The Emerging Role of Immunotherapy for the Treatment of Sarcoma

Abstract

Clinicians caring for patients with sarcoma founded the field of cancer immunotherapy. Despite this, contemporary success with immunotherapy for sarcoma has been limited. Here, we review immunotherapy for sarcoma including Coley’s toxins, interleukin-2, adoptive cell transfer, and checkpoint blockade. We detail recent and ongoing efforts to combine checkpoint blockade with other immune modulators, surgery, or radiation. These results, along with ongoing investigations, have identified immunotherapeutic approaches as a promising avenue for progress in advanced sarcomas.

INTRODUCTION

“The fact that in one of the reported cases when two attacks of erysipelas had accidentally occurred, the first diminishing the size of the tumor, and the second causing it to disappear, led me to try repeated inoculations.” – W. B. Coley, 1890¹

The first cancer immunotherapists were, famously, surgeons caring for patients with sarcoma. An important early advance was the discovery of the etiology of erysipelas by the 19^th century German surgeon Friedrich Fehleisen, who showed it could be produced by inoculation with cultures of Streptococcus erysipelatis². After reading reports of tumor regression in patients with sarcoma who developed erysipelas, William Coley then deliberately provoked erysipelas by inoculation¹. In some cases, this resulted in dramatic tumor shrinkage.

The mechanism of tumor regression induced by erysipelas infection escaped Coley, but was later discovered by John J. Morton, a surgeon who would go on to serve in France under Harvey Cushing during the Great War. Morton showed that rejection of murine tumors was associated with infiltration of lymphocytes and concluded, “the lymphocyte is a necessary factor in cancer immunity.”³ Fifty years later, the British surgeon E. J. Delorme showed that adoptive transfer of lymphocytes in a mouse model could inhibit carcinogen-induced sarcoma growth⁴. Collectively, these findings helped give form to the notion that lymphocytes may be used therapeutically to treat patients with advanced sarcoma.

A breakthrough came in the 1970s with the discovery of interleukin-2 (IL-2), which enabled in vitro culture of T lymphocytes^5,6. This paved the way for the modern era of immunotherapy, which began in 1988 when surgeon Steven A. Rosenberg and colleagues demonstrated adoptive cell transfer (ACT) of tumor infiltrating lymphocytes (TIL) could effectively treat advanced melanoma in studies conducted in the Surgery Branch of the National Cancer Institute (NCI)⁷. Subsequent trials showed high dose IL-2 in some patients induced durable regression of advanced renal cell carcinoma (RCC) and melanoma^8,9. Rosenberg & colleagues made other key advances in the field of T cell therapy including the use of a preparative regimen of lymphodepleting chemotherapy, which enhances the engraftment and persistence of transferred T cells¹⁰. Surgery Branch investigators also showed that specific immune responses could be directed against a known cancer antigen by gene modification of autologous lymphocytes with a T cell receptor (TCR) of choice¹¹.

A major distinction between immunotherapy and cytotoxic chemotherapy or targeted therapy for solid tumors in adults is that immune-mediated complete responses (CR) are usually durable for years or decades. For example, 27 (82%) of 33 patients with melanoma or renal cell carcinoma who had a CR after IL-2 were disease-free at a median of 86 months of follow-up¹². Fourteen (93%) of 15 patients with melanoma who had a CR after CTLA-4 blockade (with or without IL-2) remained disease-free at a median follow-up of 83 months¹³. In two trials evaluating ACT of TIL for melanoma, CRs were observed in 22–24% of patients, and 90% of those responses were durable at long follow-up^14,15. These results show the immune system is capable of sterilizing, or at least permanently suppressing, the entire burden of metastatic disease in some patients.

Sarcomas are a heterogeneous group of mesenchymal neoplasms of bone and soft tissue origin and include more than 70 distinct subtypes. Approximately 15,000 cases of soft tissue and bone sarcomas are diagnosed annually in the U.S. Despite primary combined modality therapy, 25–50% of patients develop recurrent and/or metastatic disease^16,17. For patients with unresectable advanced disease, chemotherapy remains the standard of care. However, complete responses are rare and the median survival in the metastatic setting is 15–20 months^18–21. Standard cytotoxic chemotherapy agents such as doxorubicin, ifosfamide, and dacarbazine result in objective responses in 10–25% of patients, but response rate varies significantly among histologic variants, and responses are typically not durable²². Therefore, novel and effective systemic therapies are desperately needed for patients with sarcomas.

INTERLEUKIN-2 FOR SARCOMA

There have been surprisingly few studies evaluating the use of IL-2 for patients with advanced sarcoma but results from one trial suggest it might be effective. Investigators enrolled ten pediatric patients with metastatic neuroblastoma, osteosarcoma, Ewing’s sarcoma or Wilms’ tumor²³. Five (50%) of 10 patients had a CR to IL-2, and those responses were durable at a median follow-up of 28 months. The study investigators reported severe but reversible toxicities.

Despite proof of efficacy in melanoma and RCC, IL-2 is not widely used for those diseases. It is dosed every 8 hours in the inpatient setting, which requires a significant commitment on the part of providers. Furthermore, perceptions regarding IL-2 toxicity, and cytokine-induced capillary leak syndrome in particular, have dampened enthusiasm for the treatment²⁴. It is worth mentioning that in experienced hands IL-2 toxicity can be manageable. One study reported 809 consecutively treated patients without a single treatment-related death²⁵.

However, in order to mitigate challenges with dosing and toxicity associated with IL-2 administration, a PEGylated recombinant form of IL-2 (Bempegaldesleukin) has been developed that is dosed every 3 weeks and is well tolerated²⁶. In one study, cytokine-related symptoms (flu-like symptoms, rash, pruritus, or hypotension) were observed primarily in cycles one and two and became significantly reduced with additional cycles. Also, there was no grade ≥3 hypotension following the implementation of hydration guidelines.

Early results from studies using Bempegaldesleukin for solid tumors are very promising. Data from the phase 1/2 PIVOT-02 study (NCT03635983) showed that Bempegaldesleukin plus nivolumab resulted in a 60% objective response rate for patients with melanoma, RCC and non-small cell lung cancer. For patients with advanced sarcoma, preliminary results of a study evaluating Bempegaldesleukin plus nivolumab were presented at ASCO in 2019, which we discuss in more detail below. More research is needed in order to determine whether non-specific immune stimulation using cytokines such as IL-2 have a role in helping to boost anti-tumor immune responses for sarcoma.

ADOPTIVE CELL TRANSFER FOR SARCOMA

Cell therapy using autologous TIL is effective in patients with melanoma, and has shown promise in other solid tumors including cervical cancer and gastrointestinal epithelial malignancies^27–29. D’Angelo and colleagues showed that TIL are present in a substantial fraction of sarcomas³⁰. While there are no reports to date of adoptive transfer of TIL for sarcoma, this approach is currently being evaluated in clinical trials at several institutions (NCT04052334, NCT03449108, NCT03935893). It is also being evaluated for commercial application by Iovance Biotherapeutics (Wardell et al. 2019 Society for ImmunoTherapy of Cancer Annual Meeting, National Harbor, MD USA).

It is possible that a relative paucity of immunogenic neoantigens will limit the efficacy of TIL for sarcoma³¹. Rather than relying on the endogenous anti-tumor specificity of TIL, an alternative cell therapy approach involves gene-modification of lymphocytes with a TCR directed against an antigen of choice in an HLA-matched patient. The potential for efficacy of such an approach has been demonstrated in synovial sarcoma. Synovial sarcoma is a translocation-associated sarcoma and has few known neoantigens; perhaps in consequence, it does not respond well to checkpoint inhibitors (discussed in more detail below). However, synovial cell sarcoma almost universally expresses the cancer-germline antigen NY-ESO-1, which is not expressed in healthy somatic tissues.

Surgery Branch, NCI investigators showed that adoptive transfer of NY-ESO-1 specific peripheral blood lymphocytes (in HLA-A02 patients) resulted in objective responses in 11 (61%) of 18 patients with synovial cell sarcomas³². Investigators at Memorial Sloan Kettering Cancer Center (MSKCC) also showed that adoptive transfer of NY-ESO-1 specific T cells resulted in objective responses in 6 (50%) of 12 patients with synovial cell sarcoma³³. MSKCC investigators are currently evaluating this approach for patients with myxoid/round cell liposarcomas, which also express NY-ESO-1³⁴. There are dozens of other cancer germline antigens that could theoretically be targeted in this fashion.

A major challenge for TCR therapy is identifying suitable patients, who must express both the target antigen and be HLA-matched³⁵. An alternative to TCR gene therapy, one that obviates the need for HLA matching, is chimeric antigen receptor (CAR) T cell therapy. CAR T cells have the advantage of not being restricted by HLA, but their disadvantage is that they cannot be used to target peptide-MHC complexes (and so cannot target intracellular peptides and most neoantigens). CAR T cells can however target many extracellular antigens, including glycoproteins³⁶.

Of critical importance for both TCR and CAR T cell therapy is avoiding “off tumor, on target” toxicity. In hematological malignancies, CAR T cells can be used safely because the healthy B cells that express the target antigen (CD19 or CD20) are expendable. In contrast, many of the antigens expressed in solid tumors are also present in normal healthy tissues that are not expendable. In one dramatic example of this, adoptive transfer of lymphocytes targeting the cancer germline antigen MAGE-A3 resulted in fatality because the antigen was expressed at low levels in the brain³⁷. In another case, adoptive transfer of anti-HER2 CAR T cells resulted in respiratory failure and death from cytokine storm triggered by low level recognition of ERBB2 on pulmonary epithelial cells³⁸. In another study, anti-carcinoembryonic antigen CAR T cells mediated objective responses in patients with advanced colorectal cancer, but at the cost of severe dose-limiting colitis³⁹.

Fortunately, there are tumor-exclusive cell surface antigens that can be targeted by CARs, and there are ways to engineer CAR T cells to enhance safety^36,40. A recent study evaluated anti-HER2 CAR T cells for HER2⁺ sarcoma in a dose escalation and demonstrated safety, but there were no responses⁴¹. The subsequent addition of lymphodepletion improved T cell expansion and resulted in CRs in 2 (20%) of 10 patients treated⁴². The two CRs were in rhabdomyosarcoma and osteosarcoma, which are almost always refractory to checkpoint inhibitors; both responses are ongoing at 15 and 32 months.

Cell therapy has other important limitations. The preparative regimen of lymphodepleting chemotherapy can itself have side effects including neutropenia and result in prolonged hospitalization. Cell therapy requires significant expertise and costs associated with maintaining GMP (good manufacturing practice) facilities. Limitations specific to TIL include the need for surgical resection and variation in the quality of the T cell product⁴³. Despite these limitations, cell therapy using TIL or gene-modified T cells is a promising avenue of investigation for patients with sarcoma.

CHECKPOINT BLOCKADE FOR SARCOMA

In general, sarcomas have a high rate of primary resistance to checkpoint blockade. However, in certain sarcoma subtypes there is a clear signal for efficacy with checkpoint inhibitors. In particular, alveolar soft part sarcomas (ASPS), vascular sarcomas (including angiosarcoma and epithelioid hemangioendothelioma) and undifferentiated pleomorphic sarcomas (UPS) appear to be sensitive to PD-1/PD-L1 blockade (Figure 1). Not surprisingly, the data also suggest combinatorial approaches to immune manipulation may increase response rates.

Figure 1. Objective response to PD-1/PD-L1 blockade in sarcoma.

All sarcoma patients were treated with PD-1 or PD-L1 blocking antibodies (nivolumab, pembrolizumab, durvalumab, atezolizumab) either as monotherapy or with another agent (ipilimumab, Bempegaldesleukin, tremelimumab, axitinib, metronomic cyclophosphamide, epacadostat, sunitinib, T-VEC). Response evaluations were by RECIST v1.1 (except Somaiah et al. 2020 ASCO, who reported responses by irRC). The dotted horizontal line indicates a 20% objective response rate. The eight histologies to the left had fewer than 20 patients total across all studies; the exact number of objective responses/number treated is shown above the bar. The ten middle histologies (from Ewing sarcoma to ASPS) had more than 20 treated patients reported in the literature. The bar graph shows mean with standard deviation. Each triangle represents 6–22 patients from a single study or aggregated from multiple studies with small numbers. The 39 liposarcoma and 40 UPS patients from the SARC028 expansion cohort were split into two groups in order to reflect the weight of those large samples. To enable comparison with non-sarcoma histologies, we included data from trials evaluating anti-PD-1 or anti-PD-L1 for patients with sun-shielded melanomas (acral, mucosal or uveal), cutaneous melanoma, and Merkel cell carcinoma. Vascular sarcoma includes angiosarcoma and epithelioid hemangioendothelioma. Abbreviations: GIST (gastrointestinal stromal tumor), NOS (not otherwise specified), UPS/MFH (undifferentiated pleomorphic sarcoma / malignant fibrous histiocytoma), ASPS (alveolar soft part sarcoma).

Single-agent checkpoint inhibition

Several studies have evaluated the use of antibodies against CTLA-4, PD-1 or PD-L1 as monotherapy for sarcomas, and in general the objective response rate is very low. A pilot study of ipilimumab (anti-CTLA-4) for synovial cell sarcoma resulted in no objective responses⁴⁴. In 22 patients with sarcoma who were treated with ipilimumab in a phase I clinic, there were no objective responses⁴⁵.

SARC028 investigators were the first to evaluate PD-1/PD-L1 axis blockade for patients with sarcoma. Pembrolizumab (anti-PD-1) was used in a two-cohort, single-arm open-label phase II trial⁴⁶. Data from the SARC028 expansion cohort were recently presented at ASCO⁴⁷. Responses in UPS were observed in 9 (23%) of 40 patients, two of which were CRs (by RECIST v1.1). Responses in liposarcoma (LPS) were less frequent, occurring in 4 (10%) of 39 patients.

The Alliance A091401 study was an open-label non-comparative randomized phase II study evaluating nivolumab (anti-PD-1) with or without ipilimumab for patients with metastatic sarcoma⁴⁸. In the nivolumab monotherapy arm, 2 (5%) of 38 patients had objective responses. The responders had ASPS and leiomyosarcoma (LMS); a 3rd patient with sarcoma not otherwise specified (NOS) also had an unconfirmed partial response. In the Alliance A091401 expansion cohort, nivolumab mediated responses in 0 (0%) of 9 patients with gastrointestinal stromal tumor (GIST), 1 (8%) of 12 patients with de-differentiated LPS and 1 (8%) of 12 with UPS⁴⁹.

A phase II trial evaluating nivolumab in 12 patients with uterine LMS reported no objective responses⁵⁰. Results of a randomized trial for patients with GIST were recently reported, but no objective responses were seen in the nivolumab arm⁵¹. However, a phase II study of atezolizumab (anti-PD-L1) for ASPS reported 8 (42%) of 19 patients had a partial response (O’Sullivan Coyne et al. 2018 CTOS, Rome, Italy). These data show that, with the exception of UPS and ASPS, the clinical activity of single-agent checkpoint blockade in sarcoma is very low.

Dual checkpoint inhibition

In the nivolumab plus ipilimumab arm of the Alliance A091401 study, 6 (16%) of 38 evaluable patients had an objective response⁴⁸. A seventh patient had an unconfirmed response. In the nivolumab plus ipilimumab arm of the Alliance trial expansion cohort, responses occurred in 0 (0%) of 9 GIST, 2 (14%) of 12 de-differentiated LPS and 2 (14%) of 12 UPS⁴⁹. In a randomized trial evaluating immunotherapy for GIST, nivolumab plus ipilimumab mediated an objective response in 1 (8%) of 12 patients⁵¹. A phase II study evaluating durvalumab (anti-PD-L1) and tremelimumab (anti-CTLA-4) resulted in objective responses in 5 (50%) of 10 ASPS⁵². While limited by small numbers, these results highlight the potential importance of combinatorial immunotherapy strategies and sarcoma subtype selection for the future design of immunotherapy trials for sarcoma.

Checkpoint inhibition plus novel immunomodulatory approaches

Results of a phase II study evaluating pembrolizumab with the IDO1 inhibitor epacadostat were recently reported at ASCO⁵³. Among 29 evaluable patients, 1 PR (3%) was observed in a patient with LMS. Given the negative results of the recently published Keynote-252 study, which evaluated epacadostat for patients with melanoma, it is unclear whether IDO1 inhibition enhances anti-PD-1 therapy for solid tumors⁵⁴.

Talimogene laherparepvec (T-VEC) is an oncolytic immunotherapy derived from a modified human herpes virus designed to self-replicate within and lyse tumor cells, thereby releasing tumor antigens and promoting regional and systemic antitumor immunity. In a phase II study of pembrolizumab plus T-VEC for 20 patients with locally advanced or metastatic sarcoma, the response rate was 35%, with a median duration of response of 56 weeks⁵⁵. Of note, 2 of 3 patients with angiosarcoma and 2 of 2 with UPS had objective responses. Other responses were seen in myxofibrosarcoma (MFS), sarcoma NOS and epithelioid sarcoma. Expansion cohorts in these subtypes (except sarcoma NOS) are now open (NCT03069378), and there is also a neoadjuvant cohort for high-risk UPS/MFS.

Bempegaldesleukin, as mentioned previously, is an experimental PEGylated form of IL-2 that activates and expands natural killer and CD8⁺ T cells via CD122 agonist activity. The combination of Bempegaldesleukin and nivolumab in melanoma showed impressive efficacy with a favorable toxicity profile. Early results from an ongoing study evaluating Bempegaldesleukin plus nivolumab for patients with sarcoma was recently presented at ASCO⁵⁶. Objective responses were seen in 5 (9%) of 57 patients, including vascular sarcoma, LMS, UPS, and chondrosarcoma. Final results of the trial are still pending.

Checkpoint inhibition plus chemotherapy

A phase II study evaluated pembrolizumab with metronomic dosing of the lymphotoxic agent cyclophosphamide for patients with LMS, UPS, GIST and a few other sarcomas. An objective response occurred in 1 (2%) of 50 patients treated, in a patient with a solitary fibrous tumor⁵⁷. It is unclear whether the combination of systemic chemotherapy with checkpoint inhibitors will be effective for patients with sarcoma.

Checkpoint inhibitors combined with TKI

The first study evaluating a combination of checkpoint inhibitors with tyrosine kinase inhibitors (TKI) in patients with sarcoma used ipilimumab with dasatinib for twenty patients with advanced GIST along with eight patients with other sarcomas⁵⁸. There were no objective responses in the study. This result may reflect the choice of TKI in a heavily pre-treated GIST population. Whether this strategy can be effective for TKI-sensitive patients remains to be seen.

The ASPL-TFE3 fusion is universally present in ASPS and leads to HIF-1α accumulation and upregulation of proangiogenic factors⁵⁹. This was the rationale behind a phase II study of pembrolizumab with the VEGF receptor TKI axitinib for patients with advanced sarcomas including ASPS. The combination of pembrolizumab and axitinib mediated objective responses in 6 (55%) of 11 ASPS, 1 of 6 LMS and 1 of 1 epithelioid sarcomas⁶⁰. Early results from the phase I/II Immunosarc study, evaluating nivolumab plus the multi-TKI inhibitor sunitinib, showed an objective response rate of 11% (Martin Broto et al. 2020 ESMO Sarcoma & GIST Symposium, Milan Italy).

Biomarkers and Response to Immunotherapy

A major limitation of the current literature is that individual subtypes of sarcomas are quite rare, so there are few data with which to estimate the comparative effectiveness of checkpoint inhibitor treatment. Furthermore, PFS in patients with sarcomas treated with checkpoint inhibitors is brief, typically lasting 2–4 months, which reflects the fact that in most patients, the natural history of the disease is not detectably altered.

It should be noted that many of the results represented in Figure 1 were from studies that combined PD-1/PD-L1 blockade with other agents including ipilimumab, T-VEC, Bempegaldesleukin, et cetera. Nonetheless, responses are being consistently observed for certain histological subtypes of sarcoma. ASPS, vascular sarcomas and UPS/MFS have response rates that are near or above 20%. Evidence of activity also exists in de-differentiated LPS, although response rates for now are below 10%. There are also promising data from subtypes such as dedifferentiated chondrosarcoma and epithelioid sarcoma.

There is a strong correlation between tumor mutational burden (TMB) and histology-specific response (among all solid tumors) to PD-1 blocking antibodies⁶¹. This may be explained by the ability of endogenous T cells to recognize the gene products of cancer-specific mutations (neoantigens)^29,62. A high TMB thus leads to an abundance of neoantigens which can be targeted by the immune system³¹. This hypothesis also explains divergent response patterns within some types of solid tumors. For example, cutaneous melanoma, which has a large burden of neoantigens due to UV mutagenesis, is highly sensitive to checkpoint blockade. In contrast, sun-shielded melanomas including acral lentiginous, mucosal or uveal melanomas have a paucity of neoantigens and are less responsive⁶³. Mismatch repair-proficient colorectal cancer has a low TMB and is unresponsive to anti-PD-1, but mismatch repair-deficient colorectal cancer has a high TMB and is highly sensitive⁶⁰.

Sarcomas in general are associated with a low TMB. One study used 100,000 human cancer genomes and ranked 167 human tumors by TMB, showing the most highly mutated sarcoma (angiosarcoma) was only ranked 67^th overall⁶⁴. Interestingly, the three sarcoma subtypes with the highest percentage of cases with more than 20 mutations per megabase were angiosarcoma (13.4% of cases), MPNST (8.2%) and UPS (8.1%). These subtypes are typically characterized by complex genotypes and often present in older patients. For comparison, only 0.5% of Ewing sarcomas and 0.2% of liposarcoma had more than 20 mutations per megabase⁶⁴. Synovial sarcoma, which ranked 143^rd among the tumors analyzed, is a translocation-associated sarcoma that usually results from the formation of SS18-SSX fusion oncogenes⁶⁵. Consistent with its low TMB, only two responses to checkpoint blockade in patients with synovial sarcoma have been reported in the literature (Figure 1). ASPS was not included in the study, but also has a very low TMB as it is translocation-associated.

It is unclear whether PD-L1 expression is a useful biomarker in sarcoma. D’Angelo et al. showed that lymphocyte and macrophage infiltration is common in sarcoma, but PD-L1 expression is uncommon in most sarcomas³⁰. Furthermore, while PD-L1 expression is high in GIST (present in 29% of cases), GIST has a very low response rate to immunotherapy thus far (Figure 1). There has been no observed association between PD-L1 expression and overall survival. It is worth noting that in the SARC028 study, 3 (4%) of 70 evaluated tumor samples had PD-L1 positivity at the 1% threshold, and 2 of those 3 had objective responses to treatment⁴⁶. Also, Wilky et al. showed 11 of 11 ASPS tumor samples were PD-L1 positive⁶⁰.

The metabolism connection

There are several curious similarities between clear cell RCC and ASPS. Clear cell RCC is as sensitive to IL-2 and PD-1 blockade as melanoma but has a low TMB⁶⁶. RCC is characterized by mutations in VHL or other enzymes of metabolism that result in HIF accumulation⁶⁶. ASPS is a rare sarcoma of uncertain mesenchymal origin that was first described at MSKCC in 1952⁶⁷. Of all sarcoma histological subtypes, ASPS appears to be the most responsive to immunotherapy (Figure 1). ASPS is a translocation-associated sarcoma and, like RCC, has a low TMB⁶⁸.

Furthermore, it has been shown that the unbalanced ASPL-TFE3 fusion associated with ASPS leads to accumulation of HIF via dysregulated lactate metabolism⁵⁹. Intriguingly, the ASPL-TFE3 fusion found in ASPS, which arises from an unbalanced translocation, can lead to primary renal neoplasms that resemble RCC if instead the fusion results from a balanced translocation⁶⁹. Thus, clear cell RCC and ASPS share a low TMB, high response rates to immunotherapy, and are characterized by metabolic dysregulation.

It is possible that the metabolic phenotype of RCC and ASPS leads to enhanced immune infiltration of tumors. A recent study evaluated gene expression from 608 tumors across a variety of subtypes of soft tissue sarcoma (but did not include ASPS)⁷⁰. Those data were used to identify distinct sarcoma immune classes (SIC): immune-low, immune high, and vascularized. The immune low SIC had low densities of CD3⁺, CD8⁺ or CD20⁺ cells. In contrast, the vascularized SIC showed a moderate infiltration of immune cells and a high density of CD34⁺ endothelial cells. RCC and ASPS upregulate angiogenesis signals via HIF, which may promote immune cell infiltration into the tumors, where they are poised to mediate regression.

Immunotherapy strategies in the surgical patient

The curative potential of immunotherapy provides clear rationale for a complementary role of local therapy in selected cases. After all, if an immune-mediated CR is usually durable, then a near-CR followed by local therapy to extirpate any remaining disease might also achieve durable disease control. Retrospective data support this conclusion, including two studies where patients with metastatic melanoma were treated with immunotherapy and then metastasectomy for oligoprogressive immunorefractory disease^71,72. In some cases, patients were disease-free for years after surgery.

The rationale for a complementary relationship between immunotherapy and local therapy that has emerged in melanoma is largely predicated on a high frequency of robust responses. In a patient with unresectable metastatic melanoma, immunotherapy can alter the natural history of the disease and achieve systemic control, while isolated local failures are controlled with surgery or local therapy. As the efficacy of immunotherapy improves, this strategy might become more widely applicable, as patients with unresectable disease undergo “oligometastatic conversion” by immunotherapy. The potential for conversion to resectability depends on strong anti-tumor immune responses, which are frequently seen in melanoma. In sarcoma, this approach may not be useful until response rates to immunotherapy can be improved.

More exciting than the prospect of a complementary relationship between immunotherapy and locoregional therapy is the possibility of a synergistic relationship between the two. In one study, 17 matched UPS samples taken from before and after radiotherapy showed a trend towards increased density of TIL⁷³. These and other data provide rationale for combination neoadjuvant immune checkpoint inhibition and radiotherapy. This approach is currently being evaluated in the SARC032 study, a phase II trial comparing neoadjuvant Pembrolizumab plus radiotherapy versus radiotherapy alone for patients with sarcoma (NCT03092323).

Investigators at MD Anderson are evaluating neoadjuvant checkpoint blockade with nivolumab or nivolumab plus ipilimumab in patients with surgically resectable retroperitoneal de-differentiated LPS or with extremity/truncal UPS, treated with concurrent neoadjuvant radiation therapy⁷⁴. In a recent presentation, the median pathologic response in the UPS cohort was reported to be 95% and was similar between the two arms. Median pathologic response in the de-differentiated LPS cohort was more modest, at 22.5%. There were 8 patients with a pathological response ≥ 85%; 1 had a partial response, 5 had stable diseases and 2 had progressive disease.

At MSKCC, encouraging results from the phase II trial of pembrolizumab and T-VEC⁵⁵ have led to an expansion cohort that includes neoadjuvant immunotherapy for patients with UPS/MFS (NCT03069378). Additionally, in the setting of locally advanced disease, we are employing a strategy of local chemotherapy (with melphalan and dactinomycin) administered by isolated limb infusion to increase the response rate of pembrolizumab (NCT04332874). MSKCC investigators previously demonstrated in patients with melanoma that a combination of high-dose chemotherapy with checkpoint inhibition promoted an inflammatory response and sensitized the cancer both locally and systemically⁷⁵. In forthcoming trials in sarcoma, correlative studies will be aimed at characterizing the pretreatment sarcoma tumor microenvironment and changes to the immune infiltrate induced by the combinatorial approaches.

Current Clinical Context

Immunotherapy in the form of checkpoint inhibition should be a consideration in patients with inoperable locally advanced/metastatic sarcoma with specific subtypes including UPS, angiosarcoma, and ASPS. The optimal time to introduce immunotherapy in the systemic therapy treatment paradigm for these sarcoma subtypes remains unknown. In other disease settings, inferior response rates to immunotherapy have been observed when administered after chemotherapy compared to when introduced in the first line setting^76,77. Hence, consideration of immunotherapy earlier in patients with advanced UPS, angiosarcoma, or ASPS is reasonable.

Combinatorial therapies may ultimately prove more efficacious, but many trials testing these strategies exclude patients with prior checkpoint inhibition. For this reason, a review of available clinical trials should be performed before starting immunotherapy for metastatic sarcoma. Trials evaluating immunotherapy for earlier stage disease hold the promise of reduced toxicity and/or potentially improved efficacy in the (neo)adjuvant setting but are currently limited to patients at highest risk of recurrence with histologic subtypes thought to be most likely to benefit. The field of immunotherapy is rapidly evolving, and individual sarcoma histologic types are rare. In this setting, smaller, hypothesis-driven trials in homogeneous patient cohorts are likely to provide the most efficient way to identify particularly promising new approaches for patients with sarcoma.

Conclusions

Clinicians treating patients with sarcomas pioneered the field of cancer immunotherapy. Recent advances have most benefited patients whose solid tumors have a high tumor mutational burden, but there is abundant evidence showing immunotherapy can successfully treat certain neoantigen-sparse malignancies. Immunotherapy is most promising in ASPS, vascular sarcoma and UPS. Dual checkpoint inhibitor or combinatorial immune modulation may result in higher response rates. Multidisciplinary management with locoregional approaches and perhaps cancer-specific targeting are likely to be indispensable for clinical success. Biomarkers to predict response and improve patient selection will further refine clinical algorithms, but at present these are largely absent. It has been over 100 years since Coley first unleashed the immune system against sarcomas, but it does appear that immunotherapy for sarcoma may finally be coming of age.

Methods for Figure 1

We included data from published trials evaluating PD-1 or PD-L1 antibodies either as monotherapy or in combination with another agent (cyclophosphamide, axitinib, ipilimumab, epacadostat or T-VEC)^{46,48,57 50,53,60}. We also included data from unpublished trials evaluating PD-1 or PD-L1 antibodies with or without other agents that were presented at national meetings, including:

• A phase II study of atezolizumab in patients with alveolar soft part sarcoma (O’Sullivan Coyne et al. CTOS 2018 Annual Meeting, Rome, Italy)

• A randomized phase II study of nivolumab monotherapy versus nivolumab combined with ipilimumab in advanced gastrointestinal stromal tumor (Singh et al. 2019 American Society of Clinical Oncology, Chicago, USA)

• SARC028 expansion cohort (Burgess et al. 2019 American Society of Clinical Oncology, Chicago, USA)

• A phase II study evaluating Durvalumab plus Tremelimumab (Somaiah et al. 2020 American Society of Clinical Oncology, Chicago, USA)

• A multicenter phase II study of nivolumab +/− ipilimumab for patients with metastatic sarcoma (Alliance A091401): Results of expansion cohorts (Chen et al. 2020 American Society of Clinical Oncology, Chicago, USA)

• A phase II study evaluating Nivolumab plus Bempegaldesleukin (D’Angelo et al. 2019 American Society of Clinical Oncology, Chicago, USA).

• A phase I/II Immunosarc study evaluating Nivolumab plus sunitinib (Martin Broto et al. 2020 European Society for Medical Oncology Sarcoma & GIST Symposium, Milan, Italy)

For those histologies for which fewer than 20 treated patients were reported in the literature (spindle cell sarcoma to myxofibrosarcoma), all patients were aggregated into a single group and presented as a bar graph. No standard deviation is shown. Above the bar shows the exact number of objective responses (by RECIST v1.1, except the study by Somaiah et al. which used irRC) and the number of patients treated.

For those histologies with more than 20 patients treated in the literature (Ewing sarcoma to ASPS), the bar graph shows mean with standard deviation. Each triangle represents 6–22 patients from a single study or aggregated from multiple studies with small numbers (1–5 patients each). The 39 liposarcoma and 40 UPS patients from the SARC028 expansion cohort are represented by two triangles in order to reflect the weight of those samples.

For purposes of comparison, data from trials evaluating anti-PD-1 or anti-PD-L1 monotherapy for patients with sun-shielded melanoma (1 acral, 2 mucosal, 2 uveal), cutaneous melanoma and Merkel cell carcinoma are shown on the far right of Figure 1. Response rates for sun-shielded melanomas include: 23.3% for mucosal melanoma and 40.9% for cutaneous melanoma⁷⁸, 32% for acral melanoma and 23% for mucosal⁷⁹, and 3.6% and 4.7% for uveal melanoma^80,81.

Trials evaluating PD-1/PD-L1 monotherapy for cutaneous melanoma reported an ORR of 28% (NCT00730639)⁸², an ORR of 44.3% (NCT01844505, CheckMate067)⁸³, and ORR of 33.7% and 32.9% (NCT01866319, KEYNOTE-066)⁸⁴. Trials evaluating PD-1 / PD-L1 for Merkel cell carcinoma included an ORR of 56% for Pembrolizumab (NCT02267603)⁸⁵, ORR of 31.8% for Avelumab (NCT02155647)⁷⁷, and ORR of 68% for Nivolumab (NCT02488759; CheckMate 358) (Topalian et al. 2017 American Association for Cancer Research Annual Meeting, Washington DC).

Data Availability

Data generated for this manuscript are available from the corresponding author upon reasonable request.

SYNOPSIS.

Sarcomas as a class have been largely refractory to immunotherapy. However, emerging data show promising evidence of efficacy of immune-based approaches, especially for certain subtypes of sarcoma.