Synovial sarcomas (SS) are tumors of mesenchymal origin that represent 5–10% of all soft tissue sarcomas. Most SS occur as a result of a translocation between the X chromosome and chromosome 18 resulting in SS18-SSX1, SS18-SSX2, and/or SS18-SSX4 fusion proteins [1]. Current therapeutic options for primary localized SS include surgical resection, radiotherapy, and chemotherapy. For patients with advanced or recurrent disease, chemotherapies and targeted therapies have limited efficacy. To date, no immunotherapies have been approved in SS, and clinical trials with checkpoint inhibitors have not shown durable benefit in this patient population [2–4].

PD-1/PD-L1 (programmed cell death) pathway inhibitors have shown durable clinical benefit in tumor histologies that exhibit T-cell infiltration, elevated levels of PD-L1 expression, and higher levels of nonsynonymous somatic mutation burden [5]. By comparison, SS are poorly infiltrated by T cells and have marginal PD-L1 expression [6, 7]. As in other translocation-driven tumors, SS also have a low overall mutational burden. This likely contributes to a low neo-antigen burden, which may at least partially explain the paucity of intra-tumoral T cells. Furthermore, SS exhibit low copy number alterations, which might additionally contribute to low tumor antigenicity [8]. One immunogenic antigen expressed in the majority (~ 70%) of SS tumors is the cancer-testis antigen NY-ESO-1 [9–11]. Studies with either an NY-ESO-1 vaccine or adoptively transferred NY-ESO-1-specific T cells alone or in combination with interleukin (IL)-2 have demonstrated that producing cell-mediated immune responses to NY-ESO-1 is a promising strategy in SS [12–14].

The results from the initial cohort of this pilot study testing genetically modified autologous T cells specific to the NY-ESO-1 peptide SLLMWITQC in patients with advanced metastatic SS have previously been published [14]. Patients were treated with T cells engineered to express an affinity-enhanced T-cell receptor (TCR) recognizing a human leukocyte antigen (HLA)-A*02-restricted NY-ESO-1/LAGE-1a-derived peptide (NY-ESO-1 SPEAR T cells; GSK 794) following lymphodepletion with cyclophosphamide and fludarabine [14]. In this cohort, we observed 50% response rate with a durable (30.9 weeks median duration of response) in patients with high intra-tumoral NY-ESO-1 expression. Long-term persistence and greater expansion of SPEAR T cells were seen in responding patients. Lastly, the persisting SPEAR T cells in the blood had a stem cell memory phenotype, were poly-functional, and did not express exhaustion markers [14]. Three additional cohorts were opened under this study to determine the impact of low antigen expression, reduced preconditioning, and the effects of fludarabine on safety and efficacy. Here we present the response data from these three additional cohorts, along with correlative data assessing the immunologic mechanisms of response and resistance in patients with advanced metastatic SS from all four cohorts of the pilot study.

Adoptive T-cell therapy using engineered receptors has shown great promise in the treatment of hematologic malignancies, as demonstrated by recent US Food and Drug Administration approvals for CD19-directed chimeric antigen receptors (CARs) in B-cell acute lymphoblastic leukemia (ALL) and in diffuse large B-cell lymphoma (DLBCL) [19–25]. Furthermore, efficacy data in multiple myeloma with B-cell maturation antigen-directed CARs [26] and NY-ESO-1 SPEAR T-cells [15] is encouraging. To date, engineered cell therapy in non-hematopoietic solid tumors has not demonstrated benefit in significant numbers of patients. Solid tumors pose unique challenges to adoptive T-cell therapies as compared with hematopoietic malignancies. These include but are not limited to heterogeneous antigen expression, physiologic barriers to T-cell trafficking, and an immunosuppressive tumor microenvironment [27]. SS is an ideal tumor target, since the fundamental oncogenic mechanisms of this translocation-related sarcoma drive aberrant expression of NY-ESO-1 through the defective SWI/SNF complex, which results in abnormal function of polycomb repressor complexes modulating gene expression [28]. Results from this study [14] (Fig. 1e-h) establish that some of these challenges can be overcome by NY-ESO-1 SPEAR T cells.

In this report, we expand upon findings reported previously [14] and include data from three additional cohorts. We continue to report overall meaningful clinical efficacy of NY-ESO-1 SPEAR T cells in SS of 36% in this treatment-resistant population, despite meeting futility and an early stop in the cohort that lacked fludarabine. Furthermore, we addressed the impact of pre-infusion lymphodepletion on outcome. Although many differing combinations of fludarabine and cyclophosphamide have been utilized in adoptive immunotherapy clinical trials with varying outcomes (reviewed in [29]), there is no consensus on the optimal lymphodepletion regimen for adoptive T-cell therapy, and much of the available data are from early phase trials testing CAR-T cells in hematologic malignancies [29–32]. The applicability of these findings to TCR-based adoptive immunotherapies in carcinomas and sarcomas have yet to be determined. Our results indicate that a fludarabine-containing preparative regimen is needed to maximize the benefit of SPEAR T-cell therapy. Elevated levels of serum IL-7 and IL-15 following a fludarabine-containing lymphodepletion regimen appear to associate with the engraftment of adoptively transferred T cells in cohort 1. The role of IL-7 and IL-15 in contributing to CAR-T proliferation and/or engraftment has been described extensively in hematological malignancies [32]. Our findings suggest that a similar phenomenon exists in adoptive TCR-based immunotherapy in sarcoma. A preparative regimen consisting solely of cyclophosphamide is not sufficient for optimal SPEAR T-cell engraftment in the studied clinical setting, as evidenced by poor peak persistence in four out of the five patients treated in cohort 3. The only responder in this cohort was a pediatric patient who had the highest post-conditioning serum levels of IL-15. Persisting SPEAR T cells isolated from this patient, approximately 1 year after infusion, maintained ex vivo cytotoxicity, further suggesting that fludarabine itself is unlikely to be contributing to long-term anti-tumor activity (Fig. 5b). Cyclophosphamide does have a modest impact on IL-7 and IL-15 induction, but this induction appears to be sub-optimal for most patients when not used in conjunction with fludarabine. Based on our findings, we would strongly suggest a lymphodepletion regimen containing a high dose of fludarabine with cyclophosphamide for SS patients being treated with SPEAR T-cells.

We also investigated the contribution of pre-infusion antigen expression to the magnitude and duration of clinical response. Our results indicate that when given comparable lymphodepletion regimens, patients with higher intra-tumoral NY-ESO-1 expression (cohort 1) are more likely to have deeper and longer responses as compared with patients with lower NY-ESO-1 expression (cohort 2; Fig. 1). There was not an appreciable difference in peak persistence between responders and non-responders in cohort 2 (Fig. 2a), despite a preparative regimen incorporating fludarabine. This observation suggests that SPEAR T-cell engraftment is at least in part driven by antigen expression. As SPEAR T cells utilize an affinity-enhanced TCR, their proliferation should depend on the quantity of SLLMWITQC-HLA-A*2:01 and/or -HLA-A*2:06 complexes present within the tumor or in peripheral lymphoid tissue. This suggests a positive correlative relationship between overall NY-ESO-1 intra-tumoral expression and peptide-major histocompatibility complex (MHC) expression.

Patients with lower levels of pre-infusion intra-tumoral NY-ESO-1 expression still had meaningful responses (Fig. 1). Our data suggest that the differences in response rates between cohort 1 and cohort 2 are largely dependent upon target antigen expression (Fig. 1). As such, increasing NY-ESO-1 expression may be an attractive strategy to improve the magnitude and duration of response. DNA methyl transferase inhibitors (DNMTi) represent one such strategy, as they have been shown to induce cancer-testis antigen expression [33–35]. As of this publication, there is currently one open trial using engineered T cells specific to an NY-ESO-1 peptide being used in conjunction with decitabine (NCT03017131) in ovarian cancer. Additional studies will be needed to understand the contribution of intra-tumoral target antigen and response in solid malignancies.

NY-ESO-1 is an immunogenic intracellular cancer-testis antigen expressed across a variety of tumors but absent in non-malignant tissue [36]. This makes it an attractive target for TCR-based therapies, and consequently, numerous immune therapies targeting NY-ESO-1 are in clinical development [36]. A unique mechanism of resistance to adoptive cell therapy is immune escape via loss of target antigen. Antigen loss has been described with CARs in the setting of hematologic malignancies [15, 37, 38]. However, antigen loss has not been widely described in non-hematopoietic solid tumors [39]. To address this potential mechanism of resistance to our therapy, we biopsied tumors from patients enrolled on our study prior to and following infusion. Our data demonstrate that loss of NY-ESO-1 protein or mRNA in SS patients treated with SPEAR T cells is not a widespread phenomenon, possibly due to the aberrant epigenetic mechanisms in SS tumors caused by the defective SWI/SNF complex [28]. In fact several patients received second infusions after confirmation of antigen presence. Of 10 patients treated, one patient experienced a PR (duration of at least 4 weeks prior to data cut-off) and another patient had a CR (duration of at least 8 weeks prior to data cut-off). Following adoptive T-cell transfer, only three biopsies out of 20 biopsies tested had low antigen expression by IHC (Fig. 4b, Additional file 3: Figure S2). Additionally, the expression of other tumor-associated antigens PRAME and MAGE-A4 [6] remained stable across all time points, implying multiple targets are available for TCR-based therapies. Our data suggest that immune evasion via antigen loss is probably not a significant mechanism of resistance in SS patients treated with SPEAR T cells.

Loss of genes responsible for antigen presentation, including but not limited to MHC class I and β2-microglobulin, have been associated with disease progression during checkpoint inhibitor therapy in melanoma [40, 41]. To date, the prevalence of these mechanisms of resistance to checkpoint inhibition is not yet well characterized across other cancers [42, 43]. Although SPEAR T-cell therapy is distinct from checkpoint blockade, defects in or loss of antigen presentation and processing represent shared potential mechanisms of resistance. In addition to investigating target antigen expression from biopsies, we also explored these potential mechanisms of resistance. Our data indicate that transcriptional loss of expression of genes responsible for antigen processing and presentation are not implicated in progression on SPEAR T-cell therapy in SS. Loss of heterozygosity has recently been described as a significant mechanism of resistance for immunotherapy [42]. Specific loss or mutations of the HLA-A*02 allele or defects in SLLMWITQC peptide loading onto HLA-A*02 have not yet been ruled out as mechanisms of resistance in our study.

The ability of the adoptively transferred T cells to traffic to and subsequently function within the tumor microenvironment is critically important, as T cells are generally administered systemically and must traffic to metastatic disease sites. We analyzed biopsies for immune markers associated with response to and progression on immune therapy. In agreement with previous studies, pre-infusion biopsies from patients enrolled in this trial were minimally infiltrated with T cells and did not have appreciable PD-L1 expression [6, 7]. These data are consistent with the lack of efficacy observed to date with checkpoint inhibitors in SS [3]. Despite this challenge, our TCR-modified T cells were able to traffic into the tumor microenvironment in selected samples studied. These emerging data suggest that SPEAR T-cell therapy could be an attractive modality to target tumors that are poorly infiltrated by T cells, which are not suitable for other immune therapies, such as PD-1/PD-L1 blockade [27, 44].

Global leukocyte infiltration of the SS microenvironment is relatively low post infusion compared to infiltration observed in inflamed tumor types, where > 30% infiltration is often observed [45]. In addition to low amounts of infiltration, a relatively high abundance of CD163⁺ TAM was observed across all time points analyzed. These macrophages are associated with a tolerogenic M2 phenotype [46] and have been associated with poor outcomes in sarcoma and other tumors [6, 47]. Additional studies are needed to elucidate the relationship between CD163⁺ TAM and SPEAR T cells.

We are encouraged by the clinical results from this study of adoptive T-cell therapy for SS using an affinity-enhanced NY-ESO-1^c259 TCR. Analyses have revealed the contribution of intra-tumoral target antigen expression and fludarabine to T-cell engraftment. These data provide an initial rationale as to how tumors with histologies resistant to checkpoint blockade can be successfully targeted with adoptive T-cell therapy. Furthermore, these data are the first to demonstrate successful infiltration of solid tumors by SPEAR T cells, which are able to kill tumor cells. Our data also suggest that antigen loss or alterations in the expression of antigen processing proteins are not primary mechanisms of resistance. Moreover, the therapeutic efficacy may be enhanced through use of a high dose fludarabine-containing preparative lymphodepletion regimen, by promoting greater engraftment at the tumor site, and through modulation of TAM.