WO2011154008A1

WO2011154008A1 - Microrna classification of thyroid follicular neoplasia

Info

Publication number: WO2011154008A1
Application number: PCT/DK2011/050202
Authority: WO
Inventors: Finn Cilius Nielsen; Maria Rossing; Finn Noe Bennedbaek
Original assignee: Rigshospitalet; Herlev Hospital
Priority date: 2010-06-11
Filing date: 2011-06-09
Publication date: 2011-12-15
Also published as: WO2011154008A9

Abstract

The present invention relates to a method for improving the pre-operative diagnosis of thyroid nodules. mi RNA classifiers based on a specific micro RNA expression pattern are disclosed herein, which distinguishes the malignant and benign subtypes of thyroid follicular neoplasia.

Description

microRNA classification of thyroid follicular neoplasia

All patent and non-patent references cited in the application are hereby incorporated by reference in their entirety.

Field of invention

The present invention relates to a method for improving the pre-operative diagnosis of thyroid nodules, as the contemporary tests available do not sufficiently distinguish between the malignant and the benign neoplastic thyroid nodules. Classifiers based on a specific microRNA expression pattern are disclosed herein, which distinguishes the malignant and benign subtypes of thyroid follicular neoplasia. This can prove as a valuable pre-operative diagnostic tool; thus reducing the number of diagnostic operations and expediting surgery for individuals with a malignant nodule. Background of invention

The prevalence of palpable thyroid nodules is about 4-7% of the population in

Denmark. Using ultrasound to detect nodules of above 1 cm in circumference reveals a 3% prevalence of 18-22 year old people, and as high as 38% of 60-65 year old people. Of females above 40 years of age, one third will present with one or more nodules of 1 cm or above.

Diagnosis of thyroid nodules to date may be performed using one or - more often - a combination of the below:

Blood sample

- Scintillation counting using a tracer to measure ionizing radiation

Ultrasound

Ultrasound guided biopsy

- Cytology

Assessment of risk factors

- Surgical removal of all or part of the thyroid gland (thyroidectomy)

The high prevalence of thyroid nodules in Denmark and around the world leads to a high diagnostic activity, although there is no general consensus in the area on the most sensitive and specific diagnostic tool. In Denmark, more than 1500 thyroidectomies are performed annually, most due to nodular goiter and suspicion of neoplasia. Totally, 120-140 incidents of thyroid cancer are diagnosed annually in Denmark. This means that the majority of thyroidectomies are performed in excess, with both economical and personal costs. The many superfluous thryoidectomies are performed mainly due to non-conclusive biopsies or a finding of follicular neoplasia. Follicular neoplasia may prove to be either malignant (follicular thyroid carcinoma, FTC) or benign (follicular thyroid adenoma, FT A). Only the malignant subtype requires surgery, whereby an improved diagnostic answer from biopsies can help reduce the number of excess thyroidectomies.

MicroRNAs (miRNA) are small, non-coding single-stranded RNA gene products that regulate mRNA translation. miRNA profiles may offer the potential of improving the preoperative differentiation between benign and malignant tumours. WO 2008/1 17278, WO 2007/148235 and US 2008/171667 are directed to miRNA profiling for the detection of cancer, including thyroid cancer. However, none of these distinguishes between the subtypes of one specific type of thyroid cancer; such as follicular neoplasia. US 2008/044824 relates to the gene expression profile (mRNA) associated with thyroid cancer, to characterise the types of thyroid cancer (papillary, follicular, medullary and anaplastic). Using software, the miRNAs associated with the expression of target genes are found for follicular carcinoma; miR-101 , miR-30A-3p, miR-200A and miR- 199A. Thus, no direct miRNA profile or classifier is generated, and no distinction between FTC and FTA is directly addressed.

Nikiforova et al. (J. Clin. Endocrinol. Metab. May 2008, 93(5): 1600-1608) shows that a subset of 7 miRNAs are over-expressed in all tumours of follicular-cell derived carcinomas, and another subset of 7 miRNA can distinguish all types of thyroid tumours from hyperplastic nodules by their over-expression. This subset does not include FTA, and may not directly and specifically distinguish between FTC and FTA.

In WO 2008/002672, the inventors address the issue of developing a method for distinguishing between follicular thyroid carcinoma and follicular thyroid adenoma. They find that an up-regulated expression (1 .4 to 1 .8-fold compared to control) of a subset of miRNAs is correlated with the diagnosis of FTC, whereas a reduced expression of said miRNA subset is correlated with FTA (miR-192, miR-197, miR-328 and/or miR-346). However, the accuracy for successfully distinguishing between FTC and FTA is only 74%, and the inventors state that miRNAs are less useful for diagnosis due to the low sample material extracted from a fine-needle aspirate.

The present invention discloses a sensitive and specific means of distinction between follicular thyroid neoplasia subtypes, comprising follicular thyroid adenomas (benign) and follicular thyroid carcinomas (malignant). The inventors have found that a subset of specific miRNAs are differentially expressed in and associated with these subtypes of follicular thyroid neoplasia, efficiently separating the benign and the malignant subtypes of follicular thyroid neoplasia by employing miRNA classifiers capable of predicting which of the above categories or classes a certain sample obtained from an individual belongs to.

The present invention is thus directed to the development of two-way miRNA classifiers that distinguishes benign FTA from malignant FTC.

The terms distinction, differentiation, classification or characterisation of a sample is used herein as being capable of predicting with a relatively high sensitivity and specificity if a given sample of unknown diagnosis belongs to the class of benign FTA or malignant FTC. The output is given as a probability of belonging to either class of between 0-1 . The use of the herein disclosed miRNA classifiers may alleviate the need for the high number of diagnostic thyroidectomies performed on suspicion of all follicular neoplasias including the benign adenomas, and is as such useful as a stand-alone or an 'add-on' method to the existing diagnostic methods currently used for characterising thyroid nodules. Further, an early diagnosis of a malignant condition may expedite treatment of patients presenting with a malignant nodule, i.e. placing this group of patients first in line for surgery. Summary of invention

Thyroid nodules are frequent in the adult population. Efforts to improve the preoperative diagnosis of thyroid nodules are needed, in order to more efficiently distinguish benign from malignant nodules without the need for diagnostic surgery.

The expression of RNA species, such as microRNAs (miRNA) is often deregulated in malignant cells and shows a highly tissue-specific pattern. A classifier based on a RNA expression profile or signature, such as a miRNA expression profile or signature, may be an ideal diagnostic tool to differentiate the malignant from the benign thyroid tumours.

The aim of the present invention is to develop a two-way miRNA classifier, which can accurately differentiate between two subtypes of follicular thyroid neoplasms; the class of thyroid follicular adenomas (FTA) from the class of thyroid follicular carcinomas (FTC).

There is provided herein a system for the identification of a malignancy-specific signature of miRNAs that are differentially expressed relative to adenoma cells. It is also an aim to present two-way miRNA classifiers, which can accurately differentiate between thyroid follicular adenomas (FTA) and normal thyroid tissue (NT), or between thyroid follicular carcinomas (FTC) and normal thyroid tissue (NT).

The two-way miRNA classifiers disclosed herein in one embodiment distinguishes benign FTA from malignant FTC and comprises or consists of one or more of hsa-miR- 1826 and hsa-miRPIus-E1078.

The two-way miRNA classifiers disclosed herein in another embodiment distinguishes benign FTA from malignant FTC and comprises or consists of one or more of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR- 15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa- let-7i, hsa-miR-326 and hsa-miR-330-3p. In yet another embodiment, the two-way miRNA classifiers disclosed herein

distinguishes widely invasive FTC from minimally invasive FTC, and comprises or consists of one or more of hsa-miRPIus-E1001 and hsa-miR-410. In a particular embodiment, the two-way miRNA classifier distinguishes benign FTA from malignant FTC and comprises one or more of miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa- miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa- miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa- miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa- miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342- 3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR- 24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa- miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR- 631 and hsa-miR-637. The miRNA classifiers may be applied ex vivo to a sample from a thyroid nodule of a human being, in order to improve the pre-operative diagnostic prognosis. This would reduce the current large number of diagnostic thyroid operations performed and expedite the necessary operations (i.e. on malignant nodules). Accordingly, provided herein are methods for diagnosing whether a subject has, or is at risk of developing, follicular thyroid carcinoma and/or adenoma, comprising the steps of extracting RNA from a sample collected from the thyroid of an individual and analysing the miRNA expression profile or signature of said sample; in one embodiment comprising one or more miRNAs selected from the group of hsa-miR-1826 or hsa- miRPIus-E1078 and in another embodiment comprising one or more miRNAs selected from the group of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa- miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p. „

Also provided are methods for determining the need for thyroidectomy in an individual presenting with a thyroid nodule by employing the miRNA classifiers disclosed herein.

The present invention is also directed to a device for measuring the expression level of at least one miRNA according to the present invention, wherein said device is used for classifying a sample obtained from a thyroid nodule of an individual.

Also provided is a system for performing a diagnosis on an individual with a thyroid nodule, comprising means for analysing the miRNA expression profile of the thyroid nodule, in one embodiment comprising at least one miRNA according to the present invention, and means for determining if said individual has a benign or a malignant condition selected from follicular thyroid adenoma and follicular thyroid carcinoma.

The present invention is also directed to a computer program product having a computer readable medium, said computer program product providing a system for predicting the diagnosis of an individual with a thyroid nodule, said computer program product comprising means for carrying out any of the steps of any of the methods as disclosed herein.

Description of the drawings

Figure 1 : MiRNA expression in follicular carcinoma and adenoma.

A. Venn diagram showing differential and common miRNAs among follicular carcinomas (FC) and follicular adenomas (FA) in relation to normal thyroid tissue. The total number of differentially expressed miRNAs is shown in black (top), and the number of up-regulated and down-regulated miRNAs, are shown in green (middle) and red (bottom), respectively. B. The graphs show the fold change of 9 common up- regulated (green/left hand side) and 49 common down-regulated (red/right hand side) miRNAs, respectively, in relation to normal thyroid. Figure 2: Principal component analysis (example 1).

A. (a) Projection of follicular carcinomas (FC) and follicular adenomas (FA) and normal thyroid (NT) employing all miRNAs or (b) after variance and t-test filtering (p<0.01 ) leaving 179 miRNAs. NT is illustrated by 'x'.

B. (a) Projection of FA versus FC in a two group comparison (p<0.01 ). FC is illustrated by 'x'. (c) and (d) shows the relative expression of the hsa-miRPIus-E1078 and hsa- miR-1826 in the samples. The bar shows the relative expression by the color coding from -2 fold {green) to 2 fold {red).

Figure 3: Quantitative reverse transcription PCR (QRT-PCR)

QRT-PCR results illustrating relative fold changes (y-axis) of miR-221 , miR-96, miR- 182, miR-199a3p, miR-144^*, and miR-199b5p in follicular carcinoma (FC) vs. normal thyroid (NT) and follicular adenoma (FA) vs. NT, respectively. Also, relative fold change of hsa-miR-1826 and hsa-miRPIus-E1078 in FC vs. FA. Dark-green bar ('χ') illustrates QRT-PCR derived fold change and light-green bar ('ο') illustrates microarray derived fold change.

Figure 4: Heatmaps

Heatmap of "Cell Cycle" factors shows the relative expression of the predicted target mRNAs in FC and NT. Corresponding to the down-regulation of miRNAs, mRNAs encoding cell cycle factors were almost entirely increased. Of the 165 mRNAs, 154 were significantly up-regulated (P<0.05), 12 transcripts were unchanged, and 2 were up-regulated. The "Tumourigenesis" heatmap shows 49 significantly enriched transcripts. Twenty-four of the 49 mRNAs were significantly upregulated (P<0.05). Heatmap named "miR-199b-5p targets" depicts 20 putative targets with a weighted cumulative context ranking score > 80. Halves of the transcripts showed a significant upregulation in FC (P<0.05).

Figure 5: Principal component analysis (example 2)

A. Projection of follicular carcinomas (FC) and follicular adenomas (FA) and normal thyroid (NT) employing all miRNAs derived from the microarray analysis. B. Projection of FC and FA employing the expression values of only miR-1826 and miR-Eplus-1078. C. Projection of FC and follicular FA and NT employing all miRNAs derived from the qRT-PCR panels. D. Projection of FC and FA employing the expression values of the 14 miRNAs that was found to be the optimal signature for classification of FC (see example 2).

Definitions

Statistical classification is a procedure in which individual items are placed into groups based on quantitative information on one or more characteristics inherent in the items (referred to as traits, variables, characters, etc) and based on a training set of previously labeled items.

A classifier is a prediction model which may distinguish between or characterize samples by classifying a given sample into a predetermined class based on certain characteristics of said sample. A two-way classifier classifies a given sample into one of two predetermined classes, and a three-way classifier classifies a given sample into one of three predetermined classes.

'Collection media' as used herein denotes any solution suitable for collecting, storing or extracting of a sample for immediate or later retrieval of RNA from said sample. 'Deregulated' means that the expression of a gene or a gene product is altered from its normal baseline levels; comprising both up- and down-regulated.

Goiter: A swelling in the neck (just below the Adam's apple or larynx) due to an enlarged thyroid gland. Also denoted goitre (British), struma (Latin), or a bronchocele.

The term "Individual" refers to vertebrates, particular members of the mammalian species, preferably primates including humans. As used herein, 'subject' and

'individual' may be used interchangeably.

The term "Kit of parts" as used herein provides a device for measuring the expression level of at least one miRNA according to the present invention, and at least one additional component. The additional component may in one embodiment be means for extracting RNA, such as miRNA, from a sample; reagents for performing microarray analysis, reagents for performing QPCR analysis and/or instructions for use of the device and/or additional components.

The term "natural nucleotide" or "nucleotide" refers to any of the four

deoxyribonucleotides, dA, dG, dT, and dC (constituents of DNA), and the four ribonucleotides, A, G, U, and C (constituents of RNA). Each natural nucleotide comprises or essentially consists of a sugar moiety (ribose or deoxyribose), a phosphate moiety, and a natural/standard base moiety. Natural nucleotides bind to complementary nucleotides according to well-known rules of base pairing (Watson and Crick), where adenine (A) pairs with thymine (T) or uracil (U); and where guanine (G) pairs with cytosine (C), wherein corresponding base-pairs are part of complementary, anti-parallel nucleotide strands. The base pairing results in a specific hybridization between predetermined and complementary nucleotides. The base pairing is the basis by which enzymes are able to catalyze the synthesis of an oligonucleotide

complementary to the template oligonucleotide. In this synthesis, building blocks (normally the triphosphates of ribo or deoxyribo derivatives of A, T, U, C, or G) are directed by a template oligonucleotide to form a complementary oligonucleotide with the correct, complementary sequence. The recognition of an oligonucleotide sequence by its complementary sequence is mediated by corresponding and interacting bases forming base pairs. In nature, the specific interactions leading to base pairing are governed by the size of the bases and the pattern of hydrogen bond donors and acceptors of the bases. A large purine base (A or G) pairs with a small pyrimidine base (T, U or C). Additionally, base pair recognition between bases is influenced by hydrogen bonds formed between the bases. In the geometry of the Watson-Crick base pair, a six membered ring (a pyrimidine in natural oligonucleotides) is juxtaposed to a ring system composed of a fused, six membered ring and a five membered ring (a purine in natural oligonucleotides), with a middle hydrogen bond linking two ring atoms, and hydrogen bonds on either side joining functional groups appended to each of the rings, with donor groups paired with acceptor groups.

As used herein, "nucleic acid" or "nucleic acid molecule" refers to polynucleotides, such as deoxyribonucleic acid (DNA) or ribonucleic acid (RNA), oligonucleotides, fragments generated by the polymerase chain reaction (PCR), and fragments generated by any of ligation, scission, endonuclease action, and exonuclease action. Nucleic acid molecules can be composed of monomers that are naturally-occurring nucleotides (such as DNA and RNA), or analogs of naturally-occurring nucleotides (e.g. alpha- enantiomeric forms of naturally-occurring nucleotides), or a combination of both.

Modified nucleotides can have alterations in sugar moieties and/or in pyrimidine or purine base moieties. Sugar modifications include, for example, replacement of one or more hydroxyl groups with halogens, alkyl groups, amines, and azido groups, or sugars can be functionalized as ethers or esters. Moreover, the entire sugar moiety can be replaced with sterically and electronically similar structures, such as aza-sugars and carbocyclic sugar analogs. Examples of modifications in a base moiety include alkylated purines and pyrimidines, acylated purines or pyrimidines, or other well-known heterocyclic substitutes. Nucleic acid monomers can be linked by phosphodiester bonds or analogs of such linkages. Analogs of phosphodiester linkages include phosphorothioate, phosphorodithioate, phosphoroselenoate, phosphorodiselenoate, phosphoroanilothioate, phosphoranilidate, phosphoramidate, and the like. The term "nucleic acid molecule" also includes e.g. so-called "peptide nucleic acids," which comprise naturally-occurring or modified nucleic acid bases attached to a polyamide backbone. Nucleic acids can be either single stranded or double stranded. In an aspect of the present invention, 'nucleic acid' is meant to comprise antisense oligonucleotides (ASO), small inhibitory RNAs (siRNA), short hairpin RNA (shRNA) and microRNA (miRNA).

A "polypeptide" or "protein" is a polymer of amino acid residues preferably joined exclusively by peptide bonds, whether produced naturally or synthetically. The term "polypeptide" as used herein covers proteins, peptides and polypeptides, wherein said proteins, peptides or polypeptides may or may not have been post-translationally modified. Post-translational modification may for example be phosphorylation, methylation and glycosylation.

A 'probe' as used herein refers to a hybridization probe. A hybridization probe is a (single-stranded) fragment of DNA or RNA of variable length (usually 100-1000 bases long), which is used in DNA or RNA samples to detect the presence of nucleotide sequences (the DNA target) that are complementary to the sequence in the probe. The probe thereby hybridizes to single-stranded nucleic acid (DNA or RNA) whose base sequence allows probe-target base pairing due to complementarity between the probe and target. To detect hybridization of the probe to its target sequence, the probe is tagged (or labeled) with a molecular marker of either radioactive or fluorescent molecules. DNA sequences or RNA transcripts that have moderate to high sequence similarity to the probe are then detected by visualizing the hybridized probe.

Hybridization probes used in DNA microarrays refer to DNA covalently attached to an inert surface, such as coated glass slides or gene chips, and to which a mobile cDNA target is hybridized. A probe set is a collection of probes designed to interrogate a given sequence.

Thyroidectomy: A thyroidectomy involves the surgical removal of all or part of the thyroid gland. Hemi-thyroidectomy is removal of one lobe of the thyroid, partly or entirely.

Due to the imprecision of standard analytical methods, molecular weights and lengths of polymers are understood to be approximate values. When such a value is expressed as "about" X or "approximately" X, the stated value of X will be understood to be accurate to +/- 20%, such as +/- 10%, for example +/- 5%. Follicular thyroid carcinoma (FTC) and follicular carcinoma (FC) may be used interchangeably herein; and follicular thyroid adenoma (FT A) and follicular adenoma (FA) may be used interchangeably herein. Detailed description of the invention

The thyroid gland

The thyroid is one of the largest endocrine glands in the body. This gland is found in the neck inferior to the thyroid cartilage ('Adam's apple' in men) and at approximately the same level as the cricoid cartilage. The thyroid controls how quickly the body burns energy, makes proteins, and how sensitive the body should be to other hormones.

The thyroid participates in these processes by producing thyroid hormones, principally thyroxine (T₄) and triiodothyronine (T₃). These hormones regulate the rate of metabolism and affect the growth and rate of function of many other systems in the body. Iodine is an essential component of both T₃ and T₄. The thyroid also produces the hormone calcitonin, which plays a role in calcium homeostasis. The thyroid is in turn controlled by the hypothalamus and pituitary.

The thyroid is composed of spherical follicles that selectively absorb iodine (as iodide ions, ) from the blood for production of thyroid hormones. Twenty-five percent of all the body's iodide ions are in the thyroid gland. Inside the follicles, colloids rich in a protein called thyroglobulin serve as a reservoir of materials for thyroid hormone production and, to a lesser extent, act as a reservoir for the hormones themselves. The follicles are surrounded by a single layer of thyroid epithelial cells (or 'follicular cells'), which secrete T3 and T4. When the gland is not secreting T3/T4 (inactive), the epithelial cells range from low columnar to cuboidal cells. When active, the epithelial cells become tall columnar cells. Scattered among follicular cells and in spaces between the spherical follicles are another type of thyroid cell, parafollicular cells or C cells, which secrete calcitonin. Thyroxine (T4) is synthesised by the follicular cells from free tyrosine and on the tyrosine residues of the protein called thyroglobulin (TG). Iodine is captured with the "iodine trap" by the hydrogen peroxide generated by the enzyme thyroid peroxidase (TPO) and linked to the 3' and 5' sites of the benzene ring of the tyrosine residues on TG, and on free tyrosine. Upon stimulation by the thyroid-stimulating hormone (TSH), the follicular cells reabsorb TG and proteolytically cleave the iodinated tyrosines from TG, forming T4 and T3 (in T3, one iodine is absent compared to T4), and releasing them into the blood. Deiodinase enzymes convert T4 to T3. Thyroid hormone that is secreted from the gland is about 90% T4 and about 1 0% T3.

Cells of the brain are a major target for the thyroid hormones T3 and T4. Thyroid hormones play a particularly crucial role in brain maturation during fetal development. A transport protein (OATP1 C1 ) has been identified that seems to be important for T4 transport across the blood brain barrier. A second transport protein (MCT8) is important for T3 transport across brain cell membranes.

In the blood, T4 and T3 are partially bound to thyroxine-binding globulin, transthyretin and albumin. Only a very small fraction of the circulating hormone is free (unbound) - T4 0.03% and T3 0.3%. Only the free fraction has hormonal activity. As with the steroid hormones and retinoic acid, thyroid hormones cross the cell membrane and bind to intracellular receptors (a α₂, βι and β₂), which act alone, in pairs or together with the retinoid X-receptor as transcription factors to modulate DNA transcription. Up to 80% of the T4 is converted to T3 by peripheral organs such as the liver, kidney and spleen. T3 is about ten times more active than T4.

The production of thyroxine and triiodothyronine is regulated by thyroid-stimulating hormone (TSH), released by the anterior pituitary (that is in turn released as a result of TRH release by the hypothalamus). The thyroid and thyrotropes form a negative feedback loop: TSH production is suppressed when the T4 levels are high, and vice versa. The TSH production itself is modulated by thyrotropin-releasing hormone (TRH), which is produced by the hypothalamus and secreted at an increased rate in situations such as cold (in which an accelerated metabolism would generate more heat). TSH production is blunted by somatostatin (SRIH), rising levels of glucocorticoids and sex hormones (estrogen and testosterone), and excessively high blood iodide

concentration.

An additional hormone produced by the thyroid contributes to the regulation of blood calcium levels. Parafollicular cells produce calcitonin in response to hypercalcemia. Calcitonin stimulates movement of calcium into bone, in opposition to the effects of parathyroid hormone (PTH). However, calcitonin seems far less essential than PTH, as calcium metabolism remains clinically normal after removal of the thyroid, but not the parathyroids.

Thyroid nodule

Thyroid nodules are lumps which commonly arise within an otherwise normal thyroid gland. Often these abnormal growths of thyroid tissue are located at the edge of the thyroid gland so they can be felt as a lump in the throat. When they are large or when they occur in very thin individuals, they may even be seen as a lump in the front of the neck. Thyroid nodules are extremely common and almost 50% of people have had one, but they are usually only detected by a general practitioner during the course of a health examination, or through a different affliction. Only a small percentage of lumps in the neck are malignant (less than 1 %), and most thyroid nodules are benign.

Thyroid neoplasia

Neoplasia is the abnormal proliferation of cells, resulting in a structure known as a neoplasm. The growth of this clone of cells exceeds, and is uncoordinated with, that of the normal tissues around it. It usually causes a lump or tumour.

Thyroid neoplasia may be benign (adenoma) or malignant (carcinoma), with only the malignant requiring surgery.

Benign neoplasia

A thyroid adenoma, or solitary thyroid nodule, is a benign tumour of the thyroid gland. A thyroid adenoma is distinguished from a multinodular goiter of the thyroid in that an adenoma is typically solitary, and is a neoplasm resulting from a genetic mutation (or other genetic abnormality) in a single precursor cell. In contrast, a multinodular goiter is usually thought to result from a hyperplastic response of the entire thyroid gland to a stimulus, such as iodine deficiency. A thyroid adenoma may be clinically silent, or it may be a functional tumour, producing excessive thyroid hormone. In this case, it may result in symptomatic hyperthyroidism, and may be referred to as a toxic thyroid adenoma. Careful pathological examination may be necessary to distinguish a thyroid adenoma from a minimally invasive follicular thyroid carcinoma. Malignant neoplasia

Thyroid cancer is more frequent in females at a ratio of three to one. Thyroid cancer can occur in any age group, although it is most common after age 30 and its aggressiveness increases significantly in older patients. The majority of patients present with a nodule on their thyroid which typically does not cause symptoms. When a thyroid cancer begins to grow within a thyroid gland, it almost always does so within a discrete nodule within the thyroid. Scintigraphically cold nodules are more likely to be cancerous, however only a small part of the cold nodules are diagnosed as cancer. Thyroid cancer or carcinoma refers to any of four kinds of malignant tumours of the thyroid gland: papillary, follicular, medullary or anaplastic. Papillary and follicular tumours are the most common. They grow slowly and may recur, but are generally not fatal in patients under 45 years of age. Medullary tumours have a good prognosis if restricted to the thyroid gland and a poorer prognosis if metastasis occurs. Anaplastic tumours are fast-growing and respond poorly to therapy.

In Denmark, 120-140 incidents of thyroid cancer are diagnosed each year. Of these, 63% are papillary carcinomas, 18% are follicular neoplasia, 7% are medullary neoplasia, 8% are anaplastic (undifferentiated); leaving 4% designated as 'others' (including metastasis, lymphoma, squamous cell carcinoma, sarcoma). The follicular and papillary types together can be classified as "differentiated thyroid cancer". These types have a more favorable prognosis than the medullary and undifferentiated types.

Papillary thyroid carcinoma

Papillary thyroid cancer is generally the most common type of thyroid cancer. It occurs more frequently in women and presents in the 30-40 year age group. It is also the predominant cancer type in children with thyroid cancer, and in patients with thyroid cancer who have had previous radiation to the head and neck. Papillary

microcarcinoma is a subset of papillary thyroid cancer defined as measuring less than or equal to 1 cm. Papillary thyroid carcinoma commonly metastasizes to cervical lymph nodes.

Thyroglobulin can be used as a tumour marker for well-differentiated papillary thyroid cancer. HBME-1 (human mesothelial cell 1 ) staining may be useful for differentiating papillary carcinomas from follicular carcinomas; in papillary lesions it tends to be positive.

Surgical treatment includes either hemithyroidectomy (or unilateral lobectomy) or isthmectomy (removing the band of tissue (or isthmus) connecting the two lobes of the thyroid), which is sometimes indicated with minimal disease (diameter up to 1 .0 centimeters). For gross disease (diameter over 1 centimeter), total thyroidectomy, and central compartment lymph node removal is the therapy of choice. As papillary carcinoma is a multifocal disease, hemithyroidectomy may leave disease in the other lobe and total thyroidectomy reduces the risk of recurrence.

Follicular thyroid carcinoma

Follicular thyroid cancer is a form of thyroid cancer which occurs more commonly in women of over 50 years. Follicular carcinoma is considered more malignant

(aggressive) than papillary carcinoma. It occurs in a slightly older age group than papillary cancer and is also less common in children. In contrast to papillary cancer, it occurs only rarely after radiation therapy. Mortality is related to the degree of vascular invasion. Age is a very important factor in terms of prognosis. Patients over 40 have a more aggressive disease and typically the tumour does not concentrate iodine as well as in younger patients. Vascular invasion is characteristic for follicular carcinoma and therefore distant metastasis is more common. Lung, bone, brain, liver, bladder, and skin are potential sites of distant spread. Lymph node involvement is far less common than in papillary carcinoma. Unlike papillary thyroid cancer, follicular thyroid cancer is today difficult to diagnose without performing surgery because there are no characteristic changes in the way the thyroid cells look; i.e. it is not possible to accurately distinguish between follicular thyroid adenoma and carcinoma on cytological grounds. Rather, the only way to tell if a follicular cell nodule (or neoplasm) is cancer is to look at the entire capsule around the nodule and see if there is any sign of invasion. A fine needle aspiration (FNA) biopsy cannot at present distinguish cytologically between follicular adenoma, follicular carcinoma and a completely benign condition called nontoxic nodular goiter. Even a coarse needle biopsy, which is typically more accurate than a FNA, cannot always provide an answer since it is only able to differentiate between a follicular neoplasm (which includes both adenoma and carcinoma) versus nontoxic nodular goiter about 40% of the time. These biopsies can only look at individual cells and not the entire capsule. If fine needle aspiration cytology suggests follicular neoplasm, thyroid lobectomy is today performed to establish the histopathological diagnosis. This difficulty in diagnosis is one of the most frustrating areas for physicians who study thyroid disease today, because it means that surgery is most often the only way of definitively diagnosing a thyroid nodule.

It is an object of the present invention to disclose a method for more efficiently distinguishing between the malignant and benign subtypes of follicular neoplasm of the thyroid; thus improving the pre-operative diagnosis of this condition and reducing the number of diagnostic surgeries required. This is achieved by providing specific miRNA classifiers that distinguish between the benign follicular adenomas and the malignant follicular carcinomas. Treatment is usually surgical, followed by radioiodine. Unilateral hemithyroidectomy

(removal of one entire lobe of the thyroid) is uncommon due to the aggressive nature of this form of thyroid cancer, but may be indicated for achieving the diagnosis. Total thyroidectomy is almost automatic with this diagnosis. This is invariably followed by radioiodine treatment following two weeks of a low iodine diet. Occasionally treatment must be repeated if annual scans indicate remaining cancerous tissue. Minimally invasive thyroidectomy has been used in recent years in cases where the nodules are small.

Fetal adenoma (microfollicular adenomas or follicular fetal adenoma) is a subgroup of follicular neoplasms with a potential to transform into malignancy. The term 'fetal adenoma' was coined to designate certain nodular tumours of the thyroid gland, which was originally believed to arise from fetal cell rests. With an advance in knowledge, however, the concept of a fetal origin for these nodules has largely been discarded. Today it has come to designate a distinctive type of nodule, on the general features of which most observers are agreed. They begin as masses of thyroid tissue which has never reached an adult stage.

Fetal adenoma represents a distinct histopathological entity. Their malignant potential is poorly characterized, but since they exhibit a high degree (58%) of aneuploidy, they may progress to malignancy. In agreement with this assumption it is known that about 5 percent of fetal adenomas prove to be follicular cancers with careful,

histopathological study.

Hurthle cell thyroid cancer is often considered a variant of follicular cell carcinoma. Hurthle cell forms are more likely than follicular carcinomas to be bilateral and multifocal and to metastasize to lymph nodes. Like follicular carcinoma, unilateral hemithyroidectomy is performed for non-invasive disease, and total thyroidectomy for invasive disease. Follicular thyroid carcinoma - minimally or widely invasive

The distinction between follicular adenoma and carcinoma is partly based on identification of invasion or metastasis, with the recognition that minimally and widely invasive subgroups of carcinoma should be separately identified. Follicular carcinomas have been divided according to their degree of invasiveness into two major categories. Minimally invasive follicular carcinoma have limited capsular penetration and/or vascular invasion. Widely invasive follicular carcinoma have widespread infiltration of adjacent thyroid tissue and/or blood vessels. This is detailed in "WHO classification - Tumours of Endocrine Organs" (2004).

Medullary thyroid carcinoma

Medullary thyroid cancer (MTC) is a form of thyroid carcinoma which originates from the parafollicular cells (C cells), which produce the hormone calcitonin. Approximately 25% the cancer develops in families. When MTC occurs by itself it is termed familial MTC; when it coexists with tumours of the parathyroid gland and medullary component of the adrenal glands (pheochromocytoma) it is called multiple endocrine neoplasia type 2 (MEN2).

While the increased serum concentration of calcitonin is not harmful, it is useful as a marker which can be tested in blood. A second marker, carcinoembryonic antigen

(CEA), also produced by medullary thyroid carcinoma, is released into the blood and it is useful as a serum or blood tumour marker. In general measurement of serum CEA is less sensitive than serum calcitonin for detecting the presence of a tumour, but has less minute to minute variability and is therefore useful as an indicator of tumour mass. Mutations in the RET proto-oncogene ("rearranged during transfection"), located on chromosome 10, lead to the expression of a mutated receptor tyrosine kinase protein, termed RET. RET is involved in the regulation of cell growth and development and its mutation is responsible for nearly all cases of hereditary or familial medullary thyroid carcinoma.

Surgery can be effective when the condition is detected early, but a risk for recurrence remains. Unlike differentiated thyroid carcinoma, there is no role for radioiodine treatment in medullary-type disease. External beam radiotherapy should be considered for patients at high risk of regional recurrence, even after optimum surgical treatment. Also, clinical trials of several new tyrosine kinase inhibitors are now being studied.

The prognosis of MTC is poorer than that of follicular and papillary thyroid cancer when it has metastasized (spread) beyond the thyroid gland.

Anaplastic thyroid carcinoma

Anaplastic thyroid cancer (ATC) or undifferentiated thyroid cancer is a form of thyroid cancer which has a very poor prognosis due to its aggressive behaviour and resistance to cancer treatments. It rapidly invades surrounding tissues (such as the trachea). The presence of regional lymphadenopathy in older patients in whom a characteristic vesicular appearance of the nuclei is revealed would support a diagnosis of anaplastic carcinoma.

The median survival from diagnosis ranges from 3 to 7 months, with worse prognosis associated with large tumours, distant metastases, acute obstructive symptoms, and leucocytosis. In the 18-24% of patients whose tumour seems both confined to the neck and grossly resectable, complete surgical resection followed by adjuvant radiotherapy and chemotherapy could yield a 75-80% survival at 2 years. Unlike its differentiated counterparts, anaplastic thyroid cancer is highly unlikely to be curable either by surgery or by any other treatment modality, and is in fact usually unresectable due to its high propensity for invading surrounding tissues. Palliative treatment consists of radiation therapy usually combined with chemotherapy. New drugs are in clinical trials that may improve chemotherapy treatment. Diagnosing thyroid neoplasia at present

Most often the first symptom of thyroid cancer is a nodule in the thyroid region of the neck. However, many adults have small undetected nodules in their thyroids. Typically fewer than 5% of these nodules are found to be malignant. Sometimes the first sign is an enlarged lymph node. Later possible symptoms are pain in the anterior region of the neck and changes in voice. Thyroid cancer is usually found in a euthyroid patient (having normal thyroid function), but symptoms of hyperthyroidism may be associated with a large or metastatic well-differentiated tumour.

Diagnosing of thyroid nodules to date may be performed using one or - more often - a combination of the below diagnostic methods:

Scintillation counting using a tracer to measure ionizing radiation (using technetium Tc or ionizing Iodine I¹³¹ or I¹²³). 85% of nodules will be

scintigraphically 'cold'; i.e. not accumulating the tracer. Of these, 5% will be malignant. Hot nodules are signs of non-cancerous nodules.

Blood sample. Measurement of thyroid stimulating hormone (TSH) and antithyroid antibodies will help decide if there is a functional (non-cancerous) thyroid disease present. The possibility of a nodule which secretes thyroid hormone (which is less likely to be cancer) or hypothyroidism is investigated by measuring thyroid stimulating hormone (TSH), and the thyroid hormones thyroxine (T4) and triiodothyronine (T3). Tests for serum thyroid auto-antibodies are sometimes done as these may indicate autoimmune thyroid disease (which can mimic nodular disease).

Ultrasound imaging. Features that may be distinguished using ultrasound relies on an assessment from the operator, and includes relating a feature with a probability (rare to high) of malignancy. Features include amongst others lymphadenopathies, invasion on adjacent structure, poorly defined margins, cystic nodule, blood flow level and microcalcifications.

Cytology/histology of resected thyroid nodule (e.g. thyroidectomy or biopsy).

Assessment of risk factors, comprising the occurrence of thyroid cancer in the family, age below 20 or above 70 years, male gender, previous irradiation to the neck and/or head area, large nodule (>4 cm), fast growing nodule, firm or hard texture, fixation to surrounding structures, compression symptoms (hoarse voice, dysphagia, dyspnea) and regional lymphadenopathy. While the above diagnostic tools may render probable that a nodule is indeed cancerous, it is not straight forward to distinguish between the four kinds of malignant tumours of the thyroid gland (papillary, follicular, medullary or anaplastic), and further to diagnose malignant follicular thyroid cancer without performing surgery, because it is at present not possible to accurately distinguish between follicular thyroid adenoma and follicular thyroid carcinoma on cytological grounds. Indeed, diagnostic surgery is the only certain way to establish a correct diagnosis on a thyroid nodule. The method disclosed herein provides a tool for improving the pre-operative diagnosis of thyroid nodules, in particular thyroid follicular neoplasm, thus reducing the number of diagnostic surgeries required. Specific miRNA classifiers are provided that may distinguish between the benign follicular adenomas and the malignant follicular carcinomas.

The miRNA classifiers as disclosed herein may in one embodiment be used in the clinic alone (stand alone diagnostic); i.e. without employing further diagnostic methods.

In another embodiment, the miRNA classifiers as disclosed herein may be used in the clinic as an add-on or supplementary diagnostic tool or method, which improves the pre-operative diagnosis of thyroid nodules by combining the output of said miRNA classifier with the output of one or more of the above-mentioned conventional diagnostic techniques to improve the accuracy of said pre-operative diagnosis of thyroid neoplasms.

Said at least one additional diagnostic method may be selected from the group consisting of CT (X-ray computed tomography), MRI (magnetic resonance imaging), Scintillation counting, Blood sample analysis, Ultrasound imaging, Cytology, Histology and Assessment of risk factors.

In one embodiment, said at least one additional diagnostic method is use of an mRNA classifier capable of distinguishing between follicular thyroid adenoma and follicular thyroid carcinoma. Such as mRNA classifier may in one preferred embodiment be as disclosed in international patent application (PCT/DK2010/050358) entitled 'mRNA classification of thyroid follicular neoplasia'. Nucleic Acids

A nucleic acid is a biopolymeric macromolecule composed of chains of monomeric nucleotides. In biochemistry these molecules carry genetic information or form structures within cells. The most common nucleic acids are deoxyribonucleic acid (DNA) and ribonucleic acid (RNA). Each nucleotide consists of three components: a nitrogenous heterocyclic base (the nucleobase component), which is either a purine or a pyrimidine; a pentose sugar (backbone residues); and a phosphate group

(internucleoside linkers). A nucleoside consists of a nucleobase (often simply referred to as a base) and a sugar residue in the absence of a phosphate linker. Nucleic acid types differ in the structure of the sugar in their nucleotides - DNA contains 2- deoxyriboses while RNA contains ribose (where the only difference is the presence of a hydroxyl group). Also, the nitrogenous bases found in the two nucleic acid types are different: adenine, cytosine, and guanine are found in both RNA and DNA, while thymine only occurs in DNA and uracil only occurs in RNA. Other rare nucleic acid bases can occur, for example inosine in strands of mature transfer RNA. Nucleobases are complementary, and when forming base pairs, must always join accordingly:

cytosine-guanine, adenine-thymine (adenine-uracil when RNA). The strength of the interaction between cytosine and guanine is stronger than between adenine and thymine because the former pair has three hydrogen bonds joining them while the latter pair has only two. Thus, the higher the GC content of double-stranded DNA, the more stable the molecule and the higher the melting temperature.

Nucleic acids are usually either single-stranded or double-stranded, though structures with three or more strands can form. A double-stranded nucleic acid consists of two single-stranded nucleic acids held together by hydrogen bonds, such as in the DNA double helix. In contrast, RNA is usually single-stranded, but any given strand may fold back upon itself to form secondary structure as in tRNA and rRNA.

The sugars and phosphates in nucleic acids are connected to each other in an alternating chain, linked by shared oxygens, forming a phosphodiester bond. In conventional nomenclature, the carbons to which the phosphate groups attach are the 3' end and the 5' end carbons of the sugar. This gives nucleic acids polarity. The bases extend from a glycosidic linkage to the 1 ' carbon of the pentose sugar ring. Bases are joined through N-1 of pyrimidines and N-9 of purines to 1 ' carbon of ribose through Ν-β glycosyl bond. microRNA

MicroRNAs (miRNA) are single-stranded RNA molecules of about 19-25 nucleotides in length, which regulate gene expression. miRNAs are either expressed from non- protein-coding transcripts or mostly expressed from protein coding transcripts. They are processed from primary transcripts known as pri-miRNA to shorter stem-loop structures called pre-miRNA and finally to functional mature miRNA. Mature miRNA molecules are partially complementary to one or more messenger RNA (mRNA) molecules, and their main function is to inhibit gene expression. This may occur by preventing mRNA translation or increasing mRNA turnover/degradation.

The transcripts encoding miRNAs are much longer than the processed mature miRNA molecule; miRNAs are first transcribed as primary transcripts or pri-miRNA with a cap and poly-A tail by RNA polymerase II and processed to short, 70-nucleotide stem-loop structures known as pre-miRNA in the cell nucleus. This processing is performed in animals by a protein complex known as the Microprocessor complex, consisting of the ribonuclease III Drosha and the double-stranded RNA binding protein Pasha. These pre-miRNAs are then exported to the cytoplasm by Exportin-5/Ran-GTP and processed to mature miRNAs by interaction with the ribonuclease III Dicer and separation of the miRNA duplexes. The mature single-stranded miRNA is incorporated into a RNA- induced silencing complex (RlSC)-like ribonucleoprotein particle (miRNP). The RISC complex is responsible for the gene silencing observed due to miRNA expression and RNA interference. The pathway is different for miRNAs derived from intronic stem- loops; these are processed by Dicer but not by Drosha.

When Dicer cleaves the pre-miRNA stem-loop, two complementary short RNA molecules are formed, but only one is integrated into the RISC complex. This strand is known as the guide strand and is selected by the argonaute protein, the catalytically active RNase in the RISC complex, on the basis of the stability of the 5' end. The remaining strand, known as the anti-guide or passenger strand, is degraded as a RISC complex substrate. After integration into the active RISC complex, miRNAs base pair with their complementary mRNA molecules. This may induce mRNA degradation by argonaute proteins, the catalytically active members of the RISC complex, or it may inhibit mRNA translation into proteins without mRNA degradation.

The function of miRNAs appears to be mainly in gene regulation. For that purpose, an miRNA is (partly) complementary to a part of one or more mRNAs. Animal miRNAs are usually complementary to a site in the 3' UTR. The annealing of the miRNA to the mRNA then inhibits protein translation, and sometimes facilitates cleavage of the mRNA (depending on the degree of complementarity). In such cases, the formation of the double-stranded RNA through the binding of the miRNA to mRNA inhibits the mRNA transcript through a process similar to RNA interference (RNAi). Further, miRNAs may regulate gene expression post-transcriptionally at the level of

translational inhibition at P-bodies. These are regions within the cytoplasm consisting of many enzymes involved in mRNA turnover; P bodies are likely the site of miRNA action, as miRNA-targeted mRNAs are recruited to P bodies and degraded or sequestered from the translational machinery. In other cases it is believed that the miRNA complex blocks the protein translation machinery or otherwise prevents protein translation without causing the mRNA to be degraded. miRNAs may also target methylation of genomic sites which correspond to targeted mRNAs. miRNAs function in association with a complement of proteins collectively termed the miRNP (miRNA ribonucleoprotein complex).

Under a standard nomenclature system, miRNA names are assigned to experimentally confirmed miRNAs before publication of their discovery. The prefix "mir" is followed by a dash and a number, the latter often indicating order of naming. For example, mir-123 was named and likely discovered prior to mir-456. The uncapitalized "mir-" refers to the pre-miRNA, while a capitalized "miR-" refers to the mature form. miRNAs with nearly identical sequences bar one or two nucleotides are annotated with an additional lower case letter. For example, miR-123a would be closely related to miR-123b. miRNAs that are 100% identical but are encoded at different places in the genome are indicated with additional dash-number suffix: miR-123-1 and miR-123-2 are identical but are produced from different pre-miRNAs. Species of origin is designated with a three-letter prefix, e.g., hsa-miR-123 would be from human (Homo sapiens) and oar-miR-123 would be a sheep (Ovis aries) miRNA. Other common prefixes include V for viral (miRNA encoded by a viral genome) and 'd' for Drosophila miRNA. microRNAs originating from the 3' or 5' end of a pre-miRNA are denoted with a -3p or -5p suffix. (In the past, this distinction was also made with 's' (sense) and 'as' (antisense)).

An asterisk following the name indicates that the miRNA is an anti-miRNA to the miRNA without an asterisk (e.g. miR-123^* is an anti-miRNA to miR-123). When relative expression levels are known, an asterisk following the name indicates a miRNA expressed at low levels relative to the miRNA in the opposite arm of a hairpin. For example, miR-123 and miR-123^* would share a pre-miRNA hairpin, but relatively more miR-123 would be found in the cell. miRBase is the central online repository for microRNA (miRNA) nomenclature, sequence data, annotation and target prediction, and may be accessed via

http://www.mirbase.org/. The miRNA names used herein throughout can be accessed via this link, and specifics retrieved. See also Griffiths-Jones et al, "miRBase: tools for microRNA genomics", Nucleic Acids Research, 2008, Vol. 36, Database issue D154- D158.

Classifier

Classifiers are relationships between sets of input variables, usually known as features, and discrete output variables, known as classes. Classes are often centered on the key questions of who, what, where and when. A classifier can intuitively be thought of as offering an opinion about whether, for instance, an individual associated with a given feature set is a member of a given class.

In other words, a classifier is a predictive model that attempts to describe one column (the label) in terms of others (the attributes). A classifier is constructed from data where the label is known, and may be later applied to predict label values for new data where the label is unknown. Internally, a classifier is an algorithm or mathematical formula that predicts one discrete value for each input row. For example, a classifier built from a dataset of iris flowers could predict the type of a presented iris given the length and width of its petals and stamen. Classifiers may also produce probability estimates for each value of the label. For example, a classifier built from a dataset of cars could predict the probability that a specific car was built in the United States.

Sensitivity and specificity

Sensitivity and specificity are statistical measures of the performance of a binary classification test. The sensitivity (also called recall rate in some fields) measures the proportion of actual positives which are correctly identified as such (i.e. the percentage of sick people who are identified as having the condition); and the specificity measures the proportion of negatives which are correctly identified (i.e. the percentage of well people who are identified as not having the condition). They are closely related to the concepts of type I and type II errors. For any test, there is usually a trade-off between each measure. For example in a manufacturing setting in which one is testing for faults, one may be willing to risk discarding functioning components (low specificity), in order to increase the chance of identifying nearly all faulty components (high sensitivity). This trade-off can be represented graphically using a ROC curve.

Βΐ of True Positives

sermitivitv

number of True Positives number of False Negatives

A sensitivity of 100% means that the test recognizes all sick people as such. Thus in a high sensitivity test, a negative result is used to rule out the disease.

Sensitivity alone does not tell us how well the test predicts other classes (that is, about the negative cases). In the binary classification, as illustrated above, this is the corresponding specificity test, or equivalently, the sensitivity for the other classes. Sensitivity is not the same as the positive predictive value (ratio of true positives to combined true and false positives), which is as much a statement about the proportion of actual positives in the population being tested as it is about the test. The calculation of sensitivity does not take into account indeterminate test results. If a test cannot be repeated, the options are to exclude indeterminate samples from analyses (but the number of exclusions should be stated when quoting sensitivity), or, alternatively, indeterminate samples can be treated as false negatives (which gives the worst-case value for sensitivity and may therefore underestimate it). number of True e atives

¾ number of True Negatives + number of False Positives

A specificity of 100% means that the test recognizes all healthy people as healthy. Thus a positive result in a high specificity test is used to confirm the disease. The maximum is trivially achieved by a test that claims everybody healthy regardless of the true condition. Therefore, the specificity alone does not tell us how well the test recognizes positive cases. We also need to know the sensitivity of the test to the class, or equivalently, the specificities to the other classes. A test with a high specificity has a low Type I error rate.

Specificity is sometimes confused with the precision or the positive predictive value, both of which refer to the fraction of returned positives that are true positives. The distinction is critical when the classes are different sizes. A test with very high specificity can have very low precision if there are far more true negatives than true positives, and vice versa. miRNA classifier of the present invention

The miRNA classifiers according to the present invention are the relationships between sets of input variables, i.e. the miRNA expression in a sample of an individual, and discrete output variables, i.e. distinction between a benign and malignant or a benign and malignant/pre-malignant condition of the thyroid. Thus, the classifier assigns a given sample to a given class with a given probability.

Distinction, differentiation or characterisation of a sample is used herein as being capable of predicting with a high sensitivity and specificity if a given sample of unknown diagnosis belongs to one of two classes (two-way classifier), or belongs to one of three classes (three-way classifier).

In one aspect, the miRNA classifier is a two-way classifier capable of predicting with an adequate sensitivity and specificity if a given sample of unknown diagnosis belongs to the class of either

A) benign FTA or malignant FTC, or

B) widely invasive carcinoma or minimally invasive carcinoma (both thyroid follicular carcinoma), or

C) follicular neoplasia (combined group of FTA and FTC) or NT, or

D) FTA or NT, or

E) FTC or NT.

In a particular embodiment, the miRNA classifier is a two-way classifier capable of predicting with an adequate sensitivity and specificity if a given sample of unknown diagnosis belongs to the class of either benign FTA or malignant FTC. In one aspect, the miRNA classifier is a two-way classifier capable of distinguishing either

A1 ) benign FTA from malignant FTC, wherein said miRNA classifier comprises or consists of one or more of the group of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542- 5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

A2) benign FTA from malignant FTC, wherein said miRNA classifier comprises or consists of one or more of the group of hsa-miR-1826 and hsa-miRPIus-E1078, or

A3) benign FTA from malignant FTC, wherein said miRNA classifier comprises one or more of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR- 429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR- 193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR- 148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa- miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa- miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR- 631 and hsa-miR-637, or

B) widely invasive FTC from minimally invasive FTC, wherein said miRNA classifier comprises or consists of one or more of hsa-hsa-miRPIus-E1001 and hsa-miR-410,

C) follicular neoplasia (combined group of FTA and FTC) from NT, wherein said miRNA classifier comprises at least 2 miRNAs according to the present invention, or

D) malignant FTC from NT, wherein said miRNA classifier comprises at least 2 miRNAs selected from the group consisting of hsa-miR-199b-5p, hsa-miR-144^*, hsa- miR-199a-3p/hsa-miR-199b-3p, hsa-miR-199a-5p, hsa-miR-144, hsa-miR-1275, hsa- miR-153, hsa-miR-451 , hsa-miR-142-3p, hsa-miR-886-5p,hsa-miR-31 , hsa-miR-455- 3p, hsa-miR-663, hsa-miR-218, hsa-miR-486-5p, hsa-miR-100, hsa-miR-542-5p, hsa- miR-1 , hsa-miR-101 , hsa-miR-20a, hsa-miR-193a-3p, hsa-miR-223, hsa-miR-886-3p, hsa-miR-18b, hsa-miR-190, hsa-miR-99a, hsa-miR-422a, hsa-miR-365, hsa-miR-943, hsa-miR-18a, hsa-miR-299-5p, hsa-miR-26a, hsa-miR-1 06a, hsa-miR-17, hsa-miR- 708, hsa-miR-455-5p, hsa-miR-27a, hsa-miR-130a, hsa-miR-143, hsa-miR-429, hsa- miR-1 38, hsa-miR-876-5p, hsa-miR-1297, hsa-miR-202, hsa-miR-33a, hsa-miR-Ι ΟΓ, hsa-let-7i^*, hsa-miR-675, hsa-miR-1 25b, hsa-miR-1 93b, hsa-miR-204, hsa-miR-326, ebv-miR-BART6-3p, hsa-miR-142-5p, hsa-miR-146b-5p, hsa-miR-140-5p, hsa-miR-

1 184, hsa-miR-1 0a, hsa-miR-23a, hsa-let-7g, hsa-let-7i, hsa-let-7d, hsa-miR-1 9a, hsa- miR-1 39-5p, hsa-miR-638, hsa-miR-374a, hsa-miR-19b, hsa-miR-124^*, hsa-miR-450a, hsa-miR-126^*, hsa-miR-133a, hsa-miR-181 c, hsa-miR-873, hsa-miR-342-3p, hsa-miR- 61 1 , hsa-miR-186, hsa-miR-508-5p, hsa-miR-30b, hsa-miR-374b, hsa-miR-146a, hsa- miR-1 95, hsa-miR-30d, hsa-miR-660, hsa-miR-557, hsa-miR-509-3-5p, hsa-miR-23b, hsa-miR-363, hsa-miR-10b, hsa-let-7a, hsa-miR-92a, hsa-let-7a^*, hsa-miR-16, hsa- miR-361 -3p, hsa-miR-145^*, hsa-miR-92b, hsa-miR-28-5p, ebv-miR-BHRF1 -1 , hsa- miR-29a^*, hsa-miR-30a^*, hsa-miR-30a, hsa-miR-30c, ebv-miR-BHRF1 -2, hsa-miR-21 ^*, hsa-miR-299-3p, hsa-miR-1 51 -5p, hsa-miR-24, hsa-miR-126, hsa-miR-513b, hsa-miR- 193b^*, hsa-miR-424, hsa-miR-15b, ebv-miR-BART19-3p, hsa-miR-602, hsa-miR-1 25b- 2^*, hsa-miR-200c^*, hsa-miR-377, hsa-miR-1304, hsa-miR-1 301 , hsa-miR-298, hsa- miR-934, hsa-miR-24-1 ^*, hcmv-miR-UL36, hsa-miR-1827, hsa-miR-647, hsa-miR-34a, hsa-miR-374b^*, hsa-miR-574-3p, hsa-miR-220c, hsa-miR-635, hsa-miR-197, hsa-miR- 1 274b, hsa-miR-1255a, hsa-miR-1298, hsa-miR-1 248, hsa-miR-629^*, hsa-miR-744, hsa-miR-22^*, hsa-miR-340, hsa-miR-634, hsa-miR-600, hsa-miR-34b, hsa-miR-1 29^*, hsa-miR-138-Γ, hsa-miR-637, hsa-miR-21 5, hsa-miR-222, hsa-miR-597, hsa-miR- 1 82, hsa-miR-96 and hsa-miR-221 , or

E) benign FTA from NT, wherein said miRNA classifier comprises at least 2 miRNAs selected from the group consisting of hsa-miR-199b-5p, hsa-miR-144^*, hsa-miR-663, hsa-miR-199a-3p/hsa-miR-1 99b-3p, hsa-miR-142-3p, hsa-miR-1 275, hsa-miR-1 99a- 5p, hsa-miR-144, hsa-miR-31 , hsa-miR-631 , hsa-miR-422a, hsa-miR-451 , hsa-miR- 218, hsa-miR-943, hsa-miR-675, hsa-miR-708, hsa-miR-486-5p, hsa-miR-492, hsa- miR-299-5p, hsa-miR-326, hiv1 -miR-H 1 , hsa-miR-455-3p, hsa-miR-202, ebv-miR- BART6-3p, hsa-miR-20a, hsa-miR-142-5p, hsa-miR-508-5p, hsa-miR-1 00, hsa-miR- 1 7, hsa-miR-490-5p, ebv-miR-BART20-3p, hsa-miR-638, hsa-miR-1 06a, hsa-miR-223, hsa-miR-18a, hsa-miR-101 ^*, hsa-miR-18b, hsa-miR-1 01 , hsa-miR-124^*, hsa-miR- 1 93b^*, hsa-miR-1303, hsa-miR-1 0a, hsa-miR-99a, hsa-miR-557, hsa-miR-376c, hsa- miR-1 25b, hsa-miR-10b, hsa-miR-936, hsa-miR-1 90, hsa-miR-61 1 , hsa-miR-1 35a^*, hsa-miR-455-5p, hsa-miR-602, hsa-miR-24-2^*, hsa-miR-26a, hsa-miR-1 184, hsa-miR- 516a-5p, hsa-miR-27a, hsa-miR-585, hsa-miR-198, hsa-miR-873, ebv-miR-BHRF1 -2, hsa-miR-10a^*, hsa-miR-192, ebv-miR-BART8^*, hsa-miR-520d-5p, hsa-miR-24-1 ^*, hsa- miR-1201 , hsa-miR-200b, hsa-miR-1264, hsa-miR-1259, hsa-miR-130b, hsa-miR-22^*, hsa-miR-363^*, hsa-miR-220a, hsa-miR-548e, hsa-miR-138-1 ^*, hsa-miR-1274a, hsa- miR-148b, hsa-miR-339-3p, hsa-miR-542-3p, hsa-miR-141 , hsa-miR-1826, hsa-miR- 34b, hsa-miR-34a^*, hsa-miR-1227, hsa-miR-887, hsa-miR-20b^*, hsa-miR-200c, hsa- miR-34a, hsa-miR-518e, hsa-miR-22, hsa-miR-526b^*, hsa-miR-1323, hsa-miR-182, hsa-miR-1274b, hsa-miR-597, hsa-miR-96, hsa-miR-522, hsa-miR-518a-3p, hsa-miR- 301 b, hsa-miR-520h, hsa-miR-512-3p, hsa-miR-517c, hsa-miR-517b, hsa-miR-517a and ebv-miR-BART8.

In one particular embodiment, the miRNA classifier is a two-way classifier capable of distinguishing the class of benign FTA from malignant FTC, wherein said miRNA classifier comprises or consists of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa- miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p. In one embodiment, said miRNA classifier consists of hsa-miR-19a, hsa-miR-501 -3p, hsa- miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa- miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330- 3p. In one embodiment, said miRNA classifier comprises at least all miRNAs from the group consisting of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa- miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p.

In another particular embodiment, the miRNA classifier is a two-way classifier capable of distinguishing the class of benign FTA from malignant FTC, wherein said miRNA classifier comprises or consists of hsa-miR-1826 and hsa-miRPIus-E1078. In one embodiment, said miRNA classifier consists of hsa-miR-1826 and hsa-miRPIus-E1078. In one embodiment, said miRNA classifier comprises at least all miRNAs from the group consisting of hsa-miR-1826 and hsa-miRPIus-E1078.

There is provided herein a system for the identification of a malignancy-specific signature of miRNAs that are differentially expressed relative to adenoma cells.

Piatt's probabilistic outputs for Support Vector Machines (Piatt, J. in Smola, A.J, et al. (eds.) Advances in large margin classifiers. Cambridge, 2000; incorporated herein by reference) is useful for applications that require posterior class probabilities. Also incorporated by reference herein is Piatt J. Advances in Large Classifiers. Cambridge, MA: MIT Press, 1999. The output of the two-way miRNA classifier is given as a probability of belonging to either class of between 0-1 (prediction probability). If the value for a sample is 0.5, no prediction is made. A number or value of between 0.51 to 1 .0 for a given sample means that the sample is predicted to belong to the class in question, e.g. FTA; and the corresponding value of 0.0 to 0.49 for the second class in question, e.g. FTC, means that the sample is predicted not to belong to the class in question.

In one embodiment, the prediction probabilities for a sample to belong to a certain class is a number falling in the range of from 0 to 1 , such as from 0.0 to 0.1 , for example 0.1 to 0.2, such as 0.2 to 0.3, for example 0.3 to 0.4, such as 0.4 to 0.49, for example 0.5, such as 0.51 to 0.6, for example 0.6 to 0.7, such as 0.7 to 0.8, for example 0.8 to 0.9, such as 0.9 to 1 .0.

In one embodiment, the prediction probability for a sample to belong to the FTA class is a number falling in the range of from 0 to 0.49, 0.5 or from 0.51 to 1 .0. In another embodiment, the prediction probability for a sample to belong to the FTC class is a number between from 0 to 0.49, 0.5 or between from 0.51 to 1 .0.

The classifiers according to the present invention may in one embodiment consist of 2 miRNAs, such as 3 miRNAs, for example 4 miRNAs, such as 5 miRNAs, for example 6 miRNAs, such as 7 miRNAs, for example 8 miRNAs, such as 9 miRNAs, for example 10 miRNAs, such as 1 1 miRNAs, for example 12 miRNAs, such as 13 miRNAs, for example 14 miRNAs, such as 15 miRNAs, for example 16 miRNAs, such as 17 miRNAs, for example 18 miRNAs, such as 19 miRNAs, for example 20 miRNAs according to the present invention.

The classifiers according to the present invention may in one embodiment consist of 2 to 4 miRNAs, such as 4 to 6 miRNAs, for example 6 to 8 miRNAs, such as 8 to 10 miRNAs, for example 10 to 12 miRNAs, such as 12 to 14 miRNAs, for example 14 to 16 miRNAs, such as 16 to 18 miRNAs, for example 18 to 20 miRNAs, such as 20 to 25 miRNAs, for example 25 to 30 miRNAs, such as 30 to 35 miRNAs, for example 35 to 40 miRNAs, such as 40 to 50 miRNAs according to the present invention.

The classifiers according to the present invention may in one embodiment consist of less than 10 miRNAs, such as less than 9 miRNAs, for example less than 8 miRNAs, such as less than 7 miRNAs, for example less than 6 miRNAs, such as less than 5 miRNAs, for example less than 4 miRNAs, such as less than 3 miRNAs according to the present invention.

In one aspect, the present invention relates to a two-way miRNA classifier for characterising a sample obtained from a thyroid nodule of an individual, wherein said miRNA classifier

i) comprises or consists of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa- miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa- miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa- miR-330-3p and distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma, or

ii) comprises or consists of hsa-miR-1826 and/or hsa-miRPIus-E1078 and distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma, or

iii) comprises or consists of hsa-miRPIus-E1001 and/or hsa-miR-410 and

distinguishes between the classes widely invasive carcinoma and minimally invasive carcinoma of the thyroid, or

iv) comprises or consists of one or more miRNAs selected from the group

consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR- 301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa- miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa- miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR- 342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa- miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa- miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637 and distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma,

wherein said distinction is given as a prediction probability for said sample of belonging to either class, said probability being a number falling in the range of from 0 to 1 .

In one embodiment, said two-way miRNA classifier is capable of distinguishing the class of benign FTA from malignant FTC and comprises or consists of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p. In one embodiment, said two-way miRNA classifier is capable of distinguishing the class of benign FTA from malignant FTC and comprises or consists of hsa-miR-1826 and hsa-miRPIus-E1078.

In another embodiment, said two-way miRNA classifier capable of distinguishing the classes widely invasive carcinoma and minimally invasive carcinoma of the thyroid and comprises or consists of hsa-miRPIus-E1001 and hsa-miR-410.

In yet another embodiment, said two-way miRNA classifier is capable of distinguishing the class of benign FTA from malignant FTC and comprises or consists of one or more of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa- miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR- 146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa- miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR- 27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR- 30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c- 5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR- 584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

In one embodiment, the two-way miRNA classifier further comprises one or more additional miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR- 886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa- miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa- miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa- miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR- 518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

In one embodiment, the two-way miRNA classifiers further comprises one or more additional miRNAs, wherein said additional miRNAs comprise no more than 10 additional miRNAs, for example 10 additional miRNAs, such as 9 additional miRNAs, for example 8 additional miRNAs, such as 7 additional miRNAs, for example 6 additional miRNAs, such as 5 additional miRNAs, for example 4 additional miRNAs, such as 3 additional miRNAs, for example 2 additional miRNAs, such as 1 additional miRNA according to the present invention.

In one embodiment, the two-way miRNA classifiers further comprises one or more additional miRNAs, such as 1 additional miRNA, for example 2 additional miRNAs, such as 3 additional miRNA, for example 4 additional miRNAs, such as 5 additional miRNA, for example 6 additional miRNAs, such as 7 additional miRNA, for example 8 additional miRNAs, such as 9 additional miRNA, for example 10 additional miRNAs, such as 1 1 additional miRNA, for example 12 additional miRNAs, such as 13 additional miRNA, for example 14 additional miRNAs, such as 15 additional miRNA, for example 16 additional miRNAs, such as 17 additional miRNA, for example 18 additional miRNAs, such as 19 additional miRNA, for example 20 additional miRNAs according to the present invention. The two-way miRNA classifiers according to the present invention preferably comprises or consists of less than 50 miRNAs, for example less than 40 miRNAs, such as less than 30 miRNAs, for example less than 20 miRNAs, such as less than 15 miRNAs, for example less than 10 miRNAs, such as less than 5 miRNAs.

In yet another embodiment, the two-way miRNA classifier does not comprise one or more of the miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR- 886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa- miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa- miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa- miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR- 518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

The expression of each miRNA in each thyroid sample used for constructing the two- way miRNA classifiers as defined herein were determined, and the combined pattern of expression of the herein disclosed miRNAs forms the basis of the classifier model capable of predicting a diagnosis. The pattern of expression for each of two of the disclosed classifiers is shown in the table below:

Classifier miRNA Expression pattern

^°A2 hsa-miRPIus-E1078 UP / (UP&DOWN)

FTA / FTC hsa-miR-1826 UP / (UP&DOWN)

B hsa-miRPIus-E1001 DOWN / (UP&DOWN)

Minimally hsa-miR-410 UP / (UP&DOWN)

invasive /

widely invasive

carcinoma The term "UP / (UP&DOWN)" is an indicator of the general expression pattern of the miRNA in question for a given classifier, and means that the miRNA is overall up- regulated or increased in FTA / Minimally invasive carcinoma (numerator) as compared to the expression observed in FTC / widely invasive carcinoma (denominator).

The term "DOWN / (UP&DOWN)" is an indicator of the general expression pattern of the miRNA in question for a given classifier, and means that the miRNA is overall down-regulated or decreased in FTA / Minimally invasive carcinoma (numerator) as compared to the expression observed in FTC / widely invasive carcinoma

(denominator).

In an embodiment, an alteration of the expression profile or signature of one or more of the miRNAs of the two-way miRNA classifiers is associated with the sample being classified as thyroid follicular adenoma or thyroid follicular carcinoma; or as widely invasive FTC or minimally invasive FTC.

In one embodiment, an alteration of the expression profile of one or more of said miRNAs is associated with thyroid follicular adenoma or thyroid follicular carcinoma. In one embodiment, an alteration of the expression profile of one or more of said miRNAs is associated with widely invasive thyroid follicular carcinoma or minimally invasive thyroid follicular carcinoma.

In one embodiment, the present invention relates to a two-way miRNA classifier for characterising a sample obtained from a thyroid nodule of an individual, wherein said miRNA classifier comprises or consists of hsa-miR-1826 and hsa-miRPIus-E1078 and distinguishes between thyroid follicular adenoma and thyroid follicular carcinoma.

In one embodiment, said two-way miRNA classifier consists of hsa-miR-1826 and hsa- miRPIus-E1078. The performance of said specific classifier for correctly classifying a sample into either of the classes FTC or FTA may have a sensitivity of 0.83, a specificity of 0.83, a positive predictive value of 0.83 and a negative predictive value of 0.83. In one embodiment, the two-way miRNA classifier is indicative of thyroid follicular adenoma in the event that hsa-miR-1826 expression is up-regulated and/or hsa- miRPIus-E1078 expression is up-regulated. In one embodiment, the two-way miRNA classifier is indicative of thyroid follicular carcinoma in the event that hsa-miR-1826 expression is down-regulated and/or hsa- miRPIus-E1078 expression is down-regulated

In another embodiment, the present invention relates to a two-way miRNA classifier for characterising a sample obtained from a thyroid nodule of an individual, wherein said miRNA classifier comprises or consists of hsa-miRPIus-E1001 and/or hsa-miR-410 and distinguishes between minimally invasive thyroid follicular carcinoma and widely invasive thyroid follicular carcinoma. In one embodiment, said two-way miRNA classifier consists of hsa-miRPIus-E1001 and hsa-miR-410.

In another embodiment, the two-way miRNA classifier is indicative of minimally invasive thyroid follicular carcinoma in the event that hsa-miR-410 expression is up- regulated and/or hsa-miRPIus-E1001 expression is down-regulated.

In yet another embodiment, the present invention relates to a two-way miRNA classifier for characterising a sample obtained from a thyroid nodule of an individual, wherein said miRNA classifier comprises or consists of one or more of hsa-miR-19a, hsa-miR- 501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR- 374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa- miR-330-3p, and distinguishes between thyroid follicular adenoma and thyroid follicular carcinoma. In a particular embodiment, said two-way miRNA classifier consists of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p. The performance of said specific classifier for correctly classifying a sample into the class FTA may have a sensitivity of 1 .0, a specificity of 0.92, a positive predictive value of 0.92 and a negative predictive value of 1 .0. The performance of said classifier for correctly classifying a sample into the class FTC may have a sensitivity of 0.92, a specificity of 1 .0, a positive predictive value of 1.0 and a negative predictive value of 0.92.

The miRNA classifiers disclosed herein in a particular embodiment has a sensitivity of at least 70%, such as at least 75%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%. The miRNA classifiers disclosed herein in a particular embodiment has a sensitivity of between 70-75%, such as 75-80%, for example 80-85%, such as 85-90%, for example 90-95%, such as 95-100%.

The miRNA classifiers disclosed herein in a particular embodiment has a specificity of at least 70%, such as at least 75%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%.

The miRNA classifiers disclosed herein in a particular embodiment has a specificity of between 70-75%, such as 75-80%, for example 80-85%, such as 85-90%, for example 90-95%, such as 95-100%. The miRNA classifiers disclosed herein in a particular embodiment has a negative predictive value for malignancies of at least 70%, such as at least 71 %, for example at least 72%, such as at least 73%, for example at least 74%, such as at least 75%, for example at least 76%, such as at least 77%, for example at least 78%, such as at least 79%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%, for example at least 96%, such as at least 97%, for example at least 98%, such as at least 99%, for example at least 100%.

The miRNA classifiers disclosed herein in a particular embodiment has a negative predictive value for malignancies of at between 70-71 %, such as 71 -72%, for example 72-73%, such as 73-74%, for example 74-75%, such as 75-76%, for example 76-77%, such as 77-78%, for example 78-79%, such as 79-80%, for example 80-81 %, such as 81 -82%, for example 82-83%, such as 83-84%, for example 84-85%, such as 85-86%, for example 86-87%, such as 87-88%, for example 88-89%, such as 89-90%, for example 90-91 %, such as 91 -92%, for example 92-93%, such as 93-94%, for example 94-95%, such as 95-96%, for example 96-97%, such as 97-98%, for example 98-99%, such as 99-100%.

The miRNA classifiers disclosed herein in a particular embodiment has a positive predictive value for malignancies of at least 70%, such as at least 71 %, for example at least 72%, such as at least 73%, for example at least 74%, such as at least 75%, for example at least 76%, such as at least 77%, for example at least 78%, such as at least 79%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%, for example at least 96%, such as at least 97%, for example at least 98%, such as at least 99%, for example at least 100%.

The miRNA classifiers disclosed herein in a particular embodiment has a positive predictive value for malignancies of at between 70-71 %, such as 71 -72%, for example 72-73%, such as 73-74%, for example 74-75%, such as 75-76%, for example 76-77%, such as 77-78%, for example 78-79%, such as 79-80%, for example 80-81 %, such as 81 -82%, for example 82-83%, such as 83-84%, for example 84-85%, such as 85-86%, for example 86-87%, such as 87-88%, for example 88-89%, such as 89-90%, for example 90-91 %, such as 91 -92%, for example 92-93%, such as 93-94%, for example 94-95%, such as 95-96%, for example 96-97%, such as 97-98%, for example 98-99%, such as 99-100%.

Methods for diagnosis employing the miRNA classifiers of the present invention

The invention in one aspect relates to a method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising measuring the expression profile of at least two miRNAs in a sample obtained from the thyroid of said individual, wherein said miRNA is selected from the group consisting of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p; or

ii) hsa-miR-1826 and hsa-miRPIus-E1078; or

iii) hsa-miRPIus-E1001 and hsa-miR-410; or

wherein a predetermined miRNA expression level of said miRNAs is indicative of the individual having, or being at risk of developing, follicular thyroid carcinoma.

In one aspect, the present invention discloses a method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising measuring the expression profile of a group of miRNAs in a sample obtained from the thyroid of said individual, wherein said group of miRNAs consists of the group consisting of

ii) hsa-miR-1826 and hsa-miRPIus-E1078,

In one aspect, the present invention discloses a method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising measuring the expression profile of a group of miRNAs in a sample obtained from the thyroid of said individual, wherein said group of miRNAs comprises at least all miRNAs from the group consisting of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-

330-3p; or

ii) hsa-miR-1826 and hsa-miRPIus-E1078,

In one aspect, the present invention discloses a method for diagnosing if an individual has, or is at risk of developing, widely invasive thyroid follicular carcinoma, comprising measuring the expression profile of a group of miRNAs in a sample obtained from the thyroid of said individual, wherein said group of miRNAs consists of the group consisting of

i) hsa-miRPIus-E1001 and hsa-miR-410,

wherein a predetermined miRNA expression level of said miRNAs is indicative of the individual having, or being at risk of developing, widely invasive thyroid follicular carcinoma.

In one embodiment, said method further comprises the step of obtaining a sample from the thyroid of an individual, by any means as disclosed herein elsewhere. In one embodiment, said thyroid sample is a thyroid nodule sample. In one embodiment, said method further comprises the step of extracting RNA from a thyroid sample collected from an individual, by any means as disclosed herein elsewhere.

In one embodiment, said method further comprises the step of determining if said individual has, or is at risk of developing, follicular thyroid carcinoma.

The invention in a further aspect relates to a method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising the steps of: 1 ) extracting RNA from a sample collected from the thyroid of an individual,

2) analysing the miRNA expression profile of the sample, comprising at least one miRNA selected from the group consisting of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa- miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-

542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-

301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR- 146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-

200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b- 5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637,

wherein a predetermined miRNA expression profile of the at least one of said miRNAs is indicative of the individual having, or being at risk of developing, follicular thyroid carcinoma.

The invention in one embodiment relates to a method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising the steps of: i) extracting RNA from a sample collected from the thyroid of an individual, ii) analysing the expression profile of hsa-miR-1826 and hsa-miRPIus-E1078 in the sample,

wherein a predetermined miRNA expression profile of said miRNAs is indicative of the individual having, or being at risk of developing, follicular thyroid carcinoma; said predetermined miRNA expression profile being associated with a prediction according to the miRNA classifier disclosed herein.

The invention in another embodiment relates to a method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising the steps of: i) extracting RNA from a sample collected from the thyroid of an individual, ii) analysing the expression profile of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR- 17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p in the sample,

The invention in a further aspect relates to a method for performing a diagnosis on an individual with a thyroid nodule, comprising the steps of:

1 ) extracting RNA from a sample collected from the thyroid of an individual,

2) analysing the miRNA expression profile of the sample, and

3) determining if said individual has a benign or a malignant condition

selected from follicular thyroid adenoma and follicular thyroid carcinoma, wherein said miRNA expression profile comprises at least one miRNA selected from the group consisting of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa- miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR- 542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-

301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR- 146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR- 200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b- 5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

The invention in one embodiment relates to a method for performing a diagnosis on an individual with a thyroid nodule, comprising the steps of:

i) extracting RNA from a sample collected from the thyroid of an individual, ii) analysing the expression profile of hsa-miR-1826 and hsa-miRPIus-E1078 in the sample, and

iii) determining if said individual has a benign or a malignant condition selected from follicular thyroid adenoma and follicular thyroid carcinoma.

The invention in another embodiment relates to a method for performing a diagnosis on an individual with a thyroid nodule, comprising the steps of:

i) extracting RNA from a sample collected from the thyroid of an individual, ii) analysing the expression profile of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR- 17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p in the sample, and

iii) determining if said individual has a benign or a malignant condition selected from follicular thyroid adenoma and follicular thyroid carcinoma. The invention in a further embodiment relates to a method for performing a diagnosis on an individual with a thyroid nodule, comprising the steps of:

i) extracting RNA from a sample collected from the thyroid of an individual, ii) analysing the expression profile of hsa-miRPIus-E1001 and hsa-miR-410 in the sample, and

iii) determining if said individual has a widely invasive follicular thyroid

carcinoma or a minimally invasive follicular thyroid carcinoma.

In one aspect, the present invention relates to a method for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said method comprising measuring the expression level of at least one miRNA in said sample, wherein said at least one miRNA is selected from the group consisting of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa- miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa- miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-

199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR- 199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299- 3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR- 451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-

140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f-2^*, hsa-miR-148b, hsa- miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa- miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2- 3p, hsa-miR-631 and hsa-miR-637, wherein the expression level of said at least one miRNA is associated with thyroid follicular carcinoma by predicting said association according to the miRNA classifier disclosed herein.

In one embodiment, the present invention relates to a method for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said method comprising measuring the expression level of hsa-miR-1826 and hsa-miRPIus-E1078 and wherein the expression level of said miRNAs is associated with thyroid follicular carcinoma.

In one embodiment, the present invention relates to a method for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said method comprising measuring the expression level of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, and wherein the expression level of said miRNAs is associated with thyroid follicular carcinoma.

In another embodiment, the present invention relates to a method for determining the presence of a minimally invasive or widely invasive malignant condition in a sample obtained from a thyroid nodule of an individual, said method comprising measuring the expression level of hsa-miRPIus-E1001 and hsa-miR-410 and wherein the expression level of said miRNAs is associated with invasiveness of FTC.

In one aspect, the present invention relates to a method for determining the presence of a benign condition in a sample obtained from a thyroid nodule of an individual, said method comprising measuring the expression level of at least one miRNA in said sample, wherein said at least one miRNA is selected from the groups consisting of i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa- miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR- 542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR- 146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR- 200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b- 5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637,

wherein said expression level of said at least one miRNA is associated with thyroid follicular adenoma by predicting said association according to the miRNA classifier disclosed herein.

The invention in a further aspect relates to a method for determining the need for thyroidectomy in an individual presenting with a thyroid nodule, comprising the steps of:

1 ) extracting RNA from a sample collected from the thyroid of an individual,

2) analysing the miRNA expression profile of the sample,

3) determining if said individual has a benign or a malignant condition

selected from follicular thyroid adenoma and follicular thyroid carcinoma, and

4) performing thyroidectomy on the individual only if the nodule is diagnosed as follicular thyroid carcinoma,

wherein said miRNA expression profile comprises at least one miRNA selected from the group consisting of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa- miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR- 542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p

hsa-miR-1826 and hsa-miRPIus-E1078, or hsa-miRPIus-E1001 and hsa-miR-410, or

hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR- 301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR- 146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR- 200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b- 5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

The invention in a further aspect relates to a method for partitioning a group of individuals presenting with thyroid nodules, comprising the steps of:

i) extracting RNA from a sample collected from the thyroid of an individual, ii) analysing the miRNA expression profile of the sample, consisting of either a. hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa- miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

b. hsa-miR-1826 and hsa-miRPIus-E1078,

iii) determining if said individual has a benign or a malignant condition selected from follicular thyroid adenoma and follicular thyroid carcinoma, and iv) performing thyroidectomy on the group of individuals only on thyroid

nodules diagnosed as follicular thyroid carcinoma, as determined according to the miRNA classifier disclosed herein.

It follows, that any of the above-mentioned methods may comprise the step of obtaining prediction probabilities of between 0-1 .

In a further embodiment, any of the above-mentioned methods may be is used in combination with at least one additional diagnostic method.

Said at least one additional diagnostic method may in one embodiment be selected from the group consisting of CT (X-ray computed tomography), MRI (magnetic resonance imaging), Scintillation counting, Blood sample analysis, Ultrasound imaging, Cytology, Histology and Assessment of risk factors.

In one embodiment, said at least one additional diagnostic method is use of an mRNA classifier capable of distinguishing between follicular thyroid adenoma and follicular thyroid carcinoma. Such as mRNA classifier may in one preferred embodiment be as disclosed in international patent application (PCT/DK2010/050358) entitled 'mRNA classification of thyroid follicular neoplasia'.

In one embodiment, said at least one additional diagnostic method improves the sensitivity and/or specificity of the combined diagnostic outcome.

The invention in a further aspect relates to a method for expression profiling of a sample obtained from the thyroid, comprising measuring at least one miRNA selected from the group of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 , hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR- 429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa- miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa- let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa- miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa- let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa- miR-126^*, hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa- miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa- miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa-miR-519a7hsa- miR-519b-5p/hsa-miR-519c-5p/hsa-miR-522^*/hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa- miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637,

and correlating said expression profile to a clinical condition of the thyroid.

In one embodiment, said clinical condition is follicular thyroid carcinoma or follicular thyroid adenoma.

In another embodiment, said clinical condition is widely invasive follicular thyroid carcinoma or minimally invasive follicular thyroid carcinoma.

In another aspect, the present invention relates to a method for determining the prognosis of an individual with a thyroid nodule, comprising the steps of

1 ) extracting RNA from a sample collected from the thyroid of an individual,

2) analysing the miRNA expression profile of the sample,

3) determining if said individual has a malignant condition being follicular thyroid carcinoma,

ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR- 301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR- 146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR- 200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b- 5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637. A model for predicting a diagnosis by employing the miRNA classifier of the present invention

In one aspect, the present invention relates to a model for predicting the diagnosis of an individual with a thyroid nodule, comprising

i) providing a set of input data to the miRNA classifier according to the present invention, and

ii) determining if said individual has a condition selected from the group of thyroid follicular adenoma and thyroid follicular carcinoma, or the group of minimally invasive thyroid follicular carcinoma and widely invasive thyroid follicular carcinoma.

In one embodiment, said input data comprises or consists of the miRNA expression profile of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR- 320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p.

In one embodiment, said input data comprises or consists of the miRNA expression profile of hsa-miR-1826 and hsa-miRPIus-E1078. In one embodiment, said input data comprises the miRNA expression profile of hsa- miRPIus-E1001 and hsa-miR-410

In one embodiment, said input data comprises the miRNA expression profile of one or more of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR- 429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR- 193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR- 148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa- miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa- miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR- 631 and hsa-miR-637.

In a further embodiment, the model according to the present invention further comprises the miRNA expression profile of one or more additional miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa- miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR- 199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa- miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa- miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR- 519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

In one embodiment, said additional miRNAs comprise no more than 10 additional miRNAs, such as 9 additional miRNAs, for example 8 additional miRNAs, such as 7 additional miRNAs, for example 6 additional miRNAs, such as 5 additional miRNAs, for example 4 additional miRNAs, such as 3 additional miRNAs, for example 2 additional miRNAs, such as 1 additional miRNA according to the present invention.

In one embodiment, said additional miRNAs comprise 1 additional miRNA, for example 2 additional miRNAs, such as 3 additional miRNA, for example 4 additional miRNAs, such as 5 additional miRNA, for example 6 additional miRNAs, such as 7 additional miRNA, for example 8 additional miRNAs, such as 9 additional miRNA, for example 10 additional miRNAs, such as 1 1 additional miRNA, for example 12 additional miRNAs, such as 13 additional miRNA, for example 14 additional miRNAs, such as 15 additional miRNA, for example 16 additional miRNAs, such as 17 additional miRNA, for example 18 additional miRNAs, such as 19 additional miRNA, for example 20 additional miRNAs according to the present invention.

In another aspect, the present invention relates to a model for predicting the diagnosis of an individual with a thyroid nodule, comprising

ii) determining if said individual has a condition selected from the group of widely invasive thyroid follicular carcinoma and minimally invasive thyroid follicular carcinoma.

In one embodiment, said input data comprises or consists of the miRNA expression profile of hsa-miRPIus-E1001 and/or hsa-miR-410. Sample type

The sample according to the present invention is extracted from an individual and used for miRNA profiling for the subsequent diagnosis of a condition. In one embodiment, the sample comprises cells and/or tissue. The sample may be collected from an individual or a cell culture, preferably an individual. The individual may be any animal, such as a mammal, including human beings. In a preferred embodiment, the individual is a human being.

In a particular embodiment, the sample is taken from the thyroid gland of a human being, such as a thyroid gland comprising thyroid neoplasia and/or a thyroid nodule. Sample collection

In one embodiment, the sample is collected from the thyroid of an individual by any available means, such as fine-needle aspiration (FNA) using a needle with a maximum diameter of 1 mm; core needle aspiration using a needle with a maximum diameter of above 1 mm (also called coarse needle aspiration or biopsy, large needle aspiration or large core aspiration); cutting biopsy; open biopsy; a surgical sample; or any other means known to the person skilled in the art. In another embodiment, the sample is collected from an in vitro cell culture.

In a preferred embodiment, the sample is a fine-needle aspirate from an individual. The fine-needle aspiration may be performed using a needle with a diameter of between 0.2 to 1 .0 mm, such as 0.2 to 0.3 mm, for example 0.3 to 0.4 mm, such as 0.4 to 0.5 mm, for example 0.5 to 0.6 mm, such as 0.6 to 0.7 mm, for example 0.7 to 0.8 mm, such as 0.8 to 0.9 mm, for example 0.9 to 1 .0 mm in diameter.

The sample may in one preferred embodiment be extracted by the method disclosed in international patent application PCT/DK2010/050056 entitled 'Improved RNA purification method'.

The diameter of the needle is indicated by the needle gauge. Various needle lengths are available for any given gauge. Needles in common medical use range from 7 gauge (the largest) to 33 (the smallest) on the Stubs scale. Although reusable needles remain useful for some scientific applications, disposable needles are far more common in medicine. Disposable needles are embedded in a plastic or aluminium hub that attaches to the syringe barrel by means of a press-fit (Luer) or twist-on (Luer-lock) fitting.

The fine-needle aspiration is in a preferred embodiment performed using a needle gauge of between 20 to 33, such as needle gauge 20, for example needle gauge 21 , such as needle gauge 22, for example needle gauge 23, such as needle gauge 24, for example needle gauge 25, such as needle gauge 26, for example needle gauge 27, such as needle gauge 28, for example needle gauge 29, such as needle gauge 30, for example needle gauge 31 , such as needle gauge 32, for example needle gauge 33. In a particular embodiment, the gauge of the needle is 23. The fine-needle aspiration may in one embodiment be assisted, such as ultra-sound (US) guided fine-needle aspiration, x-ray guided fine-needle aspiration, endoscopic ultra-sound (EUS) guided fine-needle aspiration, Endobronchial ultrasound-guided fine- needle aspiration (EBUS), ultrasonographically guided fine-needle aspiration, stereotactically guided fine-needle aspiration, computed tomography (CT)-guided percutaneous fine-needle aspiration and palpation guided fine-needle aspiration.

The skin above the area to be biopsied may in one embodiment be swiped with an antiseptic solution and/or may be draped with sterile surgical towels. The skin, underlying fat, and muscle may in one embodiment be numbed with a local anesthetic.

After the needle is placed into the mass, cells may be withdrawn by aspiration with a syringe.

The sample extracted from an individual by any means as disclosed above may be transferred to a tube or container prior to analysis. The container may be empty, or may comprise a collection media. Collection media are disclosed herein below. The sample extracted from an individual by any means as disclosed above may be analysed essentially immediately, or it may be stored prior to analysis for a variable period of time and at various temperature ranges.

In one embodiment, the sample is stored at a temperature of between -200°C to 37°C, such as between -200 to -100°C, for example -100 to -50°C, such as -50 to -25°C, for example -25 to -10°C, such as -10 to 0°C, for example 0 to 10°C, such as 10 to 20°C, for example 20 to 30°C, such as 30 to 37°C prior to analysis.

In another embodiment, the sample is stored for between 15 minutes and 100 years prior to analysis, such as between 15 minutes and 1 hour, for example 1 to 2 hours, such as 2 to 5 hours, for example 5 to 10 hours, such as 10 to 24 hours, for example 24 hours to 48 hours, such as 48 to 72 hours, for example 72 to 96 hours, such as 4 to 7 days, such as 1 week to 2 weeks, such as 2 to 4 weeks, such as 4 weeks to 1 month, such as 1 month to 2 months, for example 2 to 3 moths, such as 3 to 4 months, for example 4 to 5 moths, such as 5 to 6 months, for example 6 to 7 moths, such as 7 to 8 months, for example 8 to 9 moths, such as 9 to 10 months, for example 10 to 1 1 moths, such as 1 1 to 12 months, for example 1 year to 2 years, such as 2 to 3 years, for example 3 to 4 years, such as 4 to 5 years, for example 5 to 6 years, such as 6 to 7 years, for example 7 to 8 years, such as 8 to 9 years, for example 9 to 10 years, such as 10 to 20 years, for example 20 to 30 years, such as 30 to 40 years, for example 40 to 50 years, such as 50 to 75 years, for example 75 to 100 years prior to analysis.

In one embodiment, the sample is extracted from an individual by fine-needle aspiration.

In one embodiment, the sample is extracted from an individual by single fine-needle aspiration.

In one embodiment, the sample is extracted from an individual by multiple fine-needle aspirations.

Said multiple fine-needle aspirations may comprise 2 fine-needle aspirations, such as 3 fine-needle aspirations, for example 4 fine-needle aspirations, such as 5 fine-needle aspirations, for example 6 fine-needle aspirations, such as 7 fine-needle aspirations, for example 8 fine-needle aspirations, such as 9 fine-needle aspirations, for example 10 fine-needle aspirations.

Said multiple fine-needle aspirations may be taken or performed consecutively, such as subsequently after each other, within minutes or a few hours, or within more than a few hours such as days in between aspiration; or may be taken or performed essentially simultaneously.

In another embodiment, the sample is extracted from an individual by coarse-needle aspiration.

In yet another embodiment, the sample is extracted from an individual by thyroid surgery.

In another embodiment, the sample is extracted from an individual by hemi- thyroidectomy. In another embodiment, the sample is extracted from an individual by thyroid biopsy.

Collection media for sample

A collection media according to the present invention is any solution suitable for collecting a sample for immediate or later analysis and/or retrieval of RNA from said sample.

In one embodiment, the collection media is an RNA preservation solution or reagent suitable for containing samples without the immediate need for cooling or freezing the sample, while maintaining RNA integrity prior to extraction of RNA from the sample. An RNA preservation solution or reagent may also be known as RNA stabilization solution or reagent or RNA recovery media, and may be used interchangeably herein. The RNA preservation solution may penetrate the harvested cells of the collected sample and retards RNA degradation to a rate dependent on the storage temperature.

The RNA preservation solution may be any commercially available solutions or it may be a solution prepared according to available protocols. The commercially available RNA preservation solutions may for example be selected from RNAIater® (Ambion and Qiagen), PreservCyt medium (Cytyc Corp),

PrepProtect™ Stabilisation Buffer (Miltenyi Biotec), Allprotect Tissue Reagent (Qiagen) and RNAprotect Cell Reagent (Qiagen). Protocols for preparing a RNA stabilizing solution may be retrieved from the internet (e.g. L.A. Clarke and M.D. Amaral: 'Protocol for RNase-retarding solution for cell samples', provided through The European Workin Group on CFTR Expression), or may be produced and/or optimized according to techniques known to the skilled person.

In another embodiment, the collection media will penetrate and lyse the cells of the sample immediately, including reagents and methods for isolating RNA from a sample that may or may not include the use of a spin column.

Said reagents and methods for isolating RNA is described herein below in the section 'analysis of sample'. Other collection media according to the present invention comprises any media such as water, sterile water, denatured water, saline solutions, buffers, PBS, TBS, Allprotect Tissue Reagent (Qiagen), cell culture media such as RPMI-1640, DMEM (Dulbecco's Modified Eagle Medium), MEM (Minimal Essential Medium), IMDM (Iscove's Modified Dulbecco's Medium), BGjB (Fitton-Jackson modification), BME (Basal Medium Eagle), Brinster's BMOC-3 Medium, CMRL Medium, C0₂-Independent Medium, F-10 and F-12 Nutrient Mixture, GMEM (Glasgow Minimum Essential Medium), IMEM (Improved Minimum Essential Medium), Leibovitz's L-15 Medium, McCoy's 5A Medium, MCDB 131 Medium, Medium 199, Opti-MEM, Waymouth's MB 752/1 , Williams' Media E, Tyrode's solution, Belyakov's solution, Hanks' solution and other cell culture media known to the skilled person, tissue preservation media such as HypoThermosol®, CryoStor™ and Steinhardt's medium and other tissue preservation media known to the skilled person. In another embodiment, said collection media is means for fixation (preservation) of said thyroid sample; a tissue fixative, such as formalin (formaldehyde) or the like.

Types of tissue fixation includes heat fixation, chemical fixation (Crosslinking fixatives - Aldehydes; Precipitating fixatives - Alcohols; Oxidising agents; Mercurials; Picrates; HOPE (Hepes-glutamic acid buffer-mediated organic solvent protection effect) Fixative), and Frozen Sections.

In one embodiment, the fixation time may be between 1 to 7 calendar days; such as 1 day, 2 days, 3 days, 4 days, 5 days, 6 days or 7 days.

It follows that the invention may in one embodiment be carried out on formalin fixed paraffin embedded tissue blocks (FFPE).

Sample analysis

After the sample is collected, it is subjected to analysis. In one embodiment, the sample is initially used for isolating or extracting RNA according to any conventional methods known in the art; followed by an analysis of the miRNA expression in said sample. Extraction of RNA

The RNA isolated from the sample may be total RNA, mRNA, microRNA, tRNA, rRNA or any type of RNA. Conventional methods and reagents for isolating RNA from a sample comprise High

Pure miRNA Isolation Kit (Roche), Trizol (Invitrogen), Guanidinium thiocyanate-phenol- chloroform extraction, PureLink™ miRNA isolation kit (Invitrogen), PureLink Micro-to- Midi Total RNA Purification System (invitrogen), RNeasy kit (Qiagen), miRNeasy kit (Qiagen), Oligotex kit (Qiagen), phenol extraction, phenol-chloroform extraction, TCA/acetone precipitation, ethanol precipitation, Column purification, Silica gel membrane purification, PureYield™ RNA Midiprep (Promega), PolyATtract System 1000 (Promega), Maxwell^® 16 System (Promega), SV Total RNA Isolation (Promega), geneMAG-RNA / DNA kit (Chemicell), TRI Reagent® (Ambion), RNAqueous Kit (Ambion), ToTALLY RNA™ Kit (Ambion), Poly(A)Purist™ Kit (Ambion) and any other methods, commercially available or not, known to the skilled person.

The RNA may be further amplified, cleaned-up, concentrated, DNase treated, quantified or otherwise analysed or examined such as by agarose gel electrophoresis or Bioanalyser analysis (Agilent) or subjected to any other post-extraction method known to the skilled person.

Methods for extracting and analysing an RNA sample are disclosed in Molecular Cloning, A Laboratory Manual (Sambrook and Russell (ed.), 3^rd edition (2001 ), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, USA.

Microarray analysis

The isolated RNA is in one embodiment analysed by microarray analysis. In one embodiment, the expression level of one or more miRNAs is determined by the microarray technique.

A microarray is a multiplex technology that consists of an arrayed series of thousands of microscopic spots of DNA oligonucleotides or antisense miRNA probes, called features, each containing picomoles of a specific oligonucleotide sequence. This can be a short section of a gene or other DNA or RNA element that are used as probes to hybridize a DNA or RNA sample (called target) under high-stringency conditions. Probe-target hybridization is usually detected and quantified by fluorescence-based detection of fluorophore-labeled targets to determine relative abundance of nucleic acid sequences in the target. In standard microarrays, the probes are attached to a solid surface by a covalent bond to a chemical matrix (via epoxy-silane, amino-silane, lysine, polyacrylamide or others). The solid surface can be glass or a silicon chip, in which case they are commonly known as gene chip. DNA arrays are so named because they either measure DNA or use DNA as part of its detection system. The DNA probe may however be a modified DNA structure such as LNA (locked nucleic acid).

In one embodiment, the microarray analysis as used herein is used to detect microRNA, known as microRNA or miRNA expression profiling. The microarray for detection of microRNA may be a microarray platform, wherein the probes of the microarray may be comprised of antisense miRNAs or DNA

oligonucleotides. In the first case, the target is a labelled sense miRNA sequence, and in the latter case the miRNA has been reverse transcribed into cDNA and labelled. The microarray for detection of microRNA may be a commercially available array platform, such as NCode™ miRNA Microarray Expression Profiling (Invitrogen), miRCURY LNA™ microRNA Arrays (Exiqon), microRNA Array (Agilent), vParattcP Microfluidic Biochip Technology (LC Sciences), MicroRNA Profiling Panels (lllumina), Geniom® Biochips (Febit Inc.), microRNA Array (Oxford Gene Technology), Custom AdmiRNA™ profiling service (Applied Biological Materials Inc.), microRNA Array (Dharmacon - Thermo Scientific), LDA TaqMan analyses (Applied Biosystems), Taqman microRNA Array (Applied Biosystems) or any other commercially available array. Microarray analysis may comprise all or a subset of the steps of RNA isolation, RNA amplification, reverse transcription, target labelling, hybridisation onto a microarray chip, image analysis and normalisation, and subsequent data analysis; each of these steps may be performed according to a manufacturers protocol such as Invitrogen, or as described herein below in Example 1 . It follows, that any of the methods as disclosed herein above e.g. for diagnosing of an individual with a thyroid nodule may further comprise one or more of the steps of: i) isolating miRNA from a sample,

ii) labelling of said miRNA,

iii) hybridising said labelled miRNA to a microarray comprising miRNA-specific probes to provide a hybridisation profile for the sample,

iv) performing data analysis to obtain a measure of the miRNA expression profile of said sample. In another embodiment, the microarray for detection of microRNA is custom made.

A probe or hybridization probe is a fragment of DNA or RNA of variable length, which is used to detect in DNA or RNA samples the presence of nucleotide sequences (the target) that are complementary to the sequence in the probe. One example is a sense miRNA sequence in a sample (target) and an antisense miRNA probe. The probe thereby hybridizes to single-stranded nucleic acid (DNA or RNA) whose base sequence allows probe-target base pairing due to complementarity between the probe and target. To detect hybridization of the probe to its target sequence, the probe or the sample is tagged (or labelled) with a molecular marker. Detection of sequences with moderate or high similarity depends on how stringent the hybridization conditions were applied— high stringency, such as high hybridization temperature and low salt in hybridization buffers, permits only hybridization between nucleic acid sequences that are highly similar, whereas low stringency, such as lower temperature and high salt, allows hybridization when the sequences are less similar. Hybridization probes used in microarrays refer to nucleotide sequences covalently attached to an inert surface, such as coated glass slides, and to which a mobile target is hybridized. Depending on the method the probe may be synthesised via phosphoramidite technology or generated by PCR amplification or cloning (older methods). To design probe sequences, a probe design algorithm may be used to ensure maximum specificity (discerning closely related targets), sensitivity (maximum hybridisation intensities) and normalised melting temperatures for uniform hybridisation. RT-QPCR

In another embodiment, the isolated RNA is analysed by quantitative ('real-time') PCR (QPCR). In one embodiment, the expression level of one or more miRNAs is determined by the quantitative polymerase chain reaction (QPCR) technique.

Real-time polymerase chain reaction, also called quantitative polymerase chain reaction (Q-PCR/qPCR/RT-QPCR) or kinetic polymerase chain reaction, is a technique based on the polymerase chain reaction, which is used to amplify and simultaneously quantify a targeted DNA molecule. It enables both detection and quantification (as absolute number of copies or relative amount when normalized to DNA input or additional normalizing genes) of a specific sequence in a DNA sample.

The procedure follows the general principle of polymerase chain reaction; its key feature is that the amplified DNA is quantified as it accumulates in the reaction in real time after each amplification cycle. Two common methods of quantification are the use of fluorescent dyes that intercalate with double-stranded DNA, and modified DNA oligonucleotide probes that fluoresce when hybridized with a complementary DNA. Frequently, real-time polymerase chain reaction is combined with reverse transcription polymerase chain reaction to quantify low abundance messenger RNA (mRNA), or miRNA, enabling a researcher to quantify relative gene expression at a particular time, or in a particular cell or tissue type.

The QPCR may be performed using chemicals and/or machines from a commercially available platform.

The QPCR may be performed using QPCR machines from any commercially available platform; such as Prism, geneAmp or StepOne Real Time PCR systems (Applied Biosystems), LightCycler (Roche), RapidCycler (Idaho Technology), MasterCycler (Eppendorf), iCycler iQ system, Chromo 4 system, CFX, MiniOpticon and Opticon systems (Bio-Rad), SmartCycler system (Cepheid), RotorGene system (Corbett

Lifescience), MX3000 and MX3005 systems (Stratagene), DNA Engine Opticon system (Qiagen), Quantica qPCR systems (Techne), InSyte and Syncrom cycler system (BioGene), DT-322 (DNA Technology), Exicycler Notebook Thermal cycler, TL998 System (lanlong), Line-Gene-K systems (Bioer Technology), or any other commercially available platform. The QPCR may be performed using chemicals from any commercially available platform, such as NCode EXPRESS qPCR or EXPRESS qPCR (Invitrogen), Taqman or SYBR green qPCR systems (Applied Biosystems), Real-Time PCR reagents (Eurogentec), iTaq mix (Bio-Rad), qPCR mixes and kits (Biosense), and any other chemicals, commercially available or not, known to the skilled person.

The QPCR reagents and detection system may be probe-based, or may be based on chelating a fluorescent chemical into double-stranded oligonucleotides.

The QPCR reaction may be performed in a tube; such as a single tube, a tube strip or a plate, or it may be performed in a microfluidic card in which the relevant probes and/or primers are already integrated. A Microfluidic card allows high throughput, parallel analysis of mRNA or miRNA expression patterns, and allows for a quick and cost-effective investigation of biological pathways. The microfluidic card may be a piece of plastic that is riddled with micro channels and chambers filled with the probes needed to translate a sample into a diagnosis. A sample in fluid form is injected into one end of the card, and capillary action causes the fluid sample to be distributed into the microchannels. The microfluidic card is then placed in an appropriate device for processing the card and reading the signal.

It follows, that any of the methods as disclosed herein above e.g. for diagnosing of an individual with a thyroid nodule may further comprise one or more of the steps of:

i) isolating miRNA from a sample,

ii) performing QPCR analysis,

iii) performing data analysis to obtain a measure of the miRNA expression profile of said sample.

Other analysis methods

In yet another embodiment, the isolated RNA is analysed by northern blotting.

In one embodiment, the expression level of one or more miRNAs is determined by the northern blot technique. A northern blot is a method used to check for the presence of a RNA sequence in a sample. Northern blotting combines denaturing agarose gel or polyacrylamide gel electrophoresis for size separation of RNA with methods to transfer the size-separated RNA to a filter membrane for probe hybridization. The hybridization probe may be made from DNA or RNA.

In yet another embodiment, the isolated RNA is analysed by nuclease protection assay.

In one embodiment, the expression level of one or more miRNAs is determined by Nuclease protection assay.

Nuclease protection assay is a technique used to identify individual RNA molecules in a heterogeneous RNA sample extracted from cells. The technique can identify one or more RNA molecules of known sequence even at low total concentration. The extracted RNA is first mixed with antisense RNA or DNA probes that are

complementary to the sequence or sequences of interest and the complementary strands are hybridized to form double-stranded RNA (or a DNA-RNA hybrid). The mixture is then exposed to ribonucleases that specifically cleave only s/ng/e-stranded RNA but have no activity against double-stranded RNA. When the reaction runs to completion, susceptible RNA regions are degraded to very short oligomers or to individual nucleotides; the surviving RNA fragments are those that were

complementary to the added antisense strand and thus contained the sequence of interest.

Device

It is also an aspect of the present invention to provide a device for measuring the expression level of at least one miRNA in a sample, wherein said device comprises or consists of at least one probe or probe set for miRNAs selected from the groups consisting of

i) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p, or ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa- miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa- miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f- 2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa- miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa-miR-519a^*/hsa- miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa- miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637,

wherein said device is used for classifying a sample obtained from a thyroid nodule of an individual.

Also disclosed is a device for measuring the expression level of at least a group of miRNAs in a sample, wherein said device consists of one or more probes or probe sets for the miRNAs consisting of the group consisting of

ii) hsa-miR-1826 and hsa-miRPIus-E1078; or

iii) hsa-miRPIus-E1001 and hsa-miR-410,

wherein said device is used for classifying a sample obtained from a thyroid nodule of an individual. In one embodiment, there is provided a device for measuring the expression level of at least one miRNA, wherein said device consists of at least one probe for hsa-miR-1826 and hsa-miRPIus-E1078; and/or hsa-miRPIus-E1001 and hsa-miR-410, and wherein said device is used for classifying a sample obtained from a thyroid nodule of an individual.

In one embodiment, the device comprises or consists of probes for hsa-miR-19a, hsa- miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa- miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p.

In one embodiment, the device comprises or consists of probes for hsa-miR-1826 and hsa-miRPIus-E1078. In one embodiment, the device comprises or consists of probes for hsa-miRPIus-E1001 and hsa-miR-410.

In one embodiment, the device comprises or consists of probes for miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa- miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR- 199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa- miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa- miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR- 519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

In one embodiment, the device may be used for distinguishing between thyroid follicular adenoma and thyroid follicular carcinoma, and/or distinguishing widely invasive from minimally invasive thyroid follicular carcinoma. In one embodiment, said device may be used with the miRNA classifier according to the present invention to classify a sample into either of the classes of thyroid follicular adenoma and thyroid follicular carcinoma. In another embodiment, said device may be used with the miRNA classifier according to the present invention to classify a sample into either of the classes of minimally invasive thyroid follicular carcinoma or widely invasive thyroid follicular carcinoma.

In another embodiment said device comprises less than 50 probes, for example less than 40 probes, such as less than 30 probes, for example less than 20 probes, such as less than 10 probes, for example less than 5 probes.

In one embodiment, said device comprises or consists of a total of 1 probe or probe set for at least one miRNA to be measured, such as 2 probes, for example 3 probes, such as 4 probes, for example 5 probes, such as 6 probes, for example 7 probes, such as 8 probes, for example 9 probes, such as 10 probes, for example 1 1 probes, such as 12 probes, for example 13 probes, such as 14 probes, for example 15 probes, such as 16 probes, for example 17 probes, such as 18 probes, for example 19 probes, such as 20 probes, for example 21 probes, such as 22 probes, for example 23 probes, such as 24 probes, for example 25 probes, such as 26 probes, for example 27 probes, such as 28 probes, for example 29 probes, such as 30 probes, for example 31 probes, such as 32 probes, for example 33 probes, such as 34 probes, for example 35 probes, such as 36 probes, for example 37 probes, such as 38 probes, for example 39 probes, such as 40 probes, for example 41 probes, such as 42 probes, for example 43 probes, such as 44 probes, for example 45 probes, such as 46 probes, for example 47 probes, such as 48 probes, for example 49 probes, such as 50 probes or probe sets for at least one miRNA of the present invention to be measured.

In one embodiment said device comprises between 1 to 2 probes or probe sets per miRNA to be measured, such as 2 to 3 probes, for example 3 to 4 probes, such as 4 to 5 probes, for example 5 to 6 probes, such as 6 to 7 probes, for example 7 to 8 probes, such as 8 to 9 probes, for example 9 to 10 probes, such as 10 to 15 probes, for example 15 to 20 probes, such as 20 to 25 probes, for example 25 to 30 probes, such as 30 to 40 probes, for example 40 to 50 probes, such as 50 to 60 probes, for example 60 to 70 probes, such as 70 to 80 probes, for example 80 to 90 probes, such as 90 to 100 probes or probe sets per miRNA of the present invention to be measured.

It follows, that there may be one probe specific to a miRNA to be measured, or more than one probe specific to a miRNA to be measured - which may be called a probe set. In one embodiment, the device comprises 1 probe per miRNA to be measured, in another embodiment, said device comprises 2 probes, such as 3 probes, for example 4 probes, such as 5 probes, for example 6 probes, such as 7 probes, for example 8 probes, such as 9 probes, for example 10 probes, such as 1 1 probes, for example 12 probes, such as 13 probes, for example 14 probes, such as 15 probes per miRNA to be measured or analysed.

In one embodiment, the device may be a microarray chip; a QPCR Micro Fluidic Card; or QPCR tubes, QPCR tubes in a strip or a QPCR plate comprising one or more probes selected from hsa-miR-1826, hsa-miRPIus-E1078, hsa-miRPIus-E1001 , hsa- miR-410, hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR- 429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR- 193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR- 148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa- miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa- miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR- 631 and hsa-miR-637.

The probes may be comprised on a solid support, on at least one bead, or in a liquid reagent comprised in a tube.

In yet another embodiment, the device may comprise one or more additional miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa- miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR- 200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR- 1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f-2^*, hsa-miR-148b, hsa- miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa-miR- 519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

Computer program product

It is a further aspect of the invention to provide a computer program product having a computer readable medium, said computer program product comprising means for carrying out any of the herein listed miRNA classifiers, models and methods.

It is a further aspect of the invention to provide a system comprising means for carrying out any of the herein listed methods.

It is an aspect of the present invention to provide a system for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of at least one miRNA in said sample, wherein the expression level of said miRNAs is associated with thyroid follicular carcinoma, wherein said at least one miRNA is selected from the group consisting of

ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR- 301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR- 146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR- 200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b- 5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

In one embodiment, there is provided a system for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of hsa-miR-1826 and hsa- miRPIus-E1078 in said sample, wherein said expression level of said miRNAs is associated with thyroid follicular carcinoma, said association being predicted according to the miRNA classifier disclosed herein.

In another embodiment, there is provided a system for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of hsa-miR-19a, hsa-miR- 501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR- 374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa- miR-330-3p in said sample, wherein said expression level of said miRNAs is associated with thyroid follicular carcinoma, said association being predicted according to the miRNA classifier disclosed herein. It is a further aspect of the present invention to provide a system for determining the presence of a benign condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of at least one miRNA in said sample, wherein said expression level of said miRNAs is associated with thyroid follicular adenoma, wherein said at least one miRNA is selected from the group consisting of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa- miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR- 542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

hsa-miR-1826 and hsa-miRPIus-E1078, or hsa-miRPIus-E1001 and hsa-miR-410, or

In one embodiment, there is provided a system for determining the presence of a benign condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of hsa-miR-1826 and hsa- miRPIus-E1078 in said sample, wherein said expression level of said miRNAs is associated with thyroid follicular adenoma, said association being predicted according to the miRNA classifier disclosed herein.

In one embodiment, there is provided a system for determining the presence of a benign condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of hsa-miR-19a, hsa-miR- 501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR- 374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa- miR-330-3p in said sample, wherein said expression level of said miRNAs is associated with thyroid follicular adenoma, said association being predicted according to the miRNA classifier disclosed herein.

In another aspect, the present invention provides a system for performing a diagnosis on an individual with a thyroid nodule, comprising:

i) means for analysing the miRNA expression profile of the thyroid nodule, and

ii) means for determining if said individual has a benign or a malignant condition selected from follicular thyroid adenoma and follicular thyroid carcinoma,

ii) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR- 301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR- 146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa- miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR- 200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR- 513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b- 5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637. In one embodiment, there is provided a system for performing a diagnosis on an individual with a thyroid nodule, comprising:

i) means for analysing the miRNA expression profile of hsa-miR-19a, hsa- miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa- miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa- let-7i, hsa-miR-326 and hsa-miR-330-3p, and

ii) means for determining if said individual has a benign or a malignant condition selected from follicular thyroid adenoma and follicular thyroid carcinoma. In one embodiment, there is provided a system for performing a diagnosis on an individual with a thyroid nodule, comprising:

i) means for analysing the miRNA expression profile of hsa-miR-1826 and hsa-miRPIus-E1078, and

ii) means for determining if said individual has a benign or a malignant condition selected from follicular thyroid adenoma and follicular thyroid carcinoma.

In another aspect, the present invention provides a computer program product having a computer readable medium, said computer program product providing a system for predicting the diagnosis of an individual with a thyroid nodule, said computer program product comprising means for carrying out any of the steps of any of the methods as disclosed herein.

In another aspect, the present invention provides a system as disclosed herein wherein the data is stored, such as stored in at least one database.

Kit-of-parts

It is also an aspect to provide a kit-of-parts comprising the device according to the present invention, and at least one additional component. In one embodiment, said additional component is means for extracting RNA such as miRNA from a sample; reagents for performing microarray analysis and/or reagents for performing QPCR analysis. In another embodiment, said kit may comprise instructions for use of the device and/or the additional components.

In a further embodiment, said kit comprises a computer program product having a computer readable medium as detailed herein elsewhere.

Examples

Example 1 : MicroRNA Expression and Classification of Thyroid Follicular Adenoma and Carcinoma - LNA miRNA microarray Abstract

Context: Due to limitations in the cytopathological classification of thyroid follicular nodules, the preoperative assessment of malignancy is a clinical problem. Since microRNAs (miRNAs) are frequently changed during malignant progression, the identification of differentially expressed miRNAs could provide new insight into malignant progression of follicular neoplasia and improve diagnosis.

Objective: We characterized miRNAs in normal thyroid, follicular adenoma and follicular carcinoma and compared these results to differentially expressed mRNAs. Moreover, we generated a miRNA-based classifier that could distinguish thyroid follicular -adenoma from -carcinoma.

Design: MiRNAs were examined in 12 follicular adenomas, 12 follicular carcinomas, and 10 normal thyroid tissue samples by microarray analysis and mRNA profiles were used to integrate mRNA and miRNA expressions. The miRNA based classifier was generated using the support vector machine algorithm with leave-one-out cross- validation.

Results: 150, 107, and 56 miRNAs were differentially expressed in carcinoma or adenoma compared to normal thyroid, and between carcinoma and adenoma, respectively. miRNAs known to target oncogenes and tumour suppressors, such as miR-96, miR-182, miR-199b-5p, miR-199a-3p, and miR-221 , were among the differentially expressed. Integration of miRNA with differentially expressed mRNAs demonstrated a significant enrichment of down-regulated miRNA seed-sites among up- regulated mRNAs. Finally, we show that two miRNAs were sufficient to differentiate between follicular adenoma and carcinoma, with a negative predicted value of 83% for malignancies.

Conclusion ; We conclude that thyroid follicular neoplasia is accompanied by major changes in miRNA expression, that may be implicated in tumour development and facilitate diagnosis of follicular carcinoma. Introduction

Thyroid nodules are found in up to 7% of the adult population (1 ). Although the majority of the nodules are benign, carcinoma of the thyroid gland has an estimated worldwide incidence of 122,000 pr. year and is the most common malignancy of the endocrine system (2). Follicular adenomas are benign, encapsulated tumours. They are 5 times more frequent than follicular carcinomas (3) and several histological variants such as macrofollicular, oncocytic, follicular adenoma with papillary hyperplasia, fetal adenoma, signet-ring cell and clear cell follicular adenoma have been described (4). Follicular carcinomas mainly occur in middle-aged euthyroid women as a painless thyroid nodule and represent 10-15% of all thyroid malignancies (3). Follicular adenoma and carcinoma are morphologically similar and carcinomas are distinguished by vascular and/or capsular invasion that are considered to be the most important signs of malignancy (5). Since the latter features may be overlooked it is generally accepted, that the sensitivity and reproducibility of the diagnosis leaves room for improvement (6). MicroRNAs (miRNAs) are small, non-coding, and single-stranded RNAs of about 22 nucleotides. Transcription units are widespread in the genome and it is estimated that the number of miRNAs may exceed 1000 (904 at present time,

http://microrna.sanqer.ac.uk/). MiRNAs regulate translation and stability of particular target messenger RNAs (mRNA) by imperfect base pairing with the mRNAs (7). In this way miRNAs have been shown to regulate about 30% of all mammalian protein-coding transcripts (8;9). The expression of miRNAs is temporally and spatially regulated. Many are important for terminal differentiation processes during particular developmental stages, but miRNAs also exhibit important functions during adult life (10). MiRNAs are moreover aberrantly expressed or lost in a variety of cancers (1 1 ). Many target transcripts encode oncogenes and tumoursuppressors and in this way, dysregulated miRNAs play a causal role in malignant progression. Not surprisingly they are therefore considered attractive candidates for classification of tumours. MiRNAs have previously been characterized in various thyroid tumours (12-16). MiR-197 and miR-346 were found to be over-expressed in follicular carcinoma in comparison to adenoma and in vitro studies revealed both miRNAs to have a significant impact on proliferation of malignant cells (16).

Since miRNAs may be connected to malignant progression and provide additional information for classification of thyroid follicular nodules, we examined the miRNA expression in follicular tumours. Numerous miRNAs were shown to be differentially expressed in carcinoma, adenoma, and between carcinoma and adenoma. A number of these corresponded to changes in putative target mRNAs, implying that they may have a functional role. Two miRNAs could differentiate between follicular adenoma and carcinoma. We infer that progression from normal thyroid tissue to follicular adenomas and further to follicular carcinoma is reflected by significant changes in miRNA expression and we propose that miRNA based classification could improve diagnosis of follicular thyroid tumours.

Materials and methods

Thyroid tissue, follicular adenoma and carcinoma

The thyroid samples originated from a consecutive series of patients with a solitary or prominent scintigraphically cold thyroid nodule operated on, and with a histological diagnosis of a follicular adenoma or follicular carcinoma. Since a uniform

histopathological evaluation was essential, the diagnosis was made by a particular pathologist specialized in thyroid pathology. All tumours were diagnosed and classified according to the WHO definition of histological criteria. Clinical data are listed in table 1 . Surgically removed thyroid samples were snap frozen at the Department of

Pathology and stored at -120° over a 5-year period. The project was approved by the Ethics Committee. Twelve follicular adenomas (FA), 12 follicular carcinomas (FC), and 10 normal thyroid (NT) specimens were included. The number of tumours was balanced to provide optimal power estimates and a similar number of samples in each diagnostic category. Median age of patients in the two groups was practically the same, 44 years in carcinoma patients and 47 years in adenoma patients. The size of the tumours was compatible with a median diameter of 4 cm in the patients with a carcinoma and 3.75 cm in the patients with an adenoma.

MicroRNA Isolation

TotalRNA and miRNAs were isolated from frozen samples using Trizol, Invitrogen. Purified RNA was subsequently quantified on a NanoDrop^® ND-1000

Spectrophotometer (NanoDrop Technologies) and examined on a Bioanalyzer Nano RNA Chip (Agilent).

RNA labeling, hybridization and microarray platform

MiRNA expression levels were determined by microarray analysis. One microgram of totalRNA was labelled with fluorescent Hy3™(sample)/Hy5™(reference-sample) dye from the miRCURY LNA™microRNA Array Power Labelling Kit (Exiqon) according to manufacturer's instructions. Using a TECAN hybridization station, labelled samples were hybridized overnight to pre-printed miRCURY LNA™ microRNA Array, v.1 1 .0 (Exiqon; Catalogue number for array V.1 1 : 208202-A), containing probes for 841 human miRNAs, catalogued in the miRBase Sequence Database (Release 1 1 .0) ( ttp://microrna. Sanger. ac. uk/), and 428 proprietary human miRPIus sequences not yet annotated in miRBase.

Image analysis and normalization of miRNA expression

Arrays were scanned in an Agilent DNA Microarray Scanner, (Agilent Technologies) and resulting images were analyzed with Genepix Pro 6.0 software (Molecular

Devices). Background intensities were subtracted from foreground intensities and within array LOESS - normalized, followed by aquantile normalization between arrays as implemented in the Limma package in Bioconducter library (17) in R version 2.10 (18). Mean values of the quadruplicate probes for each of the miRNAs were obtained and log2 ratios between sample and reference were used for further analysis.

Class comparison analysis

Class comparison was performed using the Limma package. For each miRNA standard errors and fold changes were estimated by fitting a linear model. Differentially expressed miRNAs were determined by applying empirical Bayes moderated t- statistics test (19). MiRNAs were defined to be differentially expressed if they had a Benjamini-Hochberg corrected p-value below 0.05 and an absolute fold change above 1 .5.

Comparison of predicted miRNA targets and mRNA expression data

Results from the microarray analysis of normal thyroid and follicular adenoma and carcinoma samples (Borup et al. 2010, submitted), were used for integration of mRNA and miRNA array data. Moreover, samples from normal thyroid were downloaded from the Gene Expression Omnibus (GEO), ID: E-GEOD-6004 and E-GEOD-7307.

Differentially expressed transcripts from the comparisons FC vs. NT, FA vs. NT, and FC vs. FA were derived as described (Borup et at. 2010, submitted) and miRNAs were combined with putative mRNA targets among the differentially expressed transcripts using Targetscan in the Sanger miRBase data base incorporated in PARTEK software. Only the miRNAs that exhibited an absolute regulation larger than 1 .5 fold was selected. Finally, an overview of the number of target genes in the differentially expressed list that also were targets of the differentiated miRNAs was generated. Construction of classifier

Normalized data from 24 samples consisting of the expression level of 545 different probes (377 annotated hsa-miRNAs and 168 hsa miRPIus, all with an A-value > 7, A = ½(log₂R + log₂G)) were included in the analysis. The classifier was constructed for classification of FC and FA. The validation of the classifiers performance on the training set was done by leave-one-out (LOO) cross validation (CV), that provides an unbiased estimate of the prediction error by going over all training examples in turn using the complement training set (of size 24-1 = 23) to perform training (including probe ranking, selection and model fitting) and using the last left out sample for prediction. In this way, we obtained as many predictions as there were samples in the training set. These predictions formed unbiased estimates of the prediction error quantified in terms of the confusion matrix. The training of the classifiers inside the LOO loop consisted of a univariate selection step followed by applying support vector machine learning (SVM). Probes were ranked according to their differential expression using a Student t-test with a threshold of p < 0.001 (20).

Quantitative Reverse transcription PCR

CDNA was prepared from 25ng total RNA from 34 tumour samples using TagMan^® MicroRNA Reverse Transcription Kit and TagMan^® MicroRNA Assays containing predesigned primers for miR-221 , miR-182, miR-96, miR-199a3p, miR-144^*, miR- 199b5p, and miR-1826 was added. Hsa-miR-191 was used for endogenous control. Quantitative reverse transcription PCR (QRT-PCR) reaction was performed using TagMan^® Universal PCR Master Mix No AmpEras^® UNG, according to manufactures instructions, all from Applied Biosystems. Each amplification reaction was performed in triplicate, and median value of the three cycle threshold was used for further analysis. For calculations of fold changes we used the 2^"Δ0 _Τ method (21 ). Furthermore, we validated Exiqon miRPIus probe, MirPlusE-1078, used for classification of follicular carcinoma and adenoma. For this qRT-PCR analysis we employed the Exiqon microRNA LNA™ PCR primer sets together with Universal cDNA Synthesis and SYBR® Green master mix, following the manufacturer's instructions, Exiqon.

Primers for EPIus-1078 (forward and reverse primers) were purchased from Exiqon with catalogue numbers Ί χ 206999 fwd-miRPIus-E1078' and Ί χ 206999 rev-miRPIus- E1078'.

Results

MiRNA expression in follicular carcinoma and adenoma Tumours from 19 women and 5 men were examined by miRNA microarray analysis. The median age of the patients were 44 years in the FC-group and 47 years in the FA- group and nodule size ranged from 1 .5 to 10.5 cm. Hundred and fifty (150) annotated human miRNAs - 37 up-regulated and 1 13 down-regulated miRNAs - were differentially expressed in FCs compared to normal thyroid. The fold change ranged from 3.1 to -39 fold. Due to a massive 39 fold down-regulation miR-199b-5p was regarded lost. Mir- 144^*, miR-199a3p, miR-199a-5p, and miR-144 were also strongly down-regulated and considered close to background. Among the top up-regulated miRNAs, miR-221 , miR- 96, and miR-182 exhibited fold changes of 3.1 , 2.9, and 2.6, respectively. The comparison of FA to normal thyroid tissue revealed 107 differentially expressed miRNAs. Forty two were up-regulated and 65 down-regulated. Finally the comparison of carcinoma to adenoma showed, that 56 miRNAs were differentially expressed between these groups. Twelve miRNAs were up-regulated and 44 were down- regulated in the carcinoma. In adenoma and carcinoma 5 and 4 Ebstein Barr Virus derived miRNAs were dysregulated, respectively. All but EBV-miR-BART8 were down- regulated. The five most up- or down-regulated miRNAs in each comparison are listed in Table 2. For the complete list refer to Table 5. Fifty-eight (58) miRNAs were differentially expressed in both the carcinoma and adenoma compared to normal thyroid (Figure 1 A). The overlap was marked among the down-regulated miRNAs, where 49 of the 65 down-regulated miRNAs in the adenoma were shared, compared to 9 out of 42 for the up-regulated miRNAs. With the exception of miR-199b-5p, that were progressively down-regulated in carcinoma, the shared miRNAs exhibited basically the same levels in adenoma and carcinoma compared to normal thyroid (Figure 1 B). Integration of predicted miRNA targets and mRNA expression data

Results from the global expression profiling of normal thyroid and follicular adenoma and carcinoma samples were used for integration of mRNA and miRNA array data. Totally 586, 306 and 1 17 differentially expressed transcripts from the three

comparisons FC vs. NT, FA vs. NT, and FC vs. FA, were derived as described, and miRNAs were combined with predicted seed sites among the differentially expressed transcripts using Targetscan in the PARTEK miRNA - mRNA integration software. The total number of predicted seed sites among the mRNAs were 1721 , 433 and 206, respectively. In order to examine if the population of differentially expressed miRNAs in the different groups could have an impact on the corresponding changes in mRNA expression, we determined the number of predicted seed sites that were targeted by the differentially expressed miRNAs. To constrain the analysis, only mRNAs and miRNAs that exhibited an inverse expression pattern were considered. As summarized in Table 3, we observed a significant (Fishers Exact Test p<0.01 , assuming that the total number of 904 miRNA can target 22.000 transcripts at a global level) enrichment of seed-sites corresponding to down-regulated miRNAs. The differentially expressed and down-regulated miRNAs in the FC group exhibited putative seed-sites in almost 85% of the up-regulated transcripts, that distinguished carcinoma from normal thyroid and adenoma, respectively. No significant association was found between up-regulated miRNAs and down-regulated mRNAs in either of the tested combinations. A high number of miRNAs (>25), that were upregulated in FCs vs NTs exhibited predicted seed-sites in a large number (range - 25 to 50) of the inversely expressed mRNAs. Among these miRNAs, we identified several previously described oncomirs (miR-125b, miR-30a/b/c, miR-96, and miR-101 ).Taken together; the results indicate that miRNAs may have an impact on the observed changes in the transcriptome during progression to carcinoma.

MiRNA based classification of follicular nodules

To provide an overview of the miRNA expressions across follicular neoplasia (FA and FC) and NT, we generated a PCA plot using all expressed miRNAs (Figure 2A, panel a). At this stage the two populations could be discerned reflecting the relatively large difference in miRNA expression between the groups. By filtering the expression values by a t-test, we reached a subset of 179 miRNAs (p-value < 0.01 ), where follicular neoplasia could easily be distinguished from normal thyroid tissue (Figure 2A, panel b). We subsequently attempted to separate follicular carcinoma from adenoma by employing a Student t-test for feature selection and applied the supervised learning algorithm support vector machine (SVM) and leave-one-out-cross-validation (LOOCV) to generate a classifier. The optimal signature for classification of FC and FA consist of two miRNAs, hsa-miR-1826 and hsa-miRPIus-E1078. The negative predictive value for carcinoma is 83% and the positive predictive value for carcinoma was also 83%. The SVM can be turned into a probabilistic classifier giving an estimate of the probability of the predicted class label, i.e. assess the prediction uncertainty (22). The predictive probabilities for all samples are listed in Table 4. FA sample 1 1 , although correctly classified exhibited a probability of 0.5 and the misclassified FA sample 12 had a probability for FC of 0.9 indicating that FA1 1 is highly uncertain, whereas FA12 is most likely a misdiagnosed carcinoma. The relative expression of hsa-miR-1826 and hsa- miRPIus-E1078 is shown in Figure 2B. The samples are shown in a PCA plot after variance filtering and a two group comparison (P<0.01 ) (Figure 2B, panel a) and the expression levels are illustrated by the red green color coding (Figure 2B, panel b and c). Both miRNAs are down -regulated in FCs and the relative loss of expression is remarkably similar for the two miRNAs. It was not feasible to generate a classifier that could distinguish FA and minimally invasive carcinoma. However, widely invasive carcinoma can be distinguished from minimally invasive carcinoma by the expression of hsa-miRPIus-E1001 (P<0.01 ) (average 3 fold up-regulation) and lower expression of miRNA-410 (average 5 fold) compared to the minimally invasive carcinoma. Finally, expression values of miR-1826, hsa-miRPIus-E1078, miR-221 , miR-182, miR-96, miR- 199a3p, miR-144^*, miR-199b5p, were also examined by qRT-PCR and this confirmed the microarray results (Figure 3)

Discussion

In contrast to papillary and medullary thyroid cancers, where many tumours exhibit defined mutations in oncogenes, the causal mutations leading to follicular neoplasia are incompletely understood. Consequently, efforts have been devoted towards defining biomarkers, that would allow the clinicians to distinguish carcinoma from adenoma. Here we characterized the expression of miRNAs, that may be connected to malignant progression by repressing tumour suppressors or activating oncogenes following perturbed expression (23).

The employed miRNA array platform allows detection of essentially all known miRNAs and the tumours originated from consecutively referred patients whose sex and age were in accordance with that of larger epidemiological studies. Taken together, we find that follicular adenoma and carcinoma exhibit widespread changes in their miRNA expression compared to normal thyroid. Totally 150 miRNAs were altered in the carcinoma and although there was a large overlap with the dysregulated miRNAs from adenoma, the carcinoma exhibited more than 90 miRNAs, that were significantly different from those in adenoma. We identified numerous previously undescribed thyroid miRNAs in carcinoma and adenoma and we also confirm previously reported up-regulations of miR-197, -346, -187, -221 , -222, -224, and -155 in carcinoma and up- regulation of miR-339, -210, -328, and -342 in adenoma (14;16). Of particular significance, we find that follicular carcinoma exhibited extremely low levels or absence of hsa-miR-199b-5p, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-199a-5p and hsa- miR-144 and up-regulation of miR-96 and miR-182. Moreover, the finding of multiple dys regulated Ebstein Barr Virus (EBV) miRNAs was noteworthy, since expression of EBV previously have been shown to correlate with tumour progression (24;25).

It has been debated if follicular carcinoma arises from a fetal stem cell niche or occurs via multistep mechanism from adenoma, where the cells accumulate mutations in proto-oncogenes and tumour suppressor genes similar to e.g. colorectal cancers (26). The change in miRNA patterns is compatible with both models. MiRNAs are frequently expressed at high levels in terminally differentiated tissue (27) and the fact that the tumours mainly exhibit reduced levels of miRNAs may reflect their relatedness to a fetal cell. The overlap between perturbed miRNAs among adenoma and carcinoma in combination with the gradual changes of some miRNAs, is on the other hand compatible with a multistep model. As illustrated in Figure 1 B, it is evident that the major changes in the miRNA expression are seen from NT to FA, with a minor change of expression levels from FA to FC.

Since this study is based on solid tumours it is not possible to provide causal data to substantiate the functional significance of miRNAs in tumour progression. We therefore attempted to integrate the miRNA expression data with datasets of differentially expressed mRNAs in the different groups. By matching the total number of seed-sites in the differentially expressed mRNAs with the identified miRNAs, we found that seed- sites corresponding to down-regulated miRNAs in the carcinoma were almost 8 fold over-represented in the up-regulated mRNAs. In fact a large number of the differentially expressed oncomirs had up to 50 seed sites among the differentially expressed mRNAs. Evidently, many computationally predicted seed-sites are not functional in vivo, but this should not invalidate the observed over-representation of seed sites for the carcinoma associated miRNAs, since the computational predicted non-functional seed-stites should be equally represented in the control population and the test population. Our data may therefore suggest that the perturbed miRNAs have a significant effect on the changes in gene expression patterns that are characteristic of carcinoma.

This inference is substantiated by supporting data from other systems. Two recent studies demonstrate, that miR-96 and miR-182 are negative regulators of FOX01 expression (28;29). FOXO-proteins function as tumour suppressors and are known to control cell-cycle progression and apoptosis (30;31 ). And in agreement with a putative interaction between the miRNAs and FOX01 mRNA, we find that FOX01 mRNA is down-regulated in the follicular carcinoma as opposed to normal thyroid tissue.

Moreover, miR-199b-5p (also known as miR-199b) is essentially lost in the FCs. Mir- 199b-5p regulates HES1 and down-regulation of miR-199b-5p is followed by increased metastasis from meduloblastoma (32) and the expression of the oncogene SET (protein phosphatase 2A inhibitor) in chorioncarcinoma (33). Both HES1 and SET mRNA were up-regulated in the carcinoma. Mir-199a-3p is also extensively down- regulated in FCs and FAs. Mir-199a-3p is a negative regulator of the MET oncogene (34) and this is in concordance with up-regulation of MET transcripts in the follicular carcinoma. Finally, miR-221 - one of the best characterized oncomirs (35) is predicted to target NR4A1 mRNA, which, together with NR4A3, is known to control apoptosis. In mice loss of NR4A 1 and NR4A3 causes acute leukemia (36;37). We recently found that loss of apoptotic and growth arrest factors occur during progression to follicular malignancy. Down-regulation of apoptotic factors such as NR4A1 and NR4A3 in combination with JUN, FOSB and CITED2 was observed in all cancers implying that this event could precede malignancy. The finding is moreover supported by another recent study demonstrating that NR4A1 was down-regulated in follicular carcinoma {Borup et al. 2010, submitted).

To exploit if the miRNAs could be useful to depict follicular carcinoma from adenoma, we have build a diagnostic classifier. Employing a SVM algorithm, we discovered that two miR-1826 and hsa-miRPIus-E1078 were sufficient to differentiate between the two groups. Both miRNAs are down-regulated in FC compared to FA. Their fold change is moderate and they are not exclusively expressed in the FCs, but in contrast to the majority of miRNAs their expression is relatively uniform in the two groups. Although the results should be confirmed in independent studies, the performance is acceptable, since the classifier exhibited a negative predicted value of 83% for malignancies.

From a clinical point of view, the predicted probabilities derived from each individual sample is essential, since it provides a quantitative value (on the reliability) to the extent in the accuracy of the diagnosis based on the classification. We find that most samples are classifed with high accuracy. It is evident that our two-classifier does not separate the minimal invasive carcinomas from the follicular adenomas. We therefore looked at the carcinomas exclusively and listed the miRNAs with the maximum diversity in these small subgroups. Only 6 miRNAs were differently expressed with a cut-off of p < 0.01 and only two miRNAs miR-410 (5-fold down in widely invasive) and hsa-hsa-miRPIus-E1001 (3-fold up in widely invasive) were changed more than 2 fold. Taking the limited sensitivity and reproducibility of the histopathological diagnosis into account the consistency between miRNA based classification and the pathological „.

84 diagnosis is surprisingly high. This could reflect the fact that all samples were examined by the same endocrine pathologist. Among the four mis-classified samples one exhibit a predictive probability of 0.9 for FC and it is therefore possible that this tumour could be wrongly categorized by the pathologist. Studies of the inter-observer variations amongst pathologist in assessment of follicular lesions have demonstrated an observer variation for follicular carcinoma of 27%, where the carcinomas tended to be misdiagnosed as adenomas (38;39). In a similar study an overall agreement among American and Japanese pathologist of 33% and 52%, respectively, was found (39). Taking the limitation of the current diagnostic procedures into account, we trust that miRNA-based diagnosis may carry the potential to improve diagnosis of follicular cancers.

Taken together, we conclude that thyroid follicular neoplasia is accompanied by major changes in miRNA expression that may be directly implicated in transformation and may facilitate diagnosis of follicular thyroid cancer.

TABLE 1 (ex.1 ):

Tumour size Invasiveness

Diagnosis Age (years) Sex (M/F) (_cm; max (minimally vs.

diameter) widely)

FC_01_401 32 F 6 minimal

FC_02_402 60 F 2.5 minimal

FC_03_406 32 F 1.5 minimal

FC_04_407 48 F 5 widely

FC_05_410 41 M 6 minimal

FC_06_413 75 F 4 minimal

FC_07_414 35 M 7 minimal

FC_08_417 59 F 2 widely

FC_09_421 28 M 2.5 minimal

FC_10_422 46 F 10.5 widely

FC_11_423 32 F 4 minimal

FC 12 424 63 F 3.5 minimal FA..01..403 55 F 3.5 NA

FA. .02. .404 41 M 4 NA

FA. .03. .405 65 F 4 NA

FA. .04. .408 47 F 2.5 NA

FA. .05. .409 63 F 4.5 NA

FA. .06. .411 33 F 2 NA

FA. .07. .412 37 F 3.5 NA

FA. .08. .415 46 M 3 NA

FA. .09. .416 37 F 2.5 NA

FA. .10- .418 51 F 4 NA

FA. .11. .419 65 F 4 NA

FA. .12- .420 37 ^F 4 NA

Table 1. Clinical data of patients with thyroid follicular neoplasia.

12 patients with histopathological verified follicular carcinomas (FC), (minimal and widely invasiveness), and 12 patients with follicular adenomas (FA). Non applicable (NA). The table depicts diagnosis, age, sex, tumour size, and the invasiveness of the examined tumours.

TABLE 2 (ex.1 ):

miRNA Fold change p-value Adjusted p-value hsa-mi -199b-5p -39 3,15E-31 5,98E-28 hsa-miR-144* -15,1 7,10E-09 2,50E-07 hsa-miR-199a-3p/hsa-

-12,1 l,33E-28 l,26E-25 miR-199b-3p

hsa-miR-199a-5p -9 1,66E-14 1,97E-12

FC vs NT hsa-miR-144 -6,9 3,34E-21 1,06E-18

hsa-miR-221 3,1 l,12E-05 0,0001 hsa-miR-96 2,9 5,92E-06 8,72E-05 hsa-miR-182 2,6 2,llE-06 3,46E-05 hsa-miR-597 2,4 0,0020 0,0114 hsa-miR-222 2,3 0,0007 0,0049

FA vs NT hsa-miR-199b-5p -26 4,69E-27 8,92E-24

hsa-miR-144* -8,1 6,76E-06 0,0001 hsa-miR-663 -6,2 5,48E-12 4,51E-10 hsa-miR-199a-3p/hsa-

-6,1 1,26E-18 4,80E-16 miR-199b-3p

hsa-miR-142-3p -4,6 2,34E-15 4,46E-13 ebv-mi -BA T8 5,3 0,0083 0,0489 hsa-miR-517a 4,5 0,0043 0,0285 hsa-miR-517b 4 0,0003 0,0026 hsa-miR-517c 3,9 0,0002 0,0017 hsa-miR-512-3p 3,6 0,0017 0,0137 hsa-miR-512-3p -3,2 0,0032 0,0395 hsa-miR-886-5p -3 2,58E-06 0,0003 hsa-miR-450a -3 6,92E-07 0,0001 hsa-miR-301b -2,6 5,78E-05 0,0032 hsa-miR-429 -2,4 0,0001 0,0046 hsa-miR-637 2,1 7,09E-05 0,0036

FC vs FA hsa-miR-631 2 0,0028 0,0364

hsa-miR-219-2-3p 1,8 0,0041 0,0467 hsa-miR-662 1,8 0,0017 0,0272 hsa-miR-744 1,8 2,21E-07 7,00E-05

Table 2. Dysregulated miRNAs.

The 5 most up- and down-regulated miRNAs in the three comparisons, follicular carcinomas (FC) vs. normal thyroid (NT), follicular adenomas (FA) vs. NT, and FC vs. FA are listed. MiRNAs were defined to be differentially expressed if they had a Benjamini-Hochberg corrected p-value below 0.05 and an absolute fold change above 1 .5. For the complete list please refer to Table 5.

TABLE 3 (ex.1 ):

Total target

miRNA seed-sites mRNA seed-sites

Up-regulated Down-regulated Up-regulated Down-regulated

FC vs. FA 108 98 12 91*

FA vs. NT 302 131 62 166

FC vs. NT 1451 270 53 1225*

Table 3. Representation of predicted miRNA seed-sites in differentially expressed mRNAs.

Differentially expressed transcripts from the three comparisons follicular carcinomas (FC) vs. normal thyroid (NT), follicular adenomas (FA) vs. NT, and FC vs. FA, were derived as described, and miRNAs were combined with predicted seed sites among the differentially expressed transcripts using Targetscan in the PARTEK miRNA - mRNA integration software. The number of computer predicted seed-sites in up or down-regulated mRNAs are shown in the two first columns and the number of corresponding differentially expressed miRNAs are shown in the two right columns. The relative overrepresentation in the present predictions in comparison to random occurrence. The asterisks indicate p<0.01 in a Fishers Exact Test, assuming that the total number of 904 miRNA can target 22.000 transcripts at a global level. TABLE 4 (ex.1 ):

Sample SVM predicted value

FA_01_403 0.729

FA_02_404 0.661

FA_03_405 0.836

FA_04_408* 0.445

FA_05_409 0.900

FA_06_411 0.936

FA_07_412 0.934

FA_08_415 0.874

FA_09_416 0.961

FA_10_418 0.798

FA_11_419 0.500

FA_12_420* 0.098

FC_01_401 0.043

FC_02_402 0.200

FC_03_406 0.074

FC_04_407 0.407

FC_05_410 0.182

FC_06_413* 0.672

FC_07_414 0.133

FC_08_417* 0.620

FC_09_421 0.300

FC_10_422 0.463

FC_11_423 0.223

FC 12 424 0.082 Table 4. Predictive probabilities of miRNA classifier.

Predictive probabilities for follicular adenoma (FA) and follicular carcinoma (FC) samples using SVM classifier, Cost = 10, gamma = 0.01 . ^*Misclassified samples n=2, FC: n=2).

TABLE 5 (ex.1 ):

Name Fold change Adjusted p-va

hsa-miR-199a-3p/ sa-miR-199b-3p -12.15 1.26E-25 hsa-miR-199a-5p -8.95 1.97E-12 hsa-miR-144 -6.89 1.06E-18 hsa-miR-1275 -6.66 3.42E-17 hsa-miR-153 -6.54 0.032753 hsa-miR-451 -5.94 3.54E-17 hsa-miR-142-3p -5.89 8.54E-17 hsa-miR-886-5p -5.37 3.51E-09 hsa-miR-31 -5.05 5.53E-08 hsa-miR-455-3p -4.28 3.49E-09 hsa-miR-663 -4.04 1.46E-06 hsa-miR-218 -3.79 1.15E-09 hsa-miR-486-5p -3.44 2.68E-09 hsa-miR-100 -3.34 9.10E-06 hsa-miR-542-5p -3.28 0.009873 hsa-miR-1 -3.27 0.017907 hsa-mi -101 -3.27 6.77E-11 hsa-miR-20a -3.22 1.46E-06 hsa-miR-193a-3p -3.17 1.20E-07 hsa-miR-223 -3.06 7.43E-07 hsa-miR-886-3p -3.05 0.012755 hsa-miR-18b -2.97 7.70E-06 hsa-miR-190 -2.90 0.000142 hsa-miR-99a -2.87 9.24E-05 hsa-miR-422a -2.86 2.06E-07 hsa-miR-365 -2.85 7.15E-07 hsa-miR-943 -2.84 2.88E-12 hsa-miR-18a -2.79 7.50E-09 hsa-miR-299-5p -2.77 3.20E-07 hsa-miR-26a -2.74 2.82E-08 hsa-miR-106a -2.69 1.81E-05 hsa-miR-17 -2.68 3.44E-07 hsa-miR-708 -2.67 0.001686 hsa-miR-455-5p -2.61 8.67E-06 hsa-miR-27a -2.59 2.48E-10 hsa-miR-130a -2.58 2.30E-06 hsa-miR-143 -2.53 0.000786 hsa-miR-429 -2.47 0.00158 hsa-miR-138 -2.46 0.001005 hsa-miR-876-5p -2.42 0.00101 hsa-miR-1297 -2.41 0.013685 hsa-miR-202 -2.37 1.93E-09 pa-mi R-33a -2.36 1.60E-05 hsa-miR-101* -2.34 8.71E-06 hsa-let-7i* -2.32 9.10E-05 hsa-miR-675 -2.30 3.62E-13 hsa-miR-125b -2.28 0.001714 hsa-miR-193b -2.28 0.04847 hsa-miR-204 -2.24 0.005259 hsa-miR-326 -2.22 4.98E-08 ebv-miR-BART6-3p -2.22 0.000698 hsa-miR-142-5p -2.22 6.15E-06 hsa-miR-146b-5p -2.21 0.010132 hsa-miR-140-5p -2.20 4.85E-06 hsa-miR-1184 -2.18 1.17E-10 hsa-miR-lOa -2.17 1.97E-07 hsa-miR-23a -2.16 0.009847 hsa-let-7g -2.09 0.000811 hsa-let-7i -2.07 0.001572 hsa-let-7d -2.03 0.000279 hsa-miR-19a -2.01 4.04E-05 hsa-miR-139-5p -2.01 0.000189 hsa-miR-638 -2.01 7.49E-09 hsa-miR-374a -1.99 0.000356 hsa-miR-19b -1.95 8.20E-05 hsa-miR-124* -1.95 3.20E-07 hsa-miR-450a -1.94 0.017663 hsa-miR-126* -1.92 0.00058 hsa-miR-133a -1.89 0.005376 hsa-miR-181c -1.89 0.032988 hsa-miR-873 -1.88 0.001863 hsa-miR-342-3p -1.86 0.000278 hsa-miR-611 -1.86 0.001167 hsa-miR-186 -1.84 0.004731 hsa-miR-508-5p -1.82 0.000143 hsa-miR-30b -1.78 0.022545 hsa-miR-374b -1.78 0.008055 hsa-miR-146a -1.77 0.043355 hsa-miR-195 -1.77 0.003215 hsa-miR-30d -1.72 0.028035 hsa-miR-660 -1.72 0.029816 hsa-miR-557 -1.70 4.93E-05 hsa-miR-509-3-5p -1.70 7.48E-05 hsa-miR-23b -1.69 0.000124 hsa-miR-363 -1.67 0.043737 hsa-miR-lOb -1.66 0.037252 hsa-let-7a -1.66 0.001533 hsa-mi^"R-92a -1.65 0.011147 hsa-let-7a* -1.65 0.037235 hsa-miR-16 -1.65 0.005259 hsa-miR-361-3p -1.64 0.000419 hsa-miR-145* -1.64 0.007196 hsa-miR-92b -1.62 0.03661 hsa-miR-28-5p -1.61 0.012155 ebv-miR-BHRFl-1 -1.61 9.79E-05 hsa-miR-29a* -1.60 0.036469 hsa-miR-30a* -1.58 0.012152 hsa-miR-30a -1.58 0.04409 hsa-miR-30c -1.58 0.013492 ebv-miR-BHRFl-2 -1.55 0.000182 hsa-miR-21* -1.55 0.026284 hsa-miR-299-3p -1.54 0.038226 hsa-miR-151-5p -1.53 0.03385 hsa-miR-24 -1.53 0.000473 hsa-miR-126 -1.53 0.04573 hsa-miR-513b -1.52 0.001558 hsa-miR-193b* -1.52 0.000182 hsa-miR-424 -1.52 0.01711 hsa-miR-15b -1.52 0.02452 ebv-miR-BART19-3p -1.51 0.004133 hsa-miR-602 i!ii!ii! -1.51 0.00502 hsa-miR-125b-2* 1.50 0.000693 hsa-miR-200c* 1.51 0.000476 hsa-miR-377 1.52 0.006313 hsa-miR-1304 1.52 0.011085 hsa-miR-1301 1.53 9.86E-06 hsa-miR-298 1.53 5.46E-05 hsa-miR-934 1.55 0.000116 hsa-miR-24-1* 1.56 5.62E-05 hcmv-miR-UL36 1.56 0.005348 hsa-miR-1827 1.57 0.001712 hsa-miR-647 1.60 2.60E-06 hsa-miR-34a 1.60 0.007845 hsa-miR-374b* 1.61 3.01E-06 hsa-miR-574-3p 1.62 0.000124 hsa-miR-220c 1.62 0.001223 hsa-miR-635 1.62 0.001415 hsa-miR-197 1.63 0.010352 hsa-miR-1274b 1.64 0.006094 hsa-miR-1255a 1.65 1.25E-05 hsa-miR-1298 1.66 0.031991 hsa-miR-1248 1.68 0.008192 hsa-miR-629* 1.68 0.001712 hsa-miR-744 1.72 9.03E-05 hsa-miR-22* 1.73 1.41E-11 hsa-miR-340 1.73 0.000402 hsa-miR-634 1.76 2.31E-05 hsa-miR-600 1.81 0.011618 hsa-miR-34b 1.83 5.39E-11 hsa-miR-129* 1.89 0.012755 hsa-miR-138-1* 1.94 9.18E-09 hsa-miR-637 2.14 0.000935 hsa-mi^'R-215 2.17 0.002822 hsa-miR-222 2.32 0.004918 hsa-miR-597 2.35 0.011395 hsa-miR-182 2.64 3.46E-05 hsa-miR-96 2.86 8.72E-05 hsa-miR-221 3.11 0.000142 hsa-miR-199b-5p -26.02 8.92E-24

NT

hsa-miR-144* -8.13 0.00011 hsa-miR-663 -6.24 4.51E-10 hsa-miR-199a-3p/hsa-miR-199b-3p -6.05 4.80E-16 hsa-miR-142-3p -4.57 4.46E-13 hsa-miR-1275 -4.55 4.09E-12 hsa-miR-199a-5p -4.32 1.69E-06 hsa-miR-144 -4.02 2.70E-11 hsa-miR-31 -4.00 3.46E-06 hsa-miR-631 -3.57 1.39E-05 hsa-miR-422a -3.56 4.92E-10 hsa-miR-451 -3.43 1.05E-09 hsa-miR-218 -2.99 5.22E-07 hsa-miR-943 -2.90 1.85E-12 hsa-miR-675 -2.76 1.71E-17 hsa-miR-708 -2.73 0.001692 hsa-miR-486-5p -2.72 1.39E-06 hsa-miR-492 -2.60 6.97E-07 hsa-miR-299-5p -2.58 2.02E-06 hsa-miR-326 -2.52 4.50E-10 hivl-miR-Hl -2.47 0.034265 hsa-miR-455-3p -2.47 0.000262 hsa-miR-202 -2.43 8.41E-10 ebv-miR-BART6-3p -2.42 0.000182 hsa-miR-20a -2.40 0.000411 hsa-miR-142-5p -2.38 8.86E-07 hsa-miR-508-5p -2.31 1.03E-07 hsa-mi^'R-100 -2.29 0.003571 hsa-miR-17 -2.24 3.25E-05 hsa-miR-490-5p -2.24 0.000749 ebv-miR-BART20-3p -2.23 0.001037 hsa-miR-638 -2.23 6.19E-11 hsa-miR-106a -2.22 0.000785 hsa-miR-223 -2.20 0.000698 hsa-miR-18a -2.19 1.05E-05 hsa-mi^'R-101* -2.17 5.73E-05 hsa-miR-18b -2.12 0.003077 hsa-miR-101 -2.11 3.33E-05 hsa-miR-124* -2.07 3.44E-08 hsa-miR-193b* -2.04 3.73E-10 hsa-miR-1303 -2.02 0.027106 hsa-miR-lOa -1.98 4.69E-06 hsa-miR-99a -1.96 0.022749 hsa-miR-557 -1.95 3.55E-07 hsa-miR-376c -1.92 4.69E-05 hsa-miR-125b -1.91 0.023209 hsa-miR-lOb -1.89 0.008165 hsa-miR-936 -1.88 0.000207 hsa-miR-190 -1.88 0.047 hsa-miR-611 -1.85 0.00164 hsa-miR-135a* -1.84 0.001306 hsa-miR-455-5p -1.83 0.008819 hsa-miR-602 -1.80 3.79E-05 hsa-miR-24-2* -1.78 0.008217 hsa-miR-26a -1.78 0.002293 hsa-miR-1184 -1.70 1.03E-05 hsa-miR-516a-5p -1.68 0.000695 hsa-miR-27a -1.65 0.001027 hsa-miR-585 -1.65 4.89E-05 hsa-miR-198 -1.64 7.91E-07 hsa-miR-873 -1.63 0.028521 ebv-miR-BHRFl-2 -1.61 4.89E-05 hsa-miR-lOa* -1.61 0.002289 hsa-miR-192 -1.59 0.026413 ebv-miR-BART8* -1.50 0.01913 hsa-miR-520d-5p 1.51 0.005728 hsa-miR-24-1* 1.53 0.000137 hsa-miR-1201 1.53 0.033395 hsa-miR-200b 1.54 0.01882 hsa-miR-1264 1.58 1.33E-08 hsa-miR-1259 1.59 0.004105 hsa-miR-130b 1.64 0.00014 hsa-miR-22* 1.65 4.92E-10 hsa-miR-363* 1.70 0.002369 hsa-miR-220a 1.71 0.015394 hsa-miR-548e 1.71 9.04E-10 hsa-miR-138-1* 1.73 2.47E-06 hsa-miR-1274a 1.76 0.000244 hsa-miR-148b 1.77 0.003749 hsa-miR-339-3p 1.78 0.029611 hsa-miR-542-3p 1.78 0.000886 hsa-miR-141 1.79 0.000117 hsa-miR-1826 1.83 4.51E-10 hsa-miR-34b 1.86 2.70E-11 hsa-miR-34a* 1.87 0.023541 hsa-miR-1227 1.87 0.01036 hsa-miR-887 2.02 2.10E-05 hsa-miR-20b* 2.06 0.041742 hsa-miR-200c 2.14 1.35E-06 hsa-miR-34a 2.22 3.01E-06 hsa-miR-518e 2.27 0.026774 hsa-miR-22 2.36 2.09E-06 hsa-miR-526b* 2.39 0.02499 hsa-miR-1323 2.40 0.027151 hsa-miR-182 2.40 0.00023 hsa-miR-1274b 2.44 4.49E-07 hsa-miR-597 2.48 0.008812 hsa-miR-96 2.51 0.000785 hsa-miR-522 2.52 0.022089 hsa-miR-518a-3p 2.60 0.039372 hsa-miR-301b 3.20 8.55E-05 hsa-miR-520h 3.39 0.01342 hsa-miR-512-3p 3.64 0.01369 hsa-miR-517c 3.90 0.001735 hsa-miR-517b 4.02 0.002552 hsa-miR-517a 4.54 0.028529 ebv-miR-BART8 5.32 0.048905 hsa-miR-512-3p -3.18 0.039532 hsa-miR-886-5p -3.03 0.00035 hsa-miR-450a -3.01 0.00012 hsa-miR-301b -2.64 0.003225 hsa-miR-429 -2.44 0.004625 hsa-miR-542-3p -2.40 1.06E-06 hsa-miR-130a -2.34 9.90E-05 hsa-miR-146b-5p -2.25 0.017386 hsa-miR-199a-5p -2.07 0.039532 hsa-miR-193a-3p -2.07 0.003065 hsa-miR-152 -2.02 0.021238 hsa-niiR-199a-3p/hsa-miR-199b-3p -2.01 0.003225 hsa-miR-424 -1.96 0.000196 hsa-miR-22 -1.96 0.000811 hsa-miR-146a -1.95 0.026685 hsa-miR-339-3p -1.94 0.014545 hsa-miR-365 -1.92 0.005591 hsa-let-7i *^" -1.92 0.00661 hsa-miR-363* -1.87 0.000743 hsa-miR-148a -1.87 0.046653 hsa-miR-299-3p -1.87 0.003713 hsa-let^a ¹- -1.84 0.016714 hsa-miR-200b -1.79 0.00193 hsa-miR-200c -1.78 0.00089 hsa-miR-375 -1.74 0.03496 hsa-mi -451 -1.73 0.017386 hsa-miR-144 -1.72 0.022507 hsa-let-7i -1.68 0.049082 hsa-miR-1826 -1.68 6.29E-07 hsa-miR-1201 -1.66 0.012734 hsa-miR-140-5p -1.65 0.01035 hsa-miR-126 -1.64 0.028516 hsa-miR-126* -1.63 0.024186 hsa-let-7f-2 * -1.62 0.025923 hsa-miR-148b -1.61 0.027895 sa-miR-21^1' -1.59 0.030257 hsa-mi -342-3p -1.59 0.018602 hsa-miR-27a -1.56 0.007511 hsa-miR-145* -1.56 0.031996 hsa-miR-513b -1.56 0.002828 hsa-miR-101 -1.55 0.036119 hsa-miR-26a -1.54 0.039532 hsa-miR-24 -1.53 0.001766 hsa-miR-30a *^■ -1.53 0.038 hsa-miR-377 1.53 0.012115 hsa-miR-518e 7 sa-miR-519a 7 1.55 0.027164

|-|sa-miR-519b-5p/|-isa-miR-519c-5p/

sa-niiR-522 VI'isa-miR-523 ^

hsa-miR-222* 1.56 0.023641 sa-miR-452 1.57 0.000844 hsa-miR-665 1.59 0.039099 sa-niiR-584 1.67 0.00661 hsa-miR-492 1.80 0.00605 hsa-miR-744 1.82 7.00E-05 hsa-miR-662 1.83 0.027164 hsa-miR-219-2-3p 1.84 0.046653 hsa-miR-631 2.03 0.036385 hsa-miR-637 2.10 0.003551

Table 5: Up- and down-regulated miRNAs

Complete list of up- and down-regulated miRNAs in the three comparisons, follicular carcinomas (FC) vs. normal thyroid (NT), follicular adenomas (FA) vs. NT, and FC vs. FA are listed. miRNAs were defined to be differentially expressed if they had a

Benjamini-Hochberg corrected p-value (adjusted p-value) below 0.05 and an absolute fold change above 1 .5.

Example 2: Down-Regulation of microRNAs Controlling Cell Proliferation in Follicular

Thyroid Carcinoma

Abstract

The molecular determinants of thyroid follicular nodules are incompletely understood and the preoperative assessment of malignancy is consequently a diagnostic challenge. Since microRNAs (miRNAs) are frequently changed during malignant progression, the identification of differentially expressed miRNAs could provide new insight into follicular neoplasia. We characterized the global miRNA expression in normal thyroid, follicular adenoma, and follicular carcinoma by microarray and qRT-

PCR analyses. A number of miRNAs, including miR-199b-5p, miR-144^*, miR-199b-3p, miR-199a-5p, and miR-144, were strongly down-regulated in the malignant nodules and integration of perturbed miRNAs with differentially expressed mRNAs

demonstrated a significant enrichment of seed-sites among transcripts encoding proteins involved in thyroid carcinogenesis and cell cycle control. Since the perturbed miRNAs could be implicated in the malignant progression, we employed supervised learning algorithm support vector machine and leave-one-out-cross-validation, to examine if particular miRNAs could distinguish benign and malignant nodules. Using data from the microarray platform, follicular adenoma and carcinoma could be identified with a negative predicted value of 83% for malignancies, and with data generated by the qRT-PCR-platform, classification of follicular neoplasia was feasible with a NPV of 92% for malignancies. We conclude that thyroid follicular neoplasia is accompanied by major changes in the expression of a number of miRNAs that may be implicated in malignant transformation by targeting transcripts encoding factors involved in cell cycle control. Moreover, miRNAs may be used to distinguish carcinoma from adenoma.

Introduction

Thyroid nodules are found in up to 7% of the adult population (Hegedus et al., 2003). Although the majority of the nodules are benign, carcinoma of the thyroid gland is the most common malignancy of the endocrine system (Curado and Edwards, 2007). Follicular adenomas are benign, encapsulated tumours and they are 5 times more frequent than follicular carcinomas (Faquin, 2008). Follicular carcinomas mainly occur in middle-aged euthyroid women as a painless thyroid nodule and represent 10-15% of all thyroid malignancies (Faquin, 2008). Follicular adenoma and carcinoma are morphologically similar and carcinomas are distinguished by vascular and/or capsular invasion (Schmid and Farid, 2006). Since the latter features may be overlooked it is generally accepted, that application of particular biomarkers could improve diagnosis (Franc et ai, 2003).

MicroRNAs (miRNAs) are non-coding single-stranded RNAs of about 22 nucleotides. MiRNAs regulate translation and stability of particular target messenger RNAs (mRNA) by imperfect base pairing with the mRNAs (Bartel, 2004) and it is estimated that the number of miRNAs may exceed 1000 (http://microrna.sanqer.ac.uk/). In this way miRNAs regulate about one third of the mammalian protein-coding mRNAs (Bartel, 2009;Friedman et al., 2009). The expression of miRNAs is temporally and spatially regulated. Many are important for the differentiation processes during particular developmental stages, but miRNAs also exhibit important functions in mature cells (Schmittgen, 2008). MiRNAs are moreover aberrantly expressed or lost in a variety of cancers (Rosenfeld et al., 2008). Many target-mRNAs encode oncogenes and tumoursuppressors and in this way dysregulated miRNAs may play a causal role in malignant progression. Not surprisingly miRNAs are therefore considered attractive candidates for classification of tumours. The role of miRNAs in thyroid cancer is incompletely understood. A number of miRNAs have previously been characterized in various thyroid tumours (He et al., 2005;Pallante et al., 2006;Weber et al., 2006;Chen et al., 2008;Nikiforova et al., 2008). MiR-197 and miR-346 were found to be over- expressed in follicular carcinoma in comparison to adenoma and in vitro studies suggested that both miRNAs could have a significant impact on tumour cell proliferation (Weber et al., 2006).

In the present study we employed global miRNA analysis and found a number of not previously shown miRNAs to be differentially expressed in carcinoma, adenoma, and between carcinoma and adenoma. Perturbed miRNAs were integrated with

differentially expressed mRNAs and this demonstrated a significant enrichment of seed-sites among transcripts encoding proteins involved in thyroid carcinogenesis and cell cycle control. Moreover, we find that miRNA signatures may distinguish adenoma from carcinoma with negative predicted value of 83% - 92% for malignancies depending on the technical platform. The results indicate that miRNAs may be implicated in follicular neoplasia.

Materials and methods

Thyroid tissue, follicular adenoma and carcinoma The thyroid samples originated from a consecutive series of patients with a solitary or prominent scintigraphically cold thyroid nodule operated on, and with a histological diagnosis of a follicular adenoma or follicular carcinoma. Since a uniform

histopathological evaluation was essential, the diagnosis was made by a particular pathologist specialized in thyroid pathology. All tumours were diagnosed and classified according to the WHO definition of histological criteria. Clinical data are listed in Table 1 (ex.2). Surgically removed thyroid samples were snap frozen at the Department of Pathology and stored at -120° over a 5-year period. The project was approved by the Ethics Committee and informed consent was obtained in all patients. Twelve follicular adenomas (FA), 12 follicular carcinomas (FC), and 10 normal thyroid (NT) specimens were included. NT specimens were obtained by a thyroid pathologist to ensure that the tissue derived from macro-and microscopically normal tissue adjacent to the encapsulated tumours. The number of tumours was balanced to provide optimal power estimates and a similar number of samples in each diagnostic category.

TotalRNA and DNA Isolation

TotalRNA were isolated from frozen samples using Trizol, Invitrogen. Purified RNA was subsequently quantified on a NanoDrop^® ND-1000 Spectrophotometer (NanoDrop Technologies) and examined on a Bioanalyzer Nano RNA Chip (Agilent). DNA from tissues was isolated by lysing the tissue in 20ul proteinase K and 200ul

Tris/NaCI/EDTA/SDS (TNES), followed by overnight incubation at 55C. Finally, 5M NaCI is added to the lysed tissues and DNA is precipitated by adding 200ul ice-cold 96% ethanol. Detection of point mutations

Mutation analyses, including PCR-amplification and sequencing, were performed on all 24 tumour samples for KRAS and BRAF (V600E and K601E). Primer sequences for KRAS, forward; 5'- TGTAAAACGACGGCCAGTCGATACACGTCTGCAGTCAA - 3' and reverse; 5'- CAGGAAACAGCTATGACCCTCATGAAAATGGTCAGAGA - 3', BRAF, forward; 5' - TGCTTGCTCTGATAGGAAAATG - 3' and reverse; 5' - GGATGGTAAGAATTGAGGCT - 3'. miRNA expression analyses

miRNA expression levels were determined by microarray analysis. One microgram of totalRNA was labeled with fluorescent Hy3™(sample)/Hy5™(reference-sample) dye from the miRCURY LNA™microRNA Array Power Labeling Kit (Exiqon) in accordance with the manufacturer's instructions. Using a TECAN hybridization station, labeled samples were hybridized overnight to pre-printed miRCURY LNA™ microRNA Array, v.1 1 .0 (Exiqon), containing probes for 841 human miRNAs, cataloged in the miRBase Sequence Database (Release 1 1 .0) ( HttD://microrna.sanaer. ac. uk/) . and 428 proprietary human miRPIus sequences not yet annotated in miRBase.

Furthermore, miRNA expression levels from 30 thyroid specimens (12 FC, 12 FA, and 6 NT), were generated using the miRCURY LNA™ Universal RT miRNA PCR panel I and II, V2, (Exiqon). 40ng of total RNA was reversed transcribed using the Universal cDNA synthesis kit, mixed with SYBR® Green master mix kit, and subsequently added to the pre-aliquoted miRNA PCR primer sets in two 384-well PCR plates enabling profiling of 742 human microRNAs. All reagents were from Exiqon and their recommendations were followed. Each plate contained an additional six primer sets for reference miRNAs and a set of negative controls. The amplification curves were analyzed using the Roche LC software, both for determination of Cp (Cross-over Point) and for melting curve analysis. 135 miRNA assays were successfully assessed with sufficient signal (Cp < 37, or 5 Cp less than negative control) in all samples.

Image analysis and normalization of miRNA expression

Devices). Background intensities were subtracted from foreground intensities and within array LOESS - normalized, followed by aquantile normalization between arrays as implemented in the Limma package in Bioconducter library (Gentleman et al., 2004) in R version 2.10 ( 2007). Mean values of the quadruplicate probes for each of the miRNAs were obtained and log2 ratios between sample and reference were used for further analyses.

Normalization of the results derived from the RT miRNA PCR panels is performed based on the average of the assays detected in all samples; Normalized Cp = Average samples Cp (n= 135) - assay Cp. The normalized miRNA expression values were used for generating a diagnostic classifier between FC and FA as described in "Construction of classifier". Class comparison analyses

Class comparison was performed using the Limma package. For each miRNA standard errors and fold changes were estimated by fitting a linear model. Differentially expressed miRNAs were determined by applying empirical Bayes moderated t- statistics test (Wettenhall and Smyth, 2004). MiRNAs were defined to be differentially expressed if they had a Benjamini-Hochberg corrected p-value below 0.05 and an absolute fold change above 1 .5. mRNA expression data and miRNA target predictions followed by Pathway analysis The Global mRNA expression data was based on previous work by our group (Borup et ai, 2010) and on the Gene Expression Omnibus (GEO), ID: E-GEOD-6004 and E- GEOD-7307. To predict mRNA targets, based on the observed miRNA profiles, we implemented the method developed by Gallagher et al. (Gallagher et ai, 2010). Briefly, mRNA target predictions were based on the TargetScan miRNA target prediction database in combination with the observed changes in miRNAs. Only miRNAs that exhibited an absolute change >1 .5 fold and mRNA with an average expression intensity >40 in FCs were included in the analysis. This method provides a weighted miRNA inhibitor score (sum of effects), predicting the transcripts, most likely to be regulated by miRNAs.

To examine whether these likely mRNA targets were associated to certain biological functions or pathways we analysed the most significant predicted mRNAs with

Ingenuity Pathway Analysis software package (Ingenuity Systems, www.inqeuitv.com). We set a cut-off at +/- 2.2 on our weighted inhibitor score list, resulting in 953 genes that were added to the pathway analysis. The most significant pathways were identified by the Fisher's exact test, which is a measure of the probability that the selected genes are associated with a pathway by chance alone. Heatmaps were generated as a supervised cluster analysis.

Construction of classifier

Normalized data from 24 samples consisting of the expression level of 545 different probes (377 annotated miRNAs and 168 hsa miRPIus, all with an A-value > 7, A = ½(log₂R + log₂G)) were included in the analysis. The classifier was constructed in order to classify FC and FA. The validation of the classifiers performance on the training set was done by leave-one-out (LOO) cross validation (CV), that provides an unbiased estimate of the prediction error by going over all training examples in turn using the complement training set (of size 24-1 = 23) to perform training (including probe ranking, selection and model fitting) and using the last left out sample for prediction. In this way, we obtained as many predictions as there were samples in the training set. These predictions formed unbiased estimates of the prediction error quantified in terms of the confusion matrix. The training of the classifiers inside the LOO loop consisted of a univariate selection step followed by applying support vector machine learning (SVM). Probes were ranked according to their differential expression using a Student t-test with a threshold of p < 0.001 (Dudoit S, 2003). Quantitative Reverse transcription PCR

CDNA was prepared from 25ng total RNA from 34 tumour samples using TagMan^® MicroRNA Reverse Transcription Kit and TagMan^® MicroRNA Assays containing predesigned primers for miR-221 , miR-182, miR-96, miR-199a3p, miR-144^*, miR- 199b5p, and miR-1826 was added. miR-191 was used for endogenous control.

Quantitative reverse transcription PCR (qRT-PCR) reaction was performed using

TagMan^® Universal PCR Master Mix No AmpEras^® UNG, according to manufactures instructions, all from Applied Biosystems. Each amplification reaction was performed in triplicate, and median value of the three cycle threshold was used for further analyses. For calculations of fold changes we used the 2^"Δ0 _Τ method (Schmittgen and Livak, 2008). Additionally, we validated Exiqon miRPIus probe, miRPIus-E-1078, used for classification of follicular carcinoma and adenoma. For this qRT-PCR analysis we employed the Exiqon microRNA LNA™ PCR primer sets together with Universal cDNA Synthesis and SYBR® Green master mix, following the manufactures instructions (Exiqon).

Results

Tumour samples and oncogenic mutations

The thyroid samples originated from a consecutive series of patients and included tumours from 19 women and 5 men and all tumours were classified according to the WHO definition of histological criteria. The median age was 44 years in carcinoma patients and 47 years in adenoma patients. The size of the tumours ranged from 1 .5 to 10.5 cm and the median diameter was 4 cm in the carcinoma patients and 3.75 cm in the adenoma patients (Table 1 (ex.2)). All tumours were examined for KRAS and BRAF mutations and this showed that only one carcinoma sample was positive for BRAF (Table 1 (ex.2)). Taken together, the thyroid specimens in the two diagnostic groups were comparable both with respect to the clinical features and the presence of possible oncogenic mutations.

MiRNA expression in follicular carcinoma and adenoma

The following class comparison analysis and miRNA target analysis are based on the derived microarray expression data since this platform counts the highest number of miRNAs. Class comparison analysis revealed 150 annotated and differentially expressed human miRNAs - 37 up-regulated and 1 13 down-regulated miRNAs - in FCs compared to NT. The fold change ranged from 3.1 to -39 fold. Due to the substantial 39 fold down-regulation of miR-199b-5p, this miRNA is essentially lost in FC. MiR-144^*, miR-199b-3p, miR-199a-5p, and miR-144, were also strongly reduced to almost background. Among the most up-regulated miRNAs, miR-221 , miR-96, and miR-182 exhibited fold changes of 3.1 , 2.9, and 2.6, respectively. The comparison of FA to normal thyroid tissue revealed 107 differentially expressed miRNAs. Forty two were up- regulated and 65 down-regulated. Finally the comparison of carcinoma to adenoma showed that 56 miRNAs were differentially expressed between these groups. Twelve miRNAs were up-regulated and 44 were down-regulated in the carcinoma. The five most up- and down-regulated miRNAs in each comparison are listed in Table 2 (ex.2) and the complete list of miRNAs that changed more than 2 fold are listed in Table 5 (ex.2). Fifty-eight miRNAs were differentially expressed in both the carcinoma and adenoma compared to normal thyroid (Figure 1 , panel A). The overlap was marked among the down-regulated miRNAs, where 49 of the 65 down-regulated miRNAs in the adenoma were identical, compared to 9 out of 42 for the up-regulated miRNAs. With the exception of miR-199b-5p, that were progressively down-regulated in carcinoma, the shared miRNAs exhibited basically the same levels in adenoma and carcinoma compared to normal thyroid (Figure 1 , panel B). Taken together the results imply that changes in miRNA expression predominantly occur during transition from normal thyroid epithelial cell to adenoma and not during malignant transformation. Computational identification of putative target mRNAs and pathway analysis

Results from the global expression profiling of normal thyroid and follicular adenoma and carcinoma samples were used for integration of mRNA and miRNA array data. In a preliminary analysis we simply counted predicted seed sites corresponding to the perturbed miRNAs among the differentially expressed transcripts using Targetscan in PARTEK miRNA - mRNA integration software. Only mRNAs and miRNAs that exhibited an inverse expression pattern were considered. We observed a significant enrichment (Fishers Exact Test p<0.01 , assuming that the total number of 904 miRNA can target 22.000 transcripts at a global level) of seed-sites corresponding to down- regulated miRNAs. The differentially expressed and down-regulated miRNAs in the FC group exhibited putative seed-sites in almost 85% of the up-regulated transcripts, which distinguished carcinoma from normal thyroid and adenoma, respectively. In sum, this led us to assume that the changed miRNAs could have an impact on the observed changes in the transcriptome during progression to carcinoma.

To characterize biological pathways that could be regulated by the perturbed miRNAs, we subsequently employed the previously reported ranking system for mRNA target identification (Gallagher et al., 2010). We focused on the perturbed miRNAs in FC vs. NT. Upon miRNA target identification, we submitted the highest ranked transcripts to Geneontology's and pathway analysis in the Ingenuity Software. The majority of the putative targets mRNAs encoded factors involved in "cancer". When we explored the biological functions, 165 of these transcripts encoded factors linked to cell division process (Table 3 (ex.2)). The heatmap (Fig. 4) shows the relative expression of the predicted target mRNAs in FC and NT. Corresponding to the down-regulation of miRNAs, mRNAs encoding cell cycle factors, were almost entirely increased. Of the 165 mRNAs, 154 were significantly up-regulated (P<0.05) and the fold changes ranged from 2.4 to 18 fold. 12 transcripts were unchanged and 2 were up-regulated. Hence, up-regulation of mRNAs in thyroid carcinoma may at least partly be attributed by reduced levels of corresponding miRNAs. Within the grouping "Tumourigenesis" we observed the same pattern although not as stringent as in the "Cell-Cycle" grouping. However, associating these transcripts to possible biological functions, we found 49 with a significant enrichment in tumourigenesis (Fig. 4). Twenty-four of the 49 mRNAs were significantly upregulated (P<0.05). Finally, we considered putative miRNAs-199b- 5p targets, since this miRNA was completely lost in the carcinoma. Approximately 200 mRNAs expressed in thyroid tissue contained putative seed sites for miRNA-199b-5p. However, we only considered targets with a weighted cumulative context ranking score >80 and found 20 mRNAs with a high probability to represent true miRNA-199b-5p targets. Among these, 10 were significantly upregulated (P<0.05) in FC with fold changes ranging from 1 .2 to 3.9 (Figure 4). Transcripts encoding TSPAN6 and

PPP1R2 have previously been implicated in tumour cell proliferation (Takakura et al., 2001 ;Novitskaya et al., 2010). MiRNA based classification of follicular nodules

To provide an overview of the miRNA expressions across follicular neoplasia (FA and FC) and NT, we generated a PCA plot using all expressed miRNAs (Figure 5, panel A). At this stage the two populations could be discerned reflecting the relatively large and consistent differences in miRNA expression between the groups. Filtering the expression values by a t-test, we reached a subset of 179 miRNAs (p < 0.01 ), where follicular neoplasia could easily be distinguished from normal thyroid tissue. We subsequently attempted to separate follicular carcinoma from adenoma by employing a Student t-test for feature selection and applied the supervised learning algorithm support vector machine (SVM) and leave-one-out-cross-validation (LOOCV) to generate a classifier. The optimal signature for classification of FC and FA consist of two miRNAs, miR-1826 and miR-Eplus-1078, and based on expression values of the two classification miRNAs, a PCA plot was generated (Figure 5, panel B). Both the negative predictive value and the positive predictive value for carcinoma is 83%. The SVM can be turned into a probabilistic classifier giving an estimate of the probability of the predicted class label, i.e. assess the prediction uncertainty (Piatt J, 1999). The predictive probabilities for all samples are listed in Table 4 (ex.2). FA sample 1 1 , although correctly classfied exhibited a probability of 0.5 and the misclassfied FA sample 12 had a probalility for FC of 0.9 indicating that FA1 1 is highly uncertain, whereas FA12 is most likely a misdiagnosed carcinoma. It was not feasible to generate a classifier that could distinguish FA and minimally invasive carcinoma. Even so, widely invasive carcinoma can be distinguished from minimally invasive carcinoma by the expression of miRPIus-E1001 (P<0.01 ) (average 3 fold up-regulation) and lower expression of miRNA-410 (average 5 fold) compared to the minimally invasive carcinoma. Lastly, expression values of miR-1826, miR-Eplus-1078, miR-221 , miR-

182, miR-96, miR-199b-3p, miR-144^*, miR-199b-5p, were also examined by qRT-PCR and this confirmed the microarray results (Figure 3).

To validate whether miRNAs can classify thyroid follicular malignancies by a diverse method, we examined miRNA expression levels by panels of qRT-PCR - assays in 30 (12 FC, 12 FA, and 6 NT) of the included thyroid samples. An overview based on the total number of expressed miRNAs across all samples are illustrated in a PCA plot (Figure 5, panel C). We focused on building a diagnostic classifier to differentiate between FC and FA and found a signature comprising of 14 miRNAs to be most favorable. As a consequence of a different miRNA analyzing tool, the optimal signature for classification of FC and FA consists of 14 miRNAs, miR-19a, -501 -3p, -17, -335, - 106b, -15a, -16, -374a, -542-5p, -503, -320a, -326, -330-5p, and let-7i. A PCA plot based on expression values of the 14 miRNAs is illustrated in Figure 5, panel D.

Applying this signature resulted in only one misclassified carcinoma and derived in a negative predicted value of 92% and a positive predicted value of 100%, both for malignancies.

Discussion

In contrast to papillary- and medullary thyroid cancers, where many tumours exhibit defined mutations in oncogenes, the causal mutations leading to follicular neoplasia are incompletely understood. Consequently, efforts have been devoted towards defining biomarkers, that would allow the clinicians to distinguish carcinoma from adenoma. Here we characterize the expression of miRNAs and find that a majority of the computationally predicted targets are inversely expressed and implicated in tumourigenesis and cell-cycle control (Esquela-Kerscher and Slack, 2006).

Follicular adenoma and carcinoma exhibited widespread changes in their miRNA expression compared to normal thyroid. We identified numerous previously

undescribed thyroid miRNAs including miR-199b-5p, miR-144^*, miR-199b-3p, miR- 199a-5p, miR-144, miR-96, miR-182, and miR-597, to mention the miRNAs that exhibit the largest changes in expression levels. We also confirmed previously reported up- regulations of miR-197, -346, -187, -221 , -222, -224, and -155 in carcinoma and up- regulation of miR-339, -210, -328, and -342 in adenoma (Weber et al., 2006;Nikiforova et al., 2008). Of particular significance, among the unreported thyroid miRNAs, miR- 199b-5p was found to be lost in the carcinoma. Loss of miR-199b-5p (also known as miR-199b) has previously been shown to be followed by increased metastasis from meduloblastoma (Garzia et al., 2009) and a significant decrease of miR-199b-5p has been shown in chorioncarcinoma (Chao et al., 2009). We suggest that the loss of miR- 199b-5p is reflected in the corresponding mRNA targets and in the carcinogenesis of follicular thyroid tumours. Furthermore, we observed that miR-96 was markedly upregulated in the carcinomas. In a recent study miR-96 was shown to be upregulated in urothelial carcinomas and was promising tumour marker when measured in urine (Yamada et al., 2010). In addition, the up-regulation of miR-182 in FC was noteworthy since over-expression of miR-182 also have been obseved in both malignant melanomas and gliomas (Segura et al., 2009;Jiang et al., 2010). In both cases miR- 182 was moreover associated with metastasis and poor prognosis. This study is based on solid tumours and the drawback is evident since we have no causal data to substantiate the functional significance of miRNAs in tumour

progression. On the other hand solid cancers may provide a more accurate and authentic picture of the expressed miRNAs. To further substantiate the biological significance of the deregulated miRNAs, we subsequently performed a weighted target identification that previously has been shown to identify mRNA targets with high probability, followed by a molecular pathway analysis. Additionally, we examined the expression of the putative target mRNA since -80% of miRNA regulations are reflected by changes in mRNA expression levels (Guo et al., 2010). Strikingly, the pathways analysis depicted an enrichment of transcripts encoding proteins directly involved in thyroid carcinogenesis and tumour cell proliferation and cell cycle control. Taken together, the putative target mRNAs in the different pathways were in general up- regulated, in particular among cell cycle associated mRNAs, corresponding to the reduced levels of the associated miRNAs and we therefore infer that miRNAs could have an impact on the observed changes in the transcriptome during progression to carcinoma. Invasive carcinomas are known to exhibit a high proliferative grading, and it has been proposed that the mitotic index was useful to diagnose FC (Perez-Montiel and Suster, 2008;Ghossein, 2009). In agreement with this view, we previously found that carcinomas were strongly enriched in transcripts encoding proteins involved in DNA replication and mitosis corresponding to increased number of proliferating cells. The analysis of differentially expressed transcripts provided a mechanism for cancer progression and this set of transcripts provided a highly robust molecular classifier. The finding that the perturbed miRNAs target the same biological pathway further supports the fact that increased proliferative capacity is a hallmark of follicular carcinoma. It is possible that loss of miRNAs exhibiting a negative control on the mRNAs is an early event in follicular neoplasia. The latter is supported by the fact that the majority of the miRNAs are also down-regulated in adenoma.

The tumours collected from consecutively referred patients whose sex and age were in accordance with that of larger epidemiological studies. Moreover there was no preponderance of oncogenic mutations in the tumour sets. To exploit if the miRNAs could be useful to depict follicular carcinoma from adenoma, we generated a diagnostic signature from two different technical platforms. Although the results should be confirmed in independent studies, the performance of either platforms was acceptable, since the classifiers exhibited a NPV of 83% and 92% for malignancies with the microarray and qRT-PCR based platform, respectively. The qRT-PCR platform provided a better separation of FA and FC, than the microarray platform, which is reflected by the higher accuracy. From a clinical point of view, the predicted

probabilities derived from each individual sample is essential since it provides a quantitative value (on the reliability) to the extent in the accuracy of the diagnosis based on the classification. According to this we found that most samples are classifed with high accuracy.

Since we show that miRNA-based classification of histopathological follicular thyroid specimens is possible, the next obvious step is to examine whether it is feasible to implement miRNA based classification as an additional preoperative diagnostic tool. Taking the limited sensitivity and reproducibility of the histopathological diagnosis into account, the consistency between miRNA based classification and the pathological diagnosis is surprisingly high. This could reflect the fact that all samples were examined by the same endocrine pathologist. Studies of the inter-observer variations amongst pathologists in assessment of follicular lesions have demonstrated an observer variation for follicular carcinoma of 27%, where the carcinomas tended to be misdiagnosed as adenomas (Kakudo et al., 2002;Hirokawa et al., 2002). In a similar study an overall agreement among American and Japanese pathologists of 33% and 52%, respectively, was found (Hirokawa et al., 2002).

All results considered, we conclude that thyroid follicular neoplasia is accompanied by major changes in miRNA expression that may be directly implicated in malignant transformation and may facilitate diagnosis of follicular thyroid cancer.

TABLE 1 (ex.2):

Nodule

size Invasiveness

Age KRAS point BRAF

Diagnosis Sex (M/F) (cm; max (minimally

(years) genemutation genemutation diameter) vs. widely)

FC_01 32 F 6 minimal no mutation no mutation

FC_02 60 F 2.5 minimal no mutation no mutation

FC_03 32 F 1 .5 minimal no mutation no mutation

FC_04 48 F 5 widely no mutation no mutation

FC_05 41 M 6 minimal no mutation no mutation

FC_06 75 F 4 minimal no mutation no mutation _{< ηη}

Table 1. Clinical data of patients with thyroid follicular neoplasia. Twelve patients with histopathological verified follicular carcinomas (FC), (minimal and widely invasiveness), and twelve patients with follicular adenomas (FA). The table depicts diagnosis, age, sex, tumour size, invasiveness of the examined tumours, and status of known oncogenes. All tumours were negative for KRAS point mutation and

examination of BRAF showed only one positive carcinoma sample positive for BRAF point mutation.

TABLE 2 (ex.2): Dysregulated miRNAs. Identical to Table 2 (ex.1 )!

TABLE 3 (ex.2):

Transcripts involved in cell division process

RAC1, MAPK1, PSEN1, GNAI3, YWHAE, TUBG1, SMC2, PPP3CA, PPP6C, ARL8B, PPP2R5C, PIK3CA, CUL 1, YWHAG, GSPT1, PPP2CA, RB1, MOBKL 1B, DLG1, CSNK1A 1, FKBP1A, RMI1, CBL, PAK2, RIOK3, PA2G4, HAUS1, CFL 1, MAP3K7, MFN2, SOCS5, SAC3D1, BAX, KIF2A, AIMP1, NRAS, SEL 1L, RB1CC1, PTPRO, ITGB1, RFWD2, GSK3B, NAMPT, GPI, PAWR, PRPF4, CKS2, MAD2L2, PPP1R2, DNMT1, CASP3, SRPK2, BRCC3, MBD4, PGGT1B, PAFAH1B1, CDK7, MAD1L 1, MAD1L 1, BHLHE40, SMC4, RCC1, STMN1, PPRC1, FOXN3, C1QBP, TFDP1, GTF2I, TYMS, KPNA2, RECQL, MCM7, HSPA8, CLIP1, CDC5L, PTPN11 , UHMK1, CPiEM, CCNB1, SMC1A, MAPK14, POT1, CAMK2D, ACTL6A, CCNG1, TRRAP, CKAP2, RAD1, BCAT1, RACGAP1, NDC80, MDM2, ZWINT, COL4A2, BID, MSH2, TGFB1, CENPH, TPD52L 1, CD44, HELLS, ZAK, CSE1L, ZNF655, TOPBP1, PNPT1, DNM1L, TXN, THBS1, ASNS, CKAP5, CDK6, CDH13, KIF20A, DBF4, ECT2, IGFBP3, PCNA, TOP2A, SETD8, TAF1D, CDC20, RAN, TMPO, NUSAP1, KRAS, KIF11, BUB1B, UHRF1, VEGFA, RHOA, BRCA 1, CDC42, TIPIN, GNAI1, EZH2, CDC7, SOD2, ZWILCH, ANAPC5, CDKN3, ZBTB10, BIRC5, CENPA, AURKA, RHOU, CENPF, CEP55, ASPM, ANLN, NEK2, TRIP13, ATF5, GDF15, SPP1,

DLGAP5, TPX2, TTK, NUF2, MMP9, BCL2L 1, FAS, COL 1A 1, RHOB, and CDHL

Table 3. mIRNA targets Involved In cell division process. Upon ranked miRNA target predictions and gene ontology's, 165 transcripts encoded factors linked to cell division process. Of the 165 mRNAs, 154 were significantly up-regulated (P<0.05).

Transcripts are listed according to p-value, starting with the most significant.

TABLE 4 (ex.2): Predictive probabilities of miRNA classifier. Identical to Table 4 (ex.1)!

TABLE 5 (ex.2): Complete list of miRNAs that changed more than 2-fold:

miRNA Fold change Adjustet p-value

FC vs NT hsa-mi -199b-5p -39,05 5,98E-28

hsa-miR-144* -15,06 2,50E-07

hsa-miR-199a-3p -12,15 l,26E-25

hsa-miR-199a-5p -8,95 1,97E-12

hsa-miR-144 -6,89 1,06E-18

hsa-miR-1275 -6,66 3,42E-17

hsa-miR-153 -6,54 0,032752984

hsa-miR-451 -5,94 3,54E-17

hsa-miR-142-3p -5,89 8,54E-17

hsa-miR-886-5p -5,37 3,51E-09

hsa-miR-31 -5,05 5,53E-08

hsa-miR-455-3p -4,28 3,49E-09

hsa-miR-663 -4,04 l,46E-06

hsa-miR-218 -3,79 l,15E-09

hsa-miR-486-5p -3,44 2,68E-09

hsa-miR-100 -3,34 9,10E-06

hsa-miR-542-5p -3,28 0,009873021

hsa-miR-1 -3,27 0,017907167

hsa-miR-101 -3,27 6,77E-11

hsa-miR-20a -3,22 l,46E-06

hsa-miR-193a-3p -3,17 l,20E-07

hsa-miR-223 -3,06 7,43E-07

hsa-miR-886-3p -3,05 0,012755014

hsa-miR-124* -2,07 3,44E-08 hsa-miR-193b* -2,04 3,73E-10 hsa-miR-1303 -2,02 0,027106401 hsa-miR-887 2,02 2,10E-05 hsa-miR-20b* 2,06 0,041742281 hsa-miR-200c 2,14 l,35E-06 hsa-miR-34a 2,22 3,01E-06 hsa-miR-518e 2,27 0,026773546 hsa-miR-22 2,36 2,09E-06 hsa-miR-526b* 2,39 0,024990124 hsa-miR-1323 2,40 0,027151008 hsa-miR-182 2,40 0,000229917 hsa-miR-1274b 2,44 4,49E-07 hsa-miR-597 2,48 0,008812285 hsa-miR-96 2,51 0,000785249 hsa-miR-522 2,52 0,022088776 hsa-miR-518a-3p 2,60 0,039372257 hsa-miR-301b 3,20 8,55E-05 hsa-miR-512-3p 3,64 0,013690351 hsa-miR-517a 4,54 0,02852924 ebv-miR-BART8 5,32 0,048904611

FC vs FA hsa-miR-512-3p -3,18 0,039531549 hsa-miR-886-5p -3,03 0,000349881 hsa-miR-450a -3,01 0,00011959 hsa-miR-301b -2,64 0,003225254 hsa-miR-429 -2,44 0,004624901 hsa-miR-542-3p -2,40 l,06E-06 hsa-miR-130a -2,34 9,90E-05 hsa-miR-146b-5p -2,25 0,017386477 hsa-miR-199a-5p -2,07 0,039531549 hsa-miR-193a-3p -2,07 0,003064597 hsa-miR-152 -2,02 0,021237673 hsa-miR-199a-3p -2,01 0,003225254 hsa-miR-631 2,03 0,036385401 hsa-miR-637 2,10 0,003551208 Reference List - Example 1 :

1 . Hegedus L, Bonnema SJ, Bennedbaek FN. Management of simple nodular

goiter: current status and future perspectives. Endocr Rev 2003; 24(1 ):102-132.

2. Stewart B, Kleihues P. World Cancer Report. IARC Press; Lyon, 2003: 1 1 -20.

3. Faquin WC. The thyroid gland: recurring problems in histologic and cytologic evaluation. Arch Pathol Lab Med 2008; 132(4):622-632.

4. Ronald A.DeLellis, Ricardo V Lloyd, PUH, CE. Tumours of Endocrine Organs.

World Health Organization Classification of Tumours, IARC Press, 2004.

5. Schmid KW, Farid NR. How to define follicular thyroid carcinoma? Virchows Arch 2006; 448(4):385-393.

6. Franc B, de la SP, Lange F et al. Interobserver and intraobserver reproducibility in the histopathology of follicular thyroid carcinoma. Hum Pathol 2003;

34(1 1 ):1092-1 100.

7. Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 2004; 1 16(2):281 -297.

8. Bartel DP. MicroRNAs: target recognition and regulatory functions. Cell 2009;

136(2):215-233.

9. Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalian mRNAs are conserved targets of microRNAs. Genome Res 2009; 19(1 ):92-105.

10. Schmittgen TD. Regulation of microRNA processing in development,

differentiation and cancer. J Cell Mol Med 2008; 12(5B):181 1 -1819.

1 1 . Rosenfeld N, Aharonov R, Meiri E et al. MicroRNAs accurately identify cancer tissue origin. Nat Biotechnol 2008; 26(4):462-469.

12. Chen YT, Kitabayashi N, Zhou XK, Fahey TJ, III, Scognamiglio T. MicroRNA analysis as a potential diagnostic tool for papillary thyroid carcinoma. Mod Pathol 2008; 21 ((9)):1 139-1 146.

13. He H, Jazdzewski K, Li W et al. The role of microRNA genes in papillary thyroid carcinoma. Proc Natl Acad Sci U S A 2005; 102(52):19075-19080.

14. Nikiforova MN, Tseng GC, Steward D, Diorio D, Nikiforov YE. MicroRNA

expression profiling of thyroid tumours: biological significance and diagnostic utility. J Clin Endocrinol Metab 2008; 93(5):1600-1608.

15. Pallante P, Visone R, Ferracin M et al. MicroRNA deregulation in human thyroid papillary carcinomas. Endocr Relat Cancer 2006; 13(2):497-508. 16. Weber F, Teresi RE, Broelsch CE, Frilling A, Eng C. A limited set of human MicroRNA is deregulated in follicular thyroid carcinoma. J Clin Endocrinol Metab 2006; 91 (9):3584-3591 .

17. Gentleman RC, Carey VJ, Bates DM et al. Bioconductor: open software

development for computational biology and bioinformatics. Genome Biol 2004;

5(10):R80.

18. "R Development Core Team (2007) R: A Language and Environment for

Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, ISBN 3-900051 -07-0. URL. Downloads . 2007.

Ref Type: Electronic Citation

19. Wettenhall JM, Smyth GK. limmaGUI: a graphical user interface for linear

modeling of microarray data. Bioinformatics 2004; 20(18):3705-3706.

20. Dudoit S SJPBJC. Multiple hypothesis testing in microarray experiments.

Statistical Science 2003; 18(1 ):71 -103.

21 . Schmittgen TD, Livak KJ. Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc 2008; 3(6):1 101 -1 108.

22. Piatt J. Advances in Large Classifiers. Cambridge, MA: MIT Press, 1999.

23. Esquela-Kerscher A, Slack FJ. Oncomirs - microRNAs with a role in cancer. Nat Rev Cancer 2006; 6(4):259-269.

24. Cai X, Schafer A, Lu S et al. Epstein-Barr virus microRNAs are evolutionarily conserved and differentially expressed. PLoS Pathog 2006; 2(3):e23.

25. Shimakage M, Kawahara K, Sasagawa T et al. Expression of Epstein-Barr virus in thyroid carcinoma correlates with tumour progression. Hum Pathol 2003;

34(1 1 ):1 170-1 177.

26. Takano T. Fetal cell carcinogenesis of the thyroid: a hypothesis for better

understanding of gene expression profile and genomic alternation in thyroid carcinoma. Endocr J 2004; 51 ((6)):509-515.

27. Lu J, Getz G, Miska EA et al. MicroRNA expression profiles classify human

cancers. Nature 2005; 435(7043) :834-838.

28. Guttilla IK, White BA. Coordinate regulation of FOX01 by miR-27a, miR-96, and miR-182 in breast cancer cells. J Biol Chem 2009; 284(35):23204-23216.

29. Myatt SS, Wang J, Monteiro LJ et al. Definition of microRNAs that repress

expression of the tumour suppressor gene FOX01 in endometrial cancer. Cancer

Res 2010; 70(1 ):367-377. 30. Myatt SS, Lam EW. The emerging roles of forkhead box (Fox) proteins in cancer. Nat Rev Cancer 2007; 7(1 1 ):847-859.

31 . Paik JH, Kollipara R, Chu G et al. FoxOs are lineage-restricted redundant tumour suppressors and regulate endothelial cell homeostasis. Cell 2007; 128(2):309- 323.

32. Garzia L, Andolfo I, Cusanelli E et al. MicroRNA-199b-5p impairs cancer stem cells through negative regulation of HES1 in medulloblastoma. PLoS One 2009; 4(3):e4998.

33. Chao A, Tsai CL, Wei PC et al. Decreased expression of microRNA-199b

increases protein levels of SET (protein phosphatase 2A inhibitor) in human choriocarcinoma. Cancer Lett 2009.

34. Migliore C, Petrelli A, Ghiso E et al. MicroRNAs impair MET-mediated invasive growth. Cancer Res 2008; 68(24):10128-10136.

35. Park JK, Lee EJ, Esau C, Schmittgen TD. Antisense inhibition of microRNA-21 or -221 arrests cell cycle, induces apoptosis, and sensitizes the effects of gemcitabine in pancreatic adenocarcinoma. Pancreas 2009; 38(7):e190-e199.

36. Li QX, Ke N, Sundaram R, Wong-Staal F. NR4A1 , 2, 3-an orphan nuclear

hormone receptor family involved in cell apoptosis and carcinogenesis. Histol Histopathol 2006; 21 (5):533-540.

37. Mullican SE, Zhang S, Konopleva M et al. Abrogation of nuclear receptors Nr4a3 and Nr4a1 leads to development of acute myeloid leukemia. Nat Med 2007; 13(6):730-735.

38. Kakudo K, Katoh R, Sakamoto A et al. Thyroid gland: international case

conference. Endocr Pathol 2002; 13(2):131 -134.

39. Hirokawa M, Carney JA, Goellner JR et al. Observer variation of encapsulated follicular lesions of the thyroid gland. Am J Surg Pathol 2002; 26(1 1 ):1508-1514.

Reference List - Example 2:

1 . "R Development Core Team (2007) R: A Language and Environment for

Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria,

ISBN 3-900051 -07-0. URL. Downloads. 2007. Ref Type: Electronic Citation

2. Bartel DP (2004) MicroRNAs: genomics, biogenesis, mechanism, and function.

Cell, 116, 281 -297.

3. Bartel DP (2009) MicroRNAs: target recognition and regulatory functions. Cell,

136, 215-233. 4. Borup R, Rossing M, Henao R, Yamamoto Y, Krogdahl A, Godballe C, Winther O, Kiss K, Christensen L, Hogdall E, Bennedbaek F, and Nielsen FC (2010)

Molecular signatures of thyroid follicular neoplasia. Endocr Relat Cancer, 17, 691 - 708.

5. Chao A, Tsai CL, Wei PC, Hsueh S, Chao AS, Wang CJ, Tsai CN, Lee YS, Wang

TH, and Lai CH (2009) Decreased expression of microRNA-199b increases protein levels of SET (protein phosphatase 2A inhibitor) in human

choriocarcinoma. Cancer Lett.

6. Chen YT, Kitabayashi N, Zhou XK, Fahey TJ, III, and Scognamiglio T (2008) MicroRNA analysis as a potential diagnostic tool for papillary thyroid carcinoma.

Mod Pathol, 21 , 1 139-1 146.

7. Curado MP and Edwards B (2007). Cancer Incidence in Five Continents Vol. IX.

IARC Scientific Publication.

8. Dudoit S SJPBJC (2003) Multiple hypothesis testing in microarray experiments.

Statistical Science, 18(1 ) :71 -103.

9. Esquela-Kerscher A and Slack FJ (2006) Oncomirs - microRNAs with a role in cancer. Nat Rev Cancer, 6, 259-269.

10. Faquin WC (2008) The thyroid gland: recurring problems in histologic and

cytologic evaluation. Arch Pathol Lab Med, 132, 622-632.

1 1 . Franc B, de la SP, Lange F, Hoang C, Louvel A, de RA, Vilde F, Hejblum G,

Chevret S, and Chastang C (2003) Interobserver and intraobserver reproducibility in the histopathology of follicular thyroid carcinoma. Hum Pathol, 34, 1092-1 100. 12. Friedman RC, Farh KK, Burge CB, and Bartel DP (2009) Most mammalian

mRNAs are conserved targets of microRNAs. Genome Res, 19, 92-105.

13. Gallagher IJ, Scheele C, Keller P, Nielsen AR, Remenyi J, Fischer CP, Roder K, Babraj J, Wahlestedt C, Hutvagner G, Pedersen BK, and Timmons JA (2010) Integration of microRNA changes in vivo identifies novel molecular features of muscle insulin resistance in type 2 diabetes. Genome Med, 2, 9.

14. Garzia L, Andolfo I, Cusanelli E, Marino N, Petrosino G, De MD, Esposito V, Galeone A, Navas L, Esposito S, Gargiulo S, Fattet S, Donofrio V, Cinalli G,

Brunetti A, Vecchio LD, Northcott PA, Delattre O, Taylor MD, lolascon A, and Zollo M (2009) MicroRNA-199b-5p impairs cancer stem cells through negative regulation of HES1 in medulloblastoma. PLoS One, 4, e4998.

15. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J, Hornik K, Hothorn T, Huber W, lacus S, Irizarry R, Leisch F, Li C, Maechler M, Rossini AJ, Sawitzki G, Smith C, Smyth G, Tierney L, Yang JY, and Zhang J (2004) Bioconductor: open software development for computational biology and bioinformatics. Genome Biol, 5, R80.

16. Ghossein R (2009) Problems and controversies in the histopathology of thyroid carcinomas of follicular cell origin. Arch Pathol Lab Med, 133, 683-691 .

17. Guo H, Ingolia NT, Weissman JS, and Bartel DP (2010) Mammalian microRNAs predominantly act to decrease target mRNA levels. Nature, 466, 835-840.

18. He H, Jazdzewski K, Li W, Liyanarachchi S, Nagy R, Volinia S, Calin GA, Liu CG, Franssila K, Suster S, Kloos RT, Croce CM, and de la CA (2005) The role of microRNA genes in papillary thyroid carcinoma. Proc Natl Acad Sci U S Α, Λ 02,

19075-19080.

19. Hegedus L, Bonnema SJ, and Bennedbaek FN (2003) Management of simple nodular goiter: current status and future perspectives. Endocr Rev, 24, 102-132.

20. Hirokawa M, Carney JA, Goellner JR, DeLellis RA, Heffess CS, Katoh R,

Tsujimoto M, and Kakudo K (2002) Observer variation of encapsulated follicular lesions of the thyroid gland. Am J Surg Pathol, 26, 1508-1514.

21 . Jiang L, Mao P, Song L, Wu J, Huang J, Lin C, Yuan J, Qu L, Cheng SY, and Li J (2010) miR-182 as a prognostic marker for glioma progression and patient survival. Am J Pathol, 177, 29-38.

22. Kakudo K, Katoh R, Sakamoto A, Asa S, DeLellis RA, Carney JA, Naganuma H, Kameyama K, and Takami H (2002) Thyroid gland: international case conference. Endocr Pathol, 13, 131 -134.

23. Nikiforova MN, Tseng GC, Steward D, Diorio D, and Nikiforov YE (2008)

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

24. Novitskaya V, Romanska H, Dawoud M, Jones JL, and Berditchevski F (2010) Tetraspanin CD151 regulates growth of mammary epithelial cells in three- dimensional extracellular matrix: implication for mammary ductal carcinoma in situ. Cancer Res, 70, 4698-4708.

25. Pallante P, Visone R, Ferracin M, Ferraro A, Berlingieri MT, Troncone G,

Chiappetta G, Liu CG, Santoro M, Negrini M, Croce CM, and Fusco A (2006) MicroRNA deregulation in human thyroid papillary carcinomas. Endocr Relat Cancer, 13, 497-508. 26. Perez-Montiel MD and Suster S (2008) The spectrum of histologic changes in thyroid hyperplasia: a clinicopathologic study of 300 cases. Hum Pathol, 39, 1080-1087.

27. Piatt J (1999). Advances in Large Classifiers. MIT Press, Cambridge, MA.

28. Rosenfeld N, Aharonov R, Meiri E, Rosenwald S, Spector Y, Zepeniuk M,

Benjamin H, Shabes N, Tabak S, Levy A, Lebanony D, Goren Y, Silberschein E, Targan N, Ben-Ari A, Gilad S, Sion-Vardy N, Tobar A, Feinmesser M, Kharenko O, Nativ O, Nass D, Perelman M, Yosepovich A, Shalmon B, Polak-Charcon S, Fridman E, Avniel A, Bentwich I, Bentwich Z, Cohen D, Chajut A, and Barshack I (2008) MicroRNAs accurately identify cancer tissue origin. Nat Biotechnol, 26,

462-469.

29. Schmid KW and Farid NR (2006) How to define follicular thyroid carcinoma? Virchows Arch, 448, 385-393.

30. Schmittgen TD (2008) Regulation of microRNA processing in development, differentiation and cancer. J Cell Mol Med, 12, 181 1 -1819.

31 . Schmittgen TD and Livak KJ (2008) Analyzing real-time PCR data by the

comparative C(T) method. Nat Protoc, 3, 1 101 -1 108.

32. Segura MF, Hanniford D, Menendez S, Reavie L, Zou X, varez-Diaz S,

Zakrzewski J, Blochin E, Rose A, Bogunovic D, Polsky D, Wei J, Lee P,

Belitskaya-Levy I, Bhardwaj N, Osman I, and Hernando E (2009) Aberrant miR-

182 expression promotes melanoma metastasis by repressing FOX03 and microphthalmia-associated transcription factor. Proc Natl Acad Sci U S A, 106, 1814-1819.

33. Takakura S, Kohno T, Manda R, Okamoto A, Tanaka T, and Yokota J (2001 ) Genetic alterations and expression of the protein phosphatase 1 genes in human cancers. Int J Oncol, 18, 817-824.

34. Weber F, Teresi RE, Broelsch CE, Frilling A, and Eng C (2006) A limited set of human MicroRNA is deregulated in follicular thyroid carcinoma. J Clin Endocrinol Metab, 91 , 3584-3591 .

35. Wettenhall JM and Smyth GK (2004) limmaGUI: a graphical user interface for linear modeling of microarray data. Bioinformatics, 20, 3705-3706.

36. Yamada Y, Enokida H, Kojima S, Kawakami K, Chiyomaru T, Tatarano S,

Yoshino H, Kawahara K, Nishiyama K, Seki N, and Nakagawa M (2010) MiR-96 and miR-183 detection in urine serve as potential tumor markers of urothelial carcinoma: correlation with stage and grade, and comparison with urinary cytology. Cancer Sci.

hsa-miRPIus-ElOOl gguuucggguuugaaggcagc

n

hsa-miR-410 aauauaacacagauggccugu

hsa-miR-512-3p aagugcugucauagcugagguc hsa-miR-886-5p cgggucggaguuagcucaagcgg hsa-miR-450a uuuugcgauguguuccuaauau hsa-miR-301b cagugcaaugauauugucaaagc hsa-miR-429 uaauacugucugguaaaaccgu hsa-miR-542-3p ugugacagauugauaacugaaa hsa-miR-130a cagugcaauguuaaaagggcau hsa-miR-146b-5p ugagaacugaauuccauaggcu hsa-miR-199a-5p cccaguguucagacuaccuguuc hsa-miR-193a-3p aacuggccuacaaagucccagu

hsa-miR-152 ucagugcaugacagaacuugg

hsa-miR-199a-3p/hsa-miR-199b-3p acaguagucugcacauugguua hsa-miR-424 cagcagcaauucauguuuugaa hsa-miR-22 aagcugccaguugaagaacugu hsa-miR-146a ugagaacugaauuccauggguu hsa-miR-339-3p ugagcgccucgacgacagagccg hsa-miR-365 uaaugccccuaaaaauccuuau hsa-let-7i ^ cugcgcaagcuacugccuugcu

hsa-miR-363* cggguggaucacgaugcaauuu hsa-miR-148a ucagugcacuacagaacuuugu hsa-miR-299-3p uaugugggaugguaaaccgcuu hsa-let-7a * cuauacaaucuacugucuuuc

hsa-miR-200b uaauacugccugguaaugauga hsa-miR-200c uaauacugccggguaaugaugga hsa-miR-375 uuuguucguucggcucgcguga hsa-miR-451 aaaccguuaccauuacugaguu hsa-miR-144 uacaguauagaugauguacu

hsa-let-7i ugagguaguaguuugugcuguu hsa-miR-1826 auugaucaucgacacuucgaacgcaau hsa-miR-1201 agccugauuaaacacaugcucuga hsa-miR-140-5p cagugguuuuacccuaugguag hsa-miR-126 ucguaccgugaguaauaaugcg hsa-miR-126* cauuauuacuuuugguacgcg

hsa-let-7f-2 *^■ cuauacagucuacugucuuucc hsa-mi -148b ucagugcaucacagaacuuugu lisa-miR-2 uagcuuaucagacugauguuga

hsa-miR-342-3p ucucacacagaaaucgcacccgu

hsa-miR-27a uucacaguggcuaaguuccgc

hsa-miR-145* ggauuccuggaaauacuguucu

sa-miR-513b u u ca ca aggagg ug u ca u u u a u

hsa-miR-101 uacaguacugugauaacugaa

sa-miR-26a uucaaguaauccaggauaggcu

hsa-miR-24 uggcucaguucagcaggaacag

hsa-miR-30a ^f' cuuucagucggauguuugcagc

hsa-miR-377 aucacacaaaggcaacuuuugu

hsa-miR-222* cucaguagccaguguagauccu

hsa-miR-452 aacuguuugcagaggaaacuga

hsa-miR-665 accaggaggcugaggccccu

hsa-miR-584 uuaugguuugccugggacugag

hsa-miR-492 aggaccugcgggacaagauucuu

hsa-miR-744 ugcggggcuagggcuaacagca

hsa-miR-662 ucccacguuguggcccagcag

hsa-miR-219-2-3p agaauuguggcuggacaucugu

hsa-miR-631 agaccuggcccagaccucagc

hsa-miR-637 acugggggcuuucgggcucugcgi

hsa-miR-19a ugugcaaaucuaugcaaaacuga

hsa-miR-501-3p aaugcacccgggcaaggauucu

hsa-miR-17 caaagugcuuacagugcagguag

hsa-miR-335 ucaagagcaauaacgaaaaaugu

hsa-miR-106b uaaagugcugacagugcagau

hsa-miR-15a uagcagcacauaaugguuugug

hsa-miR-16 uagcagcacguaaauauuggcg

hsa-miR-374a uuauaauacaaccugauaagug

hsa-miR-542-5p ucggggaucaucaugucacgaga

hsa-miR-503 uagcagcgggaacaguucugcag

hsa-miR-320a aaaagcuggguugagagggcga

hsa-miR-326 ccucugggcccuuccuccag

hsa-miR-330-3p gcaaagcacacggccugcagaga

n Included on Array Version 1 1 , Exiqon (Cat. No. 208202-A) (http://www.exiqon.com/)

* Primers may be purchased from Exiqon (1 x 206999 fwd-miRPIus-E1078, 1 x 206999 rev- miRPIus-E1078).

To download a GenePix Array List (GAL) file for the miRCURY LNA™ microRNA Arrays, please refer to: http://www.exiqon.com/Gal-downloads (lists , Block', 'Column', 'Row', 'ID' and 'Name' for all miRNAs included on the chip/array). A two-way miRNA classifier for characterising a sample obtained from a thyroid nodule of an individual, wherein said miRNA classifier

i) comprises or consists of hsa-miR-1826 and hsa-miRPIus-E1078 and distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma, or

ii) comprises or consists of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR- 374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa- miR-326 and hsa-miR-330-3p, and distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma, or

iii) comprises or consists of hsa-miRPIus-E1001 and hsa-miR-410 and distinguishes between the classes widely invasive thyroid follicular carcinoma and minimally invasive thyroid follicular carcinoma, or iv) comprises or consists of one or more miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR- 450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR- 152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR- 363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa- miR-126^*, hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342- 3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa- miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-

518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR- 5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa- miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2- 3p, hsa-miR-631 and hsa-miR-637 and distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma, wherein said distinction is given as a prediction probability for said sample of belonging to either class, said probability being a number falling in the range of from 0 to 1 .

The miRNA classifier according to item 1 , wherein said miRNA classifier further comprises one or more additional miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR- 301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa- miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa- miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa- miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR- 342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa- miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa- miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa- miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

The miRNA classifier according to item 2, wherein said additional miRNAs comprise no more than 10 additional miRNAs, such as 9 additional miRNAs, for example 8 additional miRNAs, such as 7 additional miRNAs, for example 6 additional miRNAs, such as 5 additional miRNAs, for example 4 additional miRNAs, such as 3 additional miRNAs, for example 2 additional miRNAs, such as 1 additional miRNA.

The miRNA classifier according to item 1 , wherein said miRNA classifier comprises or consists of less than 50 miRNAs, for example less than 40 miRNAs, such as less than 30 miRNAs, for example less than 20 miRNAs, such as less than 10 miRNAs, for example less than 5 miRNAs.

The miRNA classifier according to item 1 , wherein said miRNA classifier comprises or consists of a total of 1 miRNA, such as 2 miRNAs, for example 3 miRNAs, such as 4 miRNAs, for example 5 miRNAs, such as 6 miRNAs, for example 7 miRNAs, such as 8 miRNAs, for example 9 miRNAs, such as 10 miRNAs, for example 1 1 miRNAs, such as 12 miRNAs, for example 13 miRNAs, such as 14 miRNAs, for example 15 miRNAs, such as 16 miRNAs, for example 17 miRNAs, such as 18 miRNAs, for example 19 miRNAs, such as 20 miRNAs, for example 21 miRNAs, such as 22 miRNAs, for example 23 miRNAs, such as 24 miRNAs, for example 25 miRNAs, such as 26 miRNAs, for example 27 miRNAs, such as 28 miRNAs, for example 29 miRNAs, such as 30 miRNAs, for example 31 miRNAs, such as 32 miRNAs, for example 33 miRNAs, such as 34 miRNAs, for example 35 miRNAs, such as 36 miRNAs, for example 37 miRNAs, such as 38 miRNAs, for example 39 miRNAs, such as 40 miRNAs, for example 41 miRNAs, such as 42 miRNAs, for example 43 miRNAs, such as 44 miRNAs, for example 45 miRNAs, such as 46 miRNAs, for example 47 miRNAs, such as 48 miRNAs, for example 49 miRNAs, such as 50 miRNAs.

The miRNA classifier according to item 1 , wherein the sensitivity is at least 70%, such as at least 75%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%.

The miRNA classifier according to item 1 , wherein the specificity is at least 70%, such as at least 75%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%.

The miRNA classifier according to item 1 , wherein the prediction probability of a sample for belonging to a certain class is a number falling in the range of from 0 to 1 , such as from 0.0 to 0.1 , for example 0.1 to 0.2, such as 0.2 to 0.3, for example 0.3 to 0.4, such as 0.4 to 0.49, for example 0.5, such as 0.51 to 0.6, for example 0.6 to 0.7, such as 0.7 to 0.8, for example 0.8 to 0.9, such as 0.9 to 1 .0. The miRNA classifier according to item 1 , wherein the negative predictive value for malignancies is at least 70%, such as at least 71 %, for example at least 72%, such as at least 73%, for example at least 74%, such as at least 75%, for example at least 76%, such as at least 77%, for example at least 78%, such as at least 79%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%, for example at least 96%, such as at least 97%, for example at least 98%, such as at least 99%, for example at least 100%. The miRNA classifier according to item 1 , wherein the positive predictive value for malignancies is at least 70%, such as at least 71 %, for example at least 72%, such as at least 73%, for example at least 74%, such as at least 75%, for example at least 76%, such as at least 77%, for example at least 78%, such as at least 79%, for example at least 80%, such as at least 81 %, for example at least 82%, such as at least 83%, for example at least 84%, such as at least 85%, for example at least 86%, such as at least 87%, for example at least 88%, such as at least 89%, for example at least 90%, such as at least 91 %, for example at least 92%, such as at least 93%, for example at least 94%, such as at least 95%, for example at least 96%, such as at least 97%, for example at least 98%, such as at least 99%, for example at least 100%. The miRNA classifier according to item 1 , wherein said classifier comprises or consists of hsa-miR-1826 and hsa-miRPIus-E1078. The miRNA classifier according to items 1 and 1 1 , wherein the up-regulation of hsa-miR-1826 expression and up-regulation of hsa-miRPIus-E1078 expression is indicative of thyroid follicular adenoma. The miRNA classifier according to items 1 and 1 1 , wherein the down- regulation of hsa-miR-1826 expression and down-regulation of hsa- miRPIus-E1078 expression is indicative of thyroid follicular carcinoma. The miRNA classifier according to item 1 , wherein said classifier comprises or consists of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p. The miRNA classifier according to item 1 , 1 1 and 14, wherein an alteration of the expression profile of one or more of said miRNAs is associated with thyroid follicular adenoma or thyroid follicular carcinoma. The miRNA classifier according to item 1 , wherein said classifier comprises or consists of hsa-miRPIus-E1001 and hsa-miR-410. The miRNA classifier according to items 1 and 16, wherein an alteration of the expression profile of one or more of said miRNAs is associated with widely invasive thyroid follicular carcinoma or minimally invasive thyroid follicular carcinoma. The miRNA classifier according to items 1 and 16, wherein the up-regulation of hsa-miR-410 expression and down-regulation of hsa-miRPIus-E1001 expression is indicative of minimally invasive thyroid follicular carcinoma. The miRNA classifier according to items 1 and 16, wherein the down- regulation of hsa-miR-410 expression and up-regulation of hsa-miRPIus- E1001 expression is indicative of widely invasive thyroid follicular carcinoma. The miRNA classifier according to item 1 , wherein the expression level of one or more miRNAs is determined by the microarray technique. 21 . The miRNA classifier according to item 1 , wherein the expression level of one or more miRNAs is determined by the quantitative polymerase chain reaction (QPCR) technique.

22. The miRNA classifier according to item 1 , wherein the expression level of one or more miRNAs is determined by the northern blot technique.

23. The miRNA classifier according to item 1 , wherein the expression level of one or more miRNAs is determined by Nuclease protection assay.

24. The miRNA classifier according to item 1 , wherein the sample is extracted from an individual by fine-needle aspiration.

25. The miRNA classifier according to item 24, wherein the sample is extracted from an individual by single fine-needle aspiration.

26. The miRNA classifier according to item 24, wherein the sample is extracted from an individual by multiple fine-needle aspirations. 27. The miRNA classifier according to item 1 , wherein the sample is extracted from an individual by coarse-needle aspiration.

28. The miRNA classifier according to item 1 , wherein the sample is extracted from an individual by thyroid surgery.

29. The miRNA classifier according to item 28, wherein the sample is extracted from an individual by hemi-thyroidectomy.

30. The miRNA classifier according to item 1 , wherein the sample is extracted from an individual by thyroid biopsy.

31 . A method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising measuring the expression profile of at least two miRNAs in a sample obtained from the thyroid of said individual, wherein said miRNA is selected from the group consisting of i) hsa-miR-1826 and hsa-miRPIus-E1078, or

ii) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa- miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR- 542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, or

iii) hsa-miRPIus-E1001 and hsa-miR-410,

wherein a predetermined miRNA expression level of said miRNAs is indicative of the individual having, or being at risk of developing, follicular thyroid carcinoma. A method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising the steps of:

1 ) extracting RNA from a sample collected from the thyroid of an individual,

i) hsa-miR-1826 and hsa-miRPIus-E1078, or

ii) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa- miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa- miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f- 2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa- miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa-miR-519a^*/hsa- miR-519b-5p/hsa-miR-519c-5p/hsa-miR-522^*/hsa-miR-523^*, hsa- miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637,

wherein a predetermined miRNA expression profile of the at least one of said miRNAs is indicative of the individual having, or being at risk of developing, follicular thyroid carcinoma. A method for performing a diagnosis on an individual with a thyroid nodule, comprising the steps of:

1 ) extracting RNA from a sample collected from the thyroid of an individual,

2) analysing the miRNA expression profile of the sample, and

3) determining if said individual has a benign or a malignant condition

i) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa- miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa- miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f- 2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa- miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa-miR-519a^*/hsa- miR-519b-5p/hsa-miR-519c-5p/hsa-miR-522^*/hsa-miR-523^*, hsa- miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637. A method for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said method comprising measuring the expression level of at least one miRNA in said sample, wherein said at least one miRNA is selected from the group consisting of i) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

wherein the expression level of said at least one miRNA is associated with thyroid follicular carcinoma. 35. A method for determining the presence of a benign condition in a sample obtained from a thyroid nodule of an individual, said method comprising measuring the expression level of at least one miRNA in said sample, wherein said at least one miRNA is selected from the group consisting of i) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

wherein said expression level of said at least one miRNA is associated with thyroid follicular adenoma.

36. A method for expression profiling of a sample obtained from the thyroid, comprising measuring at least one miRNA selected from the group of i) hsa-miR-1826 and hsa-miRPIus-E1078, or hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p, or

hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa- miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa- miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f- 2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa- miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e7hsa-miR-519a7hsa- miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa- miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637,

and correlating said expression profile to a clinical condition of the thyroid. The method according to item 36, wherein said clinical condition is follicular thyroid carcinoma, follicular thyroid adenoma, widely invasive follicular thyroid carcinoma or minimally invasive follicular thyroid carcinoma. A method for determining the prognosis of an individual with a thyroid nodule, comprising the steps of

1 ) extracting RNA from a sample collected from the thyroid of an individual, 2) analysing the miRNA expression profile of the sample,

wherein said miRNA expression profile comprises at least one miRNA selected from the group consisting of hsa-miR-1826 and hsa-miRPIus-E1078, or

hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p, or

hsa-miRPIus-E1001 and hsa-miR-410, or

hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa- miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa- miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f- 2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa- miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa-miR-519a^*/hsa- miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa- miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637. The method according to any of items 31 to 38, wherein said method further comprises the step of obtaining a sample from the thyroid of an individual. The method according to any of items 31 to 38, wherein said miRNA comprises or consists of hsa-miR-1826 and hsa-miRPIus-E1078. The method according to any of items 31 to 38, wherein said miRNA comprises or consists of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa- miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa- miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa- miR-330-3p. The method according to any of items 31 to 38, wherein said miRNA comprises or consists of hsa-miRPIus-E1001 and hsa-miR-410. The method according to any of items 31 to 38, wherein said method comprises obtaining prediction probabilities of between 0-1 for said sample. The method according to any of items 31 to 38, wherein said method is used in combination with at least one additional diagnostic method. The method according to item 44, wherein said at least one additional diagnostic method is selected from the group consisting of CT (X-ray computed tomography), MRI (magnetic resonance imaging), Scintillation counting, Blood sample analysis, Ultrasound imaging, Cytology, Histology and Assessment of risk factors. The method according to item 44, wherein said at least one additional diagnostic method is use of an mRNA classifier capable of distinguishing between follicular thyroid adenoma and follicular thyroid carcinoma. The method according to item 44, wherein said at least one additional diagnostic method improves the sensitivity and/or specificity of the combined diagnostic outcome. A device for measuring the expression level of at least one miRNA in a sample, wherein said device comprises or consists of at least one probe or probe set for miRNAs selected from the group consisting of

i) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

wherein said device is used for classifying a sample obtained from a thyroid nodule of an individual. The device according to item 48, wherein said device comprises or consists of probes for hsa-miR-1826 and hsa-miRPIus-E1078. The device according to item 48, wherein said device comprises or consists of probes for hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p. The device according to item 48, wherein said device comprises or consists of probes for hsa-miRPIus-E1001 and hsa-miR-410. The device according to item 48, wherein said device comprises or consists of probes selected from the group consisting of hsa-miR-512-3p, hsa-miR- 886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa- miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa- miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa- miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa- miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa- miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa- miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c- 5p hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637. The device according to item 48, wherein said device may be used for distinguishing between thyroid follicular adenoma and thyroid follicular carcinoma, and/or distinguishing between minimally invasive thyroid follicular carcinoma and widely invasive thyroid follicular carcinoma. The device according to item 48, wherein said device may be used with the miRNA classifier according to item 1 , to classify a sample into either of the classes of thyroid follicular adenoma, thyroid follicular carcinoma, minimally invasive thyroid follicular carcinoma or widely invasive thyroid follicular carcinoma. The device according to item 48, wherein said device comprises less than 50 probes, for example less than 40 probes, such as less than 30 probes, for example less than 20 probes, such as less than 10 probes, for example less than 5 probes. The device according to item 48, wherein said device comprises or consists of a total of 1 probe, such as 2 miRNAs, for example 3 miRNAs, such as 4 miRNAs, for example 5 miRNAs, such as 6 miRNAs, for example 7 miRNAs, such as 8 miRNAs, for example 9 miRNAs, such as 10 miRNAs, for example 1 1 miRNAs, such as 12 miRNAs, for example 13 miRNAs, such as 14 miRNAs, for example 15 miRNAs, such as 16 miRNAs, for example 17 miRNAs, such as 18 miRNAs, for example 19 miRNAs, such as 20 miRNAs, for example 21 miRNAs, such as 22 miRNAs, for example 23 miRNAs, such as 24 miRNAs, for example 25 miRNAs, such as 26 miRNAs, for example 27 miRNAs, such as 28 miRNAs, for example 29 miRNAs, such as 30 miRNAs, for example 31 miRNAs, such as 32 miRNAs, for example 33 miRNAs, such as 34 miRNAs, for example 35 miRNAs, such as 36 miRNAs, for example 37 miRNAs, such as 38 miRNAs, for example 39 miRNAs, such as 40 miRNAs, for example 41 miRNAs, such as 42 miRNAs, for example 43 miRNAs, such as 44 miRNAs, for example 45 miRNAs, such as 46 miRNAs, for example 47 miRNAs, such as 48 miRNAs, for example 49 miRNAs, such as 50 miRNAs. The device according to item 48, wherein said device is a microarray chip. The device according to item 57, wherein said device is a microarray chip comprising DNA probes. The device according to item 57, wherein said device is a microarray chip comprising antisense miRNA probes. The device according to item 48, wherein said device is a QPCR

Microfluidic Card. The device according to item 48, wherein said device comprises QPCR tubes, QPCR tubes in a strip or a QPCR plate. The device according to item 48, wherein said device comprises probes on a solid support. The device according to item 48, wherein said device comprises probes on at least one bead. The device according to item 48, wherein said device comprises probes in liquid form in a tube. A kit-of-parts comprising the device of item 48, and at least one additional component. The kit according to item 65, wherein said additional component is means for extracting RNA, such as miRNA, from a sample. The kit according to item 65, wherein said additional component is reagents for performing microarray analysis. The kit according to item 65, wherein said additional component is reagents for performing QPCR analysis. The kit according to item 65, wherein said additional component is the computer program product according to item 82. The kit according to item 65, wherein said additional component is instructions for use of the device. A model for predicting the diagnosis of an individual with a thyroid nodule, comprising

i) providing a set of input data to the miRNA classifier according to item 1 , and

ii) determining if said individual has a condition selected from the group of thyroid follicular adenoma and thyroid follicular carcinoma, or the group of minimally invasive thyroid follicular carcinoma and widely invasive thyroid follicular carcinoma. The model according to item 71 , wherein said input data comprises or consists of the miRNA expression profile of hsa-miR-1826 and hsa- miRPIus-E1078. The model according to item 71 , wherein said input data comprises or consists of the miRNA expression profile of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa- miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa- miR-326 and hsa-miR-330-3p. The model according to item 71 , wherein said input data comprises the miRNA expression profile of hsa-miRPIus-E1001 and hsa-miR-410. 75. The model according to item 71 , wherein said input data comprises or consists of the miRNA expression profile of one or more of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542- 3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR- 22, hsa-miR-146a, hsa-miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR- 363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR- 200c, hsa-miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa- miR-377, hsa-miR-518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c- 5p/hsa-miR-5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and/or hsa-miR-637.

76. The model according to item 71 , wherein said input data further comprises the miRNA expression profile of one or more additional miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR- 450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa- miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR- 199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa- miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa- miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR- 140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR- 101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR- 518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR- 5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR- 584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR- 631 and hsa-miR-637.

77. The model according to item 76, wherein said additional miRNAs comprise no more than 10 additional miRNAs, such as 9 additional miRNAs, for example 8 additional miRNAs, such as 7 additional miRNAs, for example 6 additional miRNAs, such as 5 additional miRNAs, for example 4 additional miRNAs, such as 3 additional miRNAs, for example 2 additional miRNAs, such as 1 additional miRNA.

78. The model according to items 71 to 73, wherein said condition is selected from the group of thyroid follicular adenoma and thyroid follicular carcinoma.

79. A system for determining the presence of a malignant condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of at least one miRNA in said sample, wherein said expression level of said miRNAs is associated with thyroid follicular carcinoma, wherein said at least one miRNA is selected from the group consisting of

i) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

iv) hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR-450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa-miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR-199a- 3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa-miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa- miR-375, hsa-miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa- miR-1201 , hsa-miR-140-5p, hsa-miR-126, hsa-miR-126^*, hsa-let-7f- 2^*, hsa-miR-148b, hsa-miR-21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa- miR-145^*, hsa-miR-513b, hsa-miR-101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR-518e^*/hsa-miR-519a^*/hsa- miR-519b-5p/hsa-miR-519c-5p/hsa-miR-5227hsa-miR-523^*, hsa- miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR-584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR-631 and hsa-miR-637.

80. A system for determining the presence of a benign condition in a sample obtained from a thyroid nodule of an individual, said system comprising means for analysing the expression level of at least one miRNA in said sample, wherein said expression level of said miRNAs is associated with thyroid follicular adenoma, wherein said at least one miRNA is selected from the group consisting of

i) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

81 . A system for performing a diagnosis on an individual with a thyroid nodule, comprising: i) means for analysing the miRNA expression profile of the thyroid nodule, and

i) hsa-miR-1826 and hsa-miRPIus-E1078, or

iii) hsa-miRPIus-E1001 and hsa-miR-410, or

82. A computer program product having a computer readable medium, said computer program product providing a system for predicting the diagnosis of an individual with a thyroid nodule, said computer program product comprising means for carrying out any of the steps of any of the methods according to any of items 79 to 81 .

83. A system according to any of items 79 to 81 , wherein the data is stored, such as stored in at least one database.

Claims

A two-way miRNA classifier for characterising a sample obtained from a thyroid nodule of an individual,

wherein said miRNA classifier consists of the group consisting of hsa-miR-

19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-

15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-

320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p,

wherein said miRNA classifier distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma,

wherein said miRNA classifier comprises at least all miRNAs from the group consisting of hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR-106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR-330-3p, wherein said miRNA classifier distinguishes between the classes thyroid follicular adenoma and thyroid follicular carcinoma,

wherein said miRNA classifier consists of the group consisting of hsa-miR- 1826 and hsa-miRPIus-E1078,

wherein said miRNA classifier comprises at least all miRNAs from the group consisting of hsa-miR-1826 and hsa-miRPIus-E1078,

wherein said miRNA classifier consists of the group consisting of hsa- miRPIus-E1001 and hsa-miR-410,

wherein said miRNA classifier distinguishes between the classes widely invasive thyroid follicular carcinoma and minimally invasive thyroid follicular carcinoma,

wherein said miRNA classifier comprises at least all miRNAs from the group consisting of hsa-miRPIus-E1001 and hsa-miR-410,

The miRNA classifier according to any of claims 2, 4 or 6, wherein said miRNA classifier further comprises one or more additional miRNAs selected from the group consisting of hsa-miR-512-3p, hsa-miR-886-5p, hsa-miR- 450a, hsa-miR-301 b, hsa-miR-429, hsa-miR-542-3p, hsa-miR-130a, hsa- miR-146b-5p, hsa-miR-199a-5p, hsa-miR-193a-3p, hsa-miR-152, hsa-miR- 199a-3p/hsa-miR-199b-3p, hsa-miR-424, hsa-miR-22, hsa-miR-146a, hsa- miR-339-3p, hsa-miR-365, hsa-let-7i^*, hsa-miR-363^*, hsa-miR-148a, hsa- miR-299-3p, hsa-let-7a^*, hsa-miR-200b, hsa-miR-200c, hsa-miR-375, hsa- miR-451 , hsa-miR-144, hsa-let-7i, hsa-miR-1826, hsa-miR-1201 , hsa-miR- 140-5p, hsa-miR-126, hsa-miR-126^*,hsa-let-7f-2^*, hsa-miR-148b, hsa-miR- 21 ^*, hsa-miR-342-3p, hsa-miR-27a, hsa-miR-145^*, hsa-miR-513b, hsa-miR- 101 , hsa-miR-26a, hsa-miR-24, hsa-miR-30a^*, hsa-miR-377, hsa-miR- 518e7hsa-miR-519a7hsa-miR-519b-5p/hsa-miR-519c-5p/hsa-miR- 5227hsa-miR-523^*, hsa-miR-222^*, hsa-miR-452, hsa-miR-665, hsa-miR- 584, hsa-miR-492, hsa-miR-744, hsa-miR-662, hsa-miR-219-2-3p, hsa-miR- 631 and hsa-miR-637.

The miRNA classifier according to claim 7, wherein said additional miRNAs comprise 10 additional miRNAs, such as 9 additional miRNAs, for example 8 additional miRNAs, such as 7 additional miRNAs, for example 6 additional miRNAs, such as 5 additional miRNAs, for example 4 additional miRNAs, such as 3 additional miRNAs, for example 2 additional miRNAs, such as 1 additional miRNA.

The miRNA classifier according to any of claims 1 to 4, wherein an alteration of the expression profile of said miRNAs is associated with thyroid follicular adenoma or thyroid follicular carcinoma.

10. The miRNA classifier according to claims 3 and 4, wherein the up-regulation of hsa-miR-1826 expression and up-regulation of hsa-miRPIus-E1078 expression is indicative of thyroid follicular adenoma; and/or wherein the down-regulation of hsa-miR-1826 expression and down-regulation of hsa- miRPIus-E1078 expression is indicative of thyroid follicular carcinoma.

1 1 . The miRNA classifier according to claims 5 and 6, wherein an alteration of the expression profile of said miRNAs is associated with widely invasive thyroid follicular carcinoma or minimally invasive thyroid follicular carcinoma.

2. The miRNA classifier according to claims 5 and 6, wherein the up-regulation of hsa-miR-410 expression and down-regulation of hsa-miRPIus-E1001 expression is indicative of minimally invasive thyroid follicular carcinoma; and/or wherein the down-regulation of hsa-miR-410 expression and up- regulation of hsa-miRPIus-E1001 expression is indicative of widely invasive thyroid follicular carcinoma.

3. The miRNA classifier according to any of claims 1 to 6, wherein the

expression level of one or more miRNAs is determined by the microarray technique or by the quantitative polymerase chain reaction (QPCR) technique.

4. The miRNA classifier according to any of claims 1 to 6, wherein the sample is extracted from an individual by fine-needle aspiration; by coarse-needle aspiration; by thyroid surgery or by thyroid biopsy.

5. A method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising measuring the expression profile of a group of miRNAs in a sample obtained from the thyroid of said individual, wherein said group of miRNAs consists of the group consisting of iii) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p; or

iv) hsa-miR-1826 and hsa-miRPIus-E1078,

16. A method for diagnosing if an individual has, or is at risk of developing, follicular thyroid carcinoma, comprising measuring the expression profile of a group of miRNAs in a sample obtained from the thyroid of said individual, wherein said group of miRNAs comprises at least all miRNAs from the group consisting of

iii) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p; or

iv) hsa-miR-1826 and hsa-miRPIus-E1078,

17. A method for diagnosing if an individual has, or is at risk of developing, widely invasive thyroid follicular carcinoma, comprising measuring the expression profile of a group of miRNAs in a sample obtained from the thyroid of said individual, wherein said group of miRNAs consists of the group consisting of

ii) hsa-miRPIus-E1001 and hsa-miR-410,

18. The method according to any of claims 15 to 17, comprising any one or more of the steps of

1 ) obtaining a sample from the thyroid of an individual,

2) extracting RNA from a sample collected from the thyroid of an individual,

3) analysing the miRNA expression profile of the sample,

4) obtaining prediction probabilities of between 0-1 for said sample, and/or

5) determining if said individual has a condition selected from follicular thyroid adenoma and follicular thyroid carcinoma; or widely invasive thyroid follicular carcinoma and minimally invasive thyroid follicular carcinoma.

19. A device for measuring the expression level of at least a group of miRNAs in a sample, wherein said device consists of one or more probes or probe sets for the miRNAs consisting of the group consisting of

iv) hsa-miR-19a, hsa-miR-501 -3p, hsa-miR-17, hsa-miR-335, hsa-miR- 106b, hsa-miR-15a, hsa-miR-16, hsa-miR-374a, hsa-miR-542-5p, hsa-miR-503, hsa-miR-320a, hsa-let-7i, hsa-miR-326 and hsa-miR- 330-3p; or

v) hsa-miR-1826 and hsa-miRPIus-E1078; or

vi) hsa-miRPIus-E1001 and hsa-miR-410,

20. The device according to claim 19, wherein said device may be used with the miRNA classifier according to any of claims 1 to 6 to classify a sample into either of the classes of thyroid follicular adenoma and thyroid follicular carcinoma; or minimally invasive thyroid follicular carcinoma and widely invasive thyroid follicular carcinoma.

21 . The device according to claim 19, wherein said device is a microarray chip.

22. The device according to claim 19, wherein said device is a QPCR

Microfluidic Card; comprises QPCR tubes, QPCR tubes in a strip or a QPCR plate.

23. A model for predicting the diagnosis of an individual with a thyroid nodule, comprising

i) providing a set of input data to a miRNA classifier according to any of claims 1 to 6, and

ii) determining if said individual has a condition selected from the group of thyroid follicular adenoma and thyroid follicular carcinoma; or the group of minimally invasive thyroid follicular carcinoma and widely invasive thyroid follicular carcinoma.

24. A system for performing a diagnosis on an individual with a thyroid nodule, comprising:

a) means for analysing the miRNA expression profile of the thyroid nodule, and

b) means for determining if said individual has a condition selected from follicular thyroid adenoma and follicular thyroid carcinoma; or minimally invasive thyroid follicular carcinoma and widely invasive thyroid follicular carcinoma;

wherein said miRNA expression profile consists of miRNAs from the group consisting of

ii) hsa-miR-1826 and hsa-miRPIus-E1078; or

iii) hsa-miRPIus-E1001 and hsa-miR-410.

25. A computer program product having a computer readable medium, said computer program product providing a system for predicting the diagnosis of an individual with a thyroid nodule, said computer program product comprising means for carrying out any of the steps of any of the methods according to claim 24.

26. A system according to claim 24, wherein the data is stored, such as stored in at least one database.