2. Molecular Landscape of Well-Differentiated Thyroid Carcinoma (WDTC)

According to the definition provided by the World Health Organization (WHO) classification of tumors of endocrine organs ^[2], PTCs are malignant carcinomas with follicular differentiation showing a set of peculiar nuclear features and/or a papillary growth pattern. The nuclear changes of PTC can be summarized on the basis of three main categories: (1) size and shape: the nuclei may appear enlarged, elongated and overlapping; (2) nuclear membrane irregularities: the contours of the nuclei appear irregular; the nuclear grooves and pseudoinclusions are typically present; (3) chromatin changes: mainly presence of chromatin clearing and margination.

The typical appearance of PTC, the classical or conventional type (CPTC), includes the presence of both distinctive nuclear features and papillary pattern, with the tumor that is rarely totally encapsulated ^[2]. However, PTC appears in a wide range of microscopically distinct entities, the so-called PTC variants. Each variant does not only show peculiar histo-morphological and molecular characteristics, but can also assume a clinically different behavior.

After the classical type, the follicular (FVPTC) is the most frequent variant of PTC. It shows the papillary-like nuclear changes but has a predominant follicular growth pattern. FVPTC has probably been the most discussed histotype of all thyroid tumors over the last 20 years. This is presumably due to its highly variable biological, pathological and clinical manifestations. Three main groups of FVPTC can be distinguished on the basis of their features of invasiveness: noninvasive encapsulated, invasive encapsulated, and infiltrative ^[6]. The noninvasive encapsulated category has been frequently reported to be clinically indolent, to such an extent that a downgrading of this type of FVPTC—from “carcinoma” to “neoplasm”—has been proposed and obtained (see the NIFTP paragraph). A clear clinical/prognostic demarcation between encapsulated invasive and infiltrative FVPTC is hard to obtain because authors rarely make such a distinction. However, it seems that infiltrative FVPTC should be molecularly closer to CPTC ^[7].

The most frequent molecular alterations detected in PTCs occur in the MAP kinase (MAPK) signaling pathway, and mutations in the BRAF gene are found in 45–50% of all PTCs ^[8]. It is well-known that the prevalence of mutations varies according to the PTC variant, so that BRAF^V600E is detected in up to 70% of CPTCs, while its frequency is much lower in FVPTC, where it is found almost exclusively in the infiltrative forms ^[7]. The most frequent driver mutations detected in FVPTC, in up to 48% of cases, occur in the RAS family genes (NRAS, HRAS and KRAS) ^[7].

The Cancer Genome Atlas (TCGA) research network delivered a multiplatform extensive integrated molecular characterization of 496 PTCs, mainly classical and follicular variants ^[9]. This study highlights that, compared to other cancers, PTCs have a relatively stable genome with genetic mutations occurring or, better, recurring in a limited number of genes and a low mutation burden (about 0.41 non-silent mutations per megabase), so that more than one mutation rarely coexists in the same tumor. As expected, the most frequent driver mutation detected in PTC was BRAFV600E (57%), followed by point mutations in NRAS (8%) and RET fusions (7%). Mutations in HRAS, EIF1AX, KRAS, PPM1D, CHEK2, BRAF fusion, PPARG fusion and THADA fusion were cumulatively present in less than 5% of cases. NTRK fusions were also rare, being found in 2% of cases, although some authors had previously reported them in up to 3–5% of sporadic adult PTC ^[7][10]. It is worth noting that in thyroid cancer almost only NTRK1 and NTRK3 fusions have been described so far.

In addition to the fact that BRAF^V600E was prevalent in CPTC, while RAS mutations were enriched in FVPTC, the TCGA study has demonstrated that, by matching molecular results including genomic, epigenomic and proteomic data, two major categories of PTCs can be distinguished: the BRAF^V600E-like and the RAS-like PTCs. BRAF^V600E-like alterations include mutations other than the BRAF^V600E ones, as for example RET/PTC and BRAF fusions. These alterations are associated with a papillary architecture and a lower expression of thyroid-differentiation genes. On the contrary, RAS-like alterations (including RAS mutations, the BRAF^K601E mutation, PPARG fusions, THADA fusions and EIF1AX mutations) are associated with a follicular architecture and a close-to-normal expression of the genes related to thyroid differentiation ^[9]. The identification of this fundamental, deep distinction between CPTCs and FVPTCs has led many authors to wonder whether BRAF-like and RAS-like definitions could be extended to the other thyroid tumors.

For example, the tall cell variant PTC (TCVPTC, about 6% of all PTCs), an aggressive variant characterized by a high prevalence of BRAF^V600E mutation ^[11], could be easily placed among the BRAF^V600E-like tumors. The same could be said for the hobnail variant PTC. On the other hand, for more rare PTC variants, dedicated genomic and transcriptomic studies are still necessary. For instance, the solid variant (1–3% of PTCs ^[2]) has been described to harbor not only BRAF mutations, but also RET and NTRK fusions; however, this variant seems to be molecularly different from CPTC ^[12].

FTCs, follicular-patterned tumors that lack PTC nuclear alterations, are usually encapsulated and show tumor capsule and/or vascular invasion. They represent 6–10% of follicular-derived thyroid tumors ^[2]. Unlike PTCs, which cause lymph node involvement in many patients, FTCs often show hematogenous spread to distant organs (mainly to the bones and lungs). The TCGA study results on PTC have been confirmed by several subsequent studies, where FTCs were also included ^[7][13][14]. Apart from the already known high prevalence of RAS (in up to 50% of cases) ^[15] and of RAS-like mutations in FTC, a genetic similarity has been demonstrated between FTCs and encapsulated FVPTCs ^[7].

Moreover, based on transcriptional data, the existence of a third-class of tumors has been hypothesized: the so-called Non-BRAF-Non-RAS (NBNR), which are neither BRAF-like, nor RAS-like; this class of tumors might also include PPARG fusions, and EIF1AX mutations, but these evidences need to be confirmed ^[7].

Finally, the HCCs are a group of encapsulated tumors predominantly composed of oncocytic cells and characterized by capsular/vascular invasion. The classification of these neoplasms as a subtype of follicular tumors has been a hotly-debated issue; according to the current indications of WHO, HCCs are independent histo-pathological entities belonging to the DTC ^[2]. HCCs account for 3–5% of all non-medullary thyroid tumors ^[16][17]. Similarly to FTC, HCCs show a higher incidence of distant metastases compared to PTCs. The molecular frame of HCCs is completely different from that of the rest of the WDTCs, being characterized by three main types of alterations: (1) mitochondrial DNA mutations, occurring as early events in genes encoding complex I subunits; (2) point mutations recurring in genes that are not typically mutated in thyroid cancer, with the exception of few RAS and EIF1AX mutations; (3) karyotype alterations, with tumors having a near-haploid state, a polysomic state and/or duplication of chromosomes 7, 5, and 12 ^[16][17].

Secondary mutations are not common in WDTCs. TERT promoter mutations show a frequency of about 10% in PTCs and 15% in FTCs ^[7][9][18]. However, a difference across PTC variants has been observed, with tall cell PTCs reaching a frequency of 25%. Similarly, TERT promoter mutations have been reported in 15–20% of HCCs, with a higher occurrence in widely invasive (32%), versus minimally invasive (5%) tumors ^[17][18]. Since TERT promoter mutations show a higher frequency in morphologically aggressive WDTCs such as tall cell PTCs and widely invasive HCCs, they can contribute in part to the evidence demonstrating their association with poor patient prognosis. Indeed, significant associations between TERT promoter mutations and distant metastases, disease persistence and recurrence, advanced stage and also patients’ survival have been demonstrated in WDTC patients ^[19][20][21]. The influence of TERT promoter mutations, mainly C250T and C228T substitutions, on prognosis can be attributed to the creation of de novo binding sites for transcription factors able to enhance the gene transcriptional activity and promote telomerase activation ^[21].

3. Poorly Differentiated and Anaplastic Thyroid Carcinoma

Compared to WDTCs, PDTCs and ATCs are rare, accounting for 5–10% of all thyroid tumors, but their prognosis is much less favorable ^[2].

The diagnostic criteria for PDTC have been quite controversial. Two main systems are currently used: the Turin proposal and the Memorial Sloan Kettering Cancer Center (MSKCC) criteria. According to the Turin proposal, also endorsed by the WHO classification system, papillary-like nuclear features must be absent, the growth pattern is solid/trabecular/insular, and tumors should present at least one feature among high mitotic index, necrosis and convoluted nuclei ^[25]. The MSKCC criteria are less rigid; they are based on the mere presence of a high mitotic rate and/or tumor necrosis, thus PDTCs are diagnosed independently of the observed growth pattern and of PTC nuclear changes ^[26]. Depending on which diagnostic system is adopted, differences have been reported not only concerning the prevalence and the prognosis of PDTC, but also affecting their molecular status. In fact, Turin-PDTCs harbor more frequently RAS mutations, while MSKCC-PDTC are enriched with BRAF mutations ^[27]; moreover, Turin-PDTCs compared to MSKCC-PDTCs show a higher incidence of TP53, EIF1AX, PTEN and PIK3CA mutations ^[28].

ATC is a highly aggressive tumor, with distant metastases at diagnosis in 30–40% of cases ^[2]. At microscopic examination, it has a variable cellular appearance, with a complete loss of follicular differentiation. ATC is often associated with a coexisting DTC, or with a clinical history of DTC, and this suggests that clones from DTC may undergo a morphological evolution towards dedifferentiation ^[29]. In this context, molecular studies have shown partial overlapping between ATCs and their differentiated counterparts ^[30].

Recently, many authors have performed massive parallel sequencing studies on both PDTCs and ATCs to investigate in detail their molecular aspects and to find differences between these two tumors. The frequency of the main mutations and fusions detected in PDTCs and ATCs are reported in Table 1.

4. Molecular Alterations in Advanced Differentiated Cancers

The study of WDTCs with clinically aggressive behavior is crucial to understand whether specific molecular alterations could be indicative of high-risk tumors. Moreover, patients with advanced treatment-resistant tumors should be investigated for the presence of druggable alterations. The definition of “advanced” is not univocal: Table 2 summarizes the main findings so far reported on advanced PTCs and FTCs, including cases with persistent/recurrent disease, radioiodine resistant tumors, stage IV at presentation, distant metastases.

5. Molecular Markers for Targeted Therapy in Thyroid Cancer

The advances in the field of biological drugs have led to the development of targeted agents against specific molecular alterations. BRAF inhibitors alone or in combination with MEK inhibitors have shown good response rates in BRAF^V600E ATCs ^[41]. The administration of multi-target thyrosine kinase inhibitors (TKI) such as sorafenib and lenvatinib, approved in many countries, has improved the progression-free survival of radioiodine-resistant DTCs ^[42][43]. Indeed, a recent meta-analysis reports that treatment with lenvatinib in DTC patients achieved a pooled partial response rate of 69%, and a progression-free survival of 19 months ^[44]. However, prognosis in ATC patients remained poor even with TKI therapy (pooled progression-free survival was 5 months), and a complete response was rarely achieved (0.3%). Moreover, owing to its inhibitory effects against multiple targets, TKI treatment causes adverse events involving fatigue, gastrointestinal symptoms, hypertension, liver disfunction and affecting also thyroid function and metabolism ^[45].

With regard to immunotherapy, it has been demonstrated that ATCs express PD-L1, but evidences of the efficacy of immune check-point inhibitors in the treatment of thyroid cancer patients are still limited ^[29][46]. The use of the anti-PD-1 drug pembrolizumab in combination with lenvatinib is currently being evaluated in clinical trials enrolling DTC, PDTC and ATC patients (NCT02973997; NCT04731740).

Drugs targeting gene rearrangements involving thyrosine kinase receptors such as NTRK have been FDA-approved for NTRK-fused cancers, independently of the tumor type. In particular, larotrectinib selectively targets NTRK rearrangements, while entrectinib has effects also on ALK- and ROS1-altered receptors ^[47][48]. Preliminary data on thyroid cancer show that larotrectinib is effective, demonstrating an overall response rate of 75% ^[49]. In spite of the rarity of NTRK fusions in non-pediatric thyroid cancers (2–5%) ^[9][10], these targeted drugs might represent a promising strategy of treatment in advanced tumors lacking the most common molecular alterations.

Finally, FDA has recently approved two selective RET inhibitors: selpercatinib and pralsetinib. These drugs have been evaluated on RET-altered solid cancers, and approved for mutated MTC, and rearranged DTC and non-small cell lung cancer. Considering only the CCDC6/RET-positive DTCs (also known as RET/PTC1 fusion) enrolled in both clinical trials, the overall response rate for selpercatinib and pralsetinib was 100% (8/8 patients) and 83% (5/6 patients), respectively ^[50]. RET fusions in ATC are quite uncommon (see Table 1), but these results have paved the way to additional therapeutic options for a non-negligible portion of patients with advanced DTC and PDTC.

6. Copy Number Alterations, Gene Expression and microRNA in Thyroid Cancer

Alterations other than gene mutations and fusions have been investigated in thyroid cancer. However, their application in the clinical context is less straightforward and may require complex data analysis as well as procedure standardization.

The biological and clinical role of copy number alterations (CNA) in thyroid cancer has been scarcely investigated, probably due to the technical difficulties related to this type of analysis. Somatic CNAs are characterized by the loss or gain of one or more copies of a gene or a locus in tumor cells. In TCGA cohort, 27% of tumors had CNAs, and this group was enriched in PTCs lacking driver mutations or fusions ^[9]. These findings indicate that somatic CNAs can represent driver genetic events in PTC tumorigenesis. The wide spread of massive parallel sequencing platforms and the availability of targeted panels that also include CNA analysis are likely to expand knowledge in this field of study.

The expression of specific transcripts and/or microRNAs (miRNA) has been proposed for diagnostic and prognostic purposes. There are currently no mRNA transcripts that have proved sensitive or specific enough to be applied as stand-alone markers in routine applications. Thanks to the advent of multitarget platforms (e.g., the nCounter nanoString platform), the most recent scientific approaches have been investigating gene panels able to identify peculiar expression patterns, rather than single markers ^[51]. This analysis strategy, in conjunction with mutation analysis, is applied to the differential diagnosis of indeterminate thyroid nodules by diagnostic commercial tests such as the Afirma Genomic Sequencing Classifier and the Thyroseq v3 Genomic Classifier ^[52].

MiRNAs are small non-coding RNA molecules that regulate the expression of specific transcripts in an epigenetics manner. Several miRNAs have been reported as being involved in thyroid cancer pathogenesis as well as in cancer progression. Some of them (for example miR-146b, miR-221 and miR-222) have been consistently and widely reported as up-regulated in PTC by many authors, and could therefore serve as PTC markers ^[53][54][55]. It has been demonstrated that up-regulated miRNAs target suppressor genes, such as PTEN, belonging to the MAPK and the PI3K/AKT pathways ^[54]. On the contrary, suppressor miRNAs have been frequently reported as down-regulated in thyroid cancer (for instance miR-375, miR-7 and miR204) ^[55][56]. Moreover, since miRNAs are differentially expressed in thyroid tumors versus benign lesions and in different histotypes of thyroid cancer, the development of miRNA-based molecular test is an appealing strategy for the diagnostic and prognostic definition of thyroid tumors ^[57].