The quest for effective immunotherapies against malignant peripheral nerve sheath tumors: Is there hope?

Abstract

Immune-based therapies represent a new paradigm in the treatment of multiple cancers, where they have helped achieve durable and safe clinical responses in a growing subset of patients. While a wealth of information is available concerning the use of these agents in treating the more common malignancies, little has been reported about the use of immunotherapies against malignant peripheral nerve sheath tumors (MPNSTs), a rare form of soft tissue sarcoma that arises from the myelin sheaths that protect peripheral nerves. Surgical resection has been the mainstay of therapy in MPNSTs, but the recurrence rate is as high as 65%, and chemotherapy is generally ineffective. The immune contexture of MPNSTs, replete with macrophages and a varying degree of T cell infiltration, presents multiple opportunities to design meaningful therapeutic interventions. While preliminary results with macrophage-targeting strategies and oncolytic viruses are promising, identifying the subset of patients that respond to immune-based strategies will be a milestone. As part of our effort to help advance the use of immunotherapy for MPNSTs, here we describe recent insights regarding the immune contexture of MPNSTs, discuss emerging immune-based strategies, and provide a brief overview of potential biomarkers of response.

Graphical abstract

In this article, Dr. Timothy Cripe and colleagues review emerging immune-based strategies against MPNST, discuss its immune contexture, and examine the potential biomarkers of response. The authors are optimistic about the prospect of leveraging immunotherapies for MPNST patients.

Introduction

Two centuries have elapsed since Von Recklinghausen first described neurofibromatosis, a group of genetic disorders that predispose afflicted individuals to the formation of tumors on their nerve tissues.¹ While these tumors are usually benign and present with mild symptoms, malignant variants also exist, such as the aptly named Malignant Peripheral Nerve Sheath Tumors (MPNSTs). While uncommon in the general population, MPNSTs are an aggressive tumor that typically portend a poor prognosis. Surgery remains the most effective means of treating MPNSTs. However, many of these tumors are unfortunately only diagnosed at an advanced stage, where they have metastasized to distant sites in the body or are otherwise unresectable.² Alternative treatment strategies are similar to those employed for other soft tissue sarcomas, involving a combination of first-line anthracycline-based drugs and cytotoxic therapies such as the alkylating agent ifosfamide and the topoisomerase II inhibitor etoposide.³ However, these therapeutic regimens tend to have limited efficacy as MPNSTs are generally insensitive to conventional chemotherapy drugs.³ Alternative therapies are clearly needed for the management of MPNSTs.

The advent of immunotherapies, which seek to leverage the immune system to mount an antitumor response, have had a profound impact on how we approach the treatment of many cancers. The approval of 17 immunotherapy products by the FDA in the past 2 decades and the ongoing development of more than 4,720 immuno-oncology drugs in clinical testing in 2020 speaks to the vast potential of these agents.⁴ Despite these promising developments, we have yet to realize the potential benefits of immunotherapy in MPNSTs.

In this review article, we will discuss available data as it pertains to the immune contexture of MPNST, highlight emerging immunotherapies being used in its treatment, and examine potential biomarkers of response.

Immune contexture of MPNST

As most cancers develop and grow, they recruit a network of infiltrating immune cells, fibroblasts, vascular cells, and lymphatic vessels. While MPNST is no different, most cases in patients with NF1 are thought to arise from the malignant transformation of benign neurofibromas, which are already populated with immune cells. These so-called microenvironmental cells have substantial roles in cancer prognosis. The characteristics of this local immune context determine the amenability of immune-based strategies to mediate a long-term effect and potentially reinstate immuno-surveillance. The clinical impact of immune contexture on survival and response to therapy is well recognized in multiple cancers, where the interplay of numerous factors influences tumor progression in a complex and ever-changing tumor microenvironment.⁵ As such, a coordinated effort to delineate the role of each immune parameter in tumor progression is highly desirable, especially in aggressive tumors like MPNSTs where conventional therapies are ineffective.⁶ In this section, we review recent evidence regarding the role of the tumor microenvironment, key immune cell infiltrates, the cytokine/chemokine milieu, and antigen processing and presentation pathways in MPNST prognosis.

Immune cell infiltrate

Although immune cell infiltration is rare in healthy peripheral nerves where MPNSTs arise, neurofibroma, the benign precursor of NF1 MPNSTs, is enriched with immune cells. Key infiltrates include mast cells and macrophages.⁷ NF1 mutant neurofibroma resembles the microenvironment induced in peripheral nerves following injury in that there is a similar mixture of immune cells and Schwann cells that are de-differentiated into a progenitor-like state.⁸ De-differentiated Schwann cells also produce cytokines and chemokines in greater quantity than their differentiated counterparts. These include factors such as stem cell factor and colony-stimulating factor-1, which recruit mast cells and macrophages. Macrophage recruitment is further sustained through signaling of the CXCL10/CXCR3 pathway, which also contributes to the infiltration of T cells and dendritic cells. The resulting inflammatory state is thought to be a key driver for the initiation of neurofibroma (refer to Fletcher et al.⁷ for a detailed review) and potentially the development of MPNST. Although considerable progress has been made in our understanding of the neurofibroma tumor microenvironment, our understanding of the MPNST microenvironment is limited. Bioinformatic deconvolution algorithms from gene expression studies have predicted the microenvironment of MPNSTs to contain activated mast cells, protumor M2-like macrophages, resting CD4+ memory T cells, and cancer-associated fibroblasts.⁹ We discuss some recent findings that have been pivotal in shaping our understanding of immune infiltrates in MPNSTs below.

Macrophages

Tumor-associated macrophages (TAMs) are the chief regulators of the tumor microenvironment and are involved in a wide array of processes, including tumorigenesis, extracellular matrix remodeling, and response to therapy.¹⁰ They are divided into two simplified subgroups: pro-inflammatory M1-like macrophages, which have a tumor-suppressing effect, and anti-inflammatory M2-like macrophages, which have a tumor-promoting role.¹¹ While the direction of polarization is largely contextual and based on the cytokine milieu, high TAM infiltration generally correlates with a poor prognosis across a variety of cancer types.¹² Dancsok et al. found that the counts of CD68+ M1-like macrophages were, on average, three times higher than tumor-infiltrating lymphocytes in MPNST through a tissue microarray data of 77 samples.¹³ CD163+ M2 macrophage counts were almost six times higher than CD68+ M1 macrophage counts in their study. Although tissue microarray is limited in its ability to consider intratumoral heterogeneity, high M2-like macrophage infiltration suggests a highly immunosuppressive immune microenvironment in MPNSTs. These macrophages also harbor a distinct population enriched for the expression of the SPP1/osteopontin gene, a secretory glycoprotein that can bind and activate matrix metalloproteinases and promote tumorigenesis.¹⁴ Such findings may present an opportunity to leverage antitumor immune responses by exploiting the dynamic nature of macrophage polarization.

Mast cells

The role of mast cells continues to remain an enigma in MPNSTs. A recent study by Dodd et al. analyzed c-kit staining in a tissue microarray of 67 human MPNSTs and observed that mast cells were enriched in human NF1-associated MPNSTs compared to sporadic patients.¹⁵ NF1-associated MPNSTs are reported to have worse outcomes than sporadic cases.^16,17 The authors also tested if the response to the conventional chemotherapy regimen of doxorubicin/ifosfamide is altered in primary murine tumors enriched for mast cell infiltration. Two murine models that vary exclusively in intratumoral mast cell density, one with an NF1 haploinsufficient stroma and the other with an NF1 null stroma, had similar responses to treatment even though the model with the NF1 haploinsufficient stroma had higher mast cell infiltration and developed tumors at an early age. This result suggests that mast cells do not alter the response of MPNSTs to chemotherapy, but how or if mast cells perturb the response of immunotherapeutic platforms remains an open question. The authors did not test for any correlation between mast cell numbers/density with patient survival. Their findings stand in contrast to a report from Vasconcelos et al., who examined tissue from 29 MPNST patients and found a similar density and distribution of mast cells between NF1-associated and sporadic MPNST.¹⁸ More importantly, the authors showed that mast cell density did not correlate with patient survival in this study. Altogether, although implicated in tumor progression in both neurofibroma and MPNST, a clear prognostic significance of mast cells in MPNST tumor progression is yet to be established.

Antigen processing and presentation pathway

We recently reported that human major histocompatibility complex (MHC) genes HLA-A/B/C are downregulated in benign and malignant NF1-associated tumors compared to normal human Schwann cells. However, we did note that immunohistochemical staining for HLA-A/B/C and B2M proteins could vary extensively.¹⁹ HLA-A/B/C scores also correlated strongly with CD8+ cellular infiltrates. In general, we observed that MPNST had higher average HLA-A/B/C scores than plexiform neurofibromas.¹⁹ Along these lines, two separate studies by Lee and colleagues showed that the MPNST-derived cell line T265 displayed significant downregulation of MHC genes compared to normal human Schwann cells.^20,21 These authors also probed transcription factors and chaperone proteins to identify potential mechanisms involved in the downregulation of MHC genes. The expression of MHC class II is regulated by a co-activator that serves as a single regulatory factor called the class II transactivator (CTIIA or MHC2TA).²² MHC2TA transcript levels were significantly lower in T265 compared to normal human Schwann cells. The transcript levels of CD74, a chaperone protein involved in the processing and transport of MHC class II, were also found to be reduced approximately 1,000-fold in MPNST cells. Alternatively spliced RNA transcripts of CD74 were also observed in T265, unlike normal human Schwann cells, as was decreased expression of transporter-activator protein, a protein that loads peptide antigens onto MHC class I molecules. These results highlight at least two cellular mechanisms, aberrant production of transcription factor and chaperone proteins, that contribute to the downregulation of genes involved in antigen processing and presentation in MPNSTs.²¹ A multitude of other mechanisms could also be involved in the suppression of genes related to antigen processing and presentation, including but not limited to loss of epigenetic regulation, aberrant expression of transcription factors, genomic instability, or mutation. Recent evidence also suggests that loss of PRC2, a histone-modifying complex primarily involved in maintaining transcriptional silencing, is associated with decreased antigen presentation and immune signature in MPNSTs.^23,24 PRC2 loss was followed by a global loss of repressive histone marks, primarily H3K27me3, but an overall increase in global DNA methylation. This epigenetic deregulation was found to sensitize MPNSTs to NSD2, a methyltransferase inhibitor, that restored MHC expression.²³ In light of these findings, MPNSTs retaining PRC2, and consequently H3K27me3, may be more amenable to immune-based therapeutic strategies, due to their preserved immune characteristics compared to tumors with PRC2 loss and H3K27me3 loss.²⁴

T cells

In 2009, Marx et al. observed that neonatal thymectomy significantly increased the risk of MPNST development in resistant rat strains. This report was the earliest work that suggested that T cells can contribute to resistance to MPNST development.²⁵ In two subsequent studies, T cell infiltration was observed in almost half of the patient samples tested through immunohistochemistry, although the extent of infiltration varied extensively.^19,26 Low-grade tumor samples harbor higher CD4+ and CD8+ T cells than high-grade MPNST samples, while FOXP3+ regulatory T cell numbers increase with tumor progression.¹⁹ There are multiple case reports of a complete response with T cell checkpoint inhibitors in MPNSTs,^27,28,29 which will be described in detail below. These results suggest that leveraging T cells might be an attractive therapeutic strategy against MPNSTs. Unfortunately, these strategies have not been tested in clinical trials.

Cancer-associated fibroblasts (CAFs)

CAFs are a heterogeneous population of fibroblast cells of mesenchymal lineage actively involved in tumorigenesis through the production of extracellular matrix proteins, cytokines, and growth factors.^30,31 Single-cell RNA sequencing of murine models genetically engineered to develop MPNSTs through Lats1/2 deficiency showed that neurofibroma tissues retain the fibroblast population found in the epineurium and perineurium of healthy peripheral nerves.¹⁴ However, both of these subpopulations of fibroblasts were notably reduced or absent in MPNSTs. Instead, advanced-stage MPNSTs harbor a different population of activated tumor-associated fibroblasts with a distinct transcriptional signature.¹⁴ These results suggest extensive disruption of tissue architecture in peripheral nerves with tumor progression. Along these lines, histologically, MPNSTs are known to have reduced or negative staining for the CD34+ fibroblast population, unlike neurofibroma.^32,33 However, the effect of activated CAFs on the prognosis of MPNST patients, if any, has not been determined.

Role of haploinsufficient tumor microenvironment in immune cell infiltration

In physiological context, NF1 MPNSTs arise in the NF1 heterozygous background.³⁴ Although sporadic cases of MPNSTs have demonstrated that NF1 haploinsufficient stroma is not a necessity for the formation of malignant tumors, the role of the heterozygous microenvironment in shaping the immune contexture cannot be ignored. It is important to note that patients with NF1-associated MPNSTs appear to have worse outcomes than patients with sporadic MPNSTs.¹⁶ Neurofibromin, the protein product of the NF1 gene, is ubiquitously expressed and regulates multiple cell signaling pathways, including the Ras-MEK-ERK and PI3K/mTOR pathways.³⁵ This fact formed the groundwork for the hypothesis that NF1 haploinsufficiency leads to altered activity of cells in the stroma, possibly creating a permissive microenvironment for tumor growth. Along these lines, Dodd et al. observed a greater than 2-fold infiltration of CD45+ immune cells in the haploinsufficient murine tumor microenvironment compared to wild-type stroma.¹⁵ They further characterized CD45+ immune cell infiltrate in a bone marrow transplant model, where mice with NF1 null Schwann cells received either wild-type or haploinsufficient bone marrow. The authors observed a higher percentage of Cd11b+Cd11c-monocytes in mice that received a haploinsufficient tumor microenvironment, but the significance of these monocytes in MPNST is not characterized.¹⁵ Taken together, these studies implicate the haploinsufficient microenvironment for increased infiltration of myeloid cells and possibly worse prognosis than sporadic cases in MPNSTs. Along these lines, similar results regarding the role of the NF1 heterozygous tumor microenvironment have also been observed in neurofibroma, the benign precursor of MPNST. NF1 haploinsufficient bone-marrow-derived cells were sufficient for neurofibroma progression in the context of Schwann cell NF1 deficiency.³⁶ Although these studies implicated mast cells as critical mediator of tumor initiation in neurofibroma and MPNST, a direct role of mast cells in tumor progression is yet to be established.

Current immunotherapies in MPNST

There are several case reports as well as ongoing clinical trials (Table 1) of immunotherapies involving patients with MPNST, mostly in the realms of checkpoint inhibition, macrophage targeting, and oncolytic viruses (Figure 1).

Table 1.

List of MPNST immunotherapy clinical trials that are currently active and recruiting

No.	Study name	Phase	Agent	Modality	Identifier
1	PLX3397 Plus Sirolimus in Unresectable Sarcoma and Malignant Peripheral Nerve Sheath Tumors (PLX3397)	I	PLX3397	small-molecule inhibitors	NCT02584647
2	Vaccine Therapy in Treating Patients With Malignant Peripheral Nerve Sheath Tumor That Is Recurrent or Cannot Be Removed by Surgery	I	Edmonston strain measles virus genetically engineered to express neurofibromatosis type 1 (oncolytic measles virus encoding thyroidal sodium-iodide symporter [MV-NIS])	oncolytic virus	NCT02700230
3	Neoadjuvant Nivolumab Plus Ipilimumab for Newly Diagnosed Malignant Peripheral Nerve Sheath Tumor	I	nivolumab and ipilimumab	immune modulators (PD-1 and CTLA-4)	NCT04465643
4	A Study of APG-115 in Combination With Pembrolizumab in Patients With Metastatic Melanomas or Advanced Solid Tumors	Ib/II	APG-115 and pembrolizumab	immune modulator (PD-1) and chemotherapy	NCT03611868
5	B7H3 CAR-T Cell Immunotherapy for Recurrent/Refractory Solid Tumors in Children and Young Adults	I	Biological: second generation 4-1BBζ B7H3-EGFRt-DHFR (B7H3-specific CAR-T cells and CD19 specific CAR-T cells) biological: second generation 4-1BBζ B7H3-EGFRt-DHFR (selected) and a second generation 4-1BBζ CD19-Her2tG	CAR-T cells	NCT04483778
6	EGFR806 CAR-T Cell Immunotherapy for Recurrent/Refractory Solid Tumors in Children and Young Adults	I	Biological: second generation 4-1BBζ EGFR806-EGFRt biological: second generation 4-1BBζ EGFR806-EGFRt and a second generation 4 1BBζ CD19-Her2tG	CAR-T cells	NCT03618381
7	Donor Stem Cell Transplant After Chemotherapy for the Treatment of Recurrent or Refractory High-Risk Solid Tumors in Pediatric and Adolescent-Young Adults	II	allogeneic hematopoietic stem cell transplantation	stem cell transfer	NCT04530487
8	Nivolumab and Ipilimumab in Treating Patients With Rare Tumors	II	nivolumab and ipilimumab	immune modulators (PD-1 and CTLA-4)	NCT02834013

Figure 1.

Emerging immunomodulatory therapies in malignant peripheral nerve sheath tumors

Key modalities that hold promise for activating the antitumor immune response as a therapy include immune checkpoint blockade, oncolytic viruses, targeting tumor-associated macrophages and the use of epigenetic drugs to enhance MHC Class I expression.

Immune checkpoint blockade

T cells that have infiltrated tumors progressively lose their effector function due to chronic antigen stimulation, leading to an exhausted state. This exhaustion is induced and maintained by signals initiated through membrane receptors that act as a gatekeeper of the immune response, which are commonly known as “immune checkpoints.” Several immune checkpoints that negatively regulate T cell response have been identified, including PD-1, CTLA-4, LAG3, TIM3, and TIGIT. Upregulation of checkpoint ligands such as PD-L1 is common among MPNSTs, providing a strong rationale to explore the possible benefit of interrupting PD-1 and/or PD-L1 function in these cancers.^37,38 PD-L1, expressed on tumors or on antigen-presenting cells, can bind to PD-1 on T cells and inhibit its activation, thereby playing a key role in the induction of immune tolerance in the tumor microenvironment. A growing body of clinical data illustrate the considerable potential of immune checkpoint blockade in MPNST, particularly those agents designed to inhibit the PD-1/PD-L1 signaling axis. MPNST patients who were refractory to at least two previous forms of conventional therapy have been subsequently treated with the FDA-approved anti-PD-1 monoclonal antibodies pembrolizumab or nivolumab.^27,28,29 Although only four individual case studies have been reported, each patient remarkably went on to achieve a complete response; however, one patient included in the phase I trial of pembrolizumab experienced relapse.³⁹ Details regarding tumor location, previous treatment, PD-L1 expression status, treatment arm, and outcome of these cases are documented in Table 2. While these initial case studies represent a promising start for the inclusion of checkpoint inhibitors in MPNST treatment, a more thorough vetting of their potential, however warranted, may be elusive because of the relative rarity of this disease.³⁹

Table 2.

Published case reports of malignant peripheral nerve sheath tumors treated with PD-1 immune checkpoint inhibition

No	Age/sex	Tumor location (primary and metastatic)	Previous treatment	PD-L1 expression status	Genetic alterations	Immunotherapeutic intervention	Treatment arm	Outcome	Reference
1	60 yr/male	primary paravertebral tumor at T7-T8 with lung metastases	•surgical resection •epirubicin and ifosfamide	70% (2+) through immunohistochemistry	ARID1A, CDKN2A, KMT2A, NF1 and TP53 through next-generation sequencing	pembrolizumab	intravenous; 21-day cycle; 6 cycles •200 mg for cycle 1–2; •400 mg from cycle 4–6	Complete remission	Larson et al.27
2	48 yr/male	primary retroperitoneal mass with mesenteric metastases	•surgical resection •doxorubicin and ifosfamide •imatinib •eribulin	90% tumor proportion score (TPS) through immunohistochemistry (DAKO 22C3 pharm DX kit)	not available	pembrolizumab (in combination with procarbazine)	6 cycles of pembrolizumab (200 mg/3 wks/cycle) in combination with procarbazine hydrochloride (50 mg/m2 twice a day)	Complete response	Payandeh et al.28
3	45 yr/male	primary left calf nodule (peroneal nerve) with metastases in the lung and pleura	•surgical resection •doxorubicin •ifosfamide	almost 100% through immunohistochemistry	MED12, TP53, NF1, PLCG1 AND EP300 CD274/PD-L1 amplification	nivolumab (in combination with radiation therapy)	•nivolumab 3 mg/kg/2 weeks •radiotherapy to bilateral anterior pleural metastases 39 Gy (13 cycles at 3 Gy)	Complete response	Özdemir et al.29
4	22 yr/male	primary head and neck of femur with metastasis in the lungs and pelvic lymph node	•surgical resection •radiotherapy	5% (+2) of cells stained positive (Ventana PD-L1 (SP142) assay)	CDK6 copy number amplification (+1) revealed through analysis of circulating tumor DNA prior to therapy	pembrolizumab	intravenous; 21 cycles •200 mg/3 weeks	Complete metabolic response	Davis et al.85

Targeting Tumor-associated macrophages

Macrophages are the predominant immune cells found in the tumor microenvironment across nearly all sarcoma types.¹³ Macrophages are found at a high frequency in MPNSTs with the pro-tumor M2-like TAMs being the predominant phenotype.^13,40 There are multiple strategies being explored for targeting TAMs, including depletion of their total population, inducing their polarization to an antitumor phenotype, or suppressing the production of macrophage-derived factors involved in survival/proliferation. Much of the work on macrophage targeting in MPNSTs has been centered on pexidartinib (trade name TURALIO^TM), a small-molecule tyrosine kinase inhibitor of colony-stimulating factor 1 (CSF1) receptor signaling.⁴¹ CSF1 (also known as M-CSF) is a key cytokine that regulates the differentiation, proliferation, and survival of monocytes/macrophages. Under homeostatic conditions, CSF1 is mainly involved in replenishing tissue macrophages.⁴² In the context of cancer, CSF1 is required for the accumulation of TAMs.⁴³ CSF1 targeting is particularly interesting in cancer immunotherapy due to its potential to reduce CD4+FoxP3+ T cells, possibly through its inhibition of myeloid cells, leading to the potentiation of the T cell response.⁴⁴

Prada et al. targeted macrophages in the benign precursor of MPNST, neurofibroma, using pexidartinib.⁴⁰ When they treated genetically engineered murine models with this small-molecule inhibitor before the formation of neurofibroma, tumor growth was accelerated. In contrast, when they treated mice after the formation of tumors, tumor regression was observed in a subset of mice.⁴⁰ These results are consistent with the dynamic role of macrophages in the tumor microenvironment where they shift from an antitumor to a protumor phenotype as tumor growth progresses.

When mice bearing MPNST xenografts were treated with pexidartinib, it effectively decreased the TAM population and suppressed tumor growth through inhibition of CSF1R, c-KIT, and PDGFRβ.⁴⁵ Both macrophage depletion and reduction in tumor volume could be enhanced by combining pexidartinib with sirolimus, an inhibitor of PI3K/Akt pathway critical for tumor growth and survival. Sirolimus inhibits a downstream component of mTOR that regulates protein synthesis and turnover.⁴⁶ Pexidartinib can also inhibit MPNST cell proliferation in vitro. This inhibition was associated with the reduced hyperphosphorylated form of retinoblastoma protein and cyclin D1, inducing a cell-cycle arrest in these cells.⁴⁵ These results prompted a phase I clinical trial of pexidartinib and sirolimus in 18 advanced sarcoma patients, six of whom had MPNST. Sirolimus (2, 4, or 6 mg) was combined with pexidartinib (600, 800, or 1,000 mg) on a 28-day cycle orally in the trial.⁴⁷ Dose-limiting toxicity was observed in five out of 18 patients. In addition, three patients had a partial response, while nine had stable disease for over 18 weeks, including three MPNST patients. The combination also decreased the “M2-like” macrophages, suggesting the potential polarization of TAMs to an antitumor phenotype.

Initial results with pexidartinib have renewed interest in targeting macrophages in MPNSTs. Given that macrophages outnumber tumor-infiltrating lymphocytes and can interfere with standard-of-care therapies, there is a growing need to explore other macrophage-targeting strategies against this aggressive tumor.¹³ Along these lines, MPNSTs were found to have a low-level expression of macrophage-associated immune checkpoint CD47/SIRPα.¹³ CD47 is expressed ubiquitously on normal cells as a membrane protein and inhibits the phagocytosis of macrophages when it engages with SIRPα on its surface.⁴⁸ Cancer cells are adept at co-opting the CD47-SIRPα axis to evade antitumor activity, and high expression of CD47 correlates with worse prognosis in multiple solid tumors.^49,50,51 A strategic approach to harness the therapeutic benefit of targeting macrophages holds much promise for MPNST. Future studies should focus on the rational design of macrophage inhibitors in combination with other immunotherapeutic regimens in preclinical and clinical settings.

Oncolytic virotherapy

Oncolytic viruses (OVs) are either non-pathogenic wild-type viruses or genetically modified attenuated pathogenic viruses that selectively kill cancer cells directly through lysis or indirectly by stimulating an antitumor immune response. As of this writing, only one clinical trial of OVs in peripheral nerve sheath tumors has been reported.⁵² The oncolytic adenovirus ONYX-015 (Dl1520) is a chimeric Ad2/Ad5 virus featuring a deletion of E1B-55kD, a gene encoding a p53 inhibitory protein, to ensure selectivity for cancer cells that have lost p53 function.⁵³ A phase I/II trial of ONYX-015 in combination with MAP (mitomycin-C, doxorubicin, cisplatin) chemotherapy included one metastatic MPNST patient out of a total of six patients with advanced sarcomas.⁵² This MPNST patient was treated intratumorally at the highest dose level of 10¹⁰ plaque-forming units (PFU)/dose without any dose-limiting toxicity. The tumor nodule directly injected with ONYX-015 decreased in size by nearly 70%. Likewise, two non-injected small lesions shrank by 50%–75% of their initial size, culminating in a partial response that lasted for a total of 11 months.

The first-in-class and only oncolytic virotherapy approved by the FDA is talimogene laheraprepvec (a.k.a T-VEC or Imlygic) for intralesional injection of patients with melanoma. T-VEC is attenuated from wild-type herpes simplex virus type 1 (HSV1) by deletion of both copies of the gene RL1 that encodes for neurovirulence factor ICP34.5. It also has a deletion of ICP47, which during wild-type virus replication, reduces antigen presentation in infected human cells through inhibition of peptide loading into MHC class I. As a consequence of that deletion, the US11 gene is activated earlier in the virus life cycle, partly ameliorating the reduced replication that results from RL1 deletion. Apart from these modifications, the virus is engineered to express an immunostimulant cytokine GM-CSF.⁵⁴ During its seminal phase III clinical trial in melanoma, a durable response rate of 16.3% was achieved in T-VEC-treated patients, compared to 2% in patients who received only GM-CSF.⁵⁵ T-VEC treated patients also exhibited a complete response rate of 10.8%.⁵⁵ The effect of T-VEC and other OVs is not thought to be histology specific, suggesting that OVs may be useful in other cancers as well.

Preclinical testing of different OVs, including but not limited to oncolytic herpes simplex virus (oHSV), adenovirus (Ad), and measles virus (MV), has continued to demonstrate their therapeutic potential.⁵⁶ HSVs are an attractive candidate against MPNSTs due to their natural tropism for tissues in the peripheral and central nervous system. Moreover, this double-stranded DNA virus can carry a large transgene payload since 30 kb out of its large 152-kb genome is non-essential for viral infection.⁵⁷ One of the common genetic modifications used in the generation of oHSV constructs is deletion of the neurovirulence gene, ICP34.5, which encodes an inhibitory protein of the protein kinase R (PKR) pathway. PKR normally acts as an innate antiviral defense mechanism of the host cell by shutting down protein translation in virus-infected cells through phosphorylation of translation initiation factor eIF2α. The ICP34.5 protein counteracts this process by recruiting protein phosphatase 1 to dephosphorylate eIF2α, thereby restoring translation of viral proteins.⁵⁸ Most tumors have defective PKR defense pathways; thus deletion of ICP34.5 confers selectivity to oHSV for replication in cancer cells.

Aberrant activation of Ras signaling, commonly observed in MPNST, can inhibit the PKR pathway.⁵⁹ Although this might suggest that Ras activity could serve as a barometer for MPNST cell permissivity to oHSV infection, a recent study of human MPNST cell lines showed that these cells were highly sensitive to oncolysis irrespective of Ras activation status.⁶⁰ Surprisingly, virus-infected MPNST cell lines had low levels of phosphorylated eIF2α even in the absence of functional ICP34.5, likely due to dephosphorylation through other mechanisms involving cellular and viral phosphatases. This result was in contrast with permissivity of mouse MPNST cells, which were dependent on Ras activation status upon infection with oHSV G207.⁶¹ In subsequent studies, attempts were made to determine if oHSV efficacy could be enhanced by increasing the expression of the HSV entry receptor nectin-1, but despite an increase in cell-to-cell spread, there was no increase in virus yield.⁶² Safety studies with the oHSV G47Δ, which is attenuated through deletion of the ribonucleotide reductase gene ICP6 along with ICP47 and ICP34.5, were also conducted by injecting the virus directly into the sciatic nerves of athymic nude mice.⁶³ No ultrastructural abnormalities, including axonal loss, demyelination, or axonal degeneration, were observed other than focal needle track damage, which was also seen in saline injections. A separate study evaluating sciatic nerve injection of oHSV NV1023 similarly did not alter neural function in nude mice.⁶⁴

Following these results, multiple groups have generated therapeutic transgene-armed versions of OV to enhance antitumor activity in preclinical nerve sheath tumor models. G47Δ armed with an antiangiogenic factor, platelet factor 4 (PF4), shows enhanced potency in reducing tumor growth and inhibited endothelial cell migration compared to an unarmed control virus in MPNSTs.⁶⁵ An oHSV armed by insertion of TIMP3 (an inhibitor of matrix metalloproteinase-3) similarly enhanced antitumor activity by inducing morphological changes in the extracellular matrix, including marked decreases in bone-marrow-derived endothelial progenitor cells and reduction of neovascularization.⁶⁶

Combining OVs with other established therapies may also be a promising strategy for treating MPNST. Hyperactive EGFR signaling is a hallmark for many MPNSTs and other cancers, providing a potential target for therapeutic intervention. While the combination of oHSV G207 with the EGFR inhibitor erlotinib was found to be more effective than either agent acting alone in MPNST cell cultures, these findings unfortunately did not translate to xenograft studies performed in athymic nude mice.⁶⁷ Enhanced basal interferon-stimulated gene (ISG) expression was shown to contribute to resistance to Δγ₁34.5 oncolytic HSV in a subset of cell lines.⁶⁸ oHSV infection promoted activation of JAK/STAT1 pathway, upregulated ISGs, and reduced productive infection as well as virus spread in resistant cell lines. Pretreatment with a small-molecule JAK inhibitor ruxolitinib downregulated ISG expression, improved viral titer, and enhanced CD8 T cell activation in the tumor microenvironment in syngeneic mouse models.⁶⁸ Along these lines, the role of a pattern recognition receptor, stimulator of interferon genes (STING), that integrates DNA sensing activity and upregulates ISGs was explored in MPNST. STING knockdown did not have any effect on basal ISG upregulation in resistant cell lines, suggesting that STING is dispensable for basal ISG upregulation.⁶⁹ STING downregulation, however, did enhance oHSV replication and spread in these resistant cell lines.

Apart from oncolytic HSV, RNA viruses including MV and vesicular stomatitis virus (VSV) have also been tested for antitumor effects in MPNST. MV-NIS, an Edmonston vaccine strain of measles virus encoding the sodium-iodide symporter for non-invasive imaging and targeted radionuclide therapy, was found to significantly shrink tumor growth in MPNST-derived xenografts of ST88-14 and S462TY after a single intratumoral injection.⁷⁰ Similarly, the VSV-G/GFP virus, which was modified to express a fusion of native G protein and GFP for easy visualization, was oncolytically enhanced through a positive selection process involving multiple passages in S462TY and STS26T cells in vitro. This resulting VSV, termed VSV-rp30a, showed enhanced infectivity and killing potential in both of these cell lines compared to the parental VSV-G/GFP (refer to Antoszczyk and Rabkin⁵⁶ for a more comprehensive review of oncolytic virotherapy in MPNST).⁷¹

Preclinical work has laid much of the groundwork for the design of potential clinical studies of OV in MPNSTs. The limited number of tumor models tested and rarity of the cancer occurrence continue to remain as challenges for the successful clinical translation of this promising immunotherapeutic modality.

Biomarkers of response

The success of checkpoint inhibitors in the clinic marks an important milestone in our pursuit of developing immune-based strategies against cancer. Although the use of checkpoint inhibitors has led to improved outcomes across a number of cancer types, the objective response rate in soft tissue sarcomas has been reported to be only 14%.⁷² Thus, it is of great importance that we be able to identify biomarkers that distinguish the subset of the population who potentially respond and benefit from immunotherapeutic interventions. In this regard, key potential biomarkers that could predict response to immune-based strategies in MPNST include PD-L1 expression, tumor mutational burden, and the presence and status of tumor-infiltrating immune cells.

PD-L1 expression

Shurell et al. created a peripheral nerve sheath tumor microarray using 53 MPNST specimens and stained for PD-L1, PD1, and CD8 through immunohistochemistry.²⁶ 13% of the MPNST samples had at least 5% PD-L1 staining. On the other hand, 57% of the MPNST samples had at least 5% CD8 T cell infiltration in the microenvironment. No PD1 expression was detected in their study. Overall, MPNSTs had low T cell infiltration, prompting the authors to characterize the MPNST microenvironment as a non-inflamed (“cold”) phenotype. Of note in this study, no correlation was observed between PD-L1 expression or CD8+ infiltration with disease-specific or disease-free survival in primary MPNST patients. MPNSTs had more PD-L1 expression compared to normal nerves and benign lesions. This was in contrast with one of our prior studies in which we did not detect a significant difference in average PD-L1 expression in MPNSTs versus benign tumor samples.¹⁹ Considering that MPNST samples have substantial tumor heterogeneity, these inconsistent results reported in the literature further add complexity to the development of predictive biomarkers of immune checkpoint inhibition.

The presence of soluble PD-L1 in the sera of patients may also have clinical utility as a biomarker of tumor progression from plexiform neurofibroma to MPNST. PD-L1 was found to be significantly elevated in the sera of MPNST patients compared to plexiform neurofibroma.³⁸ When NF1 patients were stratified into two cohorts, NF1 patients with MPNST and without MPNST, those with MPNST had higher PD-L1 expression in their sera compared to the ones without NF1.³⁸ It is unclear, however, if soluble PD-L1 might be a detriment to response by acting as a sink for antibodies that reduce their access to tumor sites.

Tumor-infiltrating immune cells

Tumor-infiltrating immune cells are an important biomarker for immunotherapy due to their predictive and prognostic significance in multiple solid malignancies.^73,74 Cytotoxic T lymphocytes, particularly CD8 T cells, are positively correlated with patient survival.⁷³ CD4 T cells, which include helper T cells and regulatory T cells (Tregs), can have starkly different effects in the tumor microenvironment. Tregs subvert cytotoxic T cell function, while helper T cells either aid the function of CD8 effector T cells or directly eliminate cancer cells.⁷⁵ NK cells are innate lymphoid cells that are generally cytotoxic to tumor cells, depending on the balance of expression of activating and suppressing receptors and their ligands, and they function without MHC specificity. Myeloid cells are classified into two subtypes: granulocytes and mononuclear phagocytes. Monocytes, macrophages, and dendritic cells together form the mononuclear phagocytes and carry two distinct functions: phagocytosis and antigen presentation to T cells. Macrophage infiltration is associated with a worse prognosis in many cancers.⁷⁶ However, it is important to remember that these stereotypical roles can vary with cancer types as well as histological subtypes, necessitating extensive characterization in individual tumor types, more so in highly heterogeneous tumors like MPNSTs. Low-grade MPNST samples have a higher cytotoxic CD4 and CD8 lymphocyte infiltration than high-grade tumors.³⁸ In addition, a decrease in effector T cell fractions (CD8+/CD57+ and CD8+CD7–) in the peripheral blood also corresponds to increasing tumor load in NF1 patients.⁷⁷ However, CD8 expression did not correlate with disease-free survival and disease-specific survival in a small cohort of 33 MPNST patients, highlighting the significant heterogeneity of these tumor types. Along these lines, macrophage infiltration correlated with disease severity in genetically engineered murine models of MPNST.⁴⁰ Thus, although attractive, a lack of prospective and retrospective trials has stymied the potential of tumor-infiltrating immune cells as a biomarker candidate in MPNST.

Tumor mutation burden (TMB)

Tumors with a high number of mutations are likely to harbor more neoantigens, which in turn increases the probability of the cancer cells being recognized by the immune system. As such, although there are exceptions, higher TMBs correlate positively with response to immune checkpoint inhibition in multiple cancers.^78,79,80 This observation prompted the FDA to prioritize the approval of TMB as a companion diagnostic biomarker for pembrolizumab, with a definition of high TMB to be >10 mutations per megabase.^81,82 Although TMB is an emerging biomarker for response to immunotherapy, MPNSTs have yet to benefit from using TMB to predict response. Soft tissue sarcomas, in general, are reported to have a low TMB of 1–2.5 mutations/Mb.⁸³ However, in an analysis of 100,000 different cancers, 8.2% of MPNSTs had mutations of more than 20 mutations/Mb.³⁷ In a separate analysis of whole-exome sequencing data of 12 MPNST patient samples, somatic coding variants per tumor ranged from 7 to 142 with a median value of 63.⁸⁴ These results suggest that TMB could be advanced as one of the potential stratification factors of response to immunotherapy in MPNST.

Conclusion

While the success of immune-based strategies in multiple cancers emerged from a thorough characterization of tumor biology spanning multiple years, MPNST immunotherapy is still in its infancy. Even though there is certainly much to elucidate, we are optimistic that immune-based treatment strategies will provide opportunities for meaningful intervention against this deadly disease. For example, MPNSTs are replete with myeloid cells and macrophages with a pre-inclination for tumor-promoting M2-like population.¹³ Pharmaceutical agents that can re-polarize and modulate the activities of these cells may stymie tumor cell growth or aid in their eradication by other immunotherapies. Genes involved in antigen processing and presentation, including MHC, have also been shown to be downregulated in a subset of MPNST patients.²¹ De-repressing the expression of these genes, such as through the administration of epigenetic drugs, may make these tumors more susceptible to immunotherapies dependent on antigen presentation (e.g., checkpoint inhibitors). These results provide rationale to explore immunotherapy even when MHC expression is downregulated.

Tumor heterogeneity will be one of the most critical hurdles as we move forward with more immunotherapeutic regimens. In this regard, identifying biomarkers of response should be one of the top priorities to guide informed decisions for designing personalized treatments. Although no correlation was observed between PD-L1 expression or CD8+ infiltration with disease-specific or disease-free survival in primary MPNST patients, low-grade MPNST patients have higher cytotoxic T lymphocyte scores (CD8+, CD4+, FOXP3+, CD45RO+, and CD56+ infiltrates) than high-grade samples.^19,26 We have also yet to test any potential immunotherapy biomarkers in a large and standardized prospective trial. The critical candidates for future trials in MPNST are PD-L1 expression, tumor-infiltrating lymphocytes, and TMB. We have also amassed significant knowledge on using the OV in MPNST, but attempts for clinical translation have been painstakingly slow. The immunosuppressive microenvironment, an ever-present challenge in solid tumors that holds the power to diminish any immunotherapeutic intervention, will also be a significant roadblock in MPNSTs. Schwann cells have been predicted to undergo successive dedifferentiation programs to acquire a mesenchymal and neural crest-like state as a malignant stem-cell-like population during advanced stages of MPNST progression.¹⁴ The impact of current immunotherapeutic platforms on these mesenchymal stem-like cells is an open question that will critically shape the design of effective therapies. A rational combination of immunotherapeutic regimens could be a holy grail to harness the clinical benefit of immunotherapy. Overall, we are optimistic about the prospect of ultimately leveraging immune-based therapies for MPNST patients.