Perspective: The Molecular Landscape of Radioactive Iodine Refractory Differentiated Thyroid Cancer and Poorly Differentiated Thyroid Cancer

Introduction/Background

One third of differentiated thyroid cancer (DTC) patients with locoregional or distant metastases have disease that does not concentrate sufficient radioactive iodine (RAI) to achieve therapeutic benefit. ^1,2 The 10-year survival rate from time of detection of metastases is ∼56% for patients with RAI uptake, but only 10% for patients with RAI refractory (RAIR) DTC. ³ Achieving a definition of RAIR state is challenging.

The 2015 American Thyroid Association guidelines ⁴ classifies RAIR structurally evident DTC into four categories: (1) cancers that never took up RAI, as evident by lack of RAI uptake outside the thyroid bed at first post-therapeutic whole-body scan; (2) malignant tissue initially taking up RAI but later losing this ability; (3) malignant tissues taking up RAI in some but not all regions and referred to as primary or metastatic malignant tissue; and (4) malignant tissue with sufficient RAI uptake that does not alter disease progression. Based on ATA management guidelines it is not always clear how to differentiate between these different types of RAIR DTC, underscoring the need for awareness of molecular and clinicopathological patient characteristics that justify avoidance of inadequate or unnecessary radiation exposure or delays in providing effective treatment. ^5,6

The aforementioned challenges have spurred efforts to characterize the molecular landscape of advanced poorly DTC to gain more meaningful constitutive characterization. ^{7 –9}

Molecular Profile of DTC—The Genomic Landscape of Low- to Intermediate-Risk Papillary Thyroid Cancer: Lessons Learned from the Cancer Genome Atlas

In 2014, the Cancer Genome Atlas (TCGA) provided the first comprehensive genomic footprint of papillary thyroid cancer (PTC) based on whole exome DNA sequencing of 402 samples. ¹⁰ A low somatic mutation density was noted in PTC compared with other cancers and was thought to reflect the inherently indolent slow growing behavior of DTC. Mutually exclusive point mutations present in ∼75% of patients included BRAF, RAS, EIF1AX, TERT, and PIK3CA. Mutation in the BRAF pathway accounted for >60% of molecular alterations in PTC. RAS-family genes were altered in ∼13%. Gene fusions including RET/PTC, PPARγ/PAX8, NTRK1/3, and ALK accounted for about 15% of the genetic changes observed. Copy number alterations (CNA) occurred in 27% of tumors and were enriched in cases with no driver mutation or fusion suggesting that these CNAs may act as the primary driver alteration of PTC.

BRAF^V600E mutation was associated with Classic Variant PTC and Tall Cell Variant PTC, whereas RAS mutations were associated with Follicular Variant PTC. TCGA analysis also identified differential signaling consequences of driver mutations and categorized the cohort into BRAF-like and RAS-like molecular signaling classifications. Tumors driven by BRAF alterations are generally not sensitive to negative feedback from ERK to RAF and, therefore, have a high MAPK-signaling output. In contrast, tumors driven by RAS-like gene alterations signal through RAF dimers that respond to ERK feedback and lead to lower MAPK-signaling output. The phenotypic consequence of this differential signaling mechanism, which is driven by distinct gene alterations, is of critical significance given that BRAF^V600E mutated tumors are more likely to have reduced iodine uptake and refractoriness to RAI treatment.

However, RAS-like mutated tumors have generally preserved iodine uptake and metabolism. This genotype-to-phenotype correlation has generated a scoring system, the BRAF^V600E-RAS score, which provides a means to quantify the extent to which the gene expression profile of any tumor may resemble either the BRAF^V600E - or the RAS-like tumors and hence their phenotypic characteristics. ¹⁰ An important scoring system derived from the TCGA data was the Thyroid Differentiation Score (TDS) based on the expression levels of 16 thyroid genes. A higher score reflects a higher expression level of genes and a differentiated state, whereas a lower TDS reflects reduced expression of thyroid-related genes and a likely dedifferentiated state generally thought associated with aggressive cancers.

The Genomic Landscape of Advanced Poorly Differentiated Thyroid Cancer

Although only a few studies have used next-generation sequencing techniques to shed light on the molecular mechanisms regulating the pathogenesis of poorly differentiated thyroid cancer (PDTC), the results have provided important evidence indicating a stepwise progression of genomic changes serving to transition low-risk well-differentiated thyroid cancer (TC) to PDTC. The findings of these studies underscore how the genetic landscape of advanced PDTC differs from low-risk DTC and the importance of approaching these tumors differently.

Studies extant confirm that genetic instability, measured as tumor mutation burden, increases across the spectrum of disease, from well-differentiated TC to PDTC to anaplastic thyroid cancer (ATC) (on average 6/Tumor in ATC, 2/Tumor in PDTC vs. 1/Tumor in PTC). ¹¹ In terms of somatic driver mutations, PDTC harbors many driver mutations that impact both MAPK and PI3K-AKT pathways. While BRAF and RAS are considered major mutations, they occur less frequently in PDTC patients (BRAF in 33–40%, NRAS in 25% and HRAS in 4%) ^{12 –14} (Table 1). PDTCs with a RAS mutation exhibit a higher probability of distant metastases, while BRAF-mutated PDTCs are more likely to metastasize to local lymph nodes. ^7,11

A remarkable finding is the increased frequency of the telomerase reverse transcriptase (TERT) promoter mutation in 40–60% of patients, which is associated with older age and aggressiveness, likely to metastasize phenotypes. ^{7 –9} RAS and BRAF mutations are mutually exclusive in advanced TC, as indicated in the TCGA. However, TERT mutations co-occurred with BRAF and RAS mutations but were mutually exclusive with TP53. TERT promoter mutation has been associated with aggressive TC and general refractoriness to RAI treatment. ^{15 –17} Coexistence of a TERT promoter mutation with either BRAF or RAS mutation appears to have a synergistic negative effect resulting in loss of RAI avidity and poor clinical outcome. ^{18 –21} As a consequence, identification of a TERT/BRAF combination is clinically important and signals need to initiate early aggressive treatment and a close follow-up regimen.

Mutational analysis in indeterminate nodules may provide the initial determination of malignancy as well as guide the intensity of postoperative management. One such unique finding is the high prevalence of mutations in the eukaryotic translation initiation factor EIF1AX in non-ATC, occurring in 11–12% of patients, with a strong association with RAS mutations. The EIF1AX mutation has been associated with larger tumors and shorter survival in PDTCs. ⁷ More recently, coexistence of EIF1AX mutation with other driver mutations, including RAS, TERT, and TP53, has been shown to confer 100% risk of malignancy in cytologically indeterminate thyroid nodules. ²² There are two main EIF1AX mutational hotspots with about half located in exon 2 and the other half at the intron 5/exon 6 splice site. ²³

Additional mutations with higher frequencies compared with the TCGA cohort include TP53 (9%), ataxia-telangiectasia mutated (ATM) (7%) and RBI (1.8%). The TP53 gene regulates the tumor suppressor protein p53. Mutations that inactivate TP53 are considered a hallmark of ATC, which has a cumulative mutation rate of 59–65% for the TP53 gene. In advanced PDTC, mutation in TP53 occurs less commonly, at a frequency of 3–10%. ^11,14,24

Mutations identified in an advanced TC cohort associated with altered signaling pathways included Mediator of RNA polymerase II transcription subunit 12 homolog (MED12), and the RNA Binding Motif Protein 10 (RBM10), with both mutations occurring at higher frequencies than in PTC/TCGA (14% vs. 0% and 11% vs. 0.5%, respectively). Mutation in MED12 has been associated with resistance to multiple cancer therapies, by activation of the tumor growth factor-beta receptor, which leads to acquired resistance to MEK and BRAF inhibitors. ^{25 –27}

Additional mutations that occur at higher frequencies in PDTC compared with PTCs include ATM (7%) and RB1 (1.8%). ATM encodes a nuclear protein that is important for influencing events related to cell division and DNA repair. Biallelic mutations have been implicated in the condition, ataxia-telangiectasia, an autosomal recessive disorder characterized by cerebellar degeneration, telangiectasias, and immunodeficiency. Monoallelic mutation leads to increased risk of developing several malignancies, including breast, stomach, bladder, pancreas, thyroid, and lung cancers. ^{28 –30}

Epigenetic alterations also have been found at higher frequencies in advanced PDTC patients. These included ARID1B (9% vs. 1%) and genes encoding components of the switch/sucrose non-fermentable (SWI/SNF) chromatin remodeling complex (16%). Emerging data from preclinical studies and limited clinical trials suggest that alteration in SWI/SNF genes may mediate resistance to redifferentiation and might serve as biomarkers to identify patients who would not benefit from redifferentiation strategies using MAPK pathway inhibition. ^31,32

Other epigenetic alterations, including ARID2, KDM6A, SMARCB1, and ARID2, were found at lower frequencies (7%, 4%, and 1.8%, respectively). ⁸

Discussion and Future Perspectives

Knowledge of the molecular alterations in advanced TC has enabled a more complete understanding of the pathogenesis of TC, and the specific pathways involved in disease initiation and progression. Molecular changes that drive tumorigenesis can be linked to specific and predictive phenotypes of disease presentation. A crucial consequence of determining the molecular pathways and mutations influencing the genomic landscape of TC is the consideration of application of new targeted treatment strategies to specific tumors based on their underlying pathways and driver mutations.

Sorafenib and Lenvatinib target angiogenic pathways in TC and are currently standard therapies for progressive metastatic RAIR DTC. Although they have resulted in the prolongation of progression-free survival, their use is associated with major adverse events. ^33,34 Second-generation compounds have been developed that target specific driver mutations, such as BRAF^V600E , RET, ALK, and MEK. Dabrafenib and the MEK inhibitor trametinib are FDA-approved for treatment of BRAF^V600E mutated ATC and DTC. ³⁵ Recognition of the underlying oncogenic driver pathways has been critical in redifferentiation efforts to restore RAI uptake in RAIR DTC.

BRAF^V600E targeting drugs (±MEK inhibitors) have been successfully used in BRAF^V600E mutated TCs to significantly increase RAI uptake in advanced RAIR DTCs. ^36,37 Such redifferentiation strategies provide opportunity for potentially effective redeployment of radioiodine that would have been previously ineffective treatment for RAIR DTC. Mutations in the SWI/SNF family of chromatin remodeling genes as well as ERBB3 overexpression have been found to confer resistance to redifferentiation therapies using BRAF/MEK inhibitors in RAIR DTC. As a consequence of the likelihood of such resistance, these patients should be treated as having advanced TC with a more guarded prognosis that warrants earlier triage to alternative therapeutic protocols. ^31,32

Despite remarkable advances made in deciphering the molecular landscape characterizing the spectrum of TC, the identification and definition of RAI refractoriness remains a clinical challenge. A reliable strategy is needed that would accurately identify RAI refractoriness at the onset of treatment planning, before ¹³¹I therapy is given, to prevent either inadequate or unnecessary radiation exposure and to avoid any delay initiating potentially more effective treatment.

Redifferentiation trials have shown great promise for re-establishing iodine uptake in RAIR DTCs. However, larger and systematic trials are needed to help identify optimum dosing, length of treatment, and timing of RAI administration. Modifications in trials will likely be based on having achieved better understanding of the effect of mutations in the SWI/SNF family of chromatin remodeling genes on redifferentiation among different age groups, sexes, and the effect of other coexistent driver mutations. Awareness of whether other gene alterations in the same pathway as SWI/SNF may alter response to redifferentiation efforts may also need to be considered.

Advances in mapping genetic alterations in advanced thyroid cancer have improved our understanding of the underlying molecular mechanisms associated with tumorigenesis, disease progression, and response to treatment. Future initiatives should integrate available and newly developing genome data to design systematic studies that better define, characterize, and predict RAI refractoriness to derive focused and effective targeted therapeutic approaches and improved outcomes.