RET-Altered Cancers—A Tumor-Agnostic Review of Biology, Diagnosis and Targeted Therapy Activity

Simple Summary

Changes in the RET gene (like mutations or fusions) are often found in lung and thyroid cancers but are also found in other cancer types. New drugs called “selective RET inhibitors”, like selpercatinib and pralsetinib, are effective in treating tumors with RET gene changes. These drugs have been tested in “basket trials” that treat patients based on gene changes in their cancer instead of cancer type. In this review, we discuss how RET gene changes cause cancer, which cancers have these changes, what tests to use, and how well targeted therapies work.

Abstract

RET alterations, such as fusions or mutations, drive the growth of multiple tumor types. These alterations are found in canonical (lung and thyroid) and non-canonical (e.g., gastrointestinal, breast, gynecological, genitourinary, histiocytic) cancers. RET alterations are best identified via comprehensive next-generation sequencing, preferably with DNA and RNA interrogation for fusions. Targeted therapies for RET-dependent cancers have evolved from older multikinase inhibitors to selective inhibitors of RET such as selpercatinib and pralsetinib. Prospective basket trials and retrospective reports have demonstrated the activity of these drugs in a wide variety of RET-altered cancers, notably those with RET fusions. This paved the way for the first tumor-agnostic selective RET inhibitor US FDA approval in 2022. Acquired resistance to RET kinase inhibitors can take the form of acquired resistance mutations (e.g., RET G810X) or bypass alterations.

1. Introduction

Advances in molecular profiling and inclusive trial designs have transformed the therapeutic landscape for patients with cancer. Basket trials now enroll patients based primarily on the presence of oncogenic drivers, such as RET (rearranged during transfection) alterations, regardless of tumor type. The landmark 2022 US FDA regulatory approval of selpercatinib, a RET selective tyrosine kinase inhibitor (TKI), represented the seventh tissue-agnostic indication in oncology [1], following tumor site-specific approvals of selpercatinib for non-small cell lung cancer (NSCLC) and thyroid cancer. RET alterations encompass both RET fusions and RET point mutations. Although NSCLC, papillary thyroid cancer (PTC), and medullary thyroid cancer (MTC) collectively account for the majority of RET-altered cancers, a plurality of additional tumor histologies also harbor RET alterations. In this review, we discuss the biology of RET alterations and strategies to target these oncogenic drivers, including contemporaneous tumor agnostic approaches based on registrational data from the selpercatinib and pralsetinib trials.

2. RET Biology

The RET proto-oncogene is a receptor tyrosine kinase (RTK) which functions as a transmembrane glycoprotein. The RET gene resides on the long arm of chromosome 10 and contains three functional domains: an extracellular ligand-binding domain, a transmembrane domain, and an intracellular tyrosine kinase domain [2]. RET is involved in normal embryonic development of the genitourinary, neuroendocrine, and sympathetic nervous systems [3,4]. Inactivating germline RET mutations and Ret knock-out mice models have yielded valuable insights into the consequences of reduced or absent RET activity throughout development and adulthood. Inactivating RET mutations have been identified in approximately 50% of familial cases of Hirschsprung disease, a condition characterized by the absence of enteric ganglion cells in the large intestine due to failure of neural crest cells to migrate completely during embryological development [5,6]. RET mutations are also associated with congenital abnormalities of the kidney and urinary tract (CAKUT) and congenital central hypoventilation syndrome (CCHS) [7,8]. In such disease entities, genotype–phenotype correlation remains poor, with variable penetrance and a wide range of RET mutations across the entire length of the gene [5]. Gain-of-function alterations are also observed; these are most commonly associated with oncogenesis and will be discussed in greater detail later.

Unlike many other RTK ligands, RET ligands such as artemin (ARTN), glial cell line-derived neurotrophic factor (GDNF), neurturin (NRTN), and persephin (PSPN) exhibit a distinctive binding mechanism [9,10,11]. These GDNF family ligands (GFLs) bind co-receptors known as GDNF family receptors (GFRs) [12,13]. Concurrent with binding of calcium ions to RET’s cadherin-like domain, the GFL–GFR complex induces recruitment of RET through its extracellular domain, thus bringing two RET monomers in close proximity. Receptor homodimerization then promotes cross-phosphorylation of key cytoplasmic tyrosine residues within RET’s intracellular domains. These domains subsequently recruit specific adaptor proteins which activate downstream signaling cascades involved in cellular proliferation and survival, such as the MAPK, PI3K, JAK-STAT, and PKA pathways (Figure 1) [14,15,16,17]. GFL and GFR family members have distinct tissue-specific expression patterns which play a crucial role in determining their biological functions.

3. Oncogenic RET Alterations

Oncogenic activation of RET can occur via two mechanisms: (1) germline or sporadic mutations activating RET’s kinase domain or (2) chromosomal rearrangements between RET and a gene partner typically containing a dimerization domain [18,19]. Both alterations can trigger oncogenic RET ligand-independent phosphorylation leading to constitutive activation of downstream transduction cascades which then results in morphological transformation and tumor growth. RET’s transforming potential was first described in 1985 using NIH/3T3 mouse fibroblastic cell lines transfected with rearranged lymphoma DNA containing a RET coding sequence [15]. Later, RET was one of the first oncogenes identified in solid tumors, more specifically in PTCs with RET fusions [20,21].

3.1. Activating RET Mutations: MTC

Germline or somatic activating RET mutations are enriched in patients with MTC. Pathognomonic non-synonymous point mutations in RET’s kinase or extracellular domain occur in this population. Mutations in the extracellular domain, such as those involving codon C634 [22], can lead to ligand-independent dimerization. In contrast, mutations in the kinase domain, such as RET M918T and the V804 gatekeeper mutations, can trigger monomeric activation of RET [23,24]. Inherited RET activating mutations are associated with MTC, either alone (familial medullary thyroid cancer [FMTC]) or as part of the multiple endocrine neoplasia type 2A (MEN 2A) or MEN 2B syndromes [23]. All MEN 2 syndromes are associated with an increased risk of MTC, pheochromocytoma, and neuroendocrine malignancies (including pulmonary or extra-pulmonary carcinoid tumors). Whereas patients with MEN 2A face an increased incidence of parathyroid hyperplasia and adenoma, patients with MEN 2B can develop intestinal ganglioneuromas or mucosal neuromas. FMTC, MEN 2A, and MEN 2B are all associated with a high penetrance of MTC in gene carriers and a 70–100% risk of developing MTC by age 70 [23]. Of note, approximately 2% of MEN 2A and 50% of MEN 2B cases arise from de novo germline mutations [25]. In contrast, somatic activating RET mutations are seen in 40–70% of sporadic MTCs, correlate with aggressive phenotype, and are most commonly associated with RET M918T mutations [26,27,28].

3.2. RET Rearrangements: Thyroid Cancer

As opposed to RET activating mutations, RET rearrangements are only known to be acquired somatically. Chromosomal rearrangements involve the intact 3′ tyrosine kinase domain of RET with various heterologous gene loci which provide dimerization domains for the constitutive expression of RET [29,30]. Fusion partners contribute different types of motifs to the fusion protein, including coiled-coil, LIS1 homology (LisH), tryptophan–aspartate repeat (WDR), and sterile alpha motif (SAM) domains, all of which mediate dimerization and ligand-independent signaling. Most RET rearrangements are intrachromosomal events, either paracentric or pericentric, involving upstream partners on chromosome 10 [29,31]. These include CCDC6 or NCOA4 [31], the most common gene partners in RET-rearranged PTC, as well as KIF5B, which is commonly involved in NSCLC [29]. Additional intrachromosomal gene partners include ACBD5, ANKRD6, FRMD4A, or KIAA1217 [32,33,34,35,36]. Interchromosomal fusions or translocations are less frequent, but have been described with many partners, including RELCH, ERC1, and TRIM33 [32,37,38].

RET rearrangements occur as a consequence of stochastic DNA double-strand breaks (DSBs) and non-homologous end joining. DSBs can result from environmental factors, such as ionizing radiation exposure, or endogenous mechanisms, including DNA replication stress or topological stress from topoisomerase activity [39,40,41]. Breakpoints typically occur at selected regions of the RET gene, often involving intron 11, which gives rise to a fusion protein containing only the cytosolic domain of RET [42]. Non-random recurrence of RET fusion partners is predisposed by the contiguity of specific loci from coiled chromosomes in the nucleus of normal cells. Spatial proximity of fusion partners involving linearly distant genes of the same chromosome or juxtaposed regions from two different chromosomes facilitates simultaneous double-end breaks and end joining [43,44].

In unselected PTC populations, RET fusions can be identified in 10–30% of patients [31,45,46,47,48]. RET fusions are typically mutually exclusive with additional oncogenic rearrangements such as BRAF alterations [49]. RET rearrangements represent a hallmark of ionizing radiation exposure, whether it be therapeutic or accidental, and were initially described in survivors of the Chernobyl nuclear disaster [50,51]. In such populations, the prevalence of RET rearrangements approximates 40–70% and is often seen in pediatric patients, likely as a consequence of increased follicular cell proliferation in this age group [50,51,52]. Of note, additional thyroid cancer subtypes can also be associated with RET fusions although at a lower prevalence, including poorly differentiated thyroid carcinoma (5%) and anaplastic thyroid cancer (1%) [48].

3.3. RET Rearrangements: NSCLC

The first case of RET rearrangement in a lung cancer patient was reported in 2012 by Ju et al. in a 33-year-old never-smoker with lung adenocarcinoma without known driver alterations in EGFR, KRAS, and ALK. The authors reported a novel fusion between RET and KIF5B which was identified in two additional patients in a replication study, thus defining a new molecular subset in patients with NSCLC [53]. The oncogenic potential of KIF5B-RET was later demonstrated in pre-clinical models via transfection of NIH/3T3 fibroblast cell lines with KIF5B-RET, leading to anchorage-independent growth activity [29].

In NSCLC, RET rearrangements occur at a frequency of approximately 1–2% in unselected patient populations and are generally non-overlapping with alterations in other oncogenic drivers, such as EGFR, ALK, and ROS1 [46,47,48,54,55]. In this population, the most common fusion partner of RET is KIF5B (40–70%), followed by CCDC6 (15–30%) [38,56,57,58]. The KIF5B gene partner is highly specific to NSCLC [48,59]. Patients diagnosed with RET-rearranged NSCLC typically have no history of tobacco smoking and tend to be relatively younger with a median age at diagnosis of <65 years [12,48,60,61]. When analyzing clinicopathologic features, RET fusions are mostly identified in adenocarcinomas, many of the solid or signet-ring cell subtypes, and typically present as peripheral tumors on lung imaging [54,55,62,63,64]. Brain metastases are common in RET-rearranged metastatic NSCLC, with a prevalence of 25% at diagnosis and a cumulative incidence of 45% during longitudinal follow-up [65]. Of note, RET fusions have also been described as an acquired resistance mechanism in patients with EGFR-mutant NSCLC treated with osimertinib [66]. In this setting, RET fusions typically involve gene partners other than KIF5B [59].

3.4. RET Activating Mutations across Other Histologies

Whereas activating RET mutations predominantly involve MTC, such alterationsthey have also been detected in other cancer populations. In a cohort of 4871 patients screened using targeted next-generation sequencing (NGS), Kato et al. described 88 patients (1.8%) presenting a RET aberration, with RET mutations (34/88) reported more commonly than RET fusions (27/88) or amplifications (22/88). The most prevalent activating mutation was RET M918T (n = 7), which apart from MTC was also reported across a spectrum of additional neuroendocrine tumors, including paraganglioma, pheochromocytoma, and carcinoid tumors [47]. Of note, small-cell lung cancer (SCLC), a high-grade neuroendocrine tumor, has also been associated with RET activating mutations although to a far lesser degree than nearly universal TP53 and RB loss-of-function mutations [59,67].

In their study, Kato et al. identified other canonical activating RET mutations, including V804M (n = 4) and C634R (n = 2), which involved cases of breast carcinoma (C634R) or gastrointestinal malignancies (V804M) such as colorectal adenocarcinoma, gastrointestinal stromal tumor, or hepatocellular carcinoma [47]. Recurring non-canonical activating mutations included RET E511K (n = 3) which was identified in patients with anaplastic thyroid cancer, endometrial adenocarcinoma, and Merkel cell carcinoma. Additional mutations of unknown significance involving tumors other than MTC included RET R114H (n = 6), M1109I/T (n = 2), RET R525Q (n = 2), RET R600Q (n = 2), RET V706M (n = 2), RET A756V (n = 1), RET M255I (n = 1), RET R163Q (n = 1), and RET T636M (n = 1).

3.5. RET Fusions across Other Histologies

Outside of NSCLC and PTC, RET fusions have been identified in other cancer types at variable frequencies (Figure 2). Pancreatic acinar cell carcinoma (PACC), a rare subtype representing less than 2% of pancreatic malignancies [68], typically lacks mutations in KRAS, CDKN2A, and TP53, while being associated with recurrent gene rearrangements [69]. BRAF fusions represent a common oncogenic event in PACC and are present in 10–20% of cases [69]. In addition, RET rearrangements have been identified in 7.5% of patients with PACC using fluorescence in-situ hybridization (FISH), with RET and BRAF fusions being mutually exclusive [70].

RET fusions have also been found in patients with metastatic colorectal cancer [48,49,59,71], with NCOA4 and CCDC6 as the most common fusion partners. When compared to a control group of RET wild-type tumors, RET rearrangements were enriched in patients with RAS and BRAF wild-type right-sided primary tumors. RET fusions were also more common in microsatellite instability-high (MSI-high) tumors when compared to MSI-low tumors (odds ratio 12.1), while being mutually exclusive with additional genomic aberrations such as RAS and BRAF. In this study, RET fusions were independently associated with worse overall survival (OS) when compared to patients without RET fusions (OS; 14.0 vs. 38.0 months, p < 0.001, respectively) [71]. This series should be interpreted with caution as study findings predate immune checkpoint inhibitor (ICI) US FDA approval for patients with MSI-high tumors; ICI use in this subgroup is active and may impact survival outcomes.

Alterations involving the RET oncogene have also been observed in patients with breast cancer, including canonical fusions and non-canonical missense mutations or amplifications [48,49,59,72]. Preclinical models demonstrated the transforming capacity of both RET fusions and amplifications in breast cancer cell lines through activation of the MAPK and PI3K pathways. While most RET fusions involved HER2-negative breast cancers with variable estrogen receptor (ER) expression, presence of the NCOA4-RET fusion was also identified as a possible bypass pathway resistance mechanism to dual HER2-targeted therapy and aromatase inhibitor in a case of ER-positive, HER2-overexpressed breast cancer.

Salivary gland carcinoma subtypes, such as salivary intraductal carcinomas, have been associated with an enrichment for RET fusions [73]. RET rearrangements have also been documented in salivary gland mammary analogue secretory carcinomas (MASCs), which are characterized by recurrent ETV6 rearrangements, with a minority of cases involving non-NTRK fusions, such as RET [74]. Apart from ETV6-RET fusions, RET rearrangements involving additional gene partners have also been described in salivary MASC such as the VIM-RET fusion [75].

Spitzoid neoplasms, including spitzoid melanomas and atypical spitzoid tumors, are associated with RET fusions in approximately 3% of cases. RET fusions typically involve the KIF5B gene partner and are mutually exclusive with other fusions such as ROS1, NTRK1, ALK, and BRAF. Interestingly, RET fusions were also detected in 3% of Spitz nevi, which represent uncommon, benign, melanocytic lesions typically arising in children and young adults. Such finding suggests that RET fusions may occur early in the pathogenesis of spitzoid malignancies [76].

RET rearrangements have been identified in sporadic cases of pheochromocytoma, without MEN 2 syndrome association. Interestingly, those include RET fusions with 3′ gene partners such as RET-SEPTIN9 [77] and RET-GRB2 [78]. In such cases, fusion events are associated with ligand-independent recruitment of canonical RET adaptor proteins, leading to constitutive activation of downstream signaling [78].

Pediatric spindle-cell mesenchymal neoplasms represent an additional group of tumors with recurrent oncogenic fusions, including NTRK, BRAF, MET, and RET. Such malignancies present histological overlap with so-called infantile fibrosarcoma cases and classically arise in the first 2 years of life. RET-rearranged spindle cell neoplasms were described with the MYH10, KIAA1217, and CLIP2 gene partners and had clinical presentations ranging from localized resectable tumors to more aggressive metastatic disease [79].

Histiocytosis is a clonal disorder of hematopoietic cells with a wide spectrum of presentations, including Langerhans cell histiocytosis (LCH) and Erdheim–Chester disease. Histiocytic neoplasms are commonly driven by activating mutations in the MAPK pathway, including BRAF and MEK alterations. Cutaneous xanthogranuloma, a non-LCH histiocytosis, has been associated with the NCOA4–RET rearrangement [80].

Further studies using comprehensive sequencing methods have also revealed the presence of RET rearrangements in KRAS wild-type pancreatic ductal adenocarcinoma, stomach adenocarcinoma, esophageal cancer, cholangiocarcinoma, bladder carcinoma, head and neck cancer, mesothelioma, low-grade glioma, atypical lung carcinoid tumor, and chronic myeloproliferative neoplasms [48,59,81,82,83,84].

3.6. RET Amplifications

Additional cytogenetic events such as RET amplifications have been described in MTC, PTC, and anaplastic thyroid cancer [85,86], although their role in thyroid carcinogenesis remains questionable. In NSCLC, RET copy number gains (8.1%) and amplifications (2.8%; when defined as innumerable RET clusters, or ≥7 copies in >10% of tumor cells) are detected in a higher proportion of tumor samples when compared to RET rearrangements (0.7%) [87]. In a pan-cancer cohort, RET amplifications (defined as ≥6 copies of wild-type RET) were detected in 0.16% (145/91,466) of tumor samples, including 0.13% (15/11,622) of NSCLC cases [88]. RET amplifications have been observed in several other tumor types, including hepatobiliary cancers [88,89], prostate cancer [88], breast cancer [77], glioblastoma [90], stomach adenocarcinoma [91], colorectal adenocarcinoma [91], and bladder urothelial carcinoma [91]. The clinical significance of RET amplifications and their correlation with increased RET protein expression have not been fully characterized.

4. Standard Methods for Detection

Since RET-selective targeted therapies are US FDA approved or available on clinical trials in countries without approval, universal screening for RET alterations should be considered when available, including for patients with NSCLC, thyroid cancer, and those with tumor histologies known to be enriched with RET fusions [92]. Activating RET mutations can be easily detected with the use of NGS. Reverse-transcriptase polymerase chain reaction (RT-PCR) or Sanger sequencing can alternatively be used in clinical settings where NGS is not readily available. Oncogenic RET rearrangements can also be identified using alternative testing methods, including FISH [53]. RET diagnostic assays, however, demonstrate different performance characteristics, including variable sensitivity for capturing specific RET fusions and mutations [48]. Initial clinical studies predominantly used RT-PCR and FISH techniques to detect RET rearrangements in tumor samples. However, the emergence of more comprehensive sequencing platforms such as NGS offered the added potential of identifying upstream partner genes involved in RET rearrangements, as well as confirmation of pathogenicity of RET fusions, and assessment of concurrent genomic alterations [93].

4.1. Immunohistochemistry (IHC)

Given that IHC can be used to measure RET protein overexpression, studies have explored the diagnostic utility of RET IHC as a potential screening assay for oncogenic RET alterations, particularly RET fusions. In clinical practice, RET IHC is, however, seldom performed as immunostaining techniques and antibodies remain poorly standardized [48,94]. Using a cytoplasmic staining cutoff of ≥1% of tumor cells, overall sensitivity and specificity of IHC with the EPR2871 clone is 87% and 82%, respectively [48]. Sensitivity of RET IHC differs according to fusion partner, with a 100% sensitivity for KIF5B, but only 50% sensitivity for RET rearrangements involving the NCOA4 fusion partner [48]. It should be noted that wild-type RET expression can be detected in normal cells by IHC, including in parafollicular C cells [94], tracheal epithelium [95], adrenal glands [96], and colorectal tissue [97]. A study also demonstrated that RET expression can be elevated in NSCLC even in the absence of RET fusions [98]. Consequently, RET IHC may be misleading when used as a standalone test, and alternative diagnostic assays usually outperform IHC as screening methods for RET alterations [74,94]. Given these challenges, RET IHC is not recommended as a clinical screening assay for oncogenic RET alterations [94].

As previously alluded to, solid tumors associated with an enrichment in NTRK fusions, such as MASC, infantile fibrosarcoma, and spitzoid neoplasms, can alternatively be associated with RET rearrangements upon negative pan-Trk IHC, which represents a validated screening tool for NTRK rearrangement [99]. Pan-TrK negativity in the setting of tumor histologies with a higher prevalence of kinase fusions should thus trigger RET-specific diagnostic assays due to therapeutic considerations.

4.2. FISH

FISH represents an alternative method for detection of RET rearrangements which was more commonly used before DNA-based NGS became readily accessible. Break-apart FISH is typically considered to be positive if ≥10–15% of tumor cells demonstrate a rearrangement pattern characterized by separation of 2 signals flanking the RET gene when evaluating 100 non-overlapping nuclei [48,94]. While the advantages of FISH include single-cell resolution as well as rapid turnaround time, there are important limitations to this method. First, FISH analysis does not have the spatial resolution to provide information regarding exonic breakpoints or inclusion of the RET kinase domain in the RET fusion. These datapoints are critical for confirming whether the RET rearrangement is pathogenic, and therefore actionable. Second, the accuracy of FISH depends on the involved fusion partner. Break-apart FISH can detect intrachromosomal rearrangements [71] with a 100% sensitivity for KIF5B and CCDC6, but is less sensitive for NCOA4 rearrangements (67% sensitivity) [48]. Such differences are likely attributable to the close contiguity of RET and NCOA4 on chromosome 10 which can lead to a more subtle splitting pattern [100]. When considering tumor primaries, sensitivity is highest with NSCLC (100%), followed by thyroid cancer (88%) and additional tumor types (75%), with such differences likely driven by the respective prevalence of fusions such as NCOA4-RET [48].

4.3. DNA PCR-Based Assays

Sanger sequencing, also known as chain-termination sequencing, allows robust and reliable identification of known single-nucleotide alterations when present at a relative allelic frequency above 15% [94]. However, due to its inability to detect gene rearrangements, partial gene deletions, and alterations with low variant allele frequencies (VAFs), it has limited use outside of FMTC and MEN2A diagnostics, where it still plays a role as a confirmatory test.

Quantitative PCR on DNA allows reliable detection of hot-spot mutations present at VAFs as low as 1% [94]. The finite number of primers used with this technique, however, plays a key role in limiting the reliability of these assays in detecting fusions and rarer oncogenic mutations that occur outside of hot spots [94]. While NGS-based assays are preferable for detecting RET activating mutations in clinical practice, quantitative PCR can still be considered as a viable alternative if NGS is not readily available.

4.4. NGS: DNA

Advancements in DNA sequencing techniques have facilitated the detection of RET alterations across various tumor types. Targeted DNA-based NGS can be used for simultaneous detection of RET mutations and rearrangements given sufficient exonic coverage and inclusion of introns (for RET fusions). A number of gene panels are clinically available to detect canonical RET alterations, with variable detection rates for RET alterations. When evaluating the performance of MSK IMPACT (hybrid capture DNA-based NGS panel) compared to RNA-based NGS as a reference standard (MSK-Fusion) in detecting RET fusions, MSK-IMPACT was highly sensitive; however, there were select cases of rearrangements of unknown significance (i.e., non-canonical RET rearrangements) where only RNA sequencing was able to confirm the presence or absence of oncogenic RET fusion transcripts [48]. Due to its high sensitivity, DNA-based NGS also allows for the identification of somatic mutations occurring at low VAFs. Subclonal RET rearrangements, which are only present in a minority of cells, can be detected by DNA-based NGS and are thought to reflect inherent tumor heterogeneity, including cases of acquired RET alterations as a resistance mechanism to osimertinib in EGFR-mutant NSCLC [31,101,102].

Plasma-based circulating free DNA (cfDNA) allows for non-invasive detection of RET alterations and represents an alternative to tissue-based DNA sequencing. The diagnostic yield of cfDNA testing is, however, contingent on the tumor burden and DNA shedding rates at the time of liquid biopsy, with increased sensitivity in patients with progressive or metastatic disease compared to those with stable or localized disease [48]. In addition to variable sensitivity, limited specificity associated with DNA-based sequencing (i.e., detection of RET rearrangements of unknown significance) remains an issue for cfDNA-based liquid biopsy testing. The use of cfDNA, however, can capture tumor heterogeneity and dynamics of treatment responses with longitudinal testing, thus providing a valuable platform to detect the influence of clonal and subclonal mutations on tumor progression [59]. Prospective cfDNA testing can therefore provide valuable insight when evaluating on-treatment resistance mechanisms and exploring rational combination therapies.

4.5. NGS: RNA

RNA-based NGS allows for the identification of functionally definitive RET fusion transcripts, irrespective of the underlying DNA-level events. In many clinical settings, RNA-based testing is an important confirmatory assay and often considered the gold standard method for RET fusion detection [103]. In general, RNA sequencing can be integrated into clinical workflow in two different ways. In workflows that involve screening by DNA-based sequencing, RNA testing can be performed on select cases with RET rearrangements of unknown significance to provide a more definitive and functional assessment. In addition, RNA-sequencing can be performed upfront with DNA-sequencing, although this may not be necessary in the majority of cases since DNA-sequencing is quite sensitive for RET fusion detection. As formalin fixation and block preservation of tissue is associated with significant RNA degradation, quality control of preanalytical conditions is crucial to ensure results accuracy.

5. RET-Altered Cancers: Pre-Targeted Therapy Era

The prognostic significance of RET fusions remains a matter of debate, with two Chinese studies delineating RET as a biomarker of inferior OS in NSCLC, especially in female patients [55,104]. Prior to the advent of selective RET inhibition, patients with RET-rearranged NSCLC treated with conventional therapies in the United States, including chemotherapy and immunotherapy, demonstrated similar survival outcomes when compared to patients with RET wild-type NSCLC in a retrospective analysis [60]. When considering chemotherapy options, pemetrexed alone or as part of a combination regimen can achieve durable clinical benefit in patients with RET-rearranged NSCLC. In a retrospective study, median progression-free survival (PFS) on pemetrexed-inclusive regimens was 19 months in RET-rearranged NSCLC, comparable to historical data from patients with tumors harboring ALK or ROS1 fusions [105].

As patients with RET-rearranged NSCLC were not specifically excluded from registrational ICI trials, a unicentric retrospective analysis was conducted in order to characterize treatment responses in this population. A majority (81%) of RET-rearranged patients had either a PD-L1 score of 0% or low expression of PD-L1 (1–49%). In addition, tumor mutational burden (TMB) was significantly lower when compared to RET wild-type cancers (1.75 vs. 5.27 mutations/megabase, respectively), reflecting the presence of an oncogenic driver. Suboptimal outcomes were observed in this cohort, with no objective responses seen in 13 evaluable patients treated with anti-PD(L)-1 with or without anti-CTLA4 therapy. Of note, baseline immunophenotype did not correlate with responses to ICI activity, and the median PFS was 3.4 months, consistent with RET fusion-positive lung cancers being immunologically cold tumors [57].

Little data are available on the efficacy of chemoimmunotherapy in patients with RET-altered cancers. In patients with RET fusion-positive NSCLCs, the combination of platinum-based chemotherapy and pembrolizumab in the first-line setting was associated with inferior OS and PFS when compared to pralsetinib in a retrospective analysis. Prospective data will be generated by the first line AcceleRET and LIBRETTO-431 studies that randomize patients with RET fusion-positive lung cancers to selective RET inhibitor therapy or chemotherapy +/− immunotherapy.

6. Multikinase RET Inhibitors

The efficacy of multikinase inhibitors (MKIs), including cabozantinib, sorafenib, vandetanib, and lenvatinib, which all exhibit activity against RET, has been observed in vitro with variable half maximal inhibitory concentration (IC₅₀) in kinase assays [61]. Pre-clinical modeling studying RET fusion driven cancer cell lines and patient-derived xenograft (PDX) models first confirmed the activity of MKIs through decreased RET autophosphorylation and inhibition of cell proliferation [35,106,107,108,109,110,111]. Similar biological activity was replicated in models with activating RET mutations, including MTC cell lines and PDX models, and in PDX models with RET amplifications [109,111,112,113], providing insights into the druggability of RET alterations by MKIs initially designed for other indications. Prospective trials have since provided evidence of confirmed responses and sustained disease control in a proportion of RET-altered thyroid, lung, and breast cancer patients treated with RET MKIs.

6.1. Cabozantinib

Cabozantinib is a potent MKI which exhibits significant inhibitory activity against various kinases, including RET (with an IC₅₀ value of 5.2 nM), as well as ROS1, MET, VEGFR2, AXL, TIE2, and KIT [64]. Cabozantinib was approved by the US FDA for the treatment of progressive metastatic MTC in 2012, following publication of a landmark phase 3 trial in a biomarker-unselected population [114]. Correlative analyses in this population later demonstrated increased PFS benefit in patients with RET mutations when compared to wild-type RET [115]. In 2013, rapid translational research led to the first clinical validation of cabozantinib use in a biomarker-selected population with RET-altered tumors, less than 2 years following initial identification of RET fusions as oncogenic drivers in patients with NSCLC [30,32,56]. A phase 2 trial enrolling patients with NSCLC and recurrent gene fusions first demonstrated evidence of clinical benefit with cabozantinib in three patients with RET fusions, with partial responses in two patients and prolonged disease stability in one patient [32,116].

The activity of cabozantinib as a RET-directed therapy is less characterized in additional cancer sites, such as patients with RET fusion-positive PTC. In a patient with ER+/HER2+ breast cancer harboring a NCOA4-RET fusion, treatment with cabozantinib in combination with trastuzumab and exemestane led to radiological tumor regression and symptomatic improvement following initial progression on the pertuzumab, trastuzumab, and anastrozole combination [72]. Whereas it remains challenging to determine with certainty whether the response was solely attributed to cabozantinib or a combination of these therapies, addition of a RET inhibitor led to a clinical response in this patient with treatment-refractory ER+/HER2-overexpressing breast cancer presenting a NCOA4-RET fusion [72].

6.2. Vandetanib

Phase 2 trials of vandetanib in pretreated patients with advanced RET-rearranged NSCLC demonstrated variable anticancer activity with overall response rates (ORRs) ranging from 18% to 50% [117]. In a biomarker-unselected population of patients with metastatic MTC, a placebo-controlled phase 3 trial demonstrated a PFS benefit with vandetanib (21 vs. 8 months with placebo) [118,119]. Of note, the correlation between survival outcomes and RET mutational status was not clearly evaluable in post-hoc analysis due to the low number of patients with wild-type RET [120]. In the setting of NSCLC, a retrospective analysis of four randomized phase 3 trials failed to demonstrate differential benefit with vandetanib in patients with RET amplifications or copy number gains when compared to the RET wild-type population [87]. In addition, no clinical benefit was seen in patients with RET rearrangements, although such alterations were detected at a low prevalence in this population [87]. Reports studying the use of vandetanib in biomarker-selected populations otherwise remain limited outside of the NSCLC population. Although rare, acquired RET mutations mediating resistance to RET MKIs such as vandetanib have been documented in the clinical setting, including the RET V804M [121] and V804L [122] gatekeeper mutations [122,123].

6.3. Limitations of Multitargeted RET Inhibitors

Studies involving additional RET MKIs, such as sorafenib [124], lenvatinib [34], and alectinib [125], also led to objective responses in RET-altered NSCLC, although with limited activity and prohibitive toxicity profiles [126]. In comparison to targeted therapies tailored to other oncogenic drivers commonly found in solid tumors, such as sensitizing EGFR or BRAF V600E mutations, as well as ALK or ROS1 rearrangements, response rates to RET-directed therapy remained relatively modest until the recent advent of selective RET inhibitors.

Factors accounting for the discrepancy between the in vitro and clinical activity of RET MKIs include inherent pharmacokinetic properties as well as off-target toxicities limiting optimal dose intensity of RET inhibitors [61]. RET MKIs do not consistently achieve concentrations allowing for effective RET inhibition, with high IC₅₀ values for both vandetanib and cabozantinib [122]. When considering metastatic disease to the brain, outcomes on RET MKIs were suboptimal, with an intracranial response rate of 18% in a cohort of patients treated with either cabozantinib, vandetanib, ponatinib, or alectinib [65], also highlighting the need for novel RET inhibitors with improved brain penetrance. RET MKIs can also be associated with a prohibitive safety profile, predominantly through VEGFR-mediated antiangiogenic toxicities [22]. Many of the MKIs targeting RET, including cabozantinib and vandetanib, are associated with preferential inhibition of VEGF signaling, with cabozantinib possessing a 100-fold higher anti-VEGFR potency relative to its anti-RET activity [112,127,128]. Treatment-emergent side effects from off-target VEGF inhibition, including hypertension, palmar-plantar dysesthesia, and proteinuria, can limit chronic drug dosing while impacting patients’ quality of life.

With the aim to improve RET selectivity, the VEGFR-sparing RET MKI RXDX-105 was studied in a phase 1/1b trial, but only demonstrated activity against non-KIF5B-associated tumors, thus limiting further clinical development [129]. Furthermore, RXDX-105 still demonstrated off-target toxicities through inhibition of non-RET kinases, with rash, fatigue, diarrhea, and hyperbilirubinemia reported as dose-limiting toxicities [129]. Unique safety signals also included drug hypersensitivity reactions, including three cases of drug reaction with eosinophilia and systemic symptoms (DRESS), likely attributable to RXDX-105 [129].

7. Selective RET Inhibitors

Mechanistically, RET inhibitors can be classified according to the Roskoski classification [130] which considers the unique chemical characteristics of each inhibitor, as well as the size of the molecule and the tridimensional conformation of the target enzyme in complex with the inhibitor. Kinases such as RET contain highly conserved structures, such as an activation loop (A-loop), an alphaC helix, and the Asp-Phe-Gly (DFG) motif [131], which play central roles in regulating kinase activation and substrate phosphorylation [132,133]. Phosphorylation of tyrosine residues in the A-loop activates the kinase and induces the opening (A-loop-out configuration) of the substrate binding pocket, while in the inactive state this structure is in the closed (A-loop-in) configuration [134]. The DFG motif is essential for ATP binding and effective catalysis and can adopt either the inactive (DFG-Dout) or the active (DFG-Din) conformation, with this latter state allowing phosphate transfer to the substrate [132,133]. Similarly, kinase activation induces conformational changes in alphaC helix which changes to the alphaC-in active configuration from the alphaC-out inactive configuration. These conformational changes induce dramatic modifications in protein structure accessibility for inhibitor binding [130]. An active RET kinase is characterized by the DFG-Din conformation in conjunction with the alphaC-in and A-loop-out configuration, while all other configurations in these regulatory structures constitute inactive states [130].

Aside from the conformational changes described above, several amino acidic residues can also contribute to volume accessibility for inhibitor binding [135], two of the most notable being the gatekeeper and the solvent front residues [108]. Bulky gatekeeper residue mutations can block access to RET kinase back pocket and modify ATP affinity [108], while solvent front mutations can restrict access to the solvent front cleft [136]. All clinically relevant RET inhibitors are ATP-competitive and can be classified as either type I or type II kinase inhibitors based on their kinase binding mode. Type I inhibitors such as vandetanib [137], pralsetinib [138], or selpercatinib [138] display affinity for the ATP binding pocket of active state RET and exert their inhibitory action via competition with ATP [139]. Type II inhibitors, such as cabozantinib, display affinity for RET in its inactive DFG-Dout conformation. Notably, the DFG-Dout conformation displays a bigger and more accessible back pocket compared to the active kinase, which is exploited for drug binding [118,139,140]. As previously alluded to, cabozantinib and vandetanib, however, both target a spectrum of kinases apart from RET, including VEGFR, EGFR, and KIT, often at a higher potency. Novel, selective RET inhibitors, including selpercatinib and pralsetinib, were designed to overcome such limitations while retaining activity against most RET alterations, including predicted gatekeeper resistance mutations [141].

7.1. Selpercatinib (LOXO-292)

Selpercatinib (LOXO-292) is an ATP-competitive, potent, and selective RET inhibitor with favorable pharmacokinetic properties and central nervous system (CNS) coverage. In contrast to RET MKIs, selpercatinib was specifically designed to target diverse RET fusions and RET activating mutations at nanomolar potency, including predicted acquired MKI resistance mutations. In a study using engineered cell lines and PDX models harboring KIF5B-RET, RET M918T, or RET V804M alterations, selpercatinib induced significant tumor regressions across models. Two patients with RET-altered malignancies refractory to MKI therapy received selpercatinib on single-patient protocols: a patient with RET M918T-mutant medullary thyroid cancer presenting an acquired RET V804M gatekeeper mutation and a second patient with a KIF5B-RET fusion-positive NSCLC and symptomatic brain metastases. Both patients had confirmed tumor responses, including a dramatic decrease in intracranial disease for the second patient, thus providing proof-of-concept of the activity of selective RET inhibition [121].

LIBRETTO-001 (NCT03157128) was a registrational phase 1/2 open-label basket study of selpercatinib in patients with solid tumors harboring oncogenic RET alterations [37,38,84]. The recommended phase 2 dose (RP2D) of selpercatinib was identified as 160 mg orally twice daily. In the NSCLC cohort, patients with prior disease progression on platinum-based chemotherapy demonstrated an ORR of 61%, whereas in previously untreated patients, the ORR was 84%, with rapid and durable responses seen in both subgroups (median duration of response 20–29 months) [142]. Of note, responses were seen regardless of the underlying RET fusion partner [38]. In patients with brain metastases, selpercatinib had an intracranial ORR of 82% and no CNS progression events were observed in patients without baseline intracranial involvement [143,144]. The thyroid cancer cohort included both RET-mutant MTC and RET-rearranged thyroid cancer. RET-mutant MTC patients previously treated with cabozantinib and/or vandetanib had an ORR of 69%, while responses were seen in 73% of patients without prior MKI therapy [37]. In previously treated RET fusion-positive thyroid cancer patients, most of which of papillary or poorly differentiated histology, the ORR was 64%. Converging real-world data in the MTC population with comparable results have subsequently been published [145]. Results from LIBRETTO-001 led to the US FDA regulatory approval of selpercatinib for the treatment of patients with RET fusion-positive NSCLC, RET fusion-positive thyroid cancers, and RET-mutant MTC in 2022 [146].

Importantly, the trial generated data in RET-rearranged, non-NSCLC, and non-thyroid cancer patients (Table 1). The ORR with selpercatinib was 44% in 45 patients with RET fusion-positive advanced solid tumors who previously progressed on standard of care therapies [84]. In this tumor-agnostic population, objective responses were documented in patients with refractory pancreatic, salivary gland, carcinoid, breast, small intestine, colorectal, ovarian, bile duct, and mesenchymal tumors, often with significant durability (median duration of treatment of 11 months). Most patients (58%) in the study had gastrointestinal malignancies and NCOA4 was the most common fusion partner, identified in 38% of cases. When considering tumor heterogeneity, response rates were highest in the 11 patients with pancreatic cancer (ORR 55%) and the 4 patients with salivary gland primaries (ORR 50%). In contrast, the ORR was lowest in colorectal cancer patients (ORR 20%), possibly as a consequence of underlying genetic diversity and MSI [71].

Since the publication of the tumor-agnostic cohort of LIBRETTO-001, additional case reports of activity in non-NSCLC and non-thyroid RET-altered cancers have emerged, including in tumor types which were not previously represented in the LIBRETTO-001 data set. Selpercatinib led to rapid responses in a patient with non-hereditary malignant pheochromocytoma harboring a novel RET-SEPTIN9 fusion [77]. Effective treatment with selpercatinib was also documented in RET fusion-positive pulmonary large cell neuroendocrine carcinoma [147] and ER+/HER2- breast cancer [148]. In addition, cutaneous xanthogranuloma, a non-LCH histiocytosis, has been associated with the NCOA4–RET rearrangement and treatment with selpercatinib led to a dramatic clinical response of cutaneous lesions [80].

Patient inclusion in the registrational studies of selective RET inhibitors was limited to solid tumors with RET fusions for the NSCLC and tumor agnostic cohorts. Subsequent case reports have shed light on the efficacy of selpercatinib in patients with RET amplifications. Notably, a patient with RET-amplified (≥6 copies) NSCLC exhibited systemic and intracranial radiologic responses on selpercatinib following disease progression on chemoradiation and durvalumab [88]. In addition, a documented response to selpercatinib was observed in a patient with RET-amplified glioblastoma. Follow-up MRI at 3 months revealed near complete resolution of gadolinium-enhancing brain lesions, with a sustained response spanning over 8 months [90]. These reports underscore the potential of selpercatinib as a therapeutic option for patients with RET-amplified tumors, thus warranting further investigation and clinical exploration.

Across cohorts, selpercatinib’s safety profile was found to be manageable. Commonly reported grade ≥ 3 toxicities included the following: hypertension (20–22%), increased alanine aminotransferase (ALT; 11–16%), increased aspartate aminotransferase (AST; 9–13%), hyponatremia (6–8%), diarrhea (6%), and QT interval prolongation on electrocardiogram (5%) [37,84,142]. Most of these treatment-related adverse events (AEs) did not warrant dose interruptions or reductions. In fact, only 2% of patients discontinued treatment due to drug-related toxicities in LIBRETTO-001, consistent with selpercatinib’s improved selectivity profile [37,38].

Other treatment-emergent toxicities have, however, been identified in the setting of selpercatinib therapy. Prior use of ICI was associated with an increased risk of hypersensitivity reaction to selpercatinib [149]. In addition, cases of spontaneous chylous effusions have been documented retrospectively, occurring in 7% of patients treated with selpercatinib. Both chylothorax and chylous ascites have been identified, with 76–80% of patients requiring additional therapeutic drainage after initial diagnostic thoraco- or paracentesis. Prospective recognition of such a side effect is crucial in clinical practice in order to avoid misattribution of chylous effusions to cancer progressive events [150]. Selpercatinib has also been associated with treatment-emergent hypothyroidism through off-target inhibition of type 2 iodothyronine deiodinase, which mediates active triiodothyronine (T3) synthesis from the inactive thyroxine (T4) form. Substituting levothyroxine (synthetic T4) with liothyronine (synthetic T3) supplementation in athyreotic patients normalized T3 levels and restored euthyroidism in this population [151].

7.2. Pralsetinib (BLU-667)

Pralsetinib (BLU-667) is a second, potent, ATP-competitive, selective RET TKI which has been approved by the FDA for the treatment of RET fusion-positive NSCLC and RET-altered thyroid cancers based on the results of the registrational ARROW trial [152,153]. In this multicohort, open-label, phase 1/2 trial, patients with locally advanced or metastatic solid tumors harboring a RET alteration were treated with pralsetinib as part of a dose-escalation design, with 400 mg once daily identified as the RP2D. In the NSCLC cohort, patients with prior platinum-based chemotherapy had an ORR of 59% and the ORR was 72% in previously untreated patients. In patients with measurable brain metastases at baseline, the intracranial response rate was 70% [58]. Real-world data later confirmed the superiority of pralsetinib over first-line ICI plus or minus chemotherapy in RET fusion-positive NSCLC [154]. The thyroid cancer cohort of ARROW included patients with RET-mutant MTC, either previously untreated or with progression on cabozantinib and/or vandetanib, as well as patients with previously treated RET-rearranged thyroid cancer. The following ORRs were observed: 71% in treatment-naïve MTC patients, 60% in MTC patients with prior cabozantinib and/or vandetanib, and 89% in previously treated RET fusion-positive thyroid cancer patients [153].

In patients with non-NSCLC, non-thyroid cancers, the tumor agnostic cohort of the ARROW trial demonstrated an ORR of 57% across histologies (Table 1). Pan-cancer efficacy was seen regardless of solid tumor types, including in epidermal (pancreatic, bile duct, head and neck, gastric) and mesenchymal (soft tissue sarcoma) tumors [36]. Of note, both complete and partial responses were observed in four pancreatic carcinoma cases (17% of the cohort). Such findings are in line with additional case reports of dramatic responses to pralsetinib in RET-rearranged pancreatic cancer, a tumor site associated with intrinsic refractoriness to most therapeutics [155]. Reponses to pralsetinib in RET-rearranged tumors have also been documented in patients with sarcomatoid carcinoma [156], as well as triple-negative breast cancer [157].

Table 1. Efficacy of selective RET inhibitors in non-NSCLC, non-thyroid cohorts with RET fusions.

Selpercatinib
Study	Study design	Tumor site	Patients, n	ORR, n (%)	Responses (n)	DOR range (mo)
LIBRETTO-001 trial (2022) [84]	Prospective	Pancreas	12	6 (55%)	PR (6), SD (6)	3–38 *
		Colorectal	10	2 (20%)	PR (2), SD (8)	6–13
		Salivary	4	2 (50%)	PR (2), SD (2)	6–29 *
		Unknown primary	3	1 (33%)	PR (1), SD (2)	9 *
		Breast	2	2 (100%)	CR (1), PR (1)	2–17
		Sarcoma	2	2 (100%)	PR (2)	15 *
		Xanthogranuloma	2	NE	N/A	N/A
		Carcinoid (lung)	1	1 (100%)	PR (1)	24 *
		Skin carcinoma	1	NE	N/A	N/A
		Cholangiocarcinoma	1	1 (100%)	PR (1)	6 *
		Ovarian	1	1 (100%)	PR (1)	15 *
		Pulmonary carcinosarcoma	1	NE	N/A	N/A
		Neuroendocrine (rectum)	1	NE	N/A	N/A
		Small intestine	1	1 (100%)	CR (1)	25
Durham et al. (2019) [80]	Case report	Xanthogranuloma	1	N/A	Clinical response	N/A
Kander et al. (2021) [81]	Case report	Carcinoid (lung)	1	N/A	PR	6 *
Watanabe et al. (2021) [148]	Case report	Breast, ER+/HER2-	1	N/A	CR	10 *
Mweempwa et al. (2021) [77]	Case report	Pheochromocytoma	1	N/A	PR	5 *
Arora et al. (2023) [147]	Case report	Neuroendocrine (lung)	1	N/A	PR	12 *
Pralsetinib
ARROW trial (2022) [36,58]	Prospective	Pancreas	4	4 (100%)	CR (1), PR (3)	3–27 *
		Cholangiocarcinoma	3	2 (67%)	PR (2), SD (1)	8–19
		Neuroendocrine	3	2 (67%)	PR (2), SD (1)	11–14 *
		Sarcoma	3	2 (67%)	CR (1), PR (1), SD (1)	11
		Head and neck	2	1 (50%	PR (1), SD (1)	9
		Colorectal	2	0 (0%)	SD (2)	N/A
		Small cell lung cancer	2	1 (50%)	PR (1), SD (1)	9 *
		Unknown primary	1	1 (100%)	CR (1)	5
		Stomach	1	0 (0%)	PD (1)	N/A
		Ovarian	1	0 (0%	PD (1)	N/A
		Thymic	1	0 (0%)	SD (1)	N/A
Wu et al. (2022) [156]	Case report	Sarcomatoid carcinoma (lung)	1	N/A	PR	NE
Zhang et al. (2023) [155]	Case report	Pancreas	1	N/A	PR	12 *
Zhao et al. (2023) [157]	Case report	Breast, triple negative	1	N/A	PR	8 *

ORR: objective response rate; DOR: duration of response; Mo: months; CR: complete response; PR: partial response; PD: progressive disease; NE: non-evaluable; N/A: not applicable. * Ongoing response.

Table 1. Efficacy of selective RET inhibitors in non-NSCLC, non-thyroid cohorts with RET fusions.

Selpercatinib
Study	Study design	Tumor site	Patients, n	ORR, n (%)	Responses (n)	DOR range (mo)
LIBRETTO-001 trial (2022) [84]	Prospective	Pancreas	12	6 (55%)	PR (6), SD (6)	3–38 *
		Colorectal	10	2 (20%)	PR (2), SD (8)	6–13
		Salivary	4	2 (50%)	PR (2), SD (2)	6–29 *
		Unknown primary	3	1 (33%)	PR (1), SD (2)	9 *
		Breast	2	2 (100%)	CR (1), PR (1)	2–17
		Sarcoma	2	2 (100%)	PR (2)	15 *
		Xanthogranuloma	2	NE	N/A	N/A
		Carcinoid (lung)	1	1 (100%)	PR (1)	24 *
		Skin carcinoma	1	NE	N/A	N/A
		Cholangiocarcinoma	1	1 (100%)	PR (1)	6 *
		Ovarian	1	1 (100%)	PR (1)	15 *
		Pulmonary carcinosarcoma	1	NE	N/A	N/A
		Neuroendocrine (rectum)	1	NE	N/A	N/A
		Small intestine	1	1 (100%)	CR (1)	25
Durham et al. (2019) [80]	Case report	Xanthogranuloma	1	N/A	Clinical response	N/A
Kander et al. (2021) [81]	Case report	Carcinoid (lung)	1	N/A	PR	6 *
Watanabe et al. (2021) [148]	Case report	Breast, ER+/HER2-	1	N/A	CR	10 *
Mweempwa et al. (2021) [77]	Case report	Pheochromocytoma	1	N/A	PR	5 *
Arora et al. (2023) [147]	Case report	Neuroendocrine (lung)	1	N/A	PR	12 *
Pralsetinib
ARROW trial (2022) [36,58]	Prospective	Pancreas	4	4 (100%)	CR (1), PR (3)	3–27 *
		Cholangiocarcinoma	3	2 (67%)	PR (2), SD (1)	8–19
		Neuroendocrine	3	2 (67%)	PR (2), SD (1)	11–14 *
		Sarcoma	3	2 (67%)	CR (1), PR (1), SD (1)	11
		Head and neck	2	1 (50%	PR (1), SD (1)	9
		Colorectal	2	0 (0%)	SD (2)	N/A
		Small cell lung cancer	2	1 (50%)	PR (1), SD (1)	9 *
		Unknown primary	1	1 (100%)	CR (1)	5
		Stomach	1	0 (0%)	PD (1)	N/A
		Ovarian	1	0 (0%	PD (1)	N/A
		Thymic	1	0 (0%)	SD (1)	N/A
Wu et al. (2022) [156]	Case report	Sarcomatoid carcinoma (lung)	1	N/A	PR	NE
Zhang et al. (2023) [155]	Case report	Pancreas	1	N/A	PR	12 *
Zhao et al. (2023) [157]	Case report	Breast, triple negative	1	N/A	PR	8 *

ORR: objective response rate; DOR: duration of response; Mo: months; CR: complete response; PR: partial response; PD: progressive disease; NE: non-evaluable; N/A: not applicable. * Ongoing response.

Across studies, pralsetinib was well tolerated with common grade 3–4 AEs consisting of neutropenia (13–31%), hypertension (7–17%), anemia (10–14%), lymphopenia (9–12%), and increased creatine phosphokinase (CPK; 6%) [36,58,153]. Additional serious AEs, including treatment-emergent pneumonitis (4%), were recorded on pralsetinib [153].

8. Acquired Resistance

Limited accounts of RET acquired resistance mutations were reported during the era of RET MKIs [121,122,129,158], likely indicative of suboptimal target inhibition and restricted selective pressure on RET-altered clones. Since the US FDA approval of potent and selective RET inhibitors, emergence of concurrent genetic alterations has been documented in patients with baseline RET aberrations with disease progression on selpercatinib or pralsetinib. On-target RET solvent front mutations involving the G810 residue (G810 C/S/R) represent a recurrent mechanism of resistance, leading to steric hindrance to the binding of selpercatinib [136,159]. RET G810X mutations also mediate cross-resistance to pralsetinib [138,160]. Alternative roof solvent front mutations, such as RET L730V/I, have been identified in patients with pralsetinib-resistant tumors, while retaining sensitivity to selpercatinib in preclinical models [161]. On-target mutations affecting the Y806 or V738 residues have also been identified as mediators of resistance to both selpercatinib and pralsetinib [138].

Off-target resistance mechanisms have been observed in patients treated with selpercatinib or pralsetinib, including bypass pathway activation involving MET amplifications [159]. Proof of concept clinical reports have demonstrated the activity of MET inhibitors in the setting of acquired MET amplification in patients with RET fusion-positive NSCLCs. The combination of crizotinib and selpercatinib led to responses in selpercatinib-refractory patients with acquired MET amplification. In a case report, sequential use of crizotinib and capmatinib led to a response in a patient with KIF5B-RET fusion-positive NSCLC with acquired MET amplification who had initially progressed on pralsetinib. Downstream MAPK pathway reactivation has also been involved in patients with acquired resistance to selective RET inhibitors [140,159,162], including the emergence of activating KRAS, NRAS, or BRAF mutations [162]. In addition, histological transformation to SCLC has been described as an acquired resistance mechanism to pralsetinib in patients with RET-rearranged lung adenocarcinoma [163,164].

9. Conclusions and Future Directions

In conclusion, RET fusions and RET activating mutations are therapeutically actionable oncogenic drivers which can be effectively targeted by novel and potent inhibitors selectively targeting the RET kinase. Deep and durable responses have been demonstrated both with selpercatinib and pralsetinib across a wide range of histologies, arguing for the universal screening of solid tumors for the presence of RET alterations. Alternatively, a two-step approach would be reasonable, with screening for RET alterations limited to patients without oncogenic drivers on initial testing. Next-generation RET inhibitors that potentially circumvent mechanisms of acquired resistance to selective RET inhibitors, including emergent solvent front mutations, are being explored. Ongoing clinical trials are assessing the safety and efficacy of novel RET inhibitors, including vepafestinib (NCT04683250) and LOXO-260 (NCT05241834). Additional efforts should lie in exploring the therapeutic potential of these newer-generation TKIs across different cancer types and defining the spectrum of RET alterations.