Therapeutic cancer vaccines are designed to stimulate a patient’s immune system against cancer and typically target a tumor antigen. Many approaches can be taken to elicit this anti-tumor immune response, such as a peptide vaccine incorporating tumor antigen peptide(s) combined with an adjuvant. A phase II trial of a trivalent ganglioside vaccine (GM2, GD2 and GD3) in 136 sarcoma patients observed serological GM2-specific and GD2-specific responses in 98% of treated patients, however, no difference in progression-free survival was observed between patient cohorts (NCT01141491) (figure 1).38 Similarly, in a study using a 9-mer peptide spanning the SYT-SSX junction in 21 synovial sarcoma patients, 7 patients showed a twofold increase in SYT-SSX specific CD8+T cells, but only one patient showed transient shrinkage of a metastatic lesion.56

The lack of therapeutic response using the aforementioned peptide vaccine approaches in sarcoma prompted the development of alternative vaccination strategies. Dendritic cell (DC) vaccines consist of patient-derived DCs loaded with autologous tumor lysate or tumor antigens ex vivo. An ongoing phase I study is investigating a tumor-lysate loaded monocyte-derived DC vaccine in advanced sarcoma and neuroblastoma pediatric patients (EudraCT 2014-003388-39). Forty-seven pediatric patients have been enrolled in the study, 11 with sarcoma. The treatment was safe and immunomonitoring of nine sarcoma patients (two EW sarcoma, four osteosarcoma, one synovial sarcoma and two RMS) revealed enhanced ex vivo T cell responses on vaccination.57 58 In another phase I/II trial, a tumor lysate-pulsed DC vaccine was evaluated in 37 patients with bone or STS. No adverse effects were associated with treatment, and immunological responses to the treatment included significantly elevated interferon (IFN-γ) and IL-12 in the serum, suggesting that this vaccine activated the patient’s immune response. However, despite evidence of boosted immunity, only 1/35 patients achieved a partial response to therapy.59

Another approach to DC-vaccination is to peptide-pulse DCs with a tumor antigen, forgoing the need for invasive surgical resection of tumor to isolate tumor lysates. One pilot study, including patients with advanced sarcoma that have progressed on standard-of-care chemotherapy and radiation (including synovial sarcoma, osteosarcoma and LPS), uses NY-ESO-1 peptide-pulsed DCs in combination with IL-2 and autologous lymphocytes tranduced with a NY-ESO-1-specific T-cell receptor, with or without CTLA-4 blockade (NCT02070406, NCT01697527).60 Treatments were safe but, only one synovial sarcoma patient (receiving treatment without iplimumab) progressed to complete response, despite evidence of anti-tumor activity determined by positron emission tomography imaging in 6 out of 10 patients. This lack of long-lasting response to therapy in the majority of patients could be attributed to the fact that the adoptively transferred cells were found to be short lived in peripheral circulation.

One preclinical study employs a conjugate vaccine targetting the tumor vasculature, thus overcoming the issues associated with targetting tumor antigen. Prophylactic immunization against the tumor vascular marker ED-B (extra domain-B) of fibronectin reduced murine fibrosarcoma (T241) tumor micro vessel density, increased the number of tumor infiltrating leukocytes, and generated a high anti-EDB antibody titer compared with controls.61 This promising technology can be used for the induction of immune responses against other tumor-antigens or components of the TME and may be an improvement on traditional peptide vaccines.

Although cancer vaccines for sarcoma appear to be safe and result in an immunological response in most of the patients, limited improvement in clinical outcome of patients suggests that many modifications need to be made to attain better therapeutic outcomes. Additional clinical trials using vaccines for sarcoma are listed in table 1.

OVs are self-propagating viruses that selectively infect and kill tumor cells. In addition to their direct effects, OVs promote virally encoded transgene expression (eg, cytokines), cause vascular collapse, and generally alter the TME to stimulate antitumor immune responses. Pointing toward the possibility of using OVs in sarcoma immunotherapy, a comprehensive study comparing the efficacy of four OV head to head in a panel of sarcoma cell lines showed that the rhabdoviruses Maraba and vesicular stomatitis virus (VSV) were highly effective at broadly killing sarcoma cell lines, while Reovirus and herpes simplex virus type I (HSV) were only effective in EW sarcoma cell lines, and synovial sarcoma (SW982) was resistant to infection by all viruses tested.62 This result corroborates previous findings that SW982 cells are refractory to infection by a different VSV strain.63 Results obtained with cell lines were corroborated using ex vivo infection of freshly collected human sarcoma samples where Maraba successfully infected ~95% of samples. Encouragingly, mice injected subcutaneously with the sarcoma cell line S180 responded to Maraba treatment, revealing Maraba MG1 as a promising candidate for sarcoma immunotherapy.63

When tested in vitro, H-1PV, a rodent protoparvovirus non-pathogenic in humans, effectively infected human synovial sarcoma (Sw-R), a subtype that is resistant to VSV infection, as well as osteosarcoma and EW sarcoma cell lines.63–66 However, when evaluated as a therapeutic in a xenograft model, one dose of H-1PV did not lead to tumor regression or increased overall survival, although the lack of therapeutic effect could be attributed to insufficient dosing or owing to the use of immunodeficient mice in this study as OVs work, at least in part, by inducing antitumor immunity.66

There is a rich pipeline of OVs in pre-clinical evaluation and many have progressed to clinical trials. Few have gained regulatory approval, such as the genetically engineered HSV-I Talimogene laherparepvec (T-VEC; Imlygic), which has specific virus attenuating mutations and encodes granulocyte-macrophage colony-stimulating factor (GM-CSF) in order to promote anti-tumor immune responses.67 Imlygic, was approved for treatment of advanced melanoma by the FDA in 2015. T-VEC in melanoma has led to its evaluation in other solid malignancies, including sarcoma (figure 1).

Two phase II clinical trials evaluating T-VEC are currently recruiting. The first is determining the combination of T-VEC and radiation therapy as a preoperative treatment for patients with newly diagnosed local STS (NCT02923778); the second is evaluating the efficacy of intratumoral administered T-VEC in locally advanced cutaneous angiosarcoma. Another oncolytic HSV-1 (HSV-1716) was tested in phase I clinical trial in patients with solid tumors that included eight pediatric patients with RMS, EW sarcoma, osteosarcoma, MPNST or chordoma and two adult patients with chondrosarcoma or EW sarcoma (NCT00931931). Unfortunately, while the virus was found safe, no objective responses were observed, suggesting that HSV-1 may not be ideally suited for these sarcoma subtypes.68

Pexastimogene devacirepvec (Pexa-Vec) is derived from a Wyeth vaccinia virus strain and engineered to target cancer cells. Similar to Imlygic, it is armed with GM-CSF. In a phase I trial, Pexa-Vec was safe when administered intratumorally in a cohort of RMS, Ewings sarcoma, osteosarcoma, and other STS pediatric patients (NCT01169584). A phase I/II study of Pexa-Vec in combination with metronomic cyclophosphamide in patients with advanced STS and advanced breast cancer is currently recruiting (NCT02630368).

Overall, preclinical studies suggest OV could be promising immunotherapies for the treatment of sarcoma. However, this remains to be corroborated in clinical studies. So far, OVs have had limited success as monotherapies and it is likely that OVs will require use in combination with other modalities that can overcome known resistance mechanisms, including innate antiviral responses and immunological resistance, for example.69–72 Additionaly, some OVs may be more effective than others depending on the sarcoma subtypes as indicated from comparative in vitro studies.62

Anti-tumor immunity can be promoted by antibodies targeting tumor antigens eliciting antibody-mediated or complement-mediated cytotoxicity. Surface disialoganglioside (GD2), a glycosphingolipid involved in cell proliferation, is also a known tumor associated antigen, most notably expressed in neuroblastoma, melanoma and half of all osteosarcomas and STS’.73 Widespread distribution of GD2 in neuroblastoma prompted preclinical development of GD2-targeting mAbs and led to phase I trials of humanized 3F8 (hu3F8) and FDA approval of the mAb ch14.18, also known as dinutuximab, in 2015 (figure 1).74 75 GD2 is expressed in ~50% of osteosarcomas and STS’, although with significant intratumoral heterogeneity, therefore, there is a potential of expanding the use of anti-GD2 for osteosarcoma, with clinical trials currently ongoing39 73 76 (NCT02502786, NCT02484443).

The polarization of macrophages associated to the tumor (TAMs), either directly infiltrating the TME or deriving from monocytes, is a key determinant of the outcome of anti-tumor immunity. In a broad sense, M1 macrophages typically promote anti-cancer immunity whereas M2 macrophages are usually associated with worse prognosis given they promote immunosuppression, tumor angiogenesis and metastasis. Therapies that indirectly prevent macrophage polarization toward an M2 phenotypes are currently being tested in preclinical models. For example, treatment of MCA-205 fibrosarcoma mouse model with anti-angiogenic peptide led to a reduction of M2-polarized macrophages, which was important for the therapeutic effect of the peptide. In another study, TAMs were directly targeted by treating sarcoma-bearing mice with nanoparticles containing zoledronate leading to significant tumor regression.77 78 A more favorable TAM polarization was instead achieved using TLR2 agonists that increased the M1/M2 ratio in sarcoma-bearing mice leading to decreased tumor growth.79

The CD47-SIRPα pathway promotes tumor immune escape by inhibiting macrophage phagocytosis (do not eat me signal) and is an appealing target for immunotherapy as revealed by studies showing that: (1) an abundance of SIRPα in TAMs was associated with worse prognosis in patients with synovial sarcoma and myxofibrosarcoma; and (2) high expression of CD47 correlated with poor survival in osteosarcoma patients.80 81 In an osteosarcoma mouse model, mice treated with a CD47 blocking antibody, alone or in combination with doxorubicin, had reduced tumor burden.82 Although no clinical trials are currently ongoing in sarcoma for monoclonal antibodies targeting CD47, targeting of the same pathway is under clinical evaluation in a phase I trial (NCT02890368) in solid tumors, including STS, using TTI-621, a SIRP1α-fusion protein that blocks CD47 binding to SIRP1α and simulataneously potentiated macrophage activation by binding to activating Fc receptors.83

The CSF1 receptor (CSF1R)/CSF1 axis was considered a promising targetable pathway to reduce M2 TAMs in sarcomas but a phase I/IIa clinical trial assessing the effect of small molecules targeting CSF1R signaling in solid tissue malignancies (NCT02452424) was terminated due to lack of clinical efficacy (figure 1). Evaluation of monoclonal antibodies targeting CSF1R (Emactuzumab) is still ongoing for the treatment of STS with no results posted to date (NCT02323191). In conclusion, targeting the macrophage compartment is a compelling yet still fairly unexplored avenue to potentiate anti-cancer immunity in sarcoma.

Cytokines play a key role in communication between immune cells and can be exploited to harness their effector functions. To date, IL-2 and IFN-α are the only immunostimulatory cytokines approved by the FDA for the treatment of several cancer indications including HIV-induced Kaposi’s sarcoma, a malignancy of blood and lymph vessels caused by the human herpersvirus 8.84 85 As a single agent, IFN-α cytokine therapy is most successful at high doses and in Kaposi’s sarcoma patients with higher baseline CD4+ T cell counts but is commonly used in conjunction with chemotherapies. With the advent of anti-retroviral therapy, IFN-α began to be used in lower doses in combination with these anti-retrovirals. Overall, the anti-proliferative, anti-angiogenic and anti-viral effects of IFN-α therapy are thought to induce tumor regression in HIV-induced Kaposi’s sarcoma.86 However, liposomal doxorubicin and paclitaxel are still the front line treatment for Kaposi’s sarcoma.

The transforming growth factor-beta (TGF-β) signaling pathway is known to be dysregulated in many cancers and the cytokine is notorious for its dual role of both tumor promoting and tumor suppressing activities in late stage and early-stage cancers, respectively.87 88 In the case of osteosarcoma, TGF-β was found to be increased in the serum of patients compared with healthy controls and is now seen as a driver of metastatic potential in this cancer.89 90 Furthermore, in vitro studies revealed that treatment of osteosarcoma cell lines with TGF-β induced epithelial-mesenchymal transition, providing further evidence of its role in stimulating metastasis.91 92 Decreasing levels of serum TGF-β is also seen in EW sarcoma and RMS patients during remission, and as such may have prognostic value.93 In light of these findings, if the key role of TGF- β is elucidated, there is potential in inhibiting the key components of the TGF-β pathway as a therapeutic strategy for osteosarcoma, EW sarcoma and RMS. Clinical trials of TGF-β ligand and receptor inhibitors are ongoing for many advanced stage carcinomas, melanomas and gliomas and should, therefore, be given considerable thought for its use in osteosarcoma.87 94

Since ICB has shown limited efficacy as a monotherapy for sarcoma, numerous studies are investigating ICB therapy in combination with other treatments. More specifically, one study evaluated NY-ESO-1 specific CD8 T cells of patients with NY-ESO-1+ positive tumors and found that these cells are enriched for expression of LAG-3 and PD-1.95 To address upregulation of inhibitory molecules, an ongoing study is evaluating a vaccine targeting NY-ESO-1 (called CMB305) in combination with anti-PDL1 (atezolizumab) in sarcoma (NCT02609984). In the context of OV, Newcastle disease virus (NDV) for treatment of syngeneic tumor models of melanoma induced PD-L1 upregulation of infected cells, and inhibited effector functions of tumor infiltrated T cells.71 Similarly, oncolytic vaccinia virus treatment of murine models of colon and ovarian cancer induced PD-L1 expression in cancer and immune cells.96 Suggesting immune checkpoints are a potential contributor to resistance to OV therapy, the combination of NDV and other OVs with ICB has shown to provide therapeutic benefit in various preclinical models.71 A phase 2 trial (NCT03069378) recently investigated the combination of pembrolizumab with T-VEC.97 The overall response rate was 30% (7/35) with no complete responses and 7 partial responses. The partial responses were observed in patients with UPS, myxofibrosarcoma, epithelioid sarcoma, cutaneous angiosarcoma and an unclassified sarcoma. Also, the TIL score and PD-L1 expression was higher in responders compared with the refractory group.97 An expansion cohort for this study is underway for patients with UPS, myxofibrosarcoma and angiosarcoma. A multi-center pilot study (NCT03282344) is currently evaluating the safety and efficacy of treating refractory sarcoma patients with nivolumab and NKTR-214, an IL-2 pathway agonist. Preliminary results from 50 enrolled patients showed partial responses in patients with UPS (2/10), dedifferentiated chondrosarcoma (1/10) and leiomyosarcoma (1/10) and no respondes from DDLPS and osteosarcoma patients.98 A phase II multi-arm study (NCT02815995) investigated the benefit of combining durvalumab (anti-PD-L1) and tremelimumab (anti-CTLA-4). Their primary endpoint which was progression-free survival at 3 months, was reached in 51% of patients (n=57).99 It was also observed that a high immune infiltration score could predict clinical benefit. Many of the combination ICB trials include patients with advanced sarcomas who are refractory to standard treatment and/or did not respond to previous ICB monotherapy. Due to the nature of these studies, monotherapy groups are not included, which can make it difficult to determine which component of the combination treatment improved the clinical outcome.

The clinical studies discussed show that there is a heterogeneous response of sarcoma to checkpoint blockade. Despite the progress of the most promising immunotherapeutic approach, ICB, only a small proportion of patients given this treatment have clinical responses. One consideration is that patient selection for ICB treatment is generally performed in a single analyte manner, whereby patients are selected based on expression levels of immune inhibitory molecules being targeted. As we increase our understanding of the immune mechanisms at play within the tumor, it is evident that this analysis provides an incomplete picture of the dynamic tumor immune microenvironment that contributes to the efficacy of ICB therapy. To this end, several studies have sought to establish predictive biomarkers using multigene analysis to guide treatment.

A seminal study identified an immune-related signature correlating with clinical responses from different clinical studies using pembrolizumab (KEYNOTE-180, KEYNOTE-181, KEYNOTE-158), and developed a scoring algorithm predicting patients’ response to pembrolizumab based on tumor expression of 18 IFN-γ-related immune genes. The Tumor Inflammation Signature (TIS) score measures activated but suppressed adaptive immune responses within tumors. Notably, this score has been retrospectively applied to clinical trial data, and high scores were found to be associated with response to pembrolizumab,100 101 further supporting the use TIS to guide patient selection in future ICB clinical trials for sarcoma.

Another group developed a new immune-based classification of STS based on the composition of the tumor immune microenvironment determined from gene expression profiles, and stratified 608 STS into one of five SICs with highly distinct immune profiles. Patients with tumors classified in the immune high group—characterized by high expression of genes specific to CD8+ T cells, NK cells, B cells and tertiary lymphoid structures, exhibited the largest response rate to PD-1 blockade therapy (in comparison to tumors with different immune class) in a phase I clinical trial of pembrolizumab in STS (SARC028). This study highlights the heterogeneity of sarcoma, and the need for increased understanding of immune status of sarcomas to select appropriate immunotherapy treatments for patients.22

Altogether, these data show that similarly to other cancers, in sarcoma a T cell-inflamed TME, characterized by cytotoxic effector lymphoid functions and active IFN-γ signaling, are features of patients responding to PD-1 blockade. Further elucidating the tumor immune microenvironment of sarcoma subtypes will refine patient eligibility criteria for sarcoma immunotherapy trials, and further investigation will be necessary to predict patient outcome in response to other immunotherapeutic approaches discussed in this review.