SEAP-GETNE consensus on prognostic and predictive molecular biomarkers in thyroid cancer

Abstract

Thyroid cancer is the most common endocrine malignancy and generally carries a favourable prognosis. However, a subset of cases exhibits aggressive behaviour, metastatic potential, and resistance to conventional therapies. The 2022 WHO classification introduced major updates, including refined subtypes of papillary thyroid carcinoma (PTC), recognition of high-grade follicular cell-derived carcinomas, and the introduction of grading criteria for medullary thyroid carcinoma (MTC). These revisions reflect advances in tumour biology and are essential for precise diagnosis and treatment planning.

Molecular profiling has become central to the management of thyroid cancer. Key driver mutations in follicular cell-derived tumours include BRAF V600E (common in PTC), RAS mutations (common in follicular carcinomas), and RET, NTRK, and ALK gene fusions, all of which influence prognosis and therapeutic strategies. MTC is primarily driven by RET and RAS mutations. In anaplastic thyroid carcinoma (ATC), the most aggressive subtype, evaluation of PD-L1 expression and BRAF mutations is recommended to guide treatment.

Accurate molecular analysis depends on appropriate tumour sample selection and processing. Genetic testing is particularly indicated in advanced, refractory, or metastatic disease to identify candidates for targeted therapies, which have shown significant clinical benefit. Two diagnostic strategies are proposed: a sequential single-gene approach, typically beginning with BRAF testing, or comprehensive profiling using next-generation sequencing (NGS). Multidisciplinary molecular tumour boards are strongly recommended to integrate histological, molecular, and clinical information for personalised treatment decisions.

Resumen

El cáncer de tiroides es la neoplasia endocrina más común y, por lo general, tiene un pronóstico favorable. No obstante, un subgrupo de tumores muestra un comportamiento más agresivo, con metástasis y resistencia a terapias convencionales. La clasificación de la OMS de 2022 introdujo actualizaciones, como la redefinición de varios subtipos del carcinoma papilar de tiroides (CPT), el reconocimiento del carcinoma de células foliculares de alto grado y la incorporación de criterios de gradación para el carcinoma medular de tiroides (CMT).

La caracterización molecular es fundamental en el manejo del cáncer de tiroides. En los tumores derivados de células foliculares se identifican mutaciones conductoras como BRAF V600E (frecuente en CPT), RAS (característica de los carcinomas foliculares) y fusiones en RET, NTRK y ALK, todas ellas con implicaciones pronósticas y terapéuticas. El CMT se asocia principalmente con mutaciones en RET y RAS. En el carcinoma anaplásico de tiroides (CAT), el subtipo más agresivo, se recomienda evaluar la expresión de PD-L1 y el estado mutacional en BRAF para orientar el tratamiento.

El análisis molecular debe realizarse sobre muestras bien seleccionadas y procesadas. Las pruebas genéticas están especialmente indicadas en pacientes con enfermedad avanzada, refractaria o metastásica, con el fin de identificar candidatos a terapias dirigidas con beneficios clínicos demostrados. Se contemplan dos estrategias diagnósticas: un enfoque secuencial gen a gen, iniciando con BRAF, o un perfil molecular integral mediante secuenciación masiva (NGS). Se recomienda establecer comités moleculares multidisciplinarios para integrar la información histológica, molecular y clínica, y así optimizar las decisiones terapéuticas personalizadas.

Introduction

Thyroid cancer is the most common endocrine malignancy, accounting for 4.4% of all new cancer diagnoses worldwide, with a five-year prevalence of 5.9% and an incidence of 8.3 per 100,000 population.¹ In Spain, 6495 new cases are expected in 2025 (1626 men and 4869 women). In 2020, prevalence was 17,857 men and 75,474 women, with five-year survival rates of 86% and 93%, respectively.² Although thyroid cancer generally carries a favourable prognosis, distant metastases develop in 7–23% of cases, and two-thirds of these patients become refractory to radioactive iodine (RAI). Prognostic factors include age, histology, tumour grade, capsular/vascular invasion, disease stage, distant metastases, surgical margins, extrathyroidal invasion, RAI therapy, and locoregional recurrence.^3,4

In light of the recent updates in the latest WHO Classification of Endocrine and Neuroendocrine Tumours of the thyroid gland, and the growing availability of selective targeted inhibitors directed at genomic alterations that have demonstrated significant improvements in oncological outcomes, ensuring optimal access of patients to high-quality molecular analysis has become a key priority.⁵ The aim of this expert consensus is to provide an overview on the latest advances in histological and molecular markers in thyroid cancer, and to offer recommendations on selecting the most appropriate tumour sample for adequate molecular analysis according to the patient's profile and potential therapeutic options.

Histology classification (WHO 2022)

The 5th edition of the WHO Classification of Endocrine and Neuroendocrine Tumours related to the thyroid gland⁶ introduces important modifications reflecting advances in the understanding of tumour cellular origin, cytopathological and molecular characteristics, and biological behaviour (Fig. 1). This edition provides general recommendations, including stratification into histological tumour types and subtypes, while reserving the term variant for molecular subtypes. Additionally, mitotic activity should be evaluated per square millimetre rather than per microscopic field of view, a change facilitated by the incorporation of digital pathology.

Fig. 1.

Advances in the understanding of tumour cellular origin, cytopathology, molecular characteristics, and biological behaviour.

Follicular neoplasms are stratified into three main categories: benign tumours, low-risk neoplasms, and malignant neoplasms. In the group of benign tumours, the term thyroid follicular nodular disease (FND) refers to multifocal hyperplastic/neoplastic lesions clinically designated as multinodular goitre, although they may also represent a subclinical entity. Follicular thyroid adenoma with papillary architecture is a newly recognised category of adenoma characterised by a papillary growth pattern, absence of the nuclear features of papillary thyroid carcinoma (PTC), and frequent autonomous hyperfunction, which may be associated with McCune Albright syndrome, Carney complex, and DICER1 syndrome.⁶ Within the group of low-risk neoplasms, non-invasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) now also includes the oncocytic and sub-centimetre subtypes.⁶ The hyalinising trabecular tumour is regarded as a rare, low-risk follicular-derived neoplasm, characterised by its non-invasive behaviour and by the presence of pathognomonic PAX8::GLIS gene fusions.⁶

In the group of malignant neoplasms, the invasive encapsulated follicular variant of PTC is now considered an independent tumour type, reflecting its greater morphological and molecular (RAS-like) resemblance to follicular thyroid carcinoma (FTC) than to PTC.⁶ In contrast, PTC is considered a BRAF-like neoplasm. In this category, microcarcinoma and cribriform-morular carcinoma have been removed as histological subtypes of PTC. Sub-centimetre PTCs should be subtyped using the same criteria applied to larger PTCs. The diagnostic threshold for the tall cell subtype of PTC has been tightened, requiring that ≥30% of tumour cells have a height-to-width ratio of at least 3:1. PTC subtypes are shown in Fig. 1. The term oncocytic cell should replace Hürthle cell, since the cells originally described by Karl Hürthle were in fact C cells.⁷ For oncocytic adenomas and carcinomas, ≥75% of tumour cells must have oncocytic characteristics and lack the nuclear features of PTC. Oncocytic carcinoma, FTC, and invasive encapsulated follicular variant PTC are now subtyped as minimally invasive (capsular invasion only), encapsulated angioinvasive, or widely invasive.

High-grade follicular cell-derived thyroid carcinoma has been introduced as a new tumour type that is morphologically, molecularly, and biologically intermediate between differentiated carcinomas and anaplastic carcinoma (ATC).⁶ These tumours are defined by an elevated mitotic count and the presence of tumour necrosis in the absence of anaplastic transformation. High-grade carcinomas are subdivided into two categories: a) Differentiated high-grade thyroid carcinoma, in which tumours retain the characteristic architectural and/or cytologic features of well-differentiated histotypes of follicular cell-derived carcinomas (e.g. high-grade FTC, high-grade PTC, and high-grade oncocytic thyroid carcinoma); and b) Poorly differentiated thyroid carcinoma, defined according to the Turin consensus criteria.⁸ Unlike ATC, high-grade follicular carcinomas show immunoreactivity for keratins, thyroglobulin, TTF1, and PAX8, although occasionally thyroglobulin expression may be reduced and can exhibit a distinctive dot-like paranuclear pattern. In this new classification primary thyroid squamous cell carcinoma is considered a histological subtype of ATC.^6,9

The distinction between high-grade and low-grade now also applies to thyroid C cell-derived carcinomas.⁶ High-grade medullary thyroid carcinoma (MTC) is defined by the presence of any of the following features: (a) tumour necrosis; (b) ≥5 mitoses per 2 mm²; and/or (c) Ki67 index ≥ 5%.^6,10

The category of salivary gland-type carcinomas of the thyroid now includes mucoepidermoid carcinoma, as recognised in previous editions, and secretory carcinoma, formerly known as mammary analogue secretory carcinoma.⁶ Secretory carcinoma is typically negative for thyroglobulin and is associated with specific rearrangements of the ETV6 gene.¹¹

Cribriform morular thyroid carcinoma (CMTC) is a malignant thyroid tumour type that typically occurs in familial adenomatous polyposis, although sporadic cases also occur. Due to its negativity for thyroglobulin and calcitonin and its characteristic molecular alterations involving the WNT signalling pathway, CMTC is no longer classified as a variant of PTC and has instead been placed within the group of tumours of uncertain histogenesis.^6,12

Thyroblastoma is a high-grade embryonal thyroid neoplasm recognised for the first time in this classification.⁶ It is associated with somatic mutations in the DICER1 gene and consists of primitive follicular epithelium positive for thyroglobulin, a small-cell blastemal component, and mesenchymal stroma.^13,14

This new edition of the WHO classification also includes extensive information on genetic syndromes (see below).

Molecular alterations of the different histological types of thyroid cancer

Follicular cell-derived thyroid carcinomas

The BRAF gene is one of the most extensively studied in PTC, with the V600E variant being the most common, found in approximately 40–60% of cases. The presence of BRAF V600E has been associated with unfavourable clinical features, including greater tumour aggressiveness, higher rates of lymph node metastasis, increased risk of recurrence, and a reduced response to radioactive iodine therapy. This association with aggressive behaviour, however, has been demonstrated only in univariate analyses and not confirmed in multivariate analyses.^6,15,16

Mutations in the RAS genes (NRAS, HRAS, and KRAS) are common in FTC (40–50%) and in high-grade FTC (30%) and occur in a minority of PTC cases (10–20%), with NRAS being the most frequently affected gene. These mutations result in constitutive activation of the MAPK and PI3K-AKT signalling pathways, which are crucial for cell proliferation and survival. Tumours with RAS genes mutations generally exhibit a more indolent behaviour than those with BRAF V600E mutations, although progression to more aggressive carcinomas can occur in some cases.

The RET gene encodes a receptor tyrosine kinase, and its rearrangements are hallmarks in a subset of PTC. RET rearrangements occur in approximately 10–20% of PTCs, particularly in patients with prior radiation exposure. These rearrangements, especially the NCOA4::RET translocation in radiation-associated cases, are common in the diffuse sclerosing variant of PTC, which is often associated with extrathyroidal extension, extensive cervical lymph node involvement, and distant metastasis.^17–19

Fusions involving the NTRK genes (NTRK1, NTRK2, and NTRK3) are rare and account for approximately 2% of patients, more frequently in PTC (infiltrative follicular, solid), young patients, and prior radiation exposure (mainly NTRK1). These fusions result in constitutive activation of TRK proteins, receptor tyrosine kinases that play a key role in cell growth and survival.

Finally, rearrangements of the ALK gene are infrequent in thyroid carcinoma, occurring in approximately 1–5% of PTC and are extremely rare in FTC. These rearrangements activate the ALK signalling pathway, promoting cell proliferation and resistance to apoptosis.

Medullary thyroid carcinoma (MTC)

The most frequent molecular alteration in MTC is RET mutation, which is present in virtually all cases of familial MTC and in around 50% of sporadic MTC.^20,21

The most common RET mutations are reported in exons 10, 11, 15, and 16, which correspond to the cysteine-rich and tyrosine-kinase domains of the protein. The presence of RET mutations has been associated with a more aggressive clinical course in sporadic MTC.^20,22 In particular, patients harbouring RET M918T, the most frequent mutation, have the worse prognosis.²³ Noteworthy, the percentage of RET-mutated carcinomas is increased in high-grade MTC.^23,24

Mutations in RAS genes, specially exons 2–3, are present in around 25% of sporadic MTC and are mutually exclusive with RET alterations.^20,24

About 25% of MTCs are considered RET/RAS wild-type, and other potentially actionable alterations may be identified, such as ALK translocations.²⁵

Anaplastic thyroid carcinoma (ATC)

ATC may arise de novo but is often the result of dedifferentiation from thyroid carcinoma which may account for the frequent presence of BRAF V600E and RAS mutations; however, early genetic divergence between ATC and PTC has been documented.^26,27 Genetic alterations in ATC include early and late molecular events, consistent with a multistep progression model, and the median mutational burden is higher than in other thyroid cancer histologies.⁸

BRAF V600E and RAS mutations remain the mutually exclusive main driver mutations in ATC, occurring in 29–39% and 23% of cases, respectively.^28,29 A higher BRAF V600E mutation rate has been associated with older patient age and with the presence of a PTC precursor.^29,30

ATC is associated with a higher rate of TERT promoter mutations of 73%, that tend to co-occur with BRAF or RAS mutations, and are linked to a trend towards greater mortality.³¹EIFIAX mutations occur in 9% of ATCs. Overall, 93% of EIFIAX-mutated ATCs harbour a concurrent RAS mutation.^28,31TP53 mutations are detected in 27–78% of ATCs, and can harbour alterations of NF1 and NF2 with a mutation frequency of 10% and 9%, respectively.^28,32,33

Mutations in PIK3CA and PTEN are prevalent in ATC, with frequencies of 18–44% and 14%, respectively. Furthermore, PIK3CA mutations tend to co-occur with BRAF V600E, while PTEN with NF1.^28,31,34 Mutations encoding components of the SWI/SNF nucleosome-remodelling complex, DNA mismatch repair genes, and histone methyltransferases occur at a higher frequency in advanced thyroid carcinomas.²⁸

About 25% of MTCs are considered RET/RAS wild-type, and other potentially actionable alterations may be identified, such as ALK translocations.³³

Approximately 11–28% of ATCs have been reported to express PD-L1.³⁵ PD-L1 expression assessed by immunohistochemistry (IHC) using a tumour proportion score (TPS) with a 1% cut-off has been detected in 60% of ATCs,³⁶ and has been associated with a better response to PD-1/PD-L1 inhibitors.³⁷

Familial follicular cell-derived thyroid cancer (FFCT)

Non-medullary familial thyroid cancer is classified into two groups: (a) Syndromic familial non-medullary thyroid carcinoma (SFNMTC), in which the clinical manifestations of non-thyroid neoplasms predominate, and (b) Non-syndromic familial non-medullary thyroid carcinoma (NSFNMTC), which includes a spectrum of familial syndromes characterised by a predominance of follicular cell-derived carcinomas.⁶

The SFNMTC group includes PTEN hamartoma tumour syndrome (PTEN), familial adenomatous polyposis (APC), DICER1 syndrome (DICER1), Carney complex (PRKAR1A), and Werner syndrome (WRN), all of which result from germline mutations, and also McCune-Albright syndrome, which is a somatic mosaicism of the GNAS gene.^38,39 PTEN hamartoma tumour syndrome comprises Cowden syndrome, Bannayann-Riley-Ruvalcaba syndrome, and PTEN-related Proteus syndrome. Some cases with a PTEN-like phenotype may be caused by alterations in RASAL1 or SDHx or by KLLN hypermethylation.³⁹ Immunohistochemical determinations of PTEN protein and beta-catenin can support the diagnosis of PTEN hamartoma tumour syndrome and familial adenomatous polyposis, respectively; however, genetic confirmation is mandatory.

In NSFNMTC, histological findings are usually non-specific. The WHO consensus has established that, in the absence of evidence of ionising radiation exposure or a hereditary cancer syndrome, the clinical criteria for NSFNMTC include (a) PTC in two or more first-degree relatives, and (b) follicular cell-derived thyroid carcinoma in at least three first-degree relatives. NSFNMTC is a heterogeneous group of hereditary thyroid cancers for which the underlying genes and susceptibility loci, except for CHEK2 and POT1, are not yet available or actionable for routine clinical use.^39,40 Fortunately, NSFNMTC and sporadic non-medullary thyroid carcinoma appear to share the same somatic mutations, so the same targeted therapies can be used, until more specific targeted treatments become available.⁴⁰

Other prognostic and predictive markers

Molecular alterations have been associated with more aggressive behaviour. The best characterised are mutations in the telomerase reverse transcriptase (TERT) promoter, occurring at two hotspots: C228T and C250T. These mutations enhance telomerase activity, promoting the immortalisation of malignant cell lineages.⁶ Their presence correlates with a more dedifferentiated phenotype, as their frequency increases progressively from well-differentiated follicular cell-derived carcinomas to poorly differentiated carcinomas and ultimately to ATC. Additionally, these mutations are linked to radioiodine refractoriness and a higher incidence of distant metastases.^41,42

Beyond TERT promoter mutations, other alterations involving TERT, such as aberrant promoter methylation patterns, TERT gene locus amplification, and TERT mRNA overexpression, have also been associated with adverse prognosis in well-differentiated follicular cell-derived thyroid carcinomas.^43,44

Another key molecular alteration linked to aggressive behaviour is the presence of TP53 mutations. Their frequency increases as tumour differentiation declines; for instance, approximately half of ATCs harbour TP53 mutations.^9,31 In high-grade, non-anaplastic follicular cell-derived thyroid carcinomas, these mutations have been associated with lower disease-specific survival.⁴⁵ Moreover, although TP53 mutations are exceedingly rare in MTC (∼2%),⁴⁶ recent studies have linked their presence to worse overall survival.²⁴

The appropriate selection of the adequate tumour sample for genetic testing

All procedures, practices, and environmental conditions to which the biospecimen is exposed before a laboratory analysis are known as preanalytical factors, and these can determine the accuracy of genetic testing.⁴⁷

Several publications indicate that over-fixation and delayed fixation (cold ischaemia) of tissues can compromise DNA and RNA integrity. A cold ischaemia time of up to 1 hour is considered prudent.^47,48 Specimens should be fixed using a fixative-to-tissue mass ratio of at least 4:1 (optimally 10:1) with standardised, quality-controlled 10% pH neutral phosphate-buffered formalin. Total fixation time should be no less than 6 h and no more than 24–36 h.^47,49,50 Ideally, specimen thickness should not exceed 5 mm. Therefore, immediate serial sectioning of thyroidectomy specimens after surgery should be considered and implemented according to local workflows.

The use of acid decalcification, whether before or during the fixation process, causes hydrolysis and degradation of DNA and RNA, and is therefore contraindicated for molecular analyses of nucleic acids.⁴⁷ When a nodule or capsule is heavily calcified, it is advisable to dissect and process a non-calcified viable portion of the tumour without prior decalcification, ensuring proper identification for subsequent molecular analysis.

Tissue processor maintenance requires strict adherence to the manufacturers’ guidelines and low-melting point paraffins are recommended (55–63 °C) for tissue impregnation.^47,49,50

All paraffin blocks should be stored in dry, pest-free conditions at room temperature (defined as 25 °C). Regarding the duration of paraffin block storage, the following thresholds have been demonstrated: DNA ≤ 5 years and RNA ≤ 1 year.^47,51 Nevertheless, due to the clinical evolution of the majority of these tumours, it is foreseeable that, even now, we will perform the molecular testing much later than surgery for the primary one. Therefore, when handling tumour samples at the time of thyroidectomy, it is essential to process them with the understanding that a certain percentage of patients may experience relapse or disease progression many years after surgery and may later require molecular analyses on these stored samples.

The slide/sample must accurately represent the diagnostic features of the neoplasm, while avoiding extensive areas of necrosis, inflammatory infiltrate, and fibrosis. In patients with multiples samples, the most recent tissue should be used.^51,52

It is recommended that a pathologist or trained biologist mark the area of the section containing neoplasia on the haematoxylin-eosin (H&E) slide for macrodissection/microdissection. Tumour cellularity is expressed as the percentage of neoplastic cells relative to the total number of nucleated cells in the sample.^51,53 It is advisable to make the estimation in deciles; in general a fraction of malignant cells greater than 10–20% is considered the lower acceptable limit for most molecular methods.^51,54,55 Input nucleic acid requirements for molecular testing are variable, with minimum recommendations ranging from 1 to 10 ng. Most molecular testing techniques require at least 5–10 ng DNA for reliable analysis.⁵⁴

In selected cases, material obtained by fine-needle aspiration cytology (FNAC) can be considered for molecular testing, provided that a cytopathologist confirms adequate cellularity for genetic testing.

Techniques for genetic testing assessment

Tissue assays commonly used to identify molecular alterations include both traditional single-gene methodologies, such as Sanger sequencing, IHC, break-apart fluorescence in situ hybridisation (FISH), and reverse transcription real-time polymerase chain reaction (RT-PCR), as well as multiple-gene techniques such as the increasingly available Next Generation Sequencing (NGS) in routine clinical settings.

BRAF is the most common driver mutation in thyroid cancer, with the V600E variant representing the most frequent alteration.^56,57 Regarding DNA-based single-gene methods, Sanger sequencing is time-intensive, but remains the gold standard for mutation detection (Table 1, Fig. 2). However, it is a low-sensitivity technique, highly dependent on the percentage of tumour cells in the sample. Therefore, other DNA-based single-gene assays with higher sensitivity, such as real-time PCR, are recommended for the individual testing of the BRAF V600E variant. The BRAF IHC assay serves as a surrogate marker for detecting the V600E variant. The analysis using the anti-BRAFV600E monoclonal antibody (VE1 clone) is characterised by rapid turnaround time (less than 24 h for in-house tests), cost-effectiveness, and high sensitivity and specificity^58,59 (Table 1, Fig. 2). One advantage over DNA-based methodologies is that the spatial distribution of positive cells can be visualised microscopically, allowing for a semi-quantitative assessment of mutant protein expression. Finally, DNA-based NGS offers high sensitivity for mutation detection, including low-variant allele frequencies⁶⁰ (Table 1, Fig. 2).

Table 1.

Testing methods for predictive biomarkers in thyroid cancer.

	Turnaround time	Input material	Precise annotation of variants	Analytical sensitivity	Diagnostic sensitivity
BRAF V600E and RET mutations
Sanger sequencing	1–2 weeks	High (6–8 slides)	Yes	Low	High
IHC^a	Hours	Low (1 slide)	Yes	High	Low
Real time PCR	2–3 days	Intermediate (3–5 slides)	Sometimes	High	Intermediate
DNA-based NGS	Variable (2–3 days for some amplicon assays)	Variable (low for some amplicon assays)	Yes	High	High
RET, NTRK1/2/3, and ALK fusions
IHC^b	Hours	Low (1–2 slides)	No	High	High
FISH	2 days	Variable (1 slide for RET and ALK genes and 3 slides for NTRK1/2/3 genes)	No	High	High
RT-PCR	3–4 days	High (6–8 slides)	Sometimes	High	Intermediate
DNA-based NGS	Variable (2–3 days for some amplicon assays)	Variable (low for some amplicon assays)	Yes	Lower than RNA-based NGS	High
RNA-based NGS	Variable (2–3 days for some amplicon assays)	Variable (low for some amplicon assays)	Yes	High	High

a

Only applicable to BRAF V600E.

b

Only applicable to NTRK and ALK fusions.

Abbreviations: DNA, deoxyribonucleic acid; FISH, fluorescence in situ hybridisation; IHC, immunohistochemistry; NGS, next-generation sequencing; PCR, polymerase chain reaction; RNA, ribonucleic acid; RT-PCR, reverse transcription real-time polymerase chain reaction.

Fig. 2.

Testing workflows for predictive biomarkers in thyroid cancer. Route A: Single-gene testing is performed first and complemented with NGS in cases that test negative. The order of assays is guided by the likelihood of alterations. Assay choice depends on local resources. *Applied only to detect RET mutations. #Used only as a screening tool; any positive result requires confirmation by NGS. Route B: Up-front NGS is performed. +Medullary thyroid carcinoma should be tested through this route.

RET oncogenic activation can occur through mutations and rearrangements. Regarding RET fusion approaches, the European Society for Medical Oncology (ESMO) recommendations encourage the use of upfront NGS, if available, or alternatively FISH or RT-PCR if NGS is not accessible.⁶¹ The limited sensitivity of DNA-based NGS for the detection of gene fusions makes it necessary to develop a complementary RNA-based NGS for driver-negative cases.^60,61 Briefly, FISH and NGS are effective tools regardless of the RET fusion partner, offering high diagnostic sensitivity or comprehensiveness of these assays. Conversely, RT-PCR typically provides the most limited coverage. Regarding FISH interpretation, it is widely described as challenging due to the proximity of common RET partners on chromosome 10.^62,63 Finally, RET IHC is currently not recommended as a clinical screening assay because of its low sensitivity and specificity⁶¹ (Table 1, Fig. 2). Patients diagnosed with MTC should be tested for the presence of germline RET mutations, so PCR or NGS can be carried out on buccal or blood samples. Detection of a germline RET variant necessitates family counselling. For patients who are germline RET-negative or whose germline status is unknown, somatic testing of tumour specimens should be conducted.^61,64

Rearrangements of the NTRK family genes have recently been incorporated as predictive biomarkers in a “tumour-agnostic” approach. The ESMO also recommends up-front NGS when available, or alternatively, IHC screening for tumours with a lower probability of NTRK fusions, using a histology-based triaging strategy.⁶⁵ In this context, pan-TRK IHC is considered the fastest and most cost-effective test.^66–68 However, this methodology has some limitations, including reduced sensitivity for detecting NTRK3 fusions and a high rate of false-positive results due to cross-reactivity with other driver gene alterations.^68–70 Therefore, all pan-TRK positive cases should be confirmed by NGS. Finally, for tumours with a higher likelihood of harbouring an NTRK fusion, FISH, RT-PCR, or RNA-based NGS are considered the preferred initial confirmatory techniques^60,65 (Table 1, Fig. 2).

ALK fusions are low-frequency molecular alterations found in thyroid cancer, with STRN and EML4 being the most common fusion partners. ALK IHC, FISH, and RNA-based NGS are available tools due to the high diagnostic sensitivity of these assays.^6,57 Currently, since NGS allows simultaneous analysis of multiple biomarkers, it provides a more tissue-efficient approach compared with serial single-gene assays when searching for low-frequency molecular alterations⁶⁰ (Table 1, Fig. 2).

When to perform genetic testing and the clinical applicability of molecular analysis results

The relevance of conducting adequate molecular analysis lies in its direct impact on patients’ management as, nowadays, several targeted drugs have already demonstrated a significant benefit across different oncological outcomes. Table 3 shows the main results of targeted therapies against RET, NTRK, ALK, and BRAF alterations across different histological types of thyroid cancer.⁵ Indeed, the ESMO Scale for Clinical Actionability of Molecular Targets (ESCAT) assigns tier I status to RET, NTRK, and BRAF alterations in metastatic thyroid carcinoma.⁷¹ Therefore, patients should have this molecular information in order not to miss the opportunity of receiving these treatments, so the optimal treatment will be linked to that indication.⁵ In this sense, ideally, patients who are candidates for the initiation of systemic treatment in the advanced RAR-DTC or MTC setting, as well as patients at diagnosis of ATC, should undergo genetic testing, at least including BRAF, RAS, RET, NTRK, and ALK. It is also encouraged to determine expression of PD-L1 in ATC patients (Table 2).

Table 3.

Key outcomes of targeted therapies across different histological types of thyroid cancer.

Molecular alteration	Drug	Subtype	N	ORR	PFS
BRAF V600E^85,86	Vemurafenib	DTC	26 (Naïve)	38%	18.2 m
			25 (Pre-treated)	27%	8.9 m
	Dabrafenib	DTC	26	42%	10.7 m
	Dabrafenib–Trametinib	DTC	27	48%	15.1 m
		ATC	36	56%	6.7 m
RET mutations^87–89	Selpercatinib	MTC	55 (Pretreated)	69%	NA
			88 Naïve (Phase 2)	73%	NA
			Naïve (Phase 3)	69%	NA
	Pralsetinib	MTC	53 Pretreated	60%	NA
			19 Naive	74%	NA
RET fusions^89,90	Selpercatinib	DTC	19	79%	NA
	Pralsetinib	DTC	11	91%	NA
NTRK fusions^91,92	Larotrectinib	DTC	21	86%	NA
	Entrectinib	ATC	7	29%	2.2 m
		DTC	10	50%	NA
ALK fusions^93–96	Crizotinib	ATC	1	–	>24 m
		DTC	1	–	7 m
	Alectinib	DTC	1	–	>8 m
	Lorlatinib	DTC	1	–	>7 m

Abbreviations: N, sample size; ORR, objective response rate; PFS, progression-free survival; DTC, differentiated thyroid carcinoma; ATC, anaplastic thyroid carcinoma; m, months; NA, not available; Naive, treatment-naïve.

Table 2.

When to perform somatic molecular testing in thyroid cancer subtypes.

Tumour type	Appropriate timing
MTC	Metastatic tumours before initiating first-line systemic treatment
Follicular cell-derived TC	In metastatic, radioiodine-refractory tumours before systemic treatment
DHGTC/PDTC	At diagnosis of metastatic disease
ATC	In localised or metastatic tumours before initiating systemic treatment

Abbreviations: DHGTC, differentiated high-grade thyroid carcinoma; PDTC, poorly differentiated thyroid carcinoma; ATC, anaplastic thyroid carcinoma; MTC, medullary thyroid carcinoma.

Potential novel recommendations for genetic testing in the future

Drug resistance mutations and clonal selection

NGS is particularly valuable in patients who have progressed on multikinase inhibitors (MKI) or selective RET or NTRK inhibitors, as resistance mechanisms to these therapies are heterogeneous and involve different molecular pathways, influencing subsequent treatment decisions.

Regarding RET, one key resistance mechanism to MKIs is the emergence of the gatekeeper mutation V804M. New selective inhibitors, such as selpercatinhib and pralsetinib, show activity against this and other RET gatekeeper mutations.⁷² Other off-target resistance mechanisms to MKIs include MET mutations and amplifications, and KRAS mutations.⁷³

Although the number of thyroid cancer patients progressing on selective RET inhibitors remains limited, preclinical studies have identified several resistance mechanisms. These include the appearance of Solvent Front Mutations (SFM) such as RET G810A.⁷⁴ In lung cancer patients treated with RET inhibitors, specific resistance mechanisms have been described. In a cohort of 18 RET inhibitor-resistant patients, on-target mutations were observed in only two cases (affecting RET G810S and G810C), while the majority developed off-target resistance mechanisms, including MET and KRAS amplifications.⁷⁵

For NTRK, resistance mechanisms to selective inhibitors have been associated with mutations in the NTRK1 domain and activation of IGF1R.⁷⁶ Additionally, on-target mutations in NTRK1, NTRK2, and NTRK3, as well as off-target resistance mechanisms, including BRAF V600E mutations and MET amplifications, have been observed.⁷⁷ Novel selective TRK inhibitors are in development, showing efficacy against on-target resistance mutations such as TRKA G595R/G667A/G667C in preclinical models.⁷⁸

Although there are no established treatment strategies for thyroid cancer patients progressing on selective RET or NTRK inhibitors, repeat testing may help identify patients eligible for clinical trials based on their on-target or off-target resistance mechanisms. Second-generation selective inhibitors are currently under development.⁷³

In addition, other detectable genomic alterations present in thyroid cancer are gradually being addressed with emerging targeted therapies, such as pioglitazone therapy for PAX8::PPARγ activated tumours.⁷⁹

Liquid biopsy in thyroid cancer

Liquid biopsy enables the detection of tumour biomarkers in blood or other bodily fluids, such as circulating cell-free DNA (cfDNA) or microRNA (miRNA). Currently the only blood-based biomarkers used in thyroid cancer monitoring are thyroglobulin for DTC, and calcitonin and CEA for MTC.⁹

Liquid biopsy could provide valuable insights into somatic tumour alterations without the need for repeat biopsies, enabling monitoring of changes over time in response to different treatments.⁸⁰ Therapeutically relevant mutations in thyroid cancer, including BRAF V600E and RET M918T, can be detected through ctDNA in liquid biopsy, with concordance rates ranging from 42% to 92% depending on the setting.⁸⁰ This concordance is particularly high in treatment-naïve ATC patients harbouring BRAF V600E mutations.⁸¹

Tracking these targetable mutations over time is useful for monitoring treatment response. For example, a decrease in BRAF V600E allele frequency (MAF) in patients receiving BRAF/MEK inhibitors correlates with treatment response, while an increase may indicate progression.⁸² Similarly, the resistance RET V804M mutation in patients treated with vandetanib or cabozantinib can be detected through ctDNA.⁸³

Other applications of liquid biopsy, such as the detection of circulating tumour cells (CTCs) or the assessment of miRNAs, are still under investigation and have not yet become clinically relevant in thyroid cancer. However, in the era of personalised medicine, liquid biopsy has the potential to enhance patients access to tailored therapies, especially in cases where histological material is not available, as well as for monitoring efficacy of targeted therapies and early detection of resistance mechanisms.⁸⁰

Main challenges to integrating liquid biopsy into clinical practice include the high variability among detection methods, patient selection biases in published studies, and the tumour stage at the time of blood sample collection.⁸⁴

Molecular tumour board

The predictive role of certain molecular alterations in thyroid carcinoma, combined with the availability of selective targeted inhibitors, has highlighted the need for multidisciplinary molecular tumour boards (MTB) to join the information on sequencing analysis interpretation and patients’ clinical characteristics to establish a personalised therapeutic strategy.

The team from a MTB should include, at least, experts in pathology, molecular biology, oncology and endocrinology.

To ensure the proper functioning of the MTB and guarantee the decisions taken, we propose the following checklist to guide the assessment of results obtained from each patient presented:

• -

  Date of sample collection.

• -

  Type of tumour sample (fine needle aspiration, core needle biopsy, surgery).

• -

  Tumour sample location: primary/lymph node metastases/visceral metastases.

• -

  Clinical status and prior treatments: surgery/RAI/MKI.

• -

  Tumour sample adequacy.

• -

  Type of analysis performed.

• -

  Molecular findings: (i) point mutations/copy number alterations/insertions/deletions, and gene rearrangements; (ii) genomic signatures, such as tumour mutational burden (TMB), loss of heterozygosity (LOH), microsatellite status, and homozygote repair deficiency (HRD).

• -

  Requirement for genetic counselling.

• -

  Actionability.

• -

  Treatment recommendation.

This MTB will also allow an outcome registry by tracking patient responses to recommended therapies, sharing the results with other MTB registries, performing educational meetings, promoting the development of research trials, and the incorporation of new techniques according to emerging needs detected.

Conclusion

Despite its generally good prognosis, thyroid cancer encompasses aggressive subtypes requiring specific diagnostic and therapeutic approaches. Molecular profiling is now essential in thyroid cancer, particularly for advanced, metastatic, or high-grade subtypes. The implementation of updated histological classification, adequate sample management, and integrated molecular diagnostics can optimise therapeutic outcomes and personalise patient care.

Ethics

This work did not involve animal experimentation, participation of patients or human subjects, nor does it constitute a clinical trial.

Generative AI

During the preparation of this work the author(s) used ChatGPT 4.0 to assist in drafting the abstract. After using this tool/service, the author(s) reviewed and edited the content as necessary and take(s) full responsibility for the content of the publication.

Funding

SH received research funding from Fundación Mutua Madrileña (AP18051-2022) and the Instituto de Salud Carlos III (ISCIII) (PI22/01700), co-funded by the European Union (EU). ILL received research funding from Fundación Galega de Investigación Biomédica Galicia Sur. JMC-T was supported by the Instituto de Salud Carlos III (ISCIII), Spain (grant PI23/00722), co-funded by the European Union (EU).

Conflicts of interest

IRC has received speaker honoraria from Lilly. TAG has served in scientific consultancy, speaker, and advisory roles with Lilly, Ipsen, Bayer, Johnson & Johnson, Astellas, Eisai, Novartis, MSD, BMS, and Pfizer, and has received research grants from IPSEN and Johnson & Johnson. SH receives honoraria from Roche, AstraZeneca, Pfizer, and Thermo Fisher Scientific. ILL has received honoraria for speaking from AstraZeneca, Ipsen, and Novartis; and has received grant support for congresses, conferences, and medical training sponsored by Novartis, Pierre Fabre, Roche, Lilly, and Daiichi-Sankyo. MTRC, PJF, HDQA, MRBC, CI, JMC-T have no conflicts of interest to declare.