WO2021116938A1

WO2021116938A1 - A new method for diagnosing a thyroid tumour and kit thereof

Info

Publication number: WO2021116938A1
Application number: PCT/IB2020/061699
Authority: WO
Inventors: Sebastiano FILETTI; Cosimo DURANTE; Antonella VERRIENTI; Marialuisa SPONZIELLO; Giorgio GRANI
Original assignee: Università Degli Studi Di Roma "La Sapienza"
Priority date: 2019-12-13
Filing date: 2020-12-09
Publication date: 2021-06-17
Also published as: IT201900024009A1; EP4073274A1

Abstract

A new method for the in vitro diagnosis of a thyroid tumour, comprising the steps of identifying the presence of gene mutations and identifying the expression of the levels of a specific miRNA, and relative kit.

Description

“A new method for diagnosing a thyroid tumour and kit thereof”

* * *

FIELD OF THE INVENTION The present invention relates to a new method and relative kit for diagnosing a thyroid tumour, capable of overcoming the intrinsic diagnostic limits of cytology, the current reference methodology, and personalizing the treatment of patients with thyroid tumour. BACKGROUND OF THE INVENTION

In recent decades, the epidemiology of the thyroid nodular pathology has changed, with a progressive increase in the finding of thyroid nodules in the adult population. Over 90% of nodules represent benign lesions, which will never become clinically significant. However, to rule out malignancy, laboratory tests, ultrasonography and, for a selected population of patients, needle aspiration cytology are required. This approach reliably establishes a diagnosis of malignancy or benignity in approximately 70-80% of cases. In the remaining 20-30% of cases, there is an indeterminate cytological report: a high percentage of these patients undergo surgery for diagnostic purposes only, rather than a curative intervention, since only a minority of these nodules they will be diagnosed as malignant on final histological examination. (1 )

The need to improve the diagnostic accuracy of cytology has led to the search for molecular markers that can help exclude or confirm the risk of malignancy in thyroid nodules. These markers are specific mutations of the thyroid tumour genome and/or their microRNA/mRNA expression profile. In the past decade, four molecular tests have been developed in the USA and are now commercially available for evaluating cytology samples: Afirma Genomic Sequencing Classifiers (GSC) (2), ThyroSeq v.3 Genomic Classifier (3, 4), ThyGeNEXT®/ThyraMIR ™ (5, 6), and Rosetta GX Reveal (7). Although their objectives are the same, these tests vary substantially from each other in terms of molecular markers tested (mutations, mRNA or microRNA expression); the methodology of analysis {next- generation sequencing, castPCR, RT-qPCRf, the type of sample examined (same sample used for cytological diagnosis or need for additional dedicated sampling) and their performance in terms of diagnostic accuracy. Table 1. Comparison of the characteristics of commercial clinical tests and their performance in cytologically indeterminate thyroid lesions

ThyroSeq GC (3, 14) Aflrma ThyGeNEXT®/ Rosetta GX

GSC (2) ThyraMIR™ (5, 6)^a Reveal(7)

Method NGS NGS NGS RT-qPCR castPCR RT-qPCR

Quantity 10 ng DNA 15 ng RNA 20 - 40 ng TNA 20 ng RNA 10 ng RNA 5 ng DNA

Sensitivity % 94.1 91.1 89 74

Spei ilii i

68.3

NPV % 97.3 96.1 94 92

^a ThyGeNEXT has not yet been validated. Test performance reported refers to the ThyraMIR test. Abbreviations: SNV, single nucleotide variations; NGS, next-generation sequencing; TNA, total nucleic acids; NPV, negative predictive value; PPV, positive predictive value.

The need is therefore felt for a diagnostic method that allows the disadvantages in terms of sensitivity and specificity of the currently available methods to be overcome.

OBJECT AND SUMMARY OF THE INVENTION

The object of the present invention is to provide a diagnostic method based on the determination of gene mutations and on the determination of the expression levels of a specific microRNA, which allow discrimination of patients with thyroid cancer from those with benign pathologies.

According to the invention, the above object is achieved thanks to the subject specified in the following claims, which are intended as an integral part of the present description. In one embodiment, the present invention relates to a method for the in vitro diagnosis of a thyroid tumour comprising the following steps:

(a) extracting DNA and RNA from a thyroid biological sample;

(b) generating from the DNA obtained in step (a) a first DNA library comprising a first plurality of amplicons obtained by means of multiplex PCR amplification of a first set of genes, wherein the first set of genes includes: BRAF exon 15, TSFIR exon 10, RET exons 5, 8, 10, 11, 13, 14, 15, 16, TERT promoter, ALK exons 20-29, the entire encoding region of the genes HRAS, KRAS, NRAS, EIF1AX, CHECK2, ATM, PTEN, PI3KCA, TP53, TG, DICERl, DNMT3A, MET, and SETD2 ;

(c) generating from the RNA obtained in step (a) the corresponding complementary DNA and, from the complementary DNA thus obtained, generating a second DNA library comprising a second plurality of amplicons obtained by means of multiplex PCR amplification of: (i) a second set of genes, wherein the second set of genes includes: the RET, ALK, NTRK1, NTRK3, PAX8, BRAF, THADA genes with a fusion partner as indicated in Table 6B and (ii) a third set of control genes, wherein the third set of control genes includes the genes: KRT7, KRT20, TG, NIS, TPO, TSHR, TTF1, CALCA, PTH, TBP, JUN and MTTB- (d) sequencing each amplicon obtained in steps (b) and (c); and

(e) determining the positivity of the thyroid biological sample from step (a) if at least one of the following conditions is met:

(i) at least one gene mutation has been identified in step (d) in the first DNA library; (ii) at least one gene mutation has been identified in step (d) in the second DNA library.

In one embodiment, the present invention also relates to a kit for carrying out the above-described method.

The present invention relates to a diagnostic method for distinguishing benign thyroid nodules from malignant thyroid nodules after the fine needle aspiration procedure by means of a combined molecular method. As demonstrated by the studies of the research program TCGA (The Cancer Genome Atlas), multi platform approaches that include mutational analysis of tumours and their microRNA/mRNA expression profile can help define clinically relevant thyroid tumour subclasses (8, 9). The use of a combination of different methods may offer multiple markers for diagnosing cancer in pre-surgical specimens (from fine needle aspiration) by increasing the positive predictive value compared to single markers (e.g. in cases with RAS mutation) or by identifying carcinomas pre-operatively with a greater potential for clinical aggression (e.g. in cases with multiple somatic mutations). In particular, the method subject of the present description allows identification of thyroid tumorigenesis driver mutations, and the expression of a microRNA specifically expressed at high levels in thyroid tumour tissues of follicular origin by exploiting the methods of Next-Generation Sequencing (NGS) and digital PCR, respectively. The presence of a mutation (point, indel, gene fusion) and/or the expression of the microRNA above respective threshold values is indicative of the presence of a malignant tumour. This method allows a personalized approach for treating nodular thyroid disease, improving the diagnostic accuracy of the cytology in indeterminate thyroid lesions and the prognostic stratification of patients.

DETAILED DESCRIPTION OF THE INVENTION

In the following description, numerous specific details are provided in order to allow a thorough understanding of the embodiments. The embodiments can be implemented without one or more of the specific details or with other methods, components, materials etc. In other cases, well-known structures, materials or operations are not shown or described in detail to avoid confusing aspects of the embodiments.

Reference made throughout the present disclosure to “one embodiment” or “an embodiment” indicates that a particular aspect, structure or characteristic described with reference to the embodiment is included in at least one embodiment. Therefore, forms of the expressions “in one embodiment” or “in an embodiment” at various points throughout the present description are not necessarily all referring to the same embodiment. Moreover, particular aspects, structures or characteristics can be combined in any convenient way in one or more embodiments.

The titles shown here are for convenience only and do not interpret the scope or meaning of the embodiments.

The terms “amplify” or “amplification” referring to nucleic acid sequences - as used in the present description - refer to methods that increase the amount of nucleic acid in a sample. The term “amplicon” as used in the present description refers to a nucleic acid sequence generated in vitro in an amplification reaction.

The terms “adapter” or “adapter sequence” as used in the present description refer to a small nucleic acid sequence that is bound to one end of a nucleic acid molecule in order to facilitate binding of another molecule to said nucleic acid molecule.

The terms “barcode” or “barcode sequence” as used in the present description refer to a small nucleic acid sequence that is bound to one end of a nucleic acid molecule in order to allow the univocal identification of said nucleic acid when mixed with other nucleic acid molecules.

The terms “complementary” or “complementarity” as used in the present description in reference to nucleic acid sequences refer to nitrogenous base pairing rules, well known to those skilled in the art. The pairing of nitrogenous bases does not necessarily have to be perfect between two complementary sequences to obtain a stable double strand, but unpaired nitrogenous bases may be present. The person skilled in the art has the necessary skills to empirically evaluate the stability of a double strand of nucleic acid.

The term “library” in the way it is used in the present description refers to a collection of amplicons deriving from the amplification reaction of DNA or RNA fragments to which adapters and barcodes are preferably bound.

The term “multiplex PCR” as used in the present description refers to an amplification of a plurality of nucleic acid molecules, wherein each nucleic acid molecule is amplified by means of a different primer pair from the primer pairs used for other nucleic acid molecules. It is to be understood that the multiplex PCR reaction may also be performed on amplicons obtained from a previous PCR reaction.

The terms “Next-generation sequencing” or “NGS” as used in the present description refer to any sequencing method that determines the nucleotide sequence of single nucleic acid molecules or high-throughput parallel clonally amplified sequences. In one embodiment, the relative abundance of the nucleic acid species in the library may be estimated by counting the relative number of occurrences of their related sequences in the data generated by the sequence experiment.

The term “analytical sensitivity” in the way it is used in the present description with reference to the method subject of the present invention refers to the lowest allele frequency (see below the definition of “allelic frequency”) at which it is possible to detect a variant of a pre-selected sequence in a heterogeneous sequence library.

The term “clinical sensitivity” in the way it is used in the present description with reference to the method subject of the present invention refers to a measure of the capacity of the method to correctly identify subjects affected by thyroid tumours. The sensitivity is calculated here using the mathematical formula VP/(VP+FN), wherein VP = True Positive and FN = False Negative.

The term “analytical specificity” in the way it is used in the present description with reference to the method subject of the present invention refers to a measure of the capacity of the method to distinguish an actually existing preselected sequence variant from sequencing artefacts or other closely related sequences. It is the capacity to avoid false positive detections. False positive detections may result from errors introduced into the sequence of interest during sample preparation, sequencing errors, or inadvertent sequencing of closely related sequences such as pseudo-genes or members of a gene family.

The term “clinical specificity” in the way it is used in the present description with reference to the method subject of the present invention refers to a measure of the capacity of the method to correctly identify subjects not affected by thyroid tumours. The specificity is calculated here using the mathematical formula VN/(VN + FP), wherein VN = True Negative and FP = False Positive.

The term “Positive Predictive Value (PPV)” as used in the present description refers to the probability that a subject tested positive at the diagnostic method is actually affected by a thyroid tumour. The Positive Predictive Value is calculated here using the mathematical formula VP/(VP + FP), wherein VP = True Positive and FP = False Positive.

The term “Negative Predictive Value (NPV)” as used in the present description refers to the probability that a subject tested negative at the diagnostic method is actually a subject that is not affected by a thyroid tumour. The Negative Predictive Value is calculated here using the mathematical formula VN/(VN + FN), where VN = True Negative and FN = False Negative.

The term “Allelic Frequency (AF)” as used in this description refers to a measure of the relative frequency of a mutated allele over the total of alleles sequenced, as calculated by the Torrent Suite v.5.10 software. The term “Minor Allelic Frequency (MAF)” as used in the present description refers to a measure of the frequency of the rarest variant of an allele at a genetic locus in the population, as noted by the software wANNOVAR.

The term “read depth (DP)” as used in the present description refers to the number of times a nucleotide is read during sequencing, as calculated by the software Torrent Suite v.5.10.

The term “genotype quality (GQ)” as used in this description refers to a measure of confidence that the genotype assigned to the sequenced sample is correct, as calculated by the software Torrent Suite v.5.10.

The term “strand bias (STB)” as used in this description refers to the discrepancy between the allele frequencies on the forward and reverse strands, as calculated by the software Torrent Suite v.5.10.

In one embodiment, the present invention concerns a method for the in vitro diagnosis of a thyroid tumour comprising the following steps:

(a) extracting DNA and RNA from a thyroid biological sample; (b) generating from the DNA obtained in step (a) a first DNA library comprising a first plurality of amplicons obtained by means of multiplex PCR amplification of a first set of genes, wherein the first set of genes includes: BRAF exon 15, TSHR exon 10, RET exons 5, 8, 10, 11, 13, 14, 15, 16, TERT promoter, ALK exons 20-29, the entire encoding region of the genes HRAS, KRAS, NRAS, EIF1AX, CHECK2, ATM, PTEN, PI3KCA, TP53, TG, DICERl, DNMT3A, MET, and SETD2 ;

(c) generating from the RNA obtained in step (a) the corresponding complementary DNA and, from the complementary DNA thus obtained, generating a second DNA library comprising a second plurality of amplicons obtained by means of multiplex PCR amplification of: (i) a second set of genes, wherein the second set of genes includes: the RET, ALK, NTRK1, NTRK3, PAX8, BRAF, THADA genes with a fusion partner as indicated in Table 6B and (ii) a third set of control genes, wherein the third set of control genes includes the genes: KRT7, KRT20, TG, NIS, TPO, TSHR, TTF1, CALCA, PTH, TBP, JUN and MTTB-,

(d) sequencing each amplicon obtained in steps (b) and (c); and

In one embodiment, the method comprises an additional step (f), which envisages determining, through a digital PCR reaction, the expression of miRNA- 146b-5p and of an endogenous control in the thyroid biological sample of step (a), wherein the biological sample of step (a) is positive if the expression of miRNA- 146b-5p has been identified.

In one embodiment, the step (e-i) of the method envisages determining whether: (1) the gene mutation has a minor allele frequency (MAF) < 0.005 in the population of European descent, as reported in the databases lOOOgenome, ExAC, ESP, gnomAD;

(2) the gene mutation has a read depth (DP) > 500, a genotype quality (GQ) > 30, a strand bias (STB) comprised between 0.3 and 0.7 and an allele frequency (AF) > 0.05; and

(3) the gene mutation has an allele frequency (AF) comprised between 0.05 and 0.40 or between 0.60 and 0.80, with the exclusion of all gene mutations present in the RET gene, and in the known hotspots of the BRAF and RAS genes; wherein the positivity of the sample is determined by the fulfilment of all the conditions (l)-(3).

In one embodiment, the step (e-ii) of the method envisages determining whether:

(1) at least 20,000 sequences selected from the sequences included in Table 6A and the control genes have been sequenced; (2) at least 20% of the sequenced sequences are positive for at least the

TG, NIS, TPO, TSHR, TTF1, CAECA, and 1Ή control genes; and

(3) at least 20 sequenced sequences comprise at least one mutation identified in Table 6B; wherein the positivity of the sample is determined by the fulfilment of all the conditions (l)-(3).

In one embodiment, the step (f) of the method comprises determining the quantitative expression level of the miRNA-146b-5p and of an endogenous control, preferably represented by the U6 nuclear RNA; the positivity of the sample is determined if the expression level is higher than a threshold equal to 0.1562. In one embodiment, the first DNA library obtained in step (b) of the method is generated by using the primer pairs indicated in Table 4.

In one embodiment, the second DNA library obtained in step (c) is generated by using the primer pairs indicated in Table 5. In one embodiment, the digital PCR reaction of the step (f) is carried out by using a primer pair capable of hybridizing to the miRNA-146b-5p (target sequence: UGAGAACUGAAUUCCAUAGGCU; chromosomal location on the reference genome GRCh38: chrlO: 102436520-102436541; Sequence ID in the miRBASE v.22 database: MIMAT0003475; SEQ ID No.: 3), preferably the primer pair is the one contained in the commercial kit Cat. # ID: 001097, Thermo Fisher Scientific. Preferably, the digital PCR reaction of step (f) is carried out by means of the use of two primer pairs: the first primer pair hybridizes to miRNA- 146b-5p (SEQ ID No.: 3), the second primer pair hybridizes to the endogenous control represented by the U6 nuclear RNA (target sequence: GTGCTCGCTTCGGCAGCACATATACTAAAATTGGAACGATACAGAGAA GATTAGCATGGCCCCTGCGCAAGGATGACACGCAAATTCGTGAAGCGT TCCATATTTT; chromosomal localization on the reference genome GRCh38: chrl5: 67839940-67840045; Sequence ID in the GenBank database: NR_004394; SEQ ID No.: 4), preferably the two primer pairs are those contained, respectively, in the commercial kits Cat. # ID: 001097 and Cat. # ID:001973, Thermo Fisher Scientific.

In one embodiment, step (b) comprises the additional steps of: b-i) enzymatically digesting the first plurality of amplicons obtained in step (b) to phosphorylate the ends of the first plurality of amplicons; b-ii) ligating an adapter sequence to each amplicon obtained in step (b-i), b-iii) quantifying by means of a Real Time PCR reaction each amplicon obtained in step (b-ii).

In one embodiment, step (c) comprises the additional steps of: c-i) enzymatically digesting the second plurality of amplicons obtained in step (c) to phosphorylate the ends of the second plurality of amplicons; c-ii) ligating an adapter sequence to each amplicon obtained in step (c-i), c-iii) quantifying each amplicon obtained in step (c-ii), by means of a Real Time PCR reaction.

In a preferred embodiment, steps (b-i) and (c-i) also envisage binding a barcode sequence to each amplicon obtained in steps (b-i) and (c-i), respectively. In one embodiment, each amplicon obtained, respectively, in steps (b-ii) and (c-ii) is immobilized on a respective bead having at least one nucleotide sequence complementary to the adapter sequence bound to its outer surface.

In one embodiment, prior to step (d), the beads on the surface of which at least one amplicon is present are subjected to an emulsion monoclonal PCR amplification reaction, obtaining - for each amplicon - a respective bead on the surface of which multiple copies of the amplicon are present. Preferably, during the emulsion monoclonal PCR amplification reaction a biotin molecule is bound to the 5' end of each copy of the amplicon. In one embodiment prior to step (d), beads on the surface of which multiple copies of the amplicon are present are separated from beads on which no copies of an amplicon are present.

In one embodiment, before step (b), a PCR amplification reaction of the DNA extracted in step (a) is carried out using the pair of primers having the sequences indicated in SEQ ID No.: 1 and 2.

In one embodiment, the thyroid biological sample of step (a) is a thyroid sample from thyroid fine needle aspiration (FNA), a thyroid sample from formalin-fixed paraffin-embedded (FFPE) tissue, fresh-frozen tumour tissue (FF).

In one embodiment, the thyroid tumour is selected from a papillary thyroid carcinoma, a non-invasive follicular neoplasm with papillary-like nuclear features; an angioinvasive oncocytic thyroid carcinoma; a poorly differentiated insular thyroid carcinoma; a medullary thyroid carcinoma, and an anaplastic thyroid carcinoma.

The present description also describes a method for selecting a subject, having one or more thyroid nodules with indeterminate outcome of the cytological examination, for a surgical intervention to surgically remove one or more thyroid nodules, comprising the steps of:

(a) extracting DNA and RNA from a thyroid biological sample taken from the subject; (b) generating from the DNA obtained in step (a) a first DNA library comprising a first plurality of amplicons obtained by means of multiplex PCR amplification of a first set of genes, wherein the first set of genes includes: BRAF exon 15, TSFIR exon 10, RET exons 5, 8, 10, 11, 13, 14, 15, 16, TERT promoter, ALK exons 20-29, the entire encoding region of the genes HRAS, KRAS, NRAS, EIF1AX, CHECK2, ATM, PTEN, PI3KCA, TP53, TG, DICERl, DNMT3A, MET, and SETD2 ;

(c) generating from the RNA obtained in step (a) the corresponding complementary DNA and, from the complementary DNA thus obtained, generating a second DNA library comprising a second plurality of amplicons obtained by means of multiplex PCR amplification of: (i) a second set of genes, wherein the second set of genes includes: the RET, AEK, NTRK1, NTRK3, PAX8- PPARG, BRAF, THADA genes with a fusion partner as indicated in Table 6B, and (ii) a third set of control genes, wherein the third set of control genes includes the genes: KRT7, KRT20, TG, NIS, TPO, TSHR, TTF1, CAECA , PTH, TBP, JUN and MTTB

(d) sequencing each amplicon obtained in steps (b) and (c); and

(e) selecting the subject for surgery if at least one of the following conditions is met:

(i) at least one gene mutation has been identified in step (d) in the first DNA library;

(ii) at least one gene mutation has been identified in step (d) in the second DNA library;

The method for selecting a subject, having one or more thyroid nodules with indeterminate outcome of the cytological examination, for a surgical intervention to surgically remove one or more thyroid nodules may also include a further step (f) which involves determining - through a digital PCR reaction - the expression of miRNA-146b-5p and an endogenous control in the thyroid biological sample of step (a), wherein the thyroid sample of step (a) is positive if miRNA-146b-5p expression has been identified. In one embodiment, the present invention also relates to a kit for carrying out the method for the in vitro diagnosis of a thyroid tumour subject of the present invention, comprising:

- primer pairs having the sequences indicated in Table 4, and a primer pair having the sequences indicated in SEQ ID No.: 1 and 2, - primer pairs having the sequences indicated in Table 5, and

- instructions for use.

In one embodiment, the kit may further comprise a primer pair capable of hybridizing to the SEQ ID No.: 3 sequence of the miRNA-146b-5p, preferably the primer pair is that contained in the commercial kit Cat. # ID: 001097, Thermo Fisher Scientific. It is to be understood that the kit is suitable for carrying out the method for the in vitro diagnosis of a thyroid tumour in any of the embodiments of the method described above.

In the preferred embodiment, the kit also comprises at least one primer pair capable of hybridizing to the SEQ ID No.: 4 sequence of the U6 nuclear RNA, preferably the primer pair is that contained in the commercial Cat kit. # ID:001973, Thermo Fisher Scientific.

The method and the respective kit are able to detect point mutations up to an allelic frequency of 5%, and gene fusions up to a dilution of 1:39, both when thyroid cancer cells are diluted in normal thyroid cells and when diluted in cells of different cell types, such as whole blood, showing that the presence of normal thyroid cells or blood cells does not interfere with the molecular dilutions tested. All known genetic alterations tested were accurately identified with a minimal change in the allelic frequency of the point mutations (coefficient of variation 0.01-0.02%), or a minimal change in the number of sequences that detect the presence of gene fusions (coefficient of variation 0.6-1.0). The minimum quantity of nucleic acids tolerated by this method is 1 ng, quantity at which the method is able to correctly identify the genetic alterations tested, and to classify them as positive in 100% of the tested samples, both for the samples isolated from thyroid fine needle aspirates and for samples isolated from paraffin-embedded tumour tissue.

The results of the diagnostic method subject of the present description (Table 2) in the biological samples tested (thyroid samples from thyroid fine needle aspiration) show that the present method is able to improve the diagnostic accuracy of the cytology (method currently in use in current clinical practice), reducing the number of indeterminate thyroid nodules, and reliably increasing the number of nodules that will be reclassified as benign. In the subgroup of cytologically indeterminate lesions evaluated with this method, where the prevalence of cancer was 22.5%, the method showed a maximum sensitivity values and NPV (100%), making it a powerful tool to exclude malignancy in these thyroid lesions, and to reduce the rate of unnecessary thyroid surgery: in the series of patients analysed, the method described here would have avoided 23 unnecessary surgeries, equal to 57.5% of the surgeries performed. If the estimate of the prevalence of ultrasound-detectable thyroid nodules in the adult population of the cities of Rome and Milan are considered, 358,052 and 173,817 patients, respectively, would be candidates to perform an ultrasound-guided thyroid fine needle aspiration according to the ACR TIRADS classification (13), the most limiting of the ultrasound systems. Among these nodules, those with indeterminate cytology (TIR3A and TIR3B, in the Italian cytological classification system SIAPEC), the candidates for the molecular test would be 55,498 and 26,942. Considering that only 17.2% of indeterminate cytology low- risk TIR3A thyroid nodules, and 75% of indeterminate high-risk TIR3B nodules undergo surgery, and considering the estimated malignancy rate in each category (17.5% and 42.4%, respectively), the application of this method to the adult population of Rome and Milan would have avoided 12,549 total thyroidectomy or lobectomy operations in the city of Rome, and 6,092 operations in the city of Milan (Table 3), with a strong reduction in costs for the national health system.

Table 2. Performance of this diagnostic method in clinical specimens: subgroup of 40 nodules at indeterminate cytological classes, TIR3A and

TIR3B

Abbreviations Cl, confidence interval; FP, false positive; FN, false negative; VP, true positive; VN, true negative; NPV, negative predictive value - VN/(VN+FN); PPV, positive predictive value - VP/(VP+FP); Sens., Sensitivity - VP/(VP + FN); Spec., Specificity - VN/(VN + FP).

Table 3. Estimation of the Prevalence of Thyroid Nodules detectable by ultrasound in the populations of the cities of Rome and Milan; potential candidate nodules for the diagnostic method subject of the application, and potential avoidable surgeries

¹ Guth S et al. Eur J Clin Invest. 2009 Aug; 39(8):699-706.

²Grani G et al. J Clin Endocrinol Metab. 2019 Jan 1; 104(1):95- 102. 3 Sparano C et al J Endocrinol Invest. 2019 Jan;42(l):l-6.

Comparison with other commercially available molecular tests (Table 1) shows numerous advantages, including i) a relatively low number of molecular markers tested, which results in lower overall costs (33 genes in total, split between: 14 genes for SNV/Indel; 3 genes for SNV/Indel + gene fusions; 4 genes for gene fusions; 2 genes for SNV/Indel + control genes; 10 control genes and 1 micro RNA); ii) the possibility of performing the test on the same material used for the cytological diagnosis, avoiding subjecting the patient to a second dedicated biopsy; iii) the use of highly sensitive methods both for identifying mutations (Next Generation Sequencing, NGS), and for evaluating the expression levels of the microRNA (digital PCR, dPCR), guaranteeing reliable results with the identification of mutations also if present at low frequency, multiple mutations in the same nodule or still unknown mutations and, finally, the precise quantification of microRNA levels. Finally, iv) this diagnostic method offers a better performance in terms of sensitivity and negative predictive value (equal to 100%).

The results obtained indicate that this method can be considered for its use in clinical practice to improve the management of patients with cytologically indeterminate thyroid nodules by significantly reducing the rate of unnecessary thyroid surgery. The relatively low number of genes tested and the multiplex sequencing capability also allows for the simultaneous analysis of multiple biological samples, reducing costs for each sample.

MATERIALS AND METHODS a) Sample Collection and Nucleic Acid Isolation

To establish inter- and intra-assay precision, analytical sensitivity, and the minimum amount needed to detect a mutation, the molecular test was tested on biological thyroid specimens with a known mutation from thyroid fine needle aspiration (FNA), formalin-fixed paraffin-embedded tumour specimens (FFPE), and fresh-frozen tumour tissues (FF) and cell lines. To test its clinical validity, the molecular test was used in a retrospective series of 118 thyroid nodules from 112 patients undergoing a needle aspiration procedure and subsequently total thyroidectomy or lobectomy. All subjects signed an informed consent form. The cytological samples were obtained starting from two aspirative punctures with needles of 25 or 27g, and processed with the Thin Prep 5000™ (Hologic Co., Marlborough, MA), following standard procedures. The slides obtained were fixed in 95% ethanol and stained with the Papanicolaou system for cytological diagnosis, while the remaining material was stored in Preservcyt™ for subsequent molecular analysis.

DNA and RNA were simultaneously isolated from each cytology sample or cell line using the All Prep DNA/RNA Kit (QIAGEN), from each FFPE sample using the Recover All™ Total Nucleic Acid Isolation Kit (Ambion), and from each fresh-frozen tissue using Trizol (Invitrogen), quantified by the Qubit® fluorometer (Thermo Fisher Scientific) and processed by NGS analysis and digital PCR. b) Next-generation sequencing Sequencing analysis is performed on the next generation Ion GeneStudio

S5 (Thermo Fisher Scientific) and Ion 540-chip sequencer following standard procedures starting from 15 ng of DNA and 10 ng of RNA. In particular, two custom panels that exploit AmpliSeq™ technology (Thermo Fisher Scientific) have been suitably designed using AmpliSeq Designer v.5.4.2 (Thermo Fisher Scientific) software to identify point mutations (SNV) and indel on DNA, and gene rearrangements on RNA. For each panel, the amplification primers are preferably distributed in two pools, to allow the creation of overlapping amplicons and thus favour the complete coverage of the regions to be sequenced. The distribution of the primers in two pools also avoids the possibility of cross- reactions between the primers. The first panel (DNA panel) allows amplification of a set of 19 genes specifically selected to include the main known drivers of thyroid cancer [exon 15 of BRAF, exon 10 of TSHR, exons 5, 8, 10, 11, 13, 14, 15, 16 of RET, the promoter of TERT (hotspot c.1-124 and c.1-146), the exons 20-29 of ALK and the entire coding region of the genes EIRAS, KRAS, NRAS, EIF1AX, CHECK2, ATM, PTEN, PI3KCA, TP53 ] (8, 9); genes mutated in thyroid tumour samples without a known driver and reported in the literature CTG; DICERl; DNMT3A ) (8); genes mutated in thyroid tumour samples without a known driver belonging to the case series of the present inventors {MET, SETD2) (10). The amplification primers are distributed into two pools at a 2x concentration and allow generation of 762 amplicons (Table 4). The second panel (RNA panel) includes primers for amplifying 204 gene rearrangements of seven known thyroid driver genes (RET, ALK, NTRK1, NTRK3, PAX8-PPARG, BRAF, THADA ) with 72 fusion partners and 12 genes control whose expression is evaluated: KRT7 and KRT20, epithelial cell markers; TG, NIS, TPO, TSFIR, TTF1, markers of thyroid follicular cells; CALCA, marker of parafollicular cells; PTH, marker of parathyroid cells; TBP, JUN, MTTB, endogenous markers. The amplification of the control genes is carried out both to ensure the presence of thyroid cells in the sample analysed (excluding any contamination in the fine needle aspiration step with different cell types), and to ensure an adequate number of sequences for each sample, even in the absence of gene fusions. The amplification primers are distributed into two pools at a 5x concentration and allow generation of 216 amplicons (Table 5). The total of screened genes (DNA panel + RNA panel) is 33. The complete list of genes, the type of variation screened for each and the fusion partners of the seven thyroid driver genes are reported in Tables 6 A and 6B.

Table 4. DNA primer panel

^A Genomic coordinates of the amplicon on the reference genome hgl9 (deposited in the database Assembly in NCBI).

^B ID of the amplicon (Themo Fisher Scientific). c Official name of the Gene (HGNC).

5 ^D ID of the gene (GenBank).

^E ID Transcript (GenBank, RefSeq).

^F The position 0 corresponds by convention to the first genomic position listed in column 1.

^G The number corresponds to the last genomic position listed in column 1.

Abbreviations: FW, forward (sense); RV, reverse ( antisense ); NC, non-coding

10

Table 5. RNA primer panel

5 ^D Transcript ID (Ensembl database for the codes whose prefix is ENST#, GenBank database for codes whose prefix is NM_ #).

^E The position 0 corresponds by convention to the first genomic position listed in column 1.

^F The number corresponds to the last genomic position listed in column 1.

Abbreviations: F, forward (sense); R, reverse (antisense); GF, gene fusion; GE, gene expression.

10 Table 6A. List of genes analysed in the method

Abbreviations: CTRL, control; GE, gene expression; SNV, single nucleotide variation; INDEL, small insertion/deletion; GF, gene fusion. Table 6B. Fusion Partner

* Fusion partner for two different driver genes. Stepl. AmpUSeq Libraries Construction.

For each sample, two libraries are built starting from 15 ng of DNA and 10 ng of RNA.

SteplA. DNA AmpUSeq Libraries Construction. The preparation of the DNA library, necessary for detecting point mutations and indels in 19 genes, is carried out using the enzymes supplied in the Ion AmpUSeq™ Library Kit Plus (Thermo Fisher Scientific) in five steps: i. target DNA Amplification: The GC-rich region of the TERT promoter is pre-amplified using a pair of primers (having the sequences for the sense primer 5'-AGGCCGGGCTCCCAGTGGA-3' (SEQ ID NO.: 1) and for the antisense primer 5'- TGGCCGGGGCCAGGGCTTC-3' (SEQ ID No.:2), the reference sequence being that of the TERT gene, reference sequence in GenBank

NC_000005.10) designed outside the amplified region by the TERT primers indicated in Table 4 used in the subsequent PCR-multiplex reaction. A total of 5 ng of genomic DNA is pre-amplified with 10 pmol of each primer (SEQ ID No.: 1 and 2), 5x MyTaq Reaction Buffer and 0.1 pi of MyTaq HS DNA Polymerase (Bioline) in a final volume of 25 pi. Amplification is performed in the Veriti thermal cycler (ThermoFisher Scientific) under the following conditions:

• 95°C for 30 sec, 40 cycles at 95°C for 10 sec, 66°C for 30 sec, 72°C for 20 sec and final extension at 72°C for 5 minutes. The PCR products are purified with the Nucleo Spin PCR & Gel Clean

Up kit (Macherey-Nagel) for removing contaminants such as nucleotides, primers, enzymes, salts; they are quantified with the Nanodrop spectrophotometer and diluted to a final concentration of 1 ng/mΐ. For creating each library, two PCR-multiplexes are prepared in a final volume of 10 mΐ each with Ion AmpliSeq HiFi Mix 5x, 5 ng genomic DNA, 0.25 ng of pre-amplified and purified TERT amplicon, 5 mΐ of primers (split into two pools) at a 2x concentration (Table 4). Amplification is performed in the Veriti thermal cycler (ThermoFisher Scientific) under the following conditions:

• 99°C for 2 min, 17 cycles at 99°C for 15 sec and 60°C for 4 min.

At the end of the reaction, the two multiplex PCRs of each library are combined in a single tube. ii. Enzymatic digestion of amplicons: the amplicons obtained in the previous reaction are enzymatically treated to partially digest the amplification primers and phosphorylate the ends of the amplicons themselves. This process is necessary for the subsequent ligation reaction. In particular, 2 mΐ of FuPa digestion enzyme is added to each library, which is subsequently digested under the following conditions: 50°C for 10 min, 55°C for 10 min, 60°C for 20 min. iii. Ligation of adapters and barcodes: At the two ends of each amplicon a specific adapter {Ion PI Adapter) is covalently ligated at one end and a barcode sequence at the other end The PI adapter is required for the subsequent clonal amplification reaction (see Step 2), while the barcode consists of a DNA sequence useful for the unique identification of the different samples during the multiplex sequencing reaction. Specifically, 4 pi of Switch Solution, 0.5 pi of Ion PI Adapter , 0.5 mΐ of Ion Xpress™ Barcode , 1 mΐ of nuclease-free water, 2 mΐ of DNA Ligase (Thermo Fisher Scientific) are added to each digested library and subsequently ligated under the following conditions: 22°C for 30 min, 68°C for 5 min, 72°C for 5 min. iv. Library purification: The products of the ligation are purified with paramagnetic beads Agencourt AMPure XP (solid-phase reversible immobilization technology - SPRI, Beckman Coulter) to remove excess primers, nucleotides, salts and enzymes. Specifically, each library is purified with 45 mΐ of Agencourt AMPure XP Reagent (Beckman Coulter) following the manufacturer's specifications, subjected to two washes with 75% ethanol and eluted in 50 mΐ of Low TE (Thermo Fisher Scientific). v. Library quantification: Each library is quantified by Real-Time PCR using the Ion AmpliSeq library Taqman Quantification Kit (Thermo Fisher Scientific) on the 7900 HT Fast Real Time PCR system (Thermo Fisher Scientific). In particular, the quantification takes place by using a standard curve generated with three serial dilutions (1:10) of the E. coli DH10B Control Library (68pM) at 6.8 pM, 0.68 pM and 0.068 pM, and the use of an Ion Library TaqMan® Quantification Assay probe labelled in FAM, complementary to the sequence of adapters ligated to each amplicon of the library. Control, probe and reaction mix ( Ion Library qPCR master Mix) are contained in the Ion AmpliSeq library Taqman Quantification Kit (Thermo Fisher Scientific). A total of 9 mΐ of a 1:100 dilution of each library, 9 mΐ of standards and negative controls are analysed in duplicate in a final reaction volume of 20 mΐ, with the Ion Library qPCR master Mix 2x and the Ion Library TaqMan® Quantification Assay 20x following the manufacturer’s specifications (Thermo Fisher Scientific). In particular, the thermal cycle envisages an enzymatic activation step at 99°C for 2 min, and 40 cycles at 99°C for 15 sec and 60°C for 1 min.

SteplB. RNA AmpUSeq Libraries Construction.

For the RNA library required for detecting 204 gene fusions of seven driver genes (Table 6B) and amplification of 12 control genes, 10 ng of RNA is pre-treated with DNAse to eliminate any residual and retro-transcribed genomic DNA (complementary DNA synthesis reaction, cDNA) with the Superscript IV VILO™ Master Mix Kit with ezDNase™ Enzyme in a final volume of 10 pi following the manufacturer's specifications (Thermo Fisher Scientific). In particular, 10 ng of RNA is digested with 0.5 mΐ of ezDNase enzyme , 0.5 mΐ of 10X ezDNase Buffer in a final volume of 5 mΐ at 37°C for 2 min, and subsequently retro-transcribed into cDNA by adding 2 mΐ of Superscript IV VILO™ Master Mix and 3 mΐ of nuclease-free water for 10 min at 25°C, 10 min at 50°C and 5 min at 85°C.

Subsequently, a library is generated for each sample using the enzymes provided in the kit Ion AmpUSeq™ Library Kit Plus·, (Thermo Fisher Scientific) in five steps: i. Target cDNA amplification: Two multiplex PCRs are prepared for each library in a final volume of 10 mΐ each with Ion AmpUSeq HiFi Mix 5x, 5 ng cDNA, 2 mΐ primers (preferably split into two pools) at a 5x concentration (Table 5). Amplification is performed in the Veriti thermal cycler (ThermoFisher Scientific) under the following conditions:

• 98°C for 2 min, 27 cycles at 98°C for 15 sec and 60°C for 4 min.

At the end of the reaction, the two multiplex PCRs of each library are combined in a single tube. ii. Enzymatic digestion of amplicons: the amplicons obtained in the previous reaction are enzymatically treated to partially digest the amplification primers and phosphorylate the ends of the amplicons themselves. This process is necessary for the subsequent ligation reaction. In particular, 2 mΐ of FuPa digestion enzyme is added to each library, which is subsequently digested under the following conditions: 50°C for 10 min, 55°C for 10 min, 60°C for 20 min. iii. Ligation of adapters and barcodes: At the two ends of each amplicon a specific adapter {Ion PI Adapter ) is covalently ligated at one end and a barcode sequence at the other end. The PI adapter is required for the subsequent clonal amplification reaction (see Step 2), while the barcode consists of a DNA sequence useful for the unique identification of the different samples during the multiplex sequencing reaction. Specifically, 4 pi of Switch Solution , 0.5 pi of Ion PI Adapter , 0.5 mΐ of Ion Xpress ™ Barcode , 1 mΐ of nuclease-free water, 2 mΐ of DNA Ligase (Thermo Fisher Scientific) are added to each digested library and subsequently ligated under the following conditions: 22°C for 30 min, 68°C for 5 min, 72°C for 5 min. iv. Library purification: The products of the ligation are purified with paramagnetic beads Agencourt AMPure XP (solid-phase reversible immobilization technology - SPRI, Beckman Coulter) to remove excess primers, nucleotides, salts and enzymes. Specifically, each library is purified with 45 mΐ of Agencourt AMPure XP Reagent (Beckman Coulter) following the manufacturer's specifications, subjected to two washes with 75% ethanol and eluted in 50 mΐ of Low TE (Thermo Fisher Scientific). v. Library quantification: Each library is quantified by Real-Time PCR using the Ion AmpliSeq library Taqman Quantification Kit (Thermo Fisher Scientific) on the 7900 HT Fast Real Time PCR system (Thermo Fisher Scientific). In particular, the quantification takes place by using a standard curve generated with three serial dilutions (1:10) of the E. coli DH10B Control Library (68pM) at 6.8 pM, 0.68 pM and 0.068 pM, and the use of an Ion Library TaqMan® Quantification Assay probe labelled in FAM, complementary to the sequence of adapters ligated to each amplicon of the library. Control, probe and reaction mix (Ion Library qPCR master Mix) are contained in the Ion AmpliSeq library Taqman Quantification Kit (Thermo Fisher Scientific). A total of 9 mΐ of a 1:100 dilution of each library, 9 mΐ of standards and negative controls are analysed in duplicate in a final reaction volume of 20 mΐ, with the Ion Library qPCR master Mix 2x and the Ion Library TaqMan® Quantification Assay 20x following the manufacturer’s specifications (Thermo Fisher Scientific). In particular, the thermal cycle envisages an enzymatic activation step at 99°C for 2 min, and 40 cycles at 99°C for 15 sec and 60°C for 1 min. After quantification, the RNA and DNA libraries of each sample are diluted to a final concentration of 100 pM. A total of 5 pi of each DNA library and 5 mΐ of each RNA library are mixed in two separate tubes (one for DNA and one for RNA) and then joined together in a ratio of 2.5:1 (DNA: RNA). This ratio has been optimized to ensure an adequate number of sequences for each library, and to allow the possible detection of mutations (point/indel on DNA and/or gene fusions on RNA) with a high degree of confidence (see STEP5: data analysis). The combined library pool is ready for the next step of multiplexing template preparation. Step2. Clonal amplification of single amplicons by means of emulsion

PCR.

The single amplicons of the libraries are immobilized on Ion Sphere Particles (ISP) coated with complementary sequences to the PI adapter and clonally amplified by means of an emulsion PCR on the Ion One Touch2 System instrument, and using the reagents of the Ion 540™ KU-OT2 kit (Thermo Fisher Scientific). Binding to the ISPs of the single amplicons occurs through complementarity of the ends of the amplicons (PI adapter sequences) with the sequences ligated to the ISPs. The emulsion is prepared by mixing 200 mΐ of Ion One Touch Reaction Oil and 2.4 ml of an Amplification Solution containing 80 mΐ of nuclease-free water, 120 mΐ of Ion S5 Enzyme mix , 120 mΐ of Ion Sphere Particles and 100 mΐ of library diluted to 8 pM (8 mΐ of the pool of libraries at 100 mM is diluted in 92 mΐ of nuclease-free water). The mixture is loaded onto the Ion OneTouch™ Reaction filter , which - in turn - is subsequently mounted on the Ion One Touch2 System instrument where the clonal amplification reaction is started with the following program: Ion S5:Ion 540 kit OT2. The filter allows generation of microscopic drops (micro -reactors) which, under ideal conditions, should contain a single sphere and a single amplicon in order to obtain monoclonal amplification. These micro -reactors pass through an amplification plate where emulsion PCR takes place. During emulsion PCR, biotin molecules are incorporated at the 5' end of the amplicon. Biotin will be used to isolate template ISPs by binding to Streptavidin-linked Cl Magnetic Beads (Invitrogen) during the successive enrichment step (see Step 3).

Step3. Enrichment of ISPs. At the end of the amplification reaction, only the ISPs on which each amplicon will be bound and amplified are selected. The selection is made on the Enricher System instrument (Thermo Fisher Scientific) using the reagents from the Ion 540™ Kit-OT2 kit (Thermo Fisher Scientific) following the manufacturer's specifications. In particular, the selection takes place through the use of 100 pi of magnetic beads conjugated with streptavidin ( Dynabeads ™ MyOne™ Streptavidin Cl Beads - Invitrogen), which have affinity for the 5' biotinylated ends of the amplicons. At the end of the reaction a final denaturation step with NaOH detaches the beads conjugated with streptavidin from the amplified ISPs. The enriched ISPs are ready for the final sequencing reaction.

Step4. Sequencing.

The sequencing reaction is performed on the Ion chip 540™ using the Ion Gene Studio S5 instrument (Thermo Fisher Scientific) and the reagents contained in the kit Ion 540™ Kit-012 (Thermo Fisher Scientific) following the manufacturer's specifications.

During the reaction, the enriched microspheres are deposited on the wells of a chip and treated with a single known nucleotide at a time and the polymerase. If by complementarity the incorporation of the nucleotide occurs, the binding reaction involves the release of a hydrogen ion, which leads to a modification of the pH detected by a sensor present in the well; the sequence is processed on these recordings, converting a chemical signal into a digital signal.

In particular, before loading the sample, there is an initialization step of the Ion Gene Studio S5 instrument (Thermo Fisher Scientific), necessary for the pH calibration of the solutions used for sequencing. This step takes place automatically in 40 min on the instrument after loading of the Ion A5™ Wash Solution bottle , of the cartridge containing the four separate Ion A5 ™ Sequencing Reagents cartridge and of the Ion A5™ Cleaning Solution bottle (Thermo Fisher Scientific).

During this procedure, the enriched beads are loaded onto the chip Ion chip 540™ (Thermo Fisher Scientific). The ISP beads enriched in the previous step are joined with 5 mΐ of Control Ion Sphere™ particles (Thermo Fisher Scientific), 15 mΐ of Ion S5 ™ Annealing Buffer , 20 mΐ of Ion S5™ Sequencing primer (Thermo Fisher Scientific), and placed on the thermal cycler at 95°C for 2 min and 37°C for 2 min to promote binding of the primer to the DNA strands. A total of 10 mΐ of Ion S5™ Loading Buffer is added to the solution, then the solution is loaded onto the Ion Chip 540™ (Thermo Fisher Scientific) and the chip is centrifuged for 10 min on the Ion Chip™ Minifuge (Thermo Fisher Scientific) to allow the beads to be immobilized in the microwells of the chip. The chip will then be subjected to a series of washes to remove the non-incorporated beads: two washes with a foamy solution prepared with 49 mΐ of 50% Ann baling Buffer and 1 mΐ of Foaming Solution (10% Triton®X-100 solution), each of which is interspersed with the 50% Annealing Buffer solution; two washes with the Flushing solution, prepared with an isovolume of 100% isopropanol and one with Ion S5™ Annealing Buffer·, three washes with 50% Ion S5™ Annealing Buffer. Finally, the excess liquid is removed and 6 mΐ of Ion S5™ Sequencing Polymerase is combined with 60 mΐ 50% Ion S5™ Annealing Buffer, which are loaded onto the chip. After 5 min incubation at room temperature, the chip is mounted on the Ion Gene Studio S5 instrument (Thermo Fisher Scientific) and the sequencing reaction is started. The chip used ( Ion chip 540™) consists of 151,539,288 wells and is capable of producing up to 60-80 M of sequences. This sequencing throughput allows libraries generated from 22 biological samples to be loaded onto chips, which include 22 DNA libraries and 22 RNA libraries in a 2.5:1 ratio (to generate approximately 55 M of DNA sequences and 22 M of sequences for RNA, corresponding to 2.5 M of sequences for DNA and 1 M of sequences for RNA).

Step5. Data analysis

The data from the DNA library sequencing (also called “first DNA library”) are analysed using Torrent Suite v.5.10 software with Coverage Analysis and Variant Caller plugins, and subsequently annotated with wANNOVAR software. Variants are prioritized based on their frequency in the general population, quality values and their somatic status. The positivity of the mutational analysis is defined as the detection of at least one variant which is: 1) rare in the general population of European descent (MAF<0.005 in the lOOOgenome, ExAC, ESP, gnomAD databases); 2) with a high quality base call, as defined by read depth (DP)>500, genotype quality (GQ)>30, strand bias (STB) between 0.3 and 0.7, and allelic frequency (AF)>0.05; 3) somatic, as suggested by an allele frequency between 0.05 and 0.40, or between 0.60 and 0.80. This last criterion is not applied to mutations of the RET gene, so that both somatic and germline mutations are implicated in the pathogenesis of medullary thyroid carcinoma and for the known hotspots of BRAF and RAS, so that clonal mutations with AF>0.40 are possible.

Data from the RNA library sequencing (also called “second DNA library”) are analysed by the Ion Reporter 5.12 software using the workflow for detecting gene fusions. The positivity of the mutational analysis is defined for samples that meet the following conditions: 1) a total of mapped sequences > 20,000, 2) at least 20% of the mapped sequences with positivity for thyroid cell markers CTG, NIS, TPO, TSHR, TTF1, CAECA, PTH), in order to ensure the presence of thyroid cells in the analysed sample (excluding any contamination in the fine needle aspiration step with different cell types), 3) at least 20 sequences that amplify the gene fusion point. c) Digital PCR

The digital PCR reaction (dPCR) is performed on the QuantStudio 3D Digital PCR Instrument (Thermo Fisher Scientific) to quantify the levels of miRNA-146b-5p, a microRNA expressed at high levels in thyroid tumour tissues (11, 12). Starting from 1.25 ng of total RNA, cDNA is synthesized using the High Capacity Reverse Transcription kit and specific primers for miR-146b-5p (target sequence SEQ ID No.: 3, Cat. # ID: 001097, Thermo Fisher Scientific) and U6 (endogenous control sequence target SEQ ID No.: 4, Cat. # ID: 001973, Thermo Fisher Scientific). For each sample, the dPCR reaction is prepared starting from 1.5 pi of a factor 5 dilution of the specific cDNA for miR-146b-5p, 1.5 mΐ of a factor 5 dilution of the cDNA of U6, 0.8 mΐ of the FAM-labelled probe of the miR-146b-5p (20x), 0.8 mΐ of the VIC-labelled probe of the U6 endogenous control, (20x), 8 mΐ of QuantStudio 3D Digital PCR Master Mix v2, and 3.4 mΐ of nuclease-free water (Thermo Fisher Scientific). Each reaction is loaded onto the QuantStudio 3D Digital PCR Chip v2 and incubated on the ProFlex 2x Flat PCR System instrument at 96°C for 10 min, followed by 44 cycles at 60°C for 2 min, 98°C for 30 sec, with a final extension at 60°C for 2 min. Primary data analysis is performed on the QuantStudio 3D Digital PCR Instrument (Thermo Fisher Scientific) following the manufacturer's specifications. Secondary analysis is performed using QuantStudio 3D Analysis Suite software (Thermo Fisher Scientific), which identifies the absolute copy number per pi of the miR-146b-5p and of the endogenous control. For each sample, the number of copies per mΐ of miR-146b-5p is normalized for those of the endogenous control U6. Each experiment includes a reaction blank. The analysis of the ROC curve and the area under the curve (AUC) were used to evaluate the diagnostic value of miR-146b- 5p and to search for the optimal cut-off to discriminate benign nodules from malignant nodules (identified cut-off: 0.1562). The molecular test is evaluated as positive when miRNA expression levels in the cytology sample exceeds the cut off value. d) Data analysis

The positivity of the test is defined by the presence of a mutation (point mutation and/or indel and/or gene fusion) or by the expression of miR-146b-5p above the calculated cut-off. The negativity of the test is defined by the absence of mutations (point mutation and/or indel and/or gene fusion) and by the expression of the microRNA below the calculated cut-off.

The performance of the method was evaluated in the previously described cohort of 118 cytological samples from 112 patients, and subsequently tested in the indeterminate lesion subgroup (n=40). The results of the single and combined molecular tests were compared with the results of the final histological analysis, and their ability to correctly classify the cytological samples as benign or malignant was evaluated by calculating sensitivity, negative predictive value (NPV), specificity, positive predictive value (PPV).

All statistical analyses were performed with GraphPad Prism v6.01 software (GraphPad Software Inc., San Diego, CA). The p-value < 0.05 was considered statistically significant. Continuous variables were reported as means ± standard deviation and compared by the Mann-Whitney U test.

RESULTS

1. Design of the diagnostic procedure

The present invention relates to the creation of a diagnostic method that - through a molecular test - allows distinguishing between benign thyroid nodules and malignant thyroid nodules during the fine needle method procedure. In particular, the present method allows identification of thyroid tumorigenesis driver mutations and of the expression of a microRNA specifically expressed at high levels in thyroid tumour tissues of follicular origin by exploiting highly sensitive methods such as Next-Generation Sequencing (NGS) and digital PCR, respectively. The selection of molecular markers was carefully carried out on the basis of results reported in the literature (8, 9) and of data obtained in-house (10,12), with the aim of identifying the minimum number of markers that can distinguish benign thyroid nodules from malignant thyroid nodules during the needle aspiration procedure. The presence of a mutation (point, indel, gene fusion) and/or the expression of the microRNA above an appropriately determined threshold value is indicative of the presence of a malignant tumour. This method, and the respective kit, allows a personalized approach for treating nodular thyroid disease, improving the diagnostic accuracy of cytology in indeterminate thyroid lesions and the prognostic stratification of patients.

2. Evaluation of the precision of the method and its analytical sensitivity

Six biological samples (two samples from thyroid fine needle aspiration, FNA1 and FNA2; two samples from formalin-fixed paraffin-embedded thyroid tumour tissue, FFPE1 and FFPE2; and two samples from fresh-frozen thyroid tumour tissue, FF1 and FF2) with known gene variants (SNV and Gene Fusions) were analysed in triplicate in a single experiment and in triplicate in three different experiments to evaluate, respectively, the intra-assay and inter-assay precision of the molecular test. To determine the analytical sensitivity of the molecular test, a thyroid tumour cell line derived from a patient with papillary thyroid carcinoma (BCPAP,

# Cat. ACC273, DSMZ) with two known mutations in homozygosity ( BRAF p.V600E and TP53 p.D259Y) was mixed in four different proportions with commercial human control DNA (CEPH Individual 1347-02, Thermo Fisher Scientific) negative for the BRAF p.V600E and TP53 p.D259Y mutations.

A thyroid tumour cell line derived from a patient with papillary thyroid carcinoma (TPC-1, # Cat. SCC147, Merck) with a known gene fusion (CCDC6- RET.C1R12.COSF1271) was mixed in four different proportions with a pool of RNA from 64 normal human thyroid glands distributed commercially by Clontech, negative for the gene fusion tested. The same cell line was mixed in four different proportions with RNA from whole (donor) blood to determine the minimum amount of thyroid cells needed for the molecular test.

Four biological samples (two samples from thyroid fine needle aspiration, FNA3 and FNA4; two samples from formalin-fixed and paraffin-embedded thyroid tumour tissue, FFPE3 and FFPE4) with known gene variants (one SNV, BRAF P.V600E and two Gene fusions, RET-NCOA4_COSF1496 and HOOK3- RET.H11R12.COSF1509) were tested in a single experiment using four different amounts of nucleic acids to detect the minimum amount needed to detect mutations. Intra-assay reproducibility. All known variants of the six biological samples from different tissue types (FNA, FFPE and FF) were detected in all three intra-assay replicates. The standard deviation of allele frequencies of the point mutations (BRAF p.V600E, RET p.C634R) ranges from 0.61% to 0.78%, with a coefficient of variation between 0.01 and 0.04 (Table 7A). The standard deviation of the sequences (reads) that detect the presence of gene fusions (PAX8- PPARG_COSF1215; RET-NCOA4_COSF1496) varies between 138 and 7,445 with a coefficient of variation between 0.12 and 0.31. (Table 7B).

Table 7A. Intra-assay reproducibility of point mutations

Abbreviations. DS, standard deviation; CV, coefficient of variance

Table 7B. Intra-assay reproducibility of gene fusions

Abbreviations. DS, standard deviation; CV, coefficient of variance

Inter-assay reproducibility All known variants of the six biological samples from different tissue types (FNA, FFPE and FF) were detected in all three inter-as ay replicates. The standard deviation of the allele frequencies of the point mutations (BRAF p.V600E, RET p.C634R) ranges from 0.2% to 0.85%, with a coefficient of variation between 0.01 and 0.02 (Table 8A). The standard deviation of the sequences (reads) that detect the presence of gene fusions (PAX8- PPARG_COSF1215; RET-NCOA4_COSF1496) varies between 7.563 and 34.088, with a coefficient of variation between 0.6 and 1.07 (Table 8B).

Table 8A. Intra-assay reproducibility of point mutations

Abbreviations. DS, standard deviation; CV, coefficient of variance

Table 8B. Inter-assay reproducibility of gene fusions

Abbreviations. DS, standard deviation; CV, coefficient of variance

Analytical sensitivity. To determine the analytical sensitivity of the molecular test, two cell lines with known mutations (BRAF p.V600E and TP53 p.D259Y point mutations and CCDC6-RET.C1R12.COSF1271 gene fusion) were mixed with control samples [(cells from a pool of healthy thyroids and cells from whole blood from a healthy donor (no thyroid nodules)], which were negative for the mutations tested. The test was able to detect point mutations up to an allele frequency of 5%. (Table 9). The test was able to detect gene fusions up to a dilution of 1:39 (Table 9), both when diluted in normal thyroid cells and when diluted in cells of different cell types (whole blood) (Table 10).

Table 9. Analytical Sensitivity of the method (Two DNA point mutations)

* Although mutations are not called by the VariantCaller software, they can be viewed in Integrative Genome Visualization (IGV) at a rate of 2%

Table 10. Analytical Sensitivity of the molecular test (one gene fusion on

RNA)

Determination of the minimum amount of nucleic acids needed to detect mutations. Four biological samples (two samples from thyroid fine needle aspiration, FNA3 and FNA4; two samples from formalin-fixed, paraffin- embedded thyroid tumour tissue, FFPE3 and FFPE4) were tested to identify the minimum amount of nucleic acids needed to detect point mutations and gene fusions at a quantity of nucleic acids between 1 and 10 ng. The test was able to correctly identify the genetic alterations tested (one SNV, BRAF p.V600E and two Gene Fusions, RET-NCOA4_COSF1496 and HOOK3-RET.H11R12.COSF1509) and classify them as positive in 100% of the tested samples, both for thyroid fine needle aspirate samples and samples isolated from paraffin-embedded tumour tissue, indicating that the minimum amount of nucleic acids tolerated by the molecular test is 1 ng. (Table 11 and Table 12)

Table 11. Minimum amount of DNA required to detect point mutations

FNA = DNA isolated from cytological sample from thyroid fine needle aspiration

FFPE = DNA isolated from formalin-fixed and paraffin-embedded thyroid tumour tissue sample.

Table 12. Minimum amount of RNA needed to detect gene fusions

FNA = RNA isolated from cytological sample from thyroid fine needle aspiration

FFPE = RNA isolated from formalin-fixed and paraffin-embedded thyroid tumour tissue sample.

RPM = reads per million

Molecular test evaluation in clinical samples. To date, the present inventors have evaluated the method of the present invention on a retrospective cohort of 118 nodules from 112 patients, who have undergone total thyroidectomy or lobectomy surgery. Of the 118 nodules, two were cytologically classified as TIR1 - nondiagnostic/inadequate sample (2%), 20 as TIR2 - benign sample (17%), 9 as TIR3A - indeterminate low risk malignancy sample (8%), 31 as TIR3B - indeterminate sample with high risk of malignancy (26%), 24 as TIR4 - sample suspected for malignancy (20%), 32 as TIR5 - malignant sample (27%). At the final histological examination, the following were classified as malignant: one case of the cytological category TIR1 (50%), zero cases of TIR2 (0%), one case of TIR3A (11%), eight cases of TIR3B (26%), 22 cases of TIR4 (92%), and all nodules with cytological TIR5 diagnosis (100%), corresponding to a total of 64 out of 118 nodules tested (54%). In particular, the malignant lesions included 58 papillary thyroid carcinomas, two non-invasive follicular thyroid neoplasms with papillary-like nuclear features (NIFTP), one angioinvasive oncocytic thyroid carcinoma, one poorly differentiated islet-type thyroid carcinoma, one medullary carcinoma and one anaplastic thyroid carcinoma. Among the 54 benign nodules (46%), 12 were histologically diagnosed as follicular adenomas, 11 as oncocyte adenomas and 31 as multinodular goiters (seven of which were hyperplastic nodules).

Considering the entire patient cohort, NGS analysis revealed a driver mutation in 58 of the 64 malignant nodules tested (sensitivity: 91%). A total of 42 of the 54 histologically benign nodules were negative, revealing a specificity of the analysis of 78%. The probability that a positive nodule is actually malignant is 83% (PPV); the probability that a negative nodule is benign is 87.5% (NPV).

The mir-146b-5p in 118 cytological samples showed a high expression in histologically malignant lesions compared to benign lesions (0.06922 ± 0.04940 vs 0.5981 ± 0.5901; p < 0.0001). Analysis of the ROC curve revealed an AUC of 0.8880 (Cl: 0.8282-0.9478) and an optimal cut-off value of 0.1562, reflecting good sensitivity (70%) and excellent specificity (96%). Among the microRNA- positive nodules, the probability of malignancy is 96% (PPV). The probability that a microRNA-negative nodule is benign is 73% (NPV).

Combined analysis of driver mutation identification and miR-146b-5p expression evaluation is associated with a higher sensitivity (94%) and NPV (91%).

If the subgroup of thyroid lesions with indeterminate cytological diagnosis (TIR3A and TIR3B, n = 40) is considered, and the cut-off values established on the entire cohort of nodules are applied, the performance of the molecular platform shows a sensitivity equal to 100%, a specificity of 72%, a NPV of 100% and a PPV of 53%.

Table 13 shows the performance of the molecular test in the entire cohort of nodules (n = 118). Table 3 shows the performance of the molecular test in the subgroup of cytologically indeterminate thyroid lesions (n=40).

Table 13. Molecular test performance in clinical samples: cohort of 118 nodules belonging to all cytological classes

Abbreviations: Cl = confidence interval; FP = false positive; FN = false negative; VP = true positive; VN = true negative; NPV = negative predictive value - VN/(VN+FN); PPV = positive predictive value - VP/ (VP+FP); Sens., Sensitivity = VP/(VP+FN); Spec., Specificity = VN/(VN+FP).

REFERENCES

1. Durante C, et al. The Diagnosis and Management of Thyroid Nodules: A Review. JAMA 2018319914-924. (doi:10.1001/jama.2018.0898) 2. Patel KN, et al. Performance of a Genomic Sequencing Classifier for the Preoperative Diagnosis of Cytologically Indeterminate Thyroid Nodules. JAMA 2018319914-924. (doi:10.1001/jama.2018.0898)

3. Nikiforova MN, et al. Analytical performance of the ThyroSeq v3 genomic classifier for cancer diagnosis in thyroid nodules. Cancer 2018 124 1682-1690.

(doi: 10.1002/cncr.31245)

4. Steward DL, et al. Performance of a Multigene Genomic Classifier in Thyroid Nodules With Indeterminate Cytology. JAMA Oncology 20195204.

(doi: 10.1001/j amaoncol.2018.4616)

5. Labourier E, et al. Molecular Testing for miRNA, mRNA, and DNA on Fine-Needle Aspiration Improves the Preoperative Diagnosis of Thyroid Nodules With Indeterminate Cytology. The Journal of Clinical Endocrinology & Metabolism 2015 1002743-2750. (doi: 10.1210/jc.2015-1158)

6. Banizs AB & Silverman JF. The utility of combined mutation analysis and microRNA classification in reclassifying cancer risk of cytologically indeterminate thyroid nodules. Diagnostic cytopathology 201947268-274. (doi: 10.1002/dc.24087)

7. Fithwick-Yanai G, et al. Multicentre validation of a microRNA-based assay for diagnosing indeterminate thyroid nodules utilising fine needle aspirate smears. Journal of clinical pathology 201770500-507. (doi:10.1136/jclinpath-2016-204089)

8 Agrawal N, et al. Integrated Genomic Characterization of Papillary Thyroid Carcinoma. Cell 2015 159676-690. (doi:10.1016/j.cell.2014.09.050)

9. Fanda I, et al. The Journal of clinical investigation 2016 126. (doi:10.1172/JCI85271)

10 Sponziello M, B et al. Whole exome sequencing identifies a germline MET mutation in two siblings with hereditary wild-type RET medullary thyroid cancer. Human Mutation 201839371-377. (doi:10.1002/humu.23378) 11. Celano M, et al. MicroRNAs as Biomarkers in Thyroid Carcinoma. International journal of genomics 201720176496570. (doi: 10.1155/2017/6496570)

12 Rosignolo F, et al. MicroRNA-based molecular classification of papillary thyroid carcinoma. International journal of oncology 201750 1767-1777.

(doi: 10.3892/ijo.2017.3960) 13. Grani G et al. Reducing the number of unnecessary thyroid biopsies while improving diagnostic accuracy: towards the “right” TIRADS. The Journal of Clinical Endocrinology & Metabolism 2018. (doi:10.1210/jc.2018-01674)

Claims

1. Method for the in vitro diagnosis of a thyroid tumour comprising the following steps:

(a) extracting DNA and RNA from a thyroid biological sample;

(c) generating from the RNA obtained in step (a) the corresponding complementary DNA and, from the complementary DNA thus obtained, generating a second DNA library comprising a second plurality of amplicons obtained by means of multiplex PCR amplification of: (i) a second set of genes, wherein the second set of genes includes: the RET, ALK, NTRK1, NTRK3, PAX8- PPARG, BRAF, THADA genes with a fusion partner as indicated in Table 6B and (ii) a third set of control genes, wherein the third set of control genes includes the genes: KRT7, KRT20, TG, NIS, TPO, TSHR, TTF1, CALCA, PTH, TBP, JUN and MTTB-

(d) sequencing each amplicon obtained in steps (b) and (c); and

(e) determining the positivity of the thyroid biological sample of step (a) if at least one of the following conditions is met:

(ii) at least one gene mutation has been identified in step (d) in the second DNA library.

2. Method for in vitro diagnosis of a thyroid tumour according to claim 1, wherein the method envisages an additional step (f) comprising the determination through a digital PCR reaction of the expression of miRNA-146b-5p and of an endogenous control in the thyroid biological sample of step (a), wherein the biological sample of step (a) is positive if miRNA-146b-5p expression has been identified.

3. Method for in vitro diagnosis of a thyroid tumour according to claim 1, wherein step (e-i) includes determining if: (1) the gene mutation has a minor allele frequency (MAF) < 0.005 in the population of European descent, as reported in the databases lOOOgenome, ExAC, ESP, gnomAD;

(3) the gene mutation has an allele frequency (AF) comprised between 0.05 and 0.40 or between 0.60 and 0.80, with the exclusion of all gene mutations present in the RET gene, and of the known hotspot mutations in BRAF and RAS genes; wherein the positivity of the sample is determined by the fulfilment of all the conditions (l)-(3).

4. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein the step (e-ii) includes determining if: (1) at least 20,000 sequences selected from the sequences included in

Table 6A and the control genes have been sequenced;

(2) at least 20% of the sequenced sequences are positive for at least the TG, NIS, TPO, TSHR, TTF1, CAECA, and 1Ή control genes; and

(3) at least 20 sequenced sequences comprise at least one mutation included in Table 6B; wherein the positivity of the sample is determined by the fulfilment of all the conditions (l)-(3).

5. Method for in vitro diagnosis of a thyroid tumour according to any of claims 2 to 4, wherein step (f) includes determining the level of quantitative expression of miRNA-146b-5p and of an endogenous control, preferably represented by the U6 nuclear RNA, and wherein the positivity of the sample is determined if the level of expression of miRNA-146b-5p is higher than a threshold equal to 0.1562.

6. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein the first DNA library obtained in step (b) is generated by using primer pairs indicated in Table 4.

7. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein the second DNA library obtained in step (c) is generated by using primer pairs indicated in Table 5.

8. Method for in vitro diagnosis of a thyroid tumour according to any one of claims 2 to 7, wherein the digital PCR reaction of step (f) is performed by using a primer pair capable of hybridizing to the SEQ ID No.: 3 sequence of miRNA- 146b-5p

9. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein the step (b) comprises the additional steps of:

(b-i) enzymatically digesting the first plurality of amplicons obtained in step (b) to phosphorylate the ends of the first plurality of amplicons;

(b-ii) ligating an adapter sequence to each amplicon obtained in step (b-i), (b-iii) quantifying by means of a Real Time PCR reaction each amplicon obtained in step (b-ii).

10. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein the step (c) comprises the additional steps of:

(c-i) enzymatically digesting the second plurality of amplicons obtained in step (c) to phosphorylate the ends of the second plurality of amplicons;

(c-ii) ligating an adapter sequence to each amplicon obtained in step (c-i), (c-iii) quantifying by means of a Real Time PCR reaction each amplicon obtained in step (c-ii).

11. Method for in vitro diagnosis of a thyroid tumour according to claim 9 or claim 10, wherein steps (b-i) and (c-i) also includes ligating a barcode sequence to each amplicon obtained in steps (b-i) and (c-i), respectively.

12. Method for in vitro diagnosis of a thyroid tumour according to any of claims 9 to 11, wherein each amplicon obtained in steps (b-ii) and (c-ii), respectively, is immobilized on a respective bead.

13. Method for in vitro diagnosis of a thyroid tumour according to claim 12, wherein, prior to step (d), the beads on the surface of which at least one amplicon is present are subjected to an emulsion monoclonal PCR amplification reaction, obtaining for each amplicon a respective bead on the surface of which multiple copies of the amplicon are present.

14. Method for in vitro diagnosis of a thyroid tumour according to claim 13, wherein during the emulsion monoclonal PCR amplification reaction at least one biotin molecule is bound to the 5' end of at least one copy of the amplicon.

15. Method for in vitro diagnosis of a thyroid tumour according to claim 14, wherein, prior to step (d), beads on the surface of which multiple copies of the amplicon are present are separated from beads on which no copies of an amplicon are present.

16. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein, prior to step (b), a PCR amplification reaction of the DNA extracted in step (a) is conducted using the primer pair having the sequences indicated in SEQ ID No.: 1 and 2.

17. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein the thyroid biological sample of step (a) is a thyroid sample from fine needle aspiration (FNA), a formalin-fixed paraffin- embedded (FFPE) sample of thyroid tissue, a fresh-frozen (FF) tumour tissue.

18. Method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims, wherein the thyroid tumour is selected from a papillary thyroid carcinoma, a non-invasive follicular neoplasm with papillary-like nuclear features, an angioinvasive oncocytic thyroid carcinoma, a poorly differentiated insular thyroid carcinoma, a medullary thyroid carcinoma and an anaplastic thyroid carcinoma.

19. Kit for implementing a method for in vitro diagnosis of a thyroid tumour according to any one of the preceding claims including:

- primer pairs having the sequences indicated in Table 4 and a primer pair having the sequences indicated in SEQ ID No.: 1 and 2,

- primer pairs having the sequences indicated in Table 5, and - instructions for use.

20. Kit according to claim 19, also comprising a pair of primers capable of hybridizing to the sequence SEQ ID No.: 3 of miRNA-146b-5p.

21. Kit according to claim 19 or claim 20, further comprising a pair of primers capable of hybridizing to the sequence SEQ ID No.: 4 of the U6 nuclear RNA.