New Prospects for Molecular Targets for Chordomas

Introduction

Chordomas are rare, low-grade tumors of the bone that originate in the midline axial skeleton. They comprise about 3% of primary bone tumors, 20% of primary spinal tumors, and have a predilection to occur males over females (2:1 ratio)^1–3. They are typically seen in adults 50–60 years of age, however they can occur earlier. Even though they are indolent tumors, chordomas are locally aggressive and have a strong propensity to recur². As slow growing tumors, they are radioresistant to low-dose conventionally fractionated external beam radiation and highly resistant to conventional chemotherapies. This property, together with their anatomic location adjacent to he brain stem, cranial nerves and spinal cord, results in their associated morbidity and mortality.

The majority of chordomas occur in three regions of the neuraxis: the sacrum, the skull base clivus, and the mobile spine, with an almost even distribution². Clival chordomas may be more likely to occur in younger patients (<50 years), while spinal chordomas typically present in older patients². In advanced cases, chordomas can also metastasize to body sites including lung, bone marrow, skin, and brain.

Several lines of evidence suggest that chordomas arise from the vestiges of the notochord. These include their distribution in regions that have been shown to contain remnants of notochord in the midline, such as the vertebral bodies of the spine and the clivus^1,2,4; their dependence on the transcription factor TBXT (brachyury; described below) which is known to be required for notochord differentiation in the embryo⁵; the histological similarities of benign notochordal cell tumors – which can transform into chordomas – to the embryonic notochordal cells^1,6; and the development of chordomas in a zebrafish model by overexpression of human RAS^V12 oncogene in the notochord^7,8.

Traditionally, the mainstay of treatment of chordoma is en bloc resection, although this is often associated with morbidity and is not always possible due to neurovascular involvement. Stereotactic radiosurgery, carbon ion, and proton beam radiation in the standalone or neoadjuvant/adjuvant settings are also in common use with good local control rates^2,9,10. Currently, the median survival of chordoma patients is 7.7 years across all races and genders³. Several ongoing trials are attempting to use pathway inhibitors in conjunction with the aforementioned modalities for recurrent or advanced chordomas. In addition, immunotherapies including TBXT-directed vaccination and checkpoint inhibition have been attempted. In this article, we will discuss the major pathways that have been implicated in the pathogenesis of chordoma, with an emphasis on molecular vulnerabilities that therapies on the horizon are attempting to exploit.

Histopathology and Genetics of Chordomas

Histopathologically, chordomas manifest as epitheloid “physaliferous” cells with rounded nuclei and enlarged intracellular vacuoles². Chordomas can manifest different degrees of atypia on histopathology, but three categories are recognized: 1) classical, which appear as described above, 2) dedifferentiated, which lack identifying cellular characteristics, and 3) chondroid, which have features of both chondromas and chondrosarcomas. Classical chordomas show immunoreactivity for brachyury, S100, cytokeratins, as well as the epithelial membrane antigen MUC1². The combination of brachyury and cytokeratin immunostaining is considered to be diagnostic and indeed pathognomonic for chondromas (sensitivity: 98%, specificity: 100%)¹¹.

Like other tumors, chordomas are cytogenetically heterogeneous and karyotypically diverse. They feature complex rearrangements of the genome, including chromosomal gains and deletions, genome-wide copy number changes, and chromothripsis^1,12–14. The most common deletions are observed on chromosome 3p while gains are most often seen in 7q¹. However, chordomas harbor a low overall mutational rate and a general paucity of driver genes, with about 15 somatic point mutation events/exome. This coincides with the slow growing nature of these tumors and suggests that chordoma cells are initially transformed by dosage changes in a limited number of genes and pathways. Among the mutations that are found in chordomas, the tumor suppressor PTEN is inactivated in about three-quarters of all tumors, while loss of the cell cycle regulators CDKN2A and CDKN2B is seen in up to 70% of chordomas¹. Inactivating mutations of p53 and RB are also frequently observed, as are mutations in the EGFR, VEGF, and PDGF pathways. Due to the recent availability of biologic inhibitors of these pathways, several trials are attempting to test the role of molecular targeted therapies (MTTs) in advanced or recurrent chordomas. These will be discussed in detail below.

Several studies have also documented frequent gene duplications and other mutations in the TBXT gene in both sporadic and familial chondromas, implicating the brachyury protein^15–18. Thus, brachyury is the major transcription factor that regulates chordoma formation and is detailed in the following section. Apart from TBXT, other mutations have been detected in epigenetic regulators that modulate genome-wide chromatin accessibility, including brachyury itself, by one of three mechanisms: remodeling nucleosome topology, modulating histone modifications, or via DNA methylation. These commonly mutated or deleted genes include SMARCB1, SETD2, PBRM1, EP300, KDM6A, and ARID1B^12,19. Thus, modulators of chromatin structure are also targets of therapies for chordomas, and we will address the mechanism of these agents below.

Brachyury is a driver of chordoma pathogenesis

TBXT (Brachyury) is an evolutionarily conserved T-box transcription factor that regulates gene regulation through its N-terminus which contains the T-box binding domain. It is expressed in the posterior part of the developing embryo where it is essential for many aspects of development, such as specification of the mesoderm and its derivatives, migration of mesoderm into the primitive streak, proliferation, epithelial to mesenchymal transition (EMT), as well as induction of primordial germ cell fates in mice^20–22. Mouse brachyury knockout embryos fail to develop mesodermal derivatives and are embryonic lethal by E13.5^23,24. In adults, brachyury expression is normally restricted to the testis and some parts of the thyroid^25,26.

In the embryo, the notochord is derived from the mesoderm between the third and fourth week of development in humans (post-conception days 15–30; E7.0-E8.5 in mice)²⁷. The notochord lies ventral to the spinal cord and is essential for establishment of the anterior-posterior and dorsoventral axis of the embryo. It does this by releasing and creating a gradient of sonic hedgehog signaling²⁸. In adults, the notochord has largely disappeared, except for remnants in the intervertebral disc, the vertebral bodies, and the clivus²⁹. The formation of the notochord during development also requires Brachyury expression, where it appears to specifically promote EMT but not proliferation of these cells³⁰.

Multiple lines of evidence have shown that TBXT is a critical gene in chordoma pathogenesis. Brachyury is expressed in all chordomas, as detected by RNA sequencing and immunostaining³¹. In addition, some 97% of chordoma patients harbor a germline single nucleotide polymorphism (SNP) rs2305089 (p. Gly177Asp) in the T-box domain, which leads to enhanced DNA binding and higher expression of its target genes and auto-transcription³². Furthermore, duplications of the TBXT allele are also present in familial chordomas and some sporadic chordomas^15–18. Additionally, genetic loss-of-function screens for toxicity have shown that Brachyury is the most highly enriched target in multiple chordoma lines, suggesting that chordoma cells are addicted to brachyury expression for their survival⁵. Lastly, knockdown of Brachyury with shRNA reverses the chordoma phenotype by preventing EMT and inducing growth arrest and senescence of cell lines^17,33,34. It is important to note that even though the Gly177Asp SNP is found in a large subset of the population, the vast majority of these people will not develop chordomas, suggesting that the TBXT SNP yields a cellular advantage but is itself insufficient, requiring interaction with other molecular factors to drive chordomagenesis. This is supported by experiments in zebrafish, where activation of receptor tyrosine kinases in notochordal cells could lead to chordomagenesis in vivo, but brachyury overexpression alone does not⁸.

While it is currently unclear how Brachyury contributes to transformation of notochord remnants, we are beginning to understand how brachyury enables local aggressiveness of chordomas. Sharifnia et al. showed that in chordoma cell lines and primary tumors, Brachyury binds to large clusters of enhancers that regulate developmental and cancer related genes⁵. These clusters are called super-enhancers and characteristically contain islands of H3K4 acetylation which marks active enhancers of genes. These super-enhancers were typically associated with genes that are highly expressed in chordomas, such as extracellular matrix (ECM)-related genes, brachyury, HoxA and EGFR. In this study, brachyury expression was specifically targeted with inhibitors of CDK9 and CDK7/12/13, which are known to be involved in actively transcribed genes and which associate with super-enhancers. These inhibitors included THZ1, NVP-2, dinaciclib, and alvocidib, which are all preclinical agents. Loss of Brachyury either by CDK inhibition or genetic knockout led to a strong decrease in notochord markers including SOX9 and COL2A1, loss of cell cycle progression genes, growth arrest, and cell death. Thus, targeting Brachyury with a new generation of specific CDK inhibitors appears to be an attractive avenue for future treatment of chordoma.

Brachyury has been previously shown to act as an oncogene in numerous malignancies including lung and squamous cell carcinomas, sarcomas, germ cell tumors, hematologic cancers, and brain tumors^{26,33,35–38}. It confers its oncogenic properties in cells by regulating a diverse array of target genes that hijack its developmental role in promoting stemness, migration, and EMT. In chordomas and other cancers, brachyury has been shown to upregulate expression and protein stabilization of the oncoprotein YAP, decrease the tumor suppressor PTEN, increase EMT genes such as SNAIL, and attenuate cell cycle progression in tumor initiating cells (TICs) by suppressing p21, cyclin D1, and pRb^{5,17,33,38–40}. Together this network of changes promote invasiveness of tumor cells and confers radio- and chemo-resistance. For these reasons, Brachyury has become an attractive target for pharmacotherapeutics as well as immunotherapies^2,19,41.

Epigenetic regulators in Chordomas

There are many chromatin modifications that affect gene expression in cells. Of these, H3K27 trimethylation (H3K27me3) and H3K9 dimethylation (H3K9me2) are considered to be repressive chromatin because they reduce transcription. In addition, a variant of histone H3, H3.3 – that differs by only 4 amino acids – has been shown to be selectively recruited to nucleosomes of actively transcribed genes^42,43. Conversely, H3K4me3, H3K36me3, and H3K4 acetylation (H3K4ac) are considered to be active chromatin marks as they positively regulate transcription (Figure 1A). Imbalances in histone marks, particularly H3K27 methylation, have been well established to contribute to oncogenesis. This can occur either by loss of the EZH2/PRC2 complex – which acts as an H3K27 histone methylation writer – or by overexpression and gain-of-function mutations in this complex, or by loss of antagonists such as H3K27 histone demethylases⁴⁴.

Figure 1: Cellular and epigenetic changes in chordoma and therapeutic opportunities.

A) A schematic of open and closed chromatin and associated epigenetic modulators and transcription factors in normal cells. The panel on the right shows the presence of normal homologous recombination (HR), non-homologous end-joining (NHEJ), cell cycle regulators (CDKN2A, CDKN2B, and CDK4/6), and the H3K36 methyltransferase SETD2 in these cells. B) In chordoma cells, mutations in epigenetic regulators and transcription factors alters the distribution of open and closed chromatin and lead to genome-wide alterations in the epigenetic signature. Together with mutations in the HR, CDKN2A/2B, and SETD2 pathways, this results in uncontrolled proliferation of cells. The drugs shown in panel B) are being used in trials to target the compensatory changes in these pathways in tumor cells to revert the chordoma cells to induce cell senescence or death.

Brachyury as a transcription regulator is known to interact extensively and directly with histone modulators. Indeed, during development, brachyury promotes permissive chromatin for transcription by recruitment of methyltransferases at H3K4 sites and demethylases to H3K27 sites and histone deacetylases⁴⁵ (Figure 1A). A conserved tyrosine residue in the N-terminus of the protein has been shown to be essential for its ability to recruit these histone modifiers.

Expression of brachyury within chordomas itself has been shown to be dependent upon recruitment of two H3K27 histone demethylases, KDM6A (UTX) and KDM6B (JMJD3), to the TBXT locus⁴⁶. Indeed, pharmacological inactivation with KDOBA67 or genetic deletion of these demethylases results in downregulation of Brachyury and its target genes in both chordoma cell lines as well as primary chordomas⁴⁶ (Figure 1B). The target genes include cell cycle genes resulting in a strong reduction in growth and cell viability. Moreover, this effect is due to a genome-wide loss of repressive chromatin marks such as H3K27me3 and H3K9me2 and gain of active chromatin marks such as H3K4me3 and H3.3 in the absence of any change in promoter methylation of TBXT⁴⁶. This suggests that at least in chordomas, loss of promoter methylation of brachyury does not play a large role in disease pathogenesis.

Previous studies have also shown that primary and recurrent chordomas express multiple histone deacetylases (HDACs), particularly high levels of HDAC6^47,48. Treatment of a chordoma cell line with pan-histone deacetylase inhibitors such as SAHA and LBH-589 resulted in increased cell death⁴⁸. Thus, modulation of histone modifiers with drugs that inhibit H3K4 acetylation or H3K27 trimethylation might represent an intriguing new approach to targeting chordomas (Figure 1B). Our work including unpublished data demonstrates intact repressive H3K27 trimethylation, hypotrimethylated (derepressed) H3K36 and H3K4 hypermethylation by histone mass spectrometry in chordoma as compared to a basket of cancer cell lines and a primary tumor sample⁴⁹.

In addition to the above histone modifiers, SETD2, a H3K36 methyltransferase, is found on is found on chromosome 3p, which as mentioned above is frequently deleted in chordomas¹². The H3K36me3 mark activates transcription and mediates DNA repair by homologous recombination; therefore, SETD2 acts as a tumor suppressor gene. Loss of SETD2 function has been observed in many cancers including renal, breast, lung, and in gliomas^50–53. The absence of SETD2 leads to a detectable loss of H3K36 trimethylation in chordomas lacking this gene¹². This presents another therapeutic opportunity as previous studies have shown that H3K36me3 deficient cancers can be targeted and killed by WEE1 inhibitors such as AZD1775 by deoxynucleotide triphosphate (dNTP) starvation⁵⁴ (Figure 1B).

Other epigenetic regulators that are frequently deleted in pediatric chordomas include SMARCB1^12,55, which is a unit of a large SWI/SNF complex that alters nucleosome topology by physically unwrapping DNA from histones (Figure 1A). SMARCB1 is a tumor suppressor and an antagonist of the EZH2, and its loss leads to an imbalance in H3K27 methylation⁵⁶. On the basis of this interaction, phase II clinical trials are currently being conducted in patients with SMARCB1 deleted chordomas with an inhibitor of EZH2, tazemetostat (NCT02601950) (Figure 1A; Table 1). Notably, tazemetostat treatment resulted in responses in a case series of two pediatric patients with chordomas^19,56.

Table 1:

Active immunotherapy trials including patients with chordoma

Trial Registration Number	Study Title	Type/Phase	Agent/Intervention
NCT02383498	A Randomized, Double-Blind, Phase 2 Trial of GI-6301 (Yeast-Brachyury Vaccine) Versus Placebo in Combination With Standard of Care Definitive Radiotherapy in Locally Advanced, Unresectable, Chordoma	Phase II	Biological: GI-6301 Vaccine (Yeast- Brachyury) Other: GI-6301 Placebo Radiation: Radiotherapy
NCT03595228	A Phase 2 Trial of BN-Brachyury and Radiation Therapy in Patients With Advanced Chordoma	Phase II	Biological: BN-Brachyury (brachyury vaccine) plus radiation
NCT03623854	A Signal Finding Phase 2 Study of Nivolumab (Anti-PD-1; BMS-936558; ONO-4538) and Relatlimab (Anti-LAG-3 Monoclonal Antibody; BMS-986016) in Patients With Advanced Chordoma	Phase II	Biological: Nivolumab (anti-PD-1) Biological: Relatlimab (anti-LAG-3)
NCT02989636	Phase I Safety Study of Stereotactic Radiosurgery With Concurrent and Adjuvant PD-1 Antibody Nivolumab in Subjects With Recurrent or Advanced Chordoma	Phase I	Biological: Nivolumab (anti-PD-1) Radiation: Stereotactic Radiosurgery
NCT02601950	A Phase II, Multicenter Study of the EZH2 Inhibitor Tazemetostat in Adult Subjects With INI1-Negative Tumors or Relapsed/Refractory Synovial Sarcoma	Phase II	Drug: Tazemetostat (for chordomas with SMARCB1/INI1 loss)
NCT02601937	A Phase 1 Study of the EZH2 Inhibitor Tazemetostat in Pediatric Subjects With Relapsed or Refractory INI1-Negative Tumors or Synovial Sarcoma	Phase I	Drug: Tazemetostat (for chordomas with INI1 loss)
NCT02193503	An Open Phase I Clinical Study Assessing Safety and Tolerability of MVX-ONCO-1 in Patients With Solid Tumor Who Are Not/Not Any Longer Amenable to Standard Therapy	Phase I	Other: MVX-ONCO-1 (autologous tumor cell vaccine+ GM-CSF)
NCT03349983	An Open-label Phase 1 Trial to Evaluate the Safety and Tolerability of a Modified Vaccinia Ankara (MVA) Priming Followed by Fowlpox Booster Vaccines Modified to Express Brachyury and T-cell Costimulatory Molecules (MVA-BN-Brachyury/FPV-Brachyury)	Phase I	Biological: MVA-BN-Brachyury/FPV-Brachyury (Brachyury vaccine)
NCT03886311	The TNT Protocol: A Phase 2 Study Using Talimogene Laherparepvec,Nivolumab and Trabectedin as First, Second/Third Line Therapy for Advanced Sarcoma, Including Desmoid Tumor and Chordoma	Phase II	Biological: Talimogene Laherparepvec [IMLYGIC] (oncolytic virus) Biological: Nivolumab (anti-PD-1) Drug: Trabectedin (chemotherapy inhibiting DNA binding of a family of transcription factors)
NCT03874455	Tazemetostat Expanded Access Program for Adults With Solid Tumors	---	Drug: Tazemetostat (for chordomas with SMARCA4/INI1 loss)
NCT03173950	Phase II trial of the immune checkpoint inhibitor nivolumab in patients with select rare CNS cancers	Phase II	Biological: Nivolumab (anti-PD-1)
NCT02815995	A phase II multi-arm study to test the efficacy of immunotherapeutic agents in multiple sarcoma subtypes	Phase II	Biological: Durvalumab (anti-PD-1) Biological: Tremelimumab (anti-CTLA4)

Molecular Targeted Therapies for Chordomas

Although management with conventional chemotherapeutic agents been reported for chordomas, a systematic review of literature suggests that chordomas are largely insensitive to traditional agents such as cisplatin, alkylating agents, camptothecin, or anthracyclines². Receptor tyrosine kinases have been shown to be variably mutated and/or overexpressed in chordomas in both primary tumors and cell lines and are thought to be variably important in the initiation and progression of chordomas. To date, MTTs have been attempted clinically in at least four signaling pathways¹⁹: 1) Platelet derived growth factor receptor (PDGFR) and stem cell factor receptor (KIT) – targeted by imatinib and dasatinib; 2) vascular endothelial growth factor receptor (VEGFR) – targeted by sorafenib, pazopanib, and sunitinib; 3) epidermal growth factor receptor (EGFR) and human epidermal growth factor receptor 2 (HER2/neu) – targeted by gefinitib, lapatinib, erolotinib, and cetuximab; 4) phosphatidylinositol 3-kinase (PI3K)/AKT/PTEN/mTOR pathway – targeted by sirolimus and temsirolimus (Figure 2).

Figure 2: Molecular pathways targeted in chordomas.

The three major signaling pathways that have been targeted in chordomas. Components of these pathways are frequently mutated in chordomas, resulting in overactivation that results in increased proliferation, upstream of the changes described in Figure 1. Of these pathways, inhibitors to all three receptor tyrosine kinases (RTKs) have been used in clinical trials of advanced or recurrent chordomas, as have RAS/MEK/ERK and PI3K/AKT/PTEN/mTOR inhibitors. JAK/STAT3 inhibitors have shown cell death and reduced proliferation in chordoma cell lines in vitro.

In most instances, these drugs have been used in clinical trials on patients with advanced disease, who have either local relapse or metastasis. Because of the limited number of patients involved in these trials and a lack of long-term follow up in many cases, the interpretation of efficacy of MTTs needs to be restrained. Furthermore, response to treatments has been measured on the basis of radiological criteria that are often variable between studies. Notwithstanding these caveats, PDGFR inhibitors such as imatinib have been described as a first line MTT agents, although the effects on progression-free survival and overall survival have been modest, with a median progression free survival of about 9.9 months¹⁹. In cases where imatinib therapy failed, EGFR inhibitors such as erlotinib appear to achieve a good response. A systematic review of published studies suggested that the use of monotherapy with one tyrosine-kinase inhibitor (such as PDGFR inhibitors) were an acceptable first line treatment for advanced disease with limited adverse effects¹⁹. In the same review, combination treatments were the recommendation for drug-resistant advanced chordomas, either with two tyrosine-kinase inhibitors or one tyrosine-kinase inhibitor and an mTOR inhibitor, with the exact choice of agents depending on the pathways expressed by individual patients’ tumors. Among MTTs on the horizon, one particular small-molecule inhibitor of the EGFR pathway, afitinib shows promise because of its ability to not just inhibit EGFR, but also promotes Brachyury degradation in multiple chordoma cell lines and inhibits growth of xenotransplant tumors in vivo⁵⁷ (Figure 2). This observation has led to a clinical trial (NCT03083678). Whether MTTs can be beneficial in primary chordomas, either as standalone or in combination with surgery and/or irradiation, will need to be addressed in future trials.

Activation of the STAT3 pathway and its downstream targets has also been observed in chordoma samples by multiple groups using microarray and immunohistochemistry for phospho-STAT3^58–60 (Figure 2). Several studies have shown that inhibition of phosphorylated STAT3 using small molecule inhibitors affects the viability of chordoma cell lines in vitro^60–63. While inhibitors of STAT3 still await preclinical studies in animal models, they represent another molecularly exploitable target for future treatments.

In addition to the above-mentioned genes and pathways, deletions in CDKN2A and CDKN2B are also frequently observed in chordomas^12,55. The products of these genes (INK4A, ARF, CDKN2B) are tumor suppressors that regulate the cell-cycle by regulating the activity of CDK4/6 and protect P53. The loss of these genes results in activation of CDK4/6 and the RB oncogene and uncontrolled cell division. This explains why inhibitors of CDK4/6 such as palbociclib can inhibit proliferation of chordoma cell lines and xenografts efficiently^64,65 (Figure 1). An ongoing phase II clinical trial will test the efficacy of palbociclib in chordoma (NCT03110744).

In a recent study of chordomas from patients who were refractory to MTTs, whole genome and exome sequencing identified large structural rearrangements in the genome. These were present with bi-allelic inactivation of DNA repair genes such as BRCA2, NBN, and CHEK2, and genome-wide mutational signatures and genomic instability which together pointed towards defective homologous recombination (HR) in these chordomas¹⁴. In one case of recurrent disease, treatment with poly ADP ribose polymerase (PARP) inhibitor olaparib was attempted. PARP1 is required for repair of single-stranded nicks in the genome and HR-deficient cells are particularly dependent on this pathway for maintaining genome integrity, making them vulnerable to PARP inhibition (Figure 1). In this case, olaparib led to disease stability and clinical improvement in the patient’s spinal cord symptomatology for 5 months. However, after 10 months on the drug, a PARP1 resistance mutation developed which restored its enzymatic activity even in the presence of olaparib. While it is unclear whether the defects in HR repair is a feature of chordomas generally, or a consequence of repeated radiotherapy regimens that are used in advanced chordomas such as those being studied here, this case report presents a new clinically actionable vulnerability for future MTT study.

Overall, the body of recent work identifying molecular vulnerabilities has yielded encouraging early results, however this remains a work in progress and requires additional validation, drug delivery paradigms, and potentially additional discovery.

The Chordoma Immune Microenvironment

Despite chordoma’s resistance to traditional chemotherapy and radiation, immunotherapy is a promising avenue in its treatment. The rationale for this includes immune microenvironment characterization studies demonstrating the presence of lymphocytes and macrophages, and the expression of PD-1/PD-L1 pathway proteins in both primary and recurrent chordomas^66–68. PD-L1 is constitutively expressed at low levels in chordoma cell lines and immunohistochemically evident, to a variable degree, in 95% of 78 tissue samples, and in some 43% to a high degree)^66,68. Additionally, in one study PD-L1 expression was shown to be higher on macrophages than on tumor cells, suggesting modification of the tumor microenvironment may make therapeutic sense⁶⁸.

Preclinical Studies in Immunotherapy of Chordomas

Emerging preclinical data have shown some promising results. A number of studies have begun exploring the utilization of antibody dependent cellular cytotoxicity (ADCC) in chordoma immunotherapy using monoclonal antibodies (mAb). Fujii et al. showed significantly increased ADCC lysis in chordoma cell lines co-cultured with natural killer cells in the setting of avelumab (anti-PD-L1 mAb) and cetuximab (anti-EGFR mAb)^67,69. In addition, PD-L1 upregulation further increased ADCC lysis of tumor cells. These results suggest that avelumab may be useful as monotherapy and/or in conjunction with other immune therapies capable of PD-L1 upregulation⁶⁷. Fujii et al. also demonstrated enhancement of ADCC lysis with irradiated high-affinity natural killer (haNK) effector cells, which express a high affinity variant of CD16⁶⁹. haNK cells are currently being investigated in multiple clinical trials for various cancer types (NCT03387085, NCT03853317, NCT03387111, NCT03586869). Finally, olaparib (described above) increased the sensitivity of chordoma cells to ADCC by natural killer cells with cetuximab and avelumab in vitro⁷⁰. Olaparib has been approved for BRCA mutant ovarian and breast cancer. This and other agents may have roles in future ADCC-based combination therapies.

Clinical Studies in Immunotherapy of Chordomas

Early anecdotal clinical experience with immunotherapy for chordomas has demonstrated some promising results. Migliorini et al. reported the results of a series of three patients with previously treated and relapsing chordoma who received immunotherapeutic agents. They reported 6 months of response with pembrolizumab in the first patient, 9 months of response with nivolumab (both are anti-PD-1 mAbs) in the second patient, and 19 months of response with the MVX-ONCO-1 vaccine (irradiated autologous tumor cells and allogeneic GM-CSF producing cells) in the third patient with radiographic improvement of the tumor burden and symptomatic relief of the patients’ symptoms in all 3 cases⁷¹.

Initial results from completed clinical trials also show encouraging results. Brachyury, the key regulator described above, has itself emerged as a promising target for immune therapies given its tumor cell specificity with limited normal-tissue expression, and immunogenicity^{38,39,72–75}. A phase 1 trial evaluating a yeast-brachyury vaccine, GI-6301, in multiple tumor types demonstrated enhanced brachyury-specific CD4 and/or CD8 T cell responses in two of three chordoma patients in the highest dose cohort, and four of seven patients in the second highest dose cohort⁷⁶. Importantly, some disease control was demonstrated in two chordoma patients (one partial response and one mixed response). Both clinical responders had received pre-vaccination radiation (each approximately 3 months before enrollment). This, in conjunction with preclinical evidence that radiation modulates the tumor microenvironment⁷⁷, served as the rationale for a phase 2 study assessing clinical response in patients randomized to receive radiation alone or radiation with the yeast-brachyury vaccine (NCT02383498).

Another phase 1 trial assessed a different brachyury vaccine built on the Modified Vaccinia Ankara (MVA) vector, and coding for the expression of brachyury plus three co-stimulatory molecules (B7.1, ICAM-1, and LFA-3)⁷⁸. The assessed outcomes were safety and the generation of brachyury-specific CD8⁺ and CD4⁺ T-cell responses. Six of eleven chordoma patients exhibited some brachyury-specific T-cell response before treatment. Five of these six patients saw a heightened brachyury-specific T-cell response post-vaccination, and all five patients without baseline response developed a response post-vaccination. Disappointingly, most patients did not maintain a brachyury-specific immune response. Thus, two more clinical trials (NCT03595228 and NCT03349983) have been planned to assess brachyury-specific T-cell response when the vaccine is followed by a fowlpox vector booster. Importantly, both the yeast-brachyury and MVA-brachyury vaccines seem to be well tolerated, with no dose-limiting toxicities seen in either trial^76,78.

There are additionally a number of ongoing trials evaluating the effect of immune checkpoint blockade as monotherapies and as strategic combination therapies (Table 1). NCT02989636 is a phase I trial that aims to assess the use of nivolumab in conjunction with stereotactic radiosurgery for treatment of recurrent or advanced chordomas. NCT03623854 is a phase II trial assessing the effect of a combination of anti-LAG-3 and anti-PD-1 mAbs for patients with advanced chordoma.

Additional targeted therapies that act through immunomodulatory mechanisms have been explored. For example, the EZH2 inhibitor tazemetostat (described above) suppresses many immune cell functions, downregulates chemokines, and prevents NK cell maturation. A patient with SMARCB1/INI1-negative, poorly differentiated chordoma treated with tazemetostat had a durable response lasting over 2 years, and had a significant increase of CD4 and CD8 T cell populations⁵⁶. There are ongoing adult and pediatric basket trials including chordoma patients for tazemetostat (Table 1).

In summary, there are encouraging early data supporting a role for immunotherapy in the treatment of chordoma, but these are not yet validated in efficacy trials against standard therapies. Lessons learned from other tumor types, for example melanoma and lung cancers, will likely inform both discovery efforts and the design of rational clinical trials. For example, defining the unique tumor-immune interactions in chordoma, both in the primary setting and at recurrence, will likely play a key role, and biomarkers (e.g. checkpoint expression) can be anticipated to identify patients most likely to benefit from particular immunotherapies. Finally, as with other cancer types, any immunomodulatory or tumor-specific immunotherapy will likely play a part of a multimodality treatment paradigm including surgical sampling/resection and external beam radiation, at least initially.

Synopsis:

Chordomas are malignant, highly recurrent, tumors of the midline skeleton that arise from the remnants of the notochord. Although chordomas are cytogenetically heterogeneous and karyotypically diverse, mutations are frequently seen in receptor tyrosine kinase pathways, epigenetic regulators, cell cycle genes, and the transcription factor brachyury. Whereas surgicalresectionhas traditionally been considered first- line treatment for primary chordomas, high-dose conventionally fractionated proton beam and hypofractionated carbon ion and stereotactic radiosurgery have demonstrated excellent tumor control used either as adjuvant or definitive treatment. Despite advances in radiation, the development of systemic therapy is critically important to ultimately curing this tumor.Several ongoing trials are attempting to use molecular targeted therapies for mutated pathways for recurrent and advanced chordomas and have shown promise. In addition, immunotherapies including brachyury-directed vaccination and checkpoint inhibition have also been attempted with encouraging results. In this article, we will discuss the major pathways that have been implicated in the pathogenesis of chordoma with an emphasis on molecular vulnerabilities that future therapies are attempting to exploit.

Key points:

• Chordomas are malignant, highly recurrent tumors of the midline skeleton

• The molecular changes involved in chordomas include genomic rearrangements, and alterations in receptor tyrosine kinase signaling, cell cycle regulation, epigenetic modulators, and the transcription factor brachyury

• Chordoma cells are exquisitely dependent on the transcription factor brachyury for maintenance

• While surgical resection and radiation remain first line treatments for primary chordomas, molecular targeted therapy with signaling inhibitors have shown some benefit in in recurrent chordomas

• Immunotherapies including brachyury vaccination and checkpoint inhibition have shown promising results in early studies