself and identified upregulation of miRNAs 146b, 221, and 222 in this second sample. This demonstrated the inherent disadvantage of FNAB—the sampling error. This issue is more relevant in patients with small tumors that are difficult to palpate. Although this sampling error did not occur in other patients with tumors <10 mm in size, it is reasonable to assume that in the above-mentioned case the FNAB was not performed from the tumor itself.

This small feasibility study has several limitations. First, this was a relatively small study this focused only on PTC. Second, aspiration biopsy was performed from resected thyroid glands on a side table. Although this simulates FNAB performed at the clinic, it still enables better performance. Third, the RQ level defined as the limit of normal tissue was chosen as the highest RQ value achieved by normal samples. This was set to avoid type I errors in which normal samples would be considered as malignant. This, by definition, resulted in 100% specificity by all miRNAs, and relatively low sensitivity in 3 miRNAs [21, 31, and 187]. To construct a receiver–operator characteristics curve that would define the limits of our diagnostic panel, a larger cohort of patients is required. After establishing a receiver–operator characteristics curve, we will be able to adjust our diagnostic panel to clinical requirements. All the patients in the study group had a cytology-based diagnosis that was proven 100% accurate by final histological examination; this implies that the added value of our molecular assay is yet to be determined. To show its added value on diagnosis, we need to repeat this feasibility study in lesions with undetermined cytology results, mainly follicular lesions. On the basis of this small cohort of patients, it seems that in this study, miRNAs 21, 31, and 187 do not have an additive value for the accuracy of the panel. However, in larger cohorts of patients and in different clinical situations, these three miRNAs may prove to be of benefit.

Although the role of miRNAs in PTC's carcinogenesis is not yet entirely clear, this study, in concurrence with others (13 –17), proves that there is a clear diagnostic role for miRNAs based panels. To date, FNAB accuracy is based solely on cytohistological characterization of the aspirated cells and is highly dependent on the expertise of the examiner. Follicular lesions are a subgroup of possible FNAB results that faces both the patient and the treating physician with a dilemma. On one hand, malignancy cannot be definitely diagnosed by FNA; on the other hand, the actual final malignant pathology rate ranges between 20% and 37%. An accurate diagnostic miRNA panel can bring a dramatic change in the decision process of such cases.

Future research may identify miRNAs that will predict malignancy grade, lymph node involvement, and perhaps even prognosis.

Footnotes

Acknowledgment

The study was supported by the Israel Cancer Association (no. 20100122).

Disclosure Statement

The authors declare that no competing financial interests exist.

References

Edwards

, Brown

, Wingo

, Howe

, Ward

, Ries

, Schrag

, Jamison

, Jemal

, Wu

, Friedman

, Harlan

, Warren

, Anderson

, Pickle

. 2005. Annual report to the nation on the status of cancer, 1975–2002, featuring population-based trends in cancer treatment. J Natl Cancer Inst, 97:1407–1427.

Cognetti

, Pribitkin

, Keane

. 2008. Management of the neck in differentiated thyroid cancer. Surg Oncol Clin North Am, 17:157–173.

Dean

, Gharib

. 2008. Epidemiology of thyroid nodules. Best Pract Res Clin Endocrinol Metab, 22:901–911.

Galloway

, Sardi

, Deconti

, Mitchell

Jr. , Bolton

. 1991. Changing trends in thyroid surgery: 38 years' experience. Am Surg, 57:18–20.

Mehanna

, Jain

, Morton

, Watkinson

, Shaha

. 2009. Investigating the thyroid nodule. BMJ, 13:b733.

Rosen

, He

, Umbricht

, Alexander

, Dackiw

, Zeiger

, Libutti

. 2005. A six-gene model for differentiating benign from malignant thyroid tumors on the basis of gene expression. Surgery, 138:1050–1056.

Esquela-Kerscher

, Slack

. 2006. Oncomirs: microRNAs with a role in cancer. Nat Rev, 6:259–269.

Negrini

, Ferracin

, Sabbioni

, Croce

. 2007. MicroRNAs in human cancer: from research to therapy. J Cell Sci, 120:1833–1840.

Hummel

, Hussey

, Haier

. 2010. MicroRNAs: predictors and modifiers of chemo- and radiotherapy in different tumour types. Eur J Cancer, 46:298–311.

10.

Lee

, Dutta

. 2009. MicroRNAs in cancer. Annu Rev Pathol, 4:199–227.

11.

Pallante

, Visone

, Ferracin

, Ferraro

, Berlingieri

, Troncone

, Chiappetta

, Liu

, Santoro

, Negrini

, Croce

, Fusco

. 2006. MicroRNA deregulation in human thyroid papillary carcinomas. Endocr Relat Cancer, 13:497–508.

12.

, Jazdzewski

, Li

, Liyanarachchi

, Nagy

, Volinia

, Calin

, Liu

, Franssila

, Suster

, Kloos

, Croce

, de la Chapelle

. 2005. The role of microRNA genes in papillary thyroid carcinoma. Proc Natl Acad Sci USA, 102:19075–19080.

13.

Tetzlaff

, Liu

, Xu

, Master

, Baldwin

, Tobias

, Livolsi

, Baloch

. 2007. Differential expression of miRNAs in papillary thyroid carcinoma compared to multinodular goiter using formalin fixed paraffin embedded tissues. Endocr Pathol, 18:163–173.

14.

Visone

, Russo

, Pallante

, De Martino

, Ferraro

, Leone

, Borbone

, Petrocca

, Alder

, Croce

, Fusco

. 2007. MicroRNAs (miR)-221 and miR-222, both overexpressed in human thyroid papillary carcinomas, regulate p27Kip1 protein levels and cell cycle. Endocr Relat Cancer, 14:791–798.

15.

Nikiforova

, Tseng

, Steward

, Diorio

, Nikiforov

. 2008. MicroRNA expression profiling of thyroid tumors: biological significance and diagnostic utility. J Clin Endocrinol Metab, 93:1600–1608.

16.

Chen

, Kitabayashi

, Zhou

, Fahey

3rd , Scognamiglio

. 2008. MicroRNA analysis as a potential diagnostic tool for papillary thyroid carcinoma. Mod Pathol, 21:1139–1146.

17.

Sheu

, Grabellus

, Schwertheim

, Worm

, Broecker-Preuss

, Schmid

. 2010. Differential miRNA expression profiles in variants of papillary thyroid carcinoma and encapsulated follicular thyroid tumours. Br J Cancer, 102:376–382.

18.

Cowland

, Hother

, Grønbaek

. 2007. MicroRNAs and cancer. APMIS, 115:1090–1106.

19.

Garzon

, Fabbri

, Cimmino

, Calin

, Croce

. 2006. MicroRNA expression and function in cancer. Trends Mol Med, 12:580–587.

20.

Zhang

, Pan

, Cobb

, Anderson

. 2007. MicroRNAs as oncogenes and tumor suppressors. Dev Biol, 302:1–12.

21.

Bartel

. 2004. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell, 116:281–297.

22.

Ambros

. 2003. MicroRNA pathways in flies and worms: growth, death, fat, stress, and timing. Cell, 113:673–676.

23.

Cho

. 2007. OncomiRs: the discovery and progress of microRNAs in cancers. Mol Cancer, 6:60.

24.

Felli

, Fontana

, Pelosi

, Botta

, Bonci

, Facchiano

, Liuzzi

, Lulli

, Morsilli

, Santoro

, Valtieri

, Calin

, Liu

, Sorrentino

, Croce

, Peschle

. 2005. MicroRNAs 221 and 222 inhibit normal erythropoiesis and erythroleukemic cell growth via kit receptor down-modulation. PNAS, 102:18081–18086.

25.

Galardi

, Mercatelli

, Giorda

, Massalini

, Frajese

, Ciafre’

, Farace

. 2007. miR-221 and miR-222 expression affects the proliferation potential of human prostate carcinoma cell lines by targeting p27Kip1. J Biol Chem, 282:23716–23724.

26.

Taganov

, Boldin

, Chang

, Baltimore

. 2006. NF-kappaB-dependent induction of microRNA miR-146, an inhibitor targeted to signaling proteins of innate immune responses. PNAS, 103:12481–12486.

27.

Bhaumik

, Scott

, Schokrpur

, Patil

, Campisi

, Benz

. 2008. Expression of microRNA-146 suppresses NF-kappaB activity with reduction of metastatic potential in breast cancer cells. Oncogene, 27:5643–5647.

28.

Labbaye

, Spinello

, Quaranta

, Pelosi

, Pasquini

, Petrucci

, Biffoni

, Nuzzolo

, Billi

, Foa’

, Brunetti

, Grignani

, Testa

, Peschle

. 2008. A three-step pathway comprising PLZF/miR-146a/CXCR4 controls megakaryopoiesis. Nat Cell Biol, 10:788–801.

29.

Hurst

, Edmonds

, Scott

, Benz

, Vaidya

, Welch

. 2009. Breast cancer metastasis suppressor 1 up-regulates miR-146, which suppresses breast cancer metastasis. Cancer Res, 69:1279–1283.