NTRK genes include NTRK1, NTRK2, and NTRK3. The autophosphorylation of NTRK1 tyrosine residues in the tyrosine-kinase domain increase NTRK1 activity. NTRK gene fusions are oncogenic drivers that lead to NTRK gene fusions due to intra- or inter-chromosomal rearrangements, and cytoplasmic Trk fusion proteins activate downstream signals through phosphoinositide 3-kinase (PI3K), mitogen-activated protein kinase (MAPK), and phospholipase C-γ (PLCγ) to drive tumor proliferation and spread. NTRK fusions have been found in 3-6.7% of papillary thyroid cancer (PTC) and 20% of PDTC (9).

The ALK is a transmembrane tyrosine kinase of the insulin receptor family that, when the ligand binds to its extracellular structural domain, promotes activation of multiple downstream signaling pathways, such as PI3K/AKT, MAPK, and Janus kinase (JAK)-signal transducer and activator of transcription (STAT). ALK mutations and rearrangements are most common in ATC (11.1%) and PDTC (4%), and they play a role in disease progression and aggressiveness (10).

The RET gene is a proto-oncogene that encodes the RET protein of the tyrosine kinase receptor superfamily. RET protein is a receptor tyrosine kinase, undergoing phosphorylation at several tyrosine residues, that can activate various downstream signaling pathways, such as MARK and PI3K, to induce cell proliferation. Rearrangements of the RET and other genes are common (5% - 25%) in PTC (3). The expression of thyroid-specific genes, increasing the differentiation process was suppressed by conditional activation of RET/PTC1 or RET/PTC3 (6). RET/PTC1 accounts for about 60% of RET-associated PTC, with RET/PTC3 accounting for approximately 30%. Although uncommon, RET/PTC rearrangements have been discovered in ATC and PDTC, primarily in carcinomas with a differentiated component.

Copy number increases have been found in different subtypes of thyroid cancer such as epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor A/B (PDGFRA/B), vascular endothelial growth factor receptor 1,2 (VEGFR1,2), mast/stem cell growth factor receptor kit (c-Kit) and metabotropic proto-oncogene receptor tyrosine kinase (MET). Missense mutations such as fibroblast growth factor receptor 2 (FGFR2) and FMS-like tyrosine kinase 3 (FLT3) were found in 11% and 17% of PDTC, respectively (3). The human epidermal growth factor receptor 2 (HER2) gene (ERBB2) overexpression was discovered in follicular thyroid cancer (FTC) (44%), PTC (18%), and some ATC (11). HER2 and HER3 are essential actors upstream of the signal-regulated kinase Extracellular Signal-Regulated Kinase (ERK) and AKT signaling pathways. Overexpression of HER2 and HER3 may provide a RAIR-DTC tumor escape mechanism for BRAF mutant cells treated with the BRAF inhibitor vemurafenib (12).

Many human cancer types exhibit activation of the MAPK signaling pathway, which is accomplished by activating mutations or overexpression of MAPK upstream activators such as RTKs, Ras, and Raf, leading to MEK phosphorylation followed by ERK phosphorylation. Changes in the MAPK pathway are prevalent in thyroid cancer, particularly in PTC (40–80%), ATC (10–50%), and PDTC (5–35%) (17). It mostly includes mutations in the BRAF gene, which result in cell differentiation loss and apoptosis inhibition. Furthermore, patients with BRAF mutations show hypermethylation of the TSHR gene promoter. The most frequent BRAF mutation is the V600E gene replacement, which boosts BRAF protein activity and keeps it active. It forms a monomer independent of the upstream RAS kinase, leading to persistent activation of MEK/ERK, cell differentiation loss, tumor development, and apoptosis inhibition. BRAF^V600E mutations are found in 45-50% of PTC and 36% of ATC (18).

Molecular alterations in the Wnt/β-catenin signaling pathway involved in adenomatous polyposis coli (APC), axis inhibition protein 1 (AXIN1), and catenin beta 1 (CTNNB1) contribute to thyroid tumorigenesis. Furthermore, the direct phosphorylation of glycogen synthase kinase 3β (GSK3β) activates the WNT/β-catenin pathway and the association between β-direct catenin and PAX-8 boosts its gene transcription for NIS expression. These changes become more common in ATC (66%) and PDTC (25%) (19).

Several studies have found that TGF-β is essential for the proliferation and differentiation of thyroid cells (6). The TGF-β-type II receptor complex motivates the phosphorylation of Smad2 and Smad3. Some researchers demonstrated that a BRAF mutation could increase NADPH oxidase 4 (NOX4) expression in thyroid cancer cells by the TGFβ/SMAD3 signaling pathway. NOX4-reactive oxygen species (ROS) generation suppresses NIS expression in follicular cells by interfering with the binding of the PAX8 to the NIS gene promoter. NOX4 might be used as a therapeutic target in conjunction with other MAPK-kinase inhibitors to enhance their efficacy on RAIR-DTC redifferentiation (20).

TP53 gene is an oncogene that encodes a protein involved in a variety of cellular activities, which could cause cell cycle arrest, apoptosis, senescence, DNA repair, or metabolic alterations in response to cellular stress. TP53 inactivation, its degradation mediated by p53 poly-ubiquitination, has long been thought to be a final step in tumor growth. TP53 mutations were found in around 40-80% of ATC, 10-35% of PDTC, 40% of PTC, and 22% of oncocytic FTC (21).

TERT maintains the length and stability of chromosomes by adding telomeres to the ends of chromosomes, which is of great significance to the lifespan of the body and various cellular activities. Mutations of the TERT promoter are mostly late events in the development of thyroid cancer, the incidence of which is high, especially in ATC (70%), PDTC (40%), FTC (20%), and PTC (10%) (15). TERT promoter mutations have also been shown to be strongly associated with aggressive clinicopathological characteristics and the probability of recurrence or distant metastasis (22).

The SWI/SNF complexes gene mutations have been detected in ATC (36%) and PDTC (6%) (3). SWI/SNF complexes are critical for maintaining differentiated function in thyroid cancer, and their loss imparts radioiodine refractoriness as well as resistance to MAPK inhibitor-based redifferentiation therapy (23).

The EIF1AX gene encodes an essential eukaryotic translation initiation factor. EIF1AX mutations have been reported in PDTC (11%) and ATC (9%) associated with oncogenic RAS. In advanced disease, the dramatic interplay of EIF1AX and RAS mutations shows that they may work together to induce tumor progression (24). The mechanism of EIF1AX mutation in thyroid cancer tumorigenesis and dedifferentiation still needs to be further studied.

The IDH1 mutations are frequently found in thyroid cancer, identified in ATC (11%), FTC (5%), and PDTC (1.25%). While IDH2 mutation was identified in 3% of ATC (25). However, further research is needed to identify their functions in the pathogenesis of thyroid carcinomas.

PAX8/PPARγ rearrangement is the second most common genetic alteration in 20-50% of FTC besides RAS mutation, with an incidence of 30-35%, and it is also present in a minority of FVPTC (5%). It plays a role in the control of cell proliferation and redifferentiation. PPAR agonists have been proven to trigger redifferentiation in thyroid cancer in some studies (6).

Notably, dysregulated histone acetyltransferase and HDAC activity are linked to cancer cell growth, proliferation, and differentiation. Some researchers have discovered that histone acetylation is altered in thyroid tumorigenesis and H3 histone is turned off in the progression from differentiated to undifferentiated thyroid cancer (26).

A widely used radioisotope, radioiodine, plays a critical role in the diagnosis and treatment of DTC, such as ¹²³I, ¹²⁴I, and ¹³¹I. ¹³¹I SPECT/CT has become a routine tool for visualizing the lesions and evaluating distant metastases in patients receiving radioactive iodine therapy (29). ¹²⁴I PET/CT could improve the sensitivity and spatial resolution of SPECT/CT, leading to superior diagnostic performance of post-therapy 131I-WBS. However, it is expensive and has low accessibility (30) (Table 1).

TFB is a sodium/iodide symporter substrate with similar NIS affinities to radioiodine. ¹⁸F-TFB has recently been established as a flexible PET probe for imaging the activity of human sodium/iodide symporters. As a result, ¹⁸F-TFB PET could be a valuable method for evaluating NIS expression in human diseases and be able to visualize DTC metastases in negative ¹²⁴I PET (31). Compared to conventional diagnostic WBS and SPECT-CT, ¹⁸F-TFB PET could detect more local recurrence or metastases of DTC (32). The combination of ¹⁸F-TFB PET and fluorine-18-fluorodeoxyglucose (¹⁸F-FDG) PET appears to be a feasible technique for characterizing DTC tumor presentations in terms of differentiation and, as a result, individually planning and monitoring therapy. Prospective studies evaluating the potential of ¹⁸F-TFB PET in recurrent DTC are needed in the future.

¹⁸F-FDG is well-known radiopharmaceutical glucose that is mostly carried by glucose-transporter family-1 (GLUT1), and its uptake has been reported to be influenced by the degree of tumor proliferation and differentiation. Additionally, the surface expression of GLUT is controlled by the PI3k/AKT pathway. Advanced TC with low radioiodine uptake usually had high ¹⁸F-FDG uptake. Some researchers observed that ¹⁸F-FDG showed positive uptake in 50 patients (17%) among 258 DTC patients, 39 (78%) of which did not show positive lesions on post-therapy WBS (33). ¹⁸F-FDG PET/CT might allow RR-DTC patients to classify their prognosis by revealing tumor aggressiveness. ¹⁸F-FDG PET/CT has shown good diagnostic performance in non-iodine avid DTC with a sensitivity, specificity, and accuracy of 92%, 94%, and 93%, respectively. Therefore, ¹⁸FDG PET/CT could enable clinicians in identifying individuals with RAIR-DTC and developing a treatment strategy earlier (34).

SSTRs are highly expressed in neuroendocrine tumors. But in recent studies, SSTRs have been found to be overexpressed in dedifferentiated thyroid cancer. Less differentiated carcinomas are more likely to express a wider range of SSTR subtypes, primarily subtypes 2, 3, and 5, bolstering the theories of peptide receptor-based nuclear diagnosis and treatment. SSTR1-5 activation suppresses PI3K/AKT signaling (35). Parveen et al. evaluate the value of ⁶⁸Ga-DOTANOC PET/CT in DTC with negative ¹³¹I WBS and elevated serum Tg levels. The detection of recurrent disease in DTC with a sensitivity and specificity of 78.4%, and 100%, respectively. It may also assist in the selection of possible peptide receptor radionuclide treatment candidates (36).

The integrin αvβ3 is overexpressed in the tumor vascular system. The tripeptide sequence arginine-glycine-aspartate (RGD) shows a high affinity and specificity for integrin αvβ3. Recently, a prospective study has indicated that ⁶⁸Ga-DOTA-RGD₂ PET/CT showed a better diagnostic performance in RAIR-DTC with negative post-therapy ¹³¹I-WBS with an accuracy, sensitivity, and specificity of 86.4%, 82.3%, and 100%, respectively, compared to ¹⁸F-FDG PET/CT [75%, 82.3%, 50% (37)]. Moreover, the novel radiotracer could provide the potential for the selection of eligible RAIR-DTC candidates for treatment with ¹⁷⁷Lu-DOTA-RGD₂.

PSMA is a type II transmembrane glycoprotein receptor expressed in prostate cancer cells and the endothelium of tumor-associated neovasculature in several malignancies. Similarly, PSMA expression has been observed in 62% of persistent or recurrent DTC. Some studies have shown that PSMA expression was also related to poor prognosis and that very high PSMA expression was associated with poorer PFS (38). For patients with RAIR-DTC, ⁶⁸Ga-PSMA PET/CT can be useful for staging because it could identify different types of lesions and may discover lesions that ¹⁸FDG PET/CT does not detect. Additionally, ⁶⁸Ga-PSMA might be utilized to screen patients for ¹⁷⁷Lu-PSMA targeted therapy in the future (39).

In over 90% of epithelial carcinomas, FAP is significantly expressed in cancer-associated fibroblasts. Increased FAP expression is associated with dedifferentiation and aggressiveness outcome of thyroid cancer. In some cases, ⁶⁸Ga–FAPI PET/CT revealed high activity in the metastatic DTC with elevated Tg and negative iodine scan. ⁶⁸Ga-FAPI might perform better than ¹⁸F-FDG in detecting metastatic DTC, particularly in pulmonary and lymph node metastases (40). Another research has also found that ⁶⁸Ga-DOTA-FAPI-04 PET/CT may have a good performance in the detection of lymph node metastasis and distant metastasis in 87.5% (21/24) of RAIR-DTC patients (41). More multicenter prospective studies with bigger sample sizes are needed to confirm these findings.

Galectin-3 (Gal-3) is a β-galactoside binding protein of the lectin family that is absent in normal and benign thyroid tissues but overexpressed in the cytoplasm, cell membranes, and intercellular components of DTC and ATC (42). Meanwhile, Gal-3 is a physiological target of p53 transcriptional activity, and its downregulation mediated by p53 is essential for p53-induced apoptosis. ⁸⁹Zr-DFO-GaI-3mAb detected specific and reliable uptake of human thyroid cancer xenograft in vivo. ⁸⁹Zr-DFO-F(ab’)2 anti-gal-3 exhibited specific uptake in tumor tissue, while the normal thyroid tissue had no uptake. Besides, in the absence of radioiodine uptake, specific and selective detection of thyroid tumors was achieved by targeting Gal-3 (43). Gal-3 immunoPET is still a new field of research, and these findings imply that diagnostic and clinical applications of Gal-3 targeted radiotracers for thyroid cancer need further investigation.

In a recent study, a HER2-specific PET imaging probe ⁸⁹Zr-Df-pertuzumab was developed to assess the diagnostic effectiveness in orthotopic ATC. These findings suggested that noninvasive HER2 molecular imaging offers a great potential for detecting HER2 status in ATC (11). With extensive clinical translation and use of ⁸⁹Zr-Df-pertuzumab PET, this imaging method may be able to identify the diverse levels of HER2 around the body. This suggests that this unique imaging method could identify ATC patients who may react to HER2-targeted therapy (such as pertuzumab and trastuzumab) and dynamically monitor therapeutic responses.

MKIs inhibit the activity of multiple receptor tyrosine kinases such as VEGFR, PDGFR, FGFR, and various Raf kinases, thereby suppressing tumor cell proliferation and angiogenesis. Novel MKIs have been evaluated and approved by FDA for advanced RAIR-DTC such as sorafenib, lenvatinib and cabozantinib (45–47). Other commercially available MKIs (such as anlotinib, donafenib, surufatinib, sunitinib, and pazopanib) can be considered if clinical trials are not available or appropriate ( Tables 2, 3) (48–52). MKIs have demonstrated clinical efficacy to prolong median progression-free survival (PFS), but in most cases, no significant benefit was observed in overall survival (OS), except in the SELECT study of lenvatinib, OS was significantly improved among patients aged > 65 years compared with placebo (53). Due to these multiple target effects, molecular testing does not predict clinical responses. And the off-target side effects are common and sometimes severe. The most common treatment-related adverse events (TRAEs) include diarrhoea, fatigue, hypertension, hand-foot skin reactions et al. Most adverse effects can be managed and are reversible with discontinuation. Below, we summarize the most important results of TKIs clinical trials in advanced or dedifferentiated thyroid cancer.

Donafenib, a modified form of sorafenib with a trideuterated N-methyl group, inhibits VEGFR, PDGFR, and various Raf kinases with improved molecular stability and pharmacokinetic profile (58). In the phase II dose exploratory study of donafenib for RAIR-DTC, the 300 mg arm showed clinical benefit in terms of PFS (14.98months) and ORR (13.3%) as well as tolerable safety profile (49). Phase III clinical trial to assess donafenib vs. placebo among patients with RAIR-DTC has been completed and is expected to unveil soon.

As mentioned above, MKIs do not target specific mutations, which may compromise the safety and durability. Screening molecular abnormalities and practicing genotype-tailored agent selection may boost both anti-tumor efficacy and improve safety profile. Specifically, kinase inhibitors targeted BRAF^V600E and MEK have been studied in advanced thyroid cancer (62–66).

ALK is a kinase that activates MAPK and PI3K/AKT pathways and is associated with younger age and aggressive behavior in DTC. As reported, ATC patients with ALK rearrangement responded well to ALK inhibitor crizotinib (69, 70). But the experience of the ALK inhibitor in advanced thyroid cancer is still limited and studies are needed to evaluate the efficacy and safety profile of ALK-dependent advanced thyroid cancer.

Though rarely thyroid cancers can be driven by rearrangements of the NTRK gene, selective inhibitors of TRK kinases larotrectinib or entrectinib provide clinical efficacy in patients with thyroid cancer harboring mutations or rearrangements in the NTRK genes (71–73). In a phase I/II trial, larotrectinib proved to be highly potent with 75% ORR for tumors harboring TPK-fusions including thyroid carcinoma (71). In another phase I/II trial for patients with NTRK fusion-positive solid tumors (n=54), entrectinib resulted in a favorable outcome with an ORR of 57%, including 4 (7%) of CR and 27 (50%) of PR (73). The promising efficacy and safety profile highlight NTRK inhibitors as an optional treatment for selective advanced thyroid cancer though more investigations are warranted.

Dysregulation of the PI3K pathway has been implicated in oncogenesis and tumor progression, however, buparlisib, a pan-PI3K inhibitor, failed to show the benefit of PFS in RAIR FTC and PDTC (74). As for the inhibitors of downstream mTOR, studies showed that PI3K/mTOR/Akt-mutated dedifferentiated thyroid cancer patients appeared to benefit from mTOR inhibitors, such as everolimus, sirolimus and temsirolimus (75, 76). However, given the relatively low ORR observed, the mTOR inhibitors have not been clinically used as a single agent in the treatment of advanced thyroid cancer. Notably, given that inhibition of mTORC1 may lead to MAPK pathway activation through a PI3K-dependent feedback loop, the potential of a combined therapeutic approach with mTOR and MAPK inhibitors may be underscored. In a phase II trial, 36 metastatic RAIR-DTC received treatment with the combination of sorafenib and temsirolimus. PR was observed in 8 patients (22%), SD in 21 (58%), and PD in 1 (3%); patients who received no prior systemic treatment had a better response rate (77).

Multikinase inhibitors with RET inhibitor activity, such as cabozantinib and vandetanib, have been evaluated for tumors with activating RET gene alterations including thyroid cancer, mainly in MTC. However, due to the nonselective nature of multikinase inhibitors, the safety and durability of responses to these agents are at least partially limited by off-target toxic effects. Noval generation of high selective RET alteration inhibitors, pralsetinib and selpercatinib demonstrated promising efficacy and favorable safety profile, and have been approved by FDA for RET-mutant medullary thyroid cancer and RET fusion-positive thyroid cancer (78–80). Though RET mutations occur mainly in medullary thyroid cancers and RET fusions occur rarely in follicular-derived thyroid cancers, novel RET alteration inhibitors may also alter the landscape of RET-dependent advanced thyroid cancers.

Dual targeting of the immune system in the thyroid tumor microenvironment may, in theory, tone up the clinical benefits. Several clinical trials are ongoing to evaluate dual immunotherapy, such as PD-1 inhibitors (nivolumab) plus CTLA-4 inhibitor (ipilimumab), and PD-L1 inhibitor (durvalumab) plus CTLA-4 inhibitor (tremelimumab) ( Table 4).

Immunotherapy combined with TKIs may also augment efficacy in ATC. In the preclinical model, both VEGF-A and BRAF^V600E are positively associated with upregulation of checkpoints expression, and the combination of BRAF^V600E inhibitor and anti-PD-L1 treatment reduced tumor burden significantly more than either single agent (108–110). Another preclinical study showed that anti-PD-1/PD-L1 therapy augments lenvatinib’s efficacy by favorably altering the immune microenvironment of murine ATC (111). Clinically, in a case report, an ATC patient with BRAF and PD-L1 positivity was treated with vemurafenib and nivolumab, the patient continues to be in complete remission for 20 months after initiation of treatment (112). A combination of lenvatinib and pembrolizumab also showed promising efficacy for ATC (n=6) with CR of 66.6%, SD of 16.7%, and PD of 16.6%; the median OS was 18.5 months with three ATC patients being still alive without relapse (40, 27, and 19 months) (113). More phase II studies are currently assessing the effect of combining MKIs with immune therapy, such as pembrolizumab plus lenvatinib, nivolumab plus encorafenib/binimetinib ( Table 4).

A preliminary study evaluated RAI and anti-PD-L1 agent durvalumab in recurrent/metastatic thyroid cancer based on the hypothesis that RAI can enhance the presentation of thyroid protein immunogens and the putative neoantigens may amplify the effectiveness of ICIs. In a preliminary trial, eleven recurrent/metastatic thyroid cancer patients were treated with durvalumab and RAI (100 mCi); two patients had PR, 7 had SD, and 2 had PD (114).

Albeit disappointing outcome in a phase II study of pembrolizumab combined with chemoradiotherapy as initial treatment for anaplastic thyroid cancer, other combination strategies, such as ICIs plus SBRT and ICIs plus chemotherapy are ongoing (115) ( Table 4).