Non-cytotoxic systemic treatment in malignant peripheral nerve sheath tumors (MPNST): A systematic review from bench to bedside

Background: Malignant peripheral nerve sheath tumors (MPNSTs) are aggressive soft tissue sarcomas. Once metastasized, prognosis is poor despite regular treatment with conventional cytotoxic drugs. This study reviews the preclinical and clinical results of non-cytotoxic systemic therapy in MPNST. Methods: A systematic search was performed in PubMed and Embase databases according to the PRISMA guidelines. Search terms related to ‘MPNST’, ‘targeted therapy’, ‘immunotherapy’, and ‘viral therapy’ were used. Only in vivo studies and clinical trials were included. Clinicaltrials.gov was also searched for any ongoing trials including MPNST patients. Qualitative synthesis was performed on all studies stratifying per target: membrane, cytoplasmic, nuclear, immunotherapy and oncolytic viruses, and other. In vivo studies were assessed for treatment effect on tumor growth (low/intermediate/high), survival, and metastases. Clinical trials were assessed on response rate, progression-free survival, and overall survival. Results: After full-text screening, 60 in vivo studies and 19 clinical trials were included. A total of 13 trials are ongoing and unpublished. The included trials displayed relatively poor response rates thus far, with patients achieving stable disease at best. Inhibiting cytoplasmic targets most commonly yielded high treatment effect, predominantly after mTOR inhibition. Oncolytic viruses and angiogenesis inhibition also demonstrate intermediate to high effect. Therapies including a combination of drugs were most effective in controlling tumor growth. Several ongoing trials investigate potentially promising pathways, while others have yet to be established. Conclusion: Targeting the PI3K/Akt/mTOR pathway seems most promising in the treatment of MPNSTs. Oncolytic viruses and angiogenesis inhibition represent emerging therapies that require further study. Combinations of targeted therapies are most likely key to maximize treatment effect. ,


Introduction
Malignant peripheral nerve sheath tumors (MPNSTs) are rare, but aggressive soft tissue sarcomas (STS) with high rates of recurrence and metastasis (Carli et al., 2005;Stucky et al., 2012;Valentin et al., 2016). Almost half of all cases are related to neurofibromatosis type I (NF1), while others occur sporadically or after radiation exposure (Stucky et al., 2012;Zou et al., 2009). The NF1 gene is commonly affected in MPNSTs causing the loss of neurofibromin, a Ras inhibiting enzyme (Basu et al., 1992). Ras activation results in the downstream activation of Ras pathways, leading to upregulation of mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K) (Endo et al., 2013). However, loss of neurofibromin alone is not enough to cause an MPNST (Kluwe et al., 1999). Research over the last three decades has implicated multiple factors in the pathogenesis of MPNSTs, including loss of function in TP53, CDKN2A, SUZ12, and PTEN genes, as well as amplification of epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor (PDGFR), and MET (Beert et al., 2011; De complex biology underlying MPNSTs, prognosis has remained poor, with 5-year survival rates ranging from 30 to 60% in patients who have undergone curative surgery of their tumor, and even lower rates in those with advanced and metastatic disease (Carli et al., 2005;Ducatman et al., 1986;Stucky et al., 2012;Valentin et al., 2016).
Surgery with wide negative margins remains the mainstay treatment for MPNST (Stucky et al., 2012;Valentin et al., 2016). Radiotherapy is commonly used either postoperatively or in a neoadjuvant setting as it improves local control, but does not affect overall survival (Bradford and Kim, 2015;Kahn et al., 2014;Stucky et al., 2012). In a study investigating neoadjuvant chemotherapy, histotype-guided treatment of four STS types, including MPNST (this cohort was treated with etoposide-ifosfamide), has not shown any benefit compared to standard anthracycline based chemotherapy (Gronchi et al., 2017). Therefore, there has thus far been no rationale for treating MPNST differently from other STS. Neoadjuvant chemotherapy could be considered for high-grade, large, and deep MPNST (Gronchi et al., 2017;Higham et al., 2017), and may allow initially inoperable patients to become operable (Carli et al., 2005;Kroep et al., 2011). However, over 10% of MPNST patients present with unresectable or metastatic disease (Carli et al., 2005;Valentin et al., 2016;Wong et al., 1998). Additionally, 40-60% of patients receiving treatment with curative intent will develop metastatic disease (Anghileri et al., 2006;Wong et al., 1998;Zehou et al., 2013).
For the whole group of STS, first line palliative chemotherapy consists of an anthracycline (doxorubicin or epirubicin) containing schedule. This might be combined doxorubicin and ifosfamide or doxorubicin monotherapy. Overall, a clinical response rate of approximately 21% has been reported for MPNST treated with combined doxorubicin and ifosfamide (Kroep et al., 2011). Adding ifosfamide to doxorubicin has improved progression-free survival (PFS), but not overall survival (OS), and comes at the cost of increased toxicity (Judson et al., 2014).
The high rates of advanced and metastatic disease and poor response to standard chemotherapy highlight the need for novel therapies in the treatment of MPNST. Targeted therapy and immunotherapy has brought new options to many other cancer types, but is not yet established in STS in general or MPNST specifically. Especially target specific, non-cytotoxic treatments are of interest as they may specifically target tumors and have limited systemic side-effects. As insights in the differences between STS subtypes are growing, more specific testing to allow for identification and subsequent personalization of treatment is necessary; however, given that MPNST represent a rare sarcoma subtype, such personalization has thus far been challenging. To better understand emerging treatment options, we pooled the available literature and performed a systematic review of non-cytotoxic systemic therapies in MPNST, aiming to guide future research efforts by identifying the most relevant targets and combinations.

Literature search
A systematic search was performed in both PubMed and Embase databases according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines, in order to identify all potentially relevant articles published from 2000 to 2018. The search string was built with the help of a professional librarian using search terms related to 'MPNST' and non-cytotoxic treatments. The exact search syntaxes for PubMed and Embase are provided in Supplemental  Table S1. Preclinical studies were included if they studied non-cytotoxic drugs on MPNSTs in vivo. Clinical studies were included if they presented results of non-cytotoxic systemic therapy specifically in MPNST patients. Articles were excluded if they were retrospective or single case studies, reviews, presented non-specific MPNST data, included data on cytotoxic drugs or drugs that were only tested in vitro, or did not provide data on tumor growth, survival, or metastases. Clinicaltrials.gov was also searched with synonyms of 'MPNST' to obtain all ongoing non-cytotoxic drug trials enrolling MPNST patients. Cross-referencing of included papers and registered trials was performed, which identified six additional papers. These studies did not include a synonym of MPNST in either their title or abstract. The initial review was carried out by two independent authors (EM, NL). Disagreements were solved through discussion, in which one additional senior author was involved (ID).

Data extraction and synthesis
Data extracted from preclinical studies included: animal model used, most effective treatment regimen studied, tissues investigated, and treatment effect on tumor growth, survival, and metastasis. The treatment effect on tumor growth was evaluated according to the mean relative tumor volume (RTV) comparing the latest mean volume measurement of the control group (C) to the mean volume of the treatment group (T) at that time point (Houghton et al., 2007;Plowman et al., 1999): T/C ≤15% represented high effect (black); T/C ≤45% but > 15% represented intermediate effect (dark gray); and T/C > 45% represented low effect (light gray, Table 1). Tumor growth was either assessed by tumor volume, weight, or area. Drugs were categorized as membrane targets, cytoplasmic targets, nuclear targets, immunotherapy and oncolytic viruses, or other targets.
Data extracted from clinical trials included: study design, number of patients, age of population, treatment regimen, and treatment effect on response rate, PFS, and OS. Study phase, country, intervention, anticipated accrual, and end date were extracted from registered unpublished trials.
Qualitative synthesis was performed summarizing data from preclinical and clinical studies according to target pathway, immunotherapy and oncolytic viruses, and a rest group.

Results
Following removal of duplicates, a total of 1938 articles and registered trials were identified in PubMed and Embase databases. Title/ abstract screening resulted in selection of 203 potentially relevant articles, of which sixty-six were selected for qualitative synthesis after full-text screening (Fig. 1). A total of sixty preclinical in vivo studies were found that used numerous genetically engineered mouse models (GEMM), (non)-cultured NF1 and sporadic patient xenografts, allografts from GEMMs, and one zebrafish model (Table 1). Nineteen trials were identified, of which six have already been published (Table 2), and thirteen are ongoing (Table 3). Fig. 2 presents the most important target pathways identified in MPNSTs.

Membrane targets -in vivo
Eight studies investigated membrane targets in vivo (Table 1). Six used receptor tyrosine kinase (RTK) targeted treatments with intermediate to high effect on tumor growth (Ki et al., 2017;Lock et al., 2016;Mo et al., 2013;Ohishi et al., 2013;Torres et al., 2011;Wu et al., 2018). The addition of verteporfin (TAZ/YAP inhibitor) to sorafenib yielded intermediate effects on tumor growth in an allograft model, while monotherapy of either drugs had significantly worse effects (Wu et al., 2018). The chemokine receptor CXCR4 stimulates cell cycle progression through PI3K and β-catenin signalling. In one in vivo study, inhibition of CXCR4 showed intermediate effect on tumor growth and increased survival of mice (Mo et al., 2013). Two in vivo studies investigated the effect of estrogen receptor blockade; one found a low effect on tumor growth (Byer et al., 2011), and another showed that the addition of a calmodulin inhibitor enhanced the effect on tumor growth (Brosius et al., 2014).

Membrane targets -trials
Four published clinical trials investigating the effect of an RTK inhibitor, of which one (Albritton et al., 2006) specifically examined MPNST patients (Table 2), were identified. None of the trials found an appreciable clinical response in MPNST patients, with only 0-20% of the patients achieving stable disease (Albritton et al., 2006;Chugh et al., 2009;Maki et al., 2009;Schuetze et al., 2016). Four additional trials were still ongoing at the time this review was written, one of which will only include MPNST patients. This study will evaluate the efficacy of the multikinase inhibitor pexidartinib in combination with mTOR inhibitor sirolimus (NCT02584647, Table 3). Multiple other trials were identified that will enroll patients with soft tissue sarcomas (NCT02584309, NCT02180867) and CD56 expressing tumors (NCT02452554) targeting additional membrane targets. One of these trials will investigate the effect of doxorubicin and ifosfamide with the addition of pazopanib, currently the only registered RTK inhibitor for STS, in a neoadjuvant setting including patients with resectable soft tissue sarcomas (NCT02180867).

Cytoplasmic targets -trials
One trial evaluating the effect of mTOR inhibition in combination with bevacizumab, a VEGF inhibitor, demonstrated stable disease in 3/ 25 patients (Widemann et al., 2016). A total of three trials that were ongoing at the time of this review were investigating the role of an mTOR inhibitor in combination with a MEK inhibitor (NCT03433183), pazopanib (NCT02601209), or heat shock protein 90 (Hsp90) inhibitor (NCT02008877, Table 3). The latter trial was completed, although its results were not yet published.

Nuclear targets -in vivo
The effect of nuclear target inhibitors was investigated in twelve studies, identifying this class of drugs to have intermediate to high effects on tumor growth (Table 1). Multiple studies found a high effect on survival (4/4 cell lines) or tumor growth (5/15 cell lines) via in vivo inhibition of several epigenetic pathways (Hirokawa et al., 2005;Kivlin et al., 2016;Lopez et al., 2015Lopez et al., , 2011Mohan et al., 2013;Nair and Schwartz, 2015;Patel et al., 2014Patel et al., , 2012Payne et al., 2018). Aurora kinase A (AURKA) is one of these epigenetic regulators, which regulates centrosome maturation and chromosome separation. Alisertib, an AURKA inhibitor was found to have a higher effect on tumor growth and survival compared to a combination of doxorubicin and ifosfamide in vivo (Payne et al., 2018).
CDK4/6 and EZH2 act via influence on the cell cycle; in vivo studies showed that their inhibition has intermediate effect on tumor growth (Perez et al., 2015;Zhang et al., 2015).
XPO1 is the main nuclear export protein and transports proteins such as survivin. One in vivo study found intermediate effect of XPO1 inhibition combined with proteasome inhibitor carfilzomib (Nair et al., 2017).

Immunotherapy and oncolytic viruses -in vivo
Next to tumor cell specific targeting, immunotherapy may also play a role in MPNST treatment. With an evolving role in other cancer types, no in vivo studies have thus far been published investigating immunotherapy regimens specifically in MPNST. Oncolytic viruses are thought to affect tumors in several ways, one of which involves the upregulation of the immune system. Eight studies investigated the effect of oncolytic viruses in MPNST in vivo (Table 1). Seven studies used an oncolytic herpes simplex virus (oHSV) with mostly intermediate to high effect (10/12 cell lines) on tumor growth (Antoszczyk et al., 2014;Currier et al., 2017;Liu et al., 2006a,b;Mahller et al., 2007Mahller et al., , 2008Maldonado et al., 2010). One study used an oncolytic measles virus (oMV) and showed high efficacy in one xenograft model, but low effect in another (Deyle et al., 2015). Almost all studies looked at survival and showed a statistically significant benefit for treatment with oncolytic viruses compared to a placebo control group. The addition of erlotinib, an EGFR inhibitor, did not significantly improve the efficacy compared to oHSV monotherapy in vivo (Mahller et al., 2007). However, additional AURKA inhibition was found to have a synergistic effect on both tumor growth and survival (Currier et al., 2017).

Immunotherapy and oncolytic viruses -trials
Two ongoing trials are investigating the role of PD1 checkpoint inhibitors (Table 3): one looks at PD1 inhibitors alone and includes MPNST patients only (NCT02691026), while the other study combines the PD1 inhibitor nivolumab with CTLA-4 inhibitor ipilimumab and includes patients with rare tumors, one of which is MPNST (NCT02834013).
No clinical trial has yet evaluated the effect of oncolytic viruses in MPNSTs. Two trials are registered of which one will use an oMV in MPNST patients only (NCT02700230) and the other, which is complete and whose results are pending, investigated the effect of an oHSV in non-central nervous system (CNS) solid tumors including MPNSTs (NCT00931931 , Table 3).

Discussion
MPNST still remains a highly aggressive sarcoma subtype with poor outcome despite regular cytotoxic treatment. Novel strategies to target metastatic MPNST and improve its outcomes, both in terms of survival as well as quality of life, are needed. In locally advanced disease, neoadjuvant treatment that can downsize the primary tumor and allow for subsequent surgical resection is also of value.
In this review, we sought to describe new approaches to treat advanced MPNST. Multiple membrane, cytoplasmic, and nuclear actors are potential targets in the therapy of MPNST, of which mTOR inhibition is most commonly investigated in vivo and has frequently resulted in high responses on tumor growth (81.3% of cell lines) and survival (100% of cell lines).
In vivo, RTK inhibitors that include VEGFR inhibition have also shown intermediate to high responses. However, monotherapy with an RTK inhibitor has not shown tumor regression clinically in MPNSTs except for a modest prolongation of median progression free survival in case of pazopanib treatment in all types STS (van der Graaf et al., 2012). Apart from two in vivo studies using cabozantinib, no other study has yet investigated the effect of MET inhibition, although it is a known contributor to malignancy in MPNSTs. RTK inhibitors targeting both the VEGF pathway as well as other pathways, or combinations with other treatment types might therefore be of interest.
Unfortunately, although MPNSTs are Ras-driven tumors, no drug has yet been found to successfully target Ras. Ras inhibitors are difficult to create due to a lack of well-defined druggable pockets and cavities on its surface (Simanshu et al., 2017). Targeting upregulated downstream targets of Ras is nevertheless possible. Besides upregulation of the PI3K/Akt/mTOR pathway, upregulation of the MAPK pathway in NF1 tumors has been described several times (Endo et al., 2013). In this review we described the potential of mTOR inhibitors, which might be increased by the current development of more specific inhibitors of elements of the mTOR pathway. Although single agent MEK inhibition has not resulted in tumor suppression (Dodd et al., 2013;Jessen et al., 2013;Kendall et al., 2016), combinations with mTOR inhibitors might prove potent in terms of anti-tumorigenic effects, but at the cost of increased toxicity (Lock et al., 2016;Malone et al., 2014). The, translationally controlled tumor protein (TCTP), a downstream effector of both the MAPK and mTOR, can be successfully inhibited leading to cell death in NF1-associated tumors (Kobayashi et al., 2014), and was found to increase mTOR activity when upregulated, indicating a positive feedback loop. In vivo studies on MPNST models are, however, still warranted. Other targets of interest identified in this review are PAK1 inhibitors (Demestre et al., 2009;Hirokawa et al., 2006;Semenova et al., 2017a,b), as well as PI3K inhibitors. ERK inhibitors are being developed as well, which may have less toxicity, but their effect on MPNST cells is still unknown (Nissan et al., 2013).
While checkpoint inhibitors are gaining interest in other types of tumors, they have yet to be extensively studied in STS. Two ongoing trials will hopefully elucidate the role of these types of drugs in MPNST (NCT02691026, NCT02834013). Oncolytic viruses are showing efficacy without severe toxicity in various cancers including MPNSTs (Chiocca and Rabkin, 2014;Lichty et al., 2014). Moreover, as demonstrated for other tumors, an additional pathway inhibitor may give a synergistic effect when combined with oncolytic viruses (Currier et al., 2017). Overall, while therapies with oncolytic viruses appear promising in MPNST, more in vivo studies are needed to better understand their role as well at the role for any treatment combinations.
The lack of progress in the treatment of MPNST is multi-factorial. First, adequate preclinical models representing both NF1-associated MPNSTs as well as sporadic MPNSTs are lacking. The causal mechanisms behind NF1-associated MPNST may differ from those in sporadic MPNST, resulting in different sensitivity for treatment. This is supported by the fact that in conventional chemotherapy, NF1 patients are known to have a lower response rate (Carli et al., 2005;Ferrari et al., 2011;Higham et al., 2017). However, only few in vivo studies show a difference in response on tumor growth between NF1 and sporadic patient-derived models, while others show no difference. Thus, clinical translation of these differences might be difficult and should ultimately be assessed in clinical trials. Second, the preclinical data have to be robust before performing a clinical trial. For example, Albritton et al. based their trial on evidence found from one in vitro study (Li et al., E. Martin, et al. Critical Reviews in Oncology / Hematology 138 (2019) 223-232 2002. It is reasonable to consider in vitro studies by themselves as weaker evidence compared to in vivo studies, and it is therefore unsurprising that such studies might not effectively translate to the clinical setting (Mak et al., 2014). Third, most studies include all types of STS since it is challenging to perform a trial in a disease as rare as MPNST. In this review, four out of the six identified studies were performed in all types of soft tissue sarcomas, for which preclinical evidence was not necessarily found in MPNSTs specifically. The investigators should however be applauded for their efforts in performing histotype subanalyses, although likely underpowered, as certain histological subtypes might well be more sensitive to a particular drug therapy than others. Finally, as suggested by the present review that is based on in vivo evidence, a combination of different drugs is likely to be more potent in MPNST patients compared to monotherapy. However, many of the published trials only investigated single targeted therapy. Unfortunately, quantitative comparison between different studies investigating different treatments in vivo was not fully feasible. To date, no tool has been established that shows high reliability of translating preclinical outcomes into clinical evidence, limiting the ability to make direct comparisons between preclinical studies. Despite the challenges in drawing quantitative comparisons across studies, assessing treatment effect by stratifying outcomes into low, intermediate, and high effect has been successfully done previously (Houghton et al., 2007). Overall, despite these limitations, to our knowledge, the current article represents the largest review to date to pool the available literature on in vivo therapies for MPSNT. By assessing various animal models and treatment regimens through a descriptive systematic review, we aimed to facilitate treatment-related decisions in patients with MPNST (Hooijmans et al., 2018). For now, such animal studies serve as the cornerstone to the advancement of therapeutics for MPNST in humans and are therefore necessary to carefully review and assess prior to initiation of human trials (Mak et al., 2014). Identification of multiple potential MPNST drugs in this review underscore fundamental principles that will guide optimization of treatment regimens in the future. For example, novel therapies should focus on improving survival while simultaneously limiting toxicity and maintaining quality of life. The utility of ultimately discovering a systemic treatment specifically targeting MPNSTs may drastically alter the course of the MPNST management, allowing for preoperative tumor reduction and potentially minimizing the need for higher doses of radiation as well as more intensive surgeries.

Conclusion
Non-cytotoxic systemic treatments have not yet demonstrated clinical efficacy for MPNST, but most promising are approaches targeting the PI3K/Akt/mTOR and VEGFR pathways, as well as utilization of oncolytic viruses. A combination of therapies will most likely be key to maximizing treatment effects. With several clinical trials now, at least in part, recruiting MPNST patients, new insights into therapeutic options for MPNST will likely result.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declarations of interest
None.

Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.critrevonc.2019.04.007.