Challenges and Strategies to Combat Resistance Mechanisms in Thyroid Cancer Therapeutics

Introduction

Advances in thyroid cancer oncogenesis have identified a number of driver alterations that are potentially targetable, including BRAF^V600E and N/H/K RAS mutations and oncogenic kinase fusions involving neurotrophin tyrosine receptor kinase (NTRK), RET, anaplastic lymphoma kinase (ALK) and ROS1. ¹ Activating driver alterations in thyroid tumors leading to oncogene addiction are almost always mutually exclusive. Clinical trials investigating tyrosine kinase inhibitors (TKIs) in advanced thyroid cancer over the past decade have shown high objective response rates (ORRs) and long progression-free survival (PFS). Nonetheless, tumor progression ultimately may occur through acquired resistance. ^{2 –5} These mechanisms of acquired resistance to TKI therapy in thyroid cancer are not fully understood but have largely been grouped into distinct categories; mutations in genes targeted specifically by TKIs and activation of bypass pathways (Table 1). ⁶

Multikinase inhibitors (MKIs) have become established therapies in several advanced thyroid cancer settings. Vandetanib and cabozantinib have demonstrated clinical activity in randomized trials in metastatic medullary thyroid cancer (MTC), leading to health authority approvals. ^4,5 Sorafenib, lenvatinib, and cabozantinib have similarly shown activity and garnered approvals in radioiodine (RAI)-refractory progressive differentiated thyroid cancer. ^2,3,7 It is thought that the activity of these therapies is largely due to vascular endothelial growth factor receptor (VEGFR) inhibition, rather than inhibition of the primary oncogenic drivers. As a class, these MKIs potently inhibit VEGFRs and other kinases. In the case of lenvatinib, other kinases targeted include RET, KIT and additional angiogenic factors, such as fibroblast growth factor receptor (FGFR) and platelet-derived growth factor receptor (PDGFR). ² While responses to VEGFR inhibition with MKIs in advanced thyroid cancers are often robust, most patients eventually experience disease progression.

Acquired resistance mechanisms responsible for disease progression on VEGFR MKI therapy are not well understood, especially in thyroid cancer. Compensatory strategies resulting in escape of tumor growth inhibition are likely multifactorial and may include upregulation of angiogenesis signaling pathways, emergence of gatekeeper mutations blocking access of the small-molecule inhibitor into the kinase adenosine triphosphate (ATP) binding pockets, and upregulation of other oncogenic pathways such as EGFR signaling. ^{8 –10} Most MKIs in clinical use are type 1 kinase inhibitors, meaning that they are ATP competitive inhibitors and occupy the active conformation of the kinase binding pocket (Fig. 1). Resistance mechanisms are those that impair the binding in the active conformation. The size of the side chain alongside the gatekeeper residue can compromise binding, a well-known example is when vandetanib is inadequate for RET V804L/M alterations. ^11,12

OPEN IN VIEWER

FIG. 1.

Representation of various forms of resistance in thyroid cancer. Steric inhibition from larger amino acid side chains prevents multikinase inhibitors (TKI; e.g., vandetanib) from successfully entering the ATP binding pocket in V804L/M (leucine/methionine) RET variants. First-generation RET inhibitors effectively block ATP binding irrespective of the larger amino acid substitutions. Solvent front mutations hinder first-generation RET inhibitor binding where second generation will overcome this challenge. ATP, adenosine triphosphate; TKI, tyrosine kinase inhibitor.

In 2018, the pan-TRK inhibitor, larotrectinib, and combination BRAF/MEK inhibitors, dabrafenib/trametinib, were the first gene selective treatments obtaining health authority approvals for patients with thyroid cancers driven by specific gene alterations. Entrectinib was approved in 2019 for solid tumors driven by NTRK gene fusions and non-small cell lung cancers (NSCLCs) harboring ROS1 fusions. Selpercatinib and pralsetinib were approved, respectively, in 2020 and 2021 for advanced thyroid cancers driven by RET alterations and RET fusion-positive NSCLC.

These new selective inhibitors have demonstrated remarkable antitumor activity, yet cases of acquired resistance emerged. ^13,14 To date, two main mechanisms of resistance to gene specific therapy have been described. The first mechanism is acquired on-target gene mutation impeding drug activity. The second involves upregulation of bypass kinase signaling pathways leading to tumor progression. Specific mechanisms of acquired resistance have been identified in only a subset of patients who have initially responded to and then progressed on RET-specific therapies, and thus, further work is needed to characterize these refractory cohorts by clinical and molecular signature.

NTRK Inhibitors

NTRK 1, 2, and 3 encode for TRK A, B and C, respectively. In development, TRK kinases play a role in neuronal differentiation and survival. ¹⁵ Oncogenic fusions involving NTRK 1–3 are not only seen in nearly all infantile fibrosarcomas and salivary gland secretory carcinomas but also occur at a low frequency across multiple solid tumors. NTRK 1 or 3 fusions drive ∼2–5% of differentiated thyroid carcinomas (DTCs) and can also be found rarely in anaplastic thyroid carcinoma (ATC). As is the case with other oncogenic kinase fusions, NTRK fusions are seen more frequently in pediatric and young adult thyroid cancers. ¹⁶

Pooled analysis of the phase I/II trials of the selective TRK inhibitor, larotrectinib, in TRK fusion-positive solid tumors demonstrated a tumor agnostic ORR of 79%. ¹⁷ Responses were durable, with a median PFS of 25.8 months. Moreover, larotrectinib was well tolerated, with grade 3 or higher treatment-related adverse events uncommon. ¹⁷ Of note, thyroid cancers and sarcomas were the two most common adult histologies enrolled in the larotrectinib studies. Focused analysis of thyroid cancer patients included 21 DTCs and 7 ATCs. In DTC, an ORR of 86% was observed. In ATCs, the ORR was 29%. ^18,19 Fewer thyroid cancer patients were enrolled in the entrectinib trials. Of five DTC patients treated, one experienced an objective response. ²⁰

Despite rapid and durable disease control in these pooled analyses, on-target mechanisms of resistance have been detected in NTRK fusion-positive cancers treated with both larotrectinib and entrectinib. TRKA G595R, TRKB G639R, and TRKC G623R mutations in the kinase solvent front domain have been found. ²¹ Less frequently, mutations in the xDFG kinase activation loop and gatekeeper domain mutations have also emerged. ^22,23 The mutations described above all lead to conformational change within the TRK kinase domain, interfering with drug binding and activity. Interestingly, these mutations are paralogous with solvent front, xDFG motif, and gatekeeper mutations seen in ALK and ROS1 fusion-positive NSCLCs treated with selective inhibitors.

Off-target bypass mechanisms of resistance involve alterations in non-NTRK oncogenes. Observed mechanisms include the acquisition of activating BRAF^V600E and KRAS G12D mutations, and MET amplification, conferring escape via MAPK pathway activation. ²² Particularly challenging is the acquisition of more than one resistance alteration identified in a subset of cases.

Therapeutic strategies for these resistant clones include cabozantinib, which has successfully overcome solvent front mutations following entrectinib resistance and second-generation TRK inhibitors repotrectinib and selitrectinib, which were designed to overcome resistance mutations acquired on first-generation inhibitor therapy. ^{24 –27} Both are compact small-molecule inhibitors that inhibit wild-type TRK kinases and overcome the steric hindrance created by solvent front mutations. Repotrectinib, an ROS1 and TRK inhibitor, was 10-fold more potent than selitrectinib in vitro and demonstrated marked antitumor effect in xenograft models. ^24,25

Early phase clinical trial data are available. TRIDENT-1 (NCT03093116) is a phase II trial investigating repotrectinib in kinase inhibitor pretreated and treatment-naive ROS1 and NTRK fusion-positive cancers. ²⁸ Selitrectinib was also evaluated in early phase study (NCT03215511). ²⁹ Early data from these next-generation TRK inhibitors, especially in NTRK fusion-positive cancers pretreated with kinase inhibitors, are encouraging. Other new agents, such as SIM1803-1A (NCT04671849) and PBI-200 (NCT04901806), are also in development.

BRAF

BRAF^V600E is the most frequent molecular alteration in thyroid cancer. Approximately 60% of all papillary thyroid carcinomas (PTCs) harbor BRAF^V600E mutations, whereas BRAF^V600E mutations are seen in ∼30% of ATCs. ^1,30 In advanced PTC, BRAF^V600E has long been a desired target for treatment but, thus far, phase II studies of BRAF-directed therapy have not yielded enough clinical activity to establish this approach as a new first-line standard of care. ^{31 –33} Combination dabrafenib and trametinib were not superior to dabrafenib monotherapy in PTC, and overall response rates for both groups were modest at 30% and 35%, respectively. ³²

This is not the case in ATC. Updated results from a basket trial of dabrafenib plus trametinib in non-melanoma solid tumors harboring BRAF^V600E mutations enrolling a 36-patient ATC cohort have now been reported. ³⁴ The confirmed ORR was 56%, with a median PFS and overall survival of 6.7 and 14.5 months, respectively. This activity in a larger number of patients confirmed the promising activity seen in the first 16 ATC patients enrolled in the study, which led to health authority approvals for the use of dabrafenib plus trametinib in BRAF^V600E mutation-positive ATC. ³⁵

In vitro studies have yielded insights into both intrinsic and acquired resistance to BRAF inhibition in thyroid cancer models. ^{36 –38} Induction of RAS activity by loss of feedback inhibition can lead to MAPK pathway activation despite BRAF suppression in cell lines demonstrating primary resistance. ³⁷ Similar reactivation of the MAPK pathway by the induction of HER3 expression in BRAF mutant thyroid cell lines emerged on vemurafenib treatment. ³⁶ In an ATC model, intrinsic resistance to BRAF inhibition was mediated by the PI3K/AKT pathway activation resulting from c-Met overexpression. ³⁸ Dual inhibition with c-MET and BRAF inhibitors overcame the resistance and led to a sustained treatment response. ³⁸

Clinical cases of acquired resistance emerging in patients with BRAF^V600E mutation-positive PTC treated with BRAF-directed therapy have now been reported. ^{39 –42} Thus far, activating mutations in K and N RAS have been identified on tumor biopsy and/or circulating tumor DNA (ctDNA) analysis in several treated patients. ⁴³ This was confirmed in larger scale analysis by Lee et al. who also outlined post-BRAF inhibitor resistance to therapy through dedifferentiation. ³⁹

RET

Other than BRAF^V600E , RET/PTC rearrangements are the most common activating genetic alterations in the MAPK pathway in thyroid cancer, ⁴⁴ and these are overrepresented in pediatric thyroid cancer where fusion genes dominate and can confer more aggressive disease. ⁴⁵ The targeted RET inhibitor, selpercatinib, has shown durable responses in thyroid cancer harboring RET alterations with treatment-naive MTC heralding ORRs of 73%, pretreated MTC ORR of 69%, and RET fusion PTC ORR of 64%. ¹³ To date, the median PFS rates are not yet defined despite fairly long median follow-up times.

Nonetheless, as with other targeted therapies, resistance on RET-specific therapy can emerge and lead to disease progression. Pralsetinib, another potent and effective selective RET inhibitor, has also demonstrated effectiveness in thyroid cancer with ORR of 71% in treatment-naive MTC patients, 60% in pretreated and 89% in RET fusion PTC patients. ⁴⁶ These first-generation selective inhibitors were designed to bind the RET kinase domain in a manner whereby they avoid steric clash with the gatekeeper V804L/M mutations. ¹¹ However, the effectiveness of these inhibitors can be severely impacted by the development of acquired resistance, which may be off- and on-target reactivation.

Determinants of resistance were explored by Rosen et al. who found that the pretreatment genomic landscape did not predict the response to selpercatinib except rarely in RAS-mediated primary resistance. ⁶ Secondary RET mutations arising from pretreatment of MKIs include known drivers V804M, G601E, K666E, and D898Y and others that can arise within the RET ATP-binding site, including G810C/S/R (described below) as well as Y806C/N and V738A. ^47,48 The former drivers appear to remain sensitive to selpercatinib, but the structural alterations whereby substitution of the solvent front G810 with bulky side-chained residues such as cysteine, serine, or arginine sterically interfere with the binding of both pralsetinib and selpercatinib. ⁶

This mechanism of action involves key van der Waals interactions, which are disrupted by Y806 and V738 mutations, in the case of selpercatinib; while hydrophobic interactions with pralsetinib are disrupted by mutation of L730 or V738. ¹⁴ Accordingly, preclinical characterization in animal models identified L730V/I mutations as strongly resistant to pralsetinib but not to selpercatinib. ⁴⁹ A possible future treatment option may be the MKI TPX-0046, which has been demonstrated preclinically to inhibit RET solvent front mutations, although it lacks inhibitory activity against gatekeeper mutations ⁵⁰ and is currently undergoing early clinical investigation in the phase 1 SWORD-1 trial (ClinicalTrials.gov Identifier: NCT04161391).

However, there are now emerging reports of acquired resistance by activation of target-bypass pathways, such as NTRK3 fusion, ⁵¹ and MET or KRAS amplification. ⁵² Lin et al. performed biopsies post-RET inhibition in RET fusion-positive NSCLC to characterize resistance mechanisms. They identified solvent front mutations and separately MET and KRAS amplification following treatment as the mechanisms for resistance. ⁴⁸ Following this, novel single patient protocols were developed to develop a proof-of-concept study on NSCLC patients with secondary MET amplification following RET inhibition for their RET fusion-positive cancers. They describe some clinical and mixed radiological improvement when adding the MET inhibitor crizotinib to selpercatinib. ⁵² Combination therapy yields concern about additive toxicities, but in that small cohort of four patients following a slow dose escalation, this was successfully managed. Ultimately, MAPK activation is central to acquired and primary resistance following RET inhibition, and a combination cocktail of more specific inhibitors rather than sequential therapy may yield better results. ⁶

Selpercatinib can overcome both hereditary and acquired RET V804M gatekeeper mutation, and the description by Solomon et al. of the G810 solvent front mutations demonstrates for the first time acquired “on-target” resistance to selective RET inhibition in patients. ⁴⁷ Novel compounds with different structures encompassing a rigid macrocyclic structure that demonstrates in vitro and in vivo inhibitory activity against RET and including solvent front-mediated resistance. ⁵⁰ Pralsetinib is also well tolerated and demonstrates durable efficacy in malignancies harboring RET alterations including thyroid. ⁴⁶

Both pralsetinib and selpercatinib mutations at both the solvent front mutations and hinge have been identified. As discussed above, while their novel binding technique inhibits the gatekeeper mutations, they were still subject to other resistance. ¹¹ Here, non-gatekeeper mutations in the solvent front, hinge, and (beta)2 shared resistance to both drugs. In vitro studies confirmed that there was complete cross-inhibition of both drugs demonstrating similar mechanisms of resistance development. ¹¹ Unfortunately, this implies utilizing the alternative following clinical progression would be ineffective. ⁵³ Fortunately, the next generation of RET-specific inhibitors designed to overcome on-target acquired resistance are already in first-in-human phase I study (NCT04683250; NCT05241834).

The concept of resistance in RET mutant thyroid cancers is very familiar since the trials with the MKI vandetanib demonstrated resistance to patients harboring the 804M mutation ¹² and then L881V aberration. ⁵⁴ The significance of mutations in the 804 codon has recently become questionable with the history suggesting based on penetrance carriers there is a high lifetime risk for MTC development, but latest data suggest that the penetrance is closer to 5%. ⁵⁵ Irrespective there are still patients with metastatic disease who have now been able to respond to selpercatinib therapy where vandetanib utility was previously not beneficial.

In lung cancer, various mechanisms for vandetanib resistance have been discussed including new RET mutants yielding increased ATP affinity and autophosphorylation, ⁵⁶ and in general, mechanisms for progression and resistance with vandetanib may involve Yes-Associated Protein (YAP), which in in vitro models when inhibited overcomes vandetanib resistance. ⁵⁷ Cross-inhibition between MKIs and selective TKIs will continue to be important as the upfront identification of resistance mechanisms will lead to the direct use of specific inhibitors. For example, nintedanib, an MKI similar to vandetanib, is effective in inhibiting the vandetanib-resistant G810A mutant as well as the L881V mutant found in familial MTC. ⁵⁴

ALK Fusions

Chromosomal translocations generating gene fusions involving the ALK gene such as EML4-ALK are a rare occurrence in thyroid cancer (0.8% and 4% of PTC and poorly differentiated thyroid carcinoma, respectively). ^1,58 Given the success of ALK-TKIs in NSCLC, ⁵⁹ ALK fusions represent an attractive therapeutic target for this subgroup of thyroid cancer patients. Case reports have shown whole-genome sequencing in aggressive PTCs identifying ALK as a possible therapeutic target ⁶⁰ and others extended this with successful crizotinib treatment in thyroid cancers harboring ALK rearrangements. ^61,62 Similar to other oncogenic drivers, ultimate resistance is driven by various mechanisms including ALK secondary mutations and copy number gain and off-target mechanisms where there is amplification of bypass tracks (EGFR, KRAS, MET). ⁶³

To date, two case studies report the real-world experience of ALK-TKI use in a patient with EML4-ALK-positive metastatic RAI-refractory thyroid cancer. ^60,64 The first study reported the off-label administration of first-generation ALK-TKI, crizotinib, to a patient with RAI-refractory EML4-ALK-positive metastatic tall cell variant PTC, which demonstrated intrinsic resistance to the inhibitor (stable appearance on computed tomography [CT] of cervical, mediastinal, and lung metastases after 6 months of treatment). ⁶⁰ The second study reported administration of successive ALK-TKIs to a patient with EML4-ALK-positive RAI-refractory follicular variant PTC. The patient initially experienced a clinical response on crizotinib, but disease progression became evident following 8 months of treatment.

Initiation of lenvatinib failed to control metastatic spread, before the patient was switched the third-generation ALK-TKI, lorlatinib, on which partial response was ongoing at the time of publication 7 months post-commencement of treatment (61% decrease from baseline in target lesions on CT). ⁶⁴ Lorlatinib has the broadest coverage of ALK resistance mutations and can cross the blood–brain barrier. Future trials can be informed by the considerable experience, which has been garnered from over a decade of ALK-TKI use in the treatment of NSCLC. Despite the efficacy of ALK inhibitors in NSCLC, almost all lung cancers harboring ALK gene rearrangements develop acquired resistance to crizotinib within 1–2 years.

A number of mechanisms underlying acquired resistance to crizotinib have been reported and include secondary ALK mutations such as L1196M (“gatekeeper” mutation), F1174L, C1156Y, G1202R, S1206Y, and G1269A; as well as off-target mechanisms such as EGFR activation by upregulated amphiregulin or TGF-a, ^{65 –68} activation of HER family proteins (EGFR, HER2, and HER3) by upregulated EGF or NRG1, ^69,70 P2Y purinergic receptor family activation; ⁶⁹ KIT ^71,72 and MET ^{73 –75} gene amplifications; and KRAS mutations. ⁷⁶ Results from a recent study also suggest that high mutational burden and mutations afflicting DNA repair genes such as TP53 in ALK-positive NSCLC confer rapid acquired resistance to crizotinib. ⁷⁷

Identification of Resistance and Causal Driver Alterations

Acquired resistance occurs when patients have initially responded to the TKI, but tumor progression occurs as per RECIST v1.1 criteria. Identifying resistance mutations early is crucial for clinicians' preparation for second-line therapy to minimize the extent of cancer progression. However, the current means of obtaining tumoral DNA for genetic testing—biopsy by fine needle aspiration (FNA), is painful and expensive. Furthermore, FNA is increasingly inadequate for the complex genomic analyses tracking evolving resistance mechanisms described above. Accessibility to genomic profiling is mixed, and many patients are not able to have comprehensive molecular profiling initially, let alone at multiple time points along treatment.

ctDNA is a novel tool exploiting the release of DNA into the circulation by cancers, which can be detected by highly sensitive assays for cancer-related genetic markers: the so-called “liquid biopsies.” Liquid biopsies of ctDNA are simple blood samples and are relatively easy to perform and endow clinicians with the capacity for frequent sampling necessary to track tumor evolution (Fig. 2). Liquid biopsies have already been proven a powerful clinical tool for certain cancers, most notably NSCLC, where EGFR mutation testing can be used by the clinician to guide a patient to therapy. They have been effective in detecting new resistance mutations in colorectal cancers, which when picked up earlier can lead to initiation of alternative therapies. ^{78 –80} Thus, renewed interest in the role of liquid biopsies in detecting common somatic mutations in thyroid cancer has emerged.

OPEN IN VIEWER

FIG. 2.

Figurative description for options of re-biopsy following development of acquired resistance. NGS, next-generation sequencing; PCR, polymerase chain reaction; qPCR, quantitative polymerase chain reaction; WT, wild type.

BRAF^V600E mutations were identified through ctDNA in 42.1% (24/57) of patients with PTC and known tumoral mutation using digital droplet polymerase chain reaction, with a higher prevalence in more high-risk disease (12/16). ⁸¹ In identifying RET ctDNA, an analysis of 34 patients post-selpercatinib, which initially included 13 mutations and 21 fusions, showed a 50% reduction in allelic frequency in 79% of sampled patients. ⁸² The presence of preoperative ctDNA in MTC has also predicted disease outcomes. ⁸³ Thus, monitoring of these frequent known mutations is feasible and ongoing (reviewed in Wijewardene et al. ⁸⁴ ). While ctDNA studies have shown moderate success, ultimately if ctDNA sensitivity is too low, tissue re-biopsy of the progressing lesion can be performed.

Conclusions

Interrogating the molecular profile of advanced thyroid carcinomas has become commonplace in many international tertiary centers with the goal to identify a druggable target. PFS is curtailed by developing acquired resistance. To minimize this therapeutic liability, clinicians must be anticipatory in identifying the drivers and characterizing mechanisms of on-target resistance.

The well-tolerated selective RET kinase inhibitors have been a highlight in the history of thyroid cancer TKIs. Despite their novel binding approach, concealing them from gatekeeper mutations, several mechanisms of resistance have emerged mostly involving gene mutations and bypass tracts centered around the activation of the MAPK pathway. ALK and NTRK inhibitors have already reached several generations of development to broaden the scope of inhibition and combat resistance. Re-biopsy (either tissue or liquid) can characterize the molecular signature of each resistant lesion, thus should be performed where possible.

However, we recognize that multiple resistance mechanisms can emerge in a single patient complicating the strategy for targeted therapies. Clear evidence of the benefit for characterizing the drivers and mechanisms of resistance is seen in small clinical cases series, but ideally this would become commonplace for all patients with established pathways for alternative therapies. Molecular evolution of these tumors post-TKI treatment is a challenge, but understanding these mechanisms can allow clinicians to pivot strategically to provide a high quality of care.