Pembrolizumab in Advanced Soft Tissue and Bone Sarcomas: Results of SARC028, A Multicentre, Single arm, Phase 2 Trial

Abstract

Background:

Patients with advanced sarcomas have a poor prognosis and few treatment options improve overall survival. Chemotherapy and targeted therapies offer short-lived disease control. We assessed pembrolizumab (P), an anti-PD-1 antibody, for safety and activity in patients with advanced soft tissue (STS) and bone (BS) sarcomas.

Methods:

In this non-randomized, open-label, phase 2 study, we enrolled patients 18 years or older (12 years or older if they had BS), who had received at least one prior systemic therapy, had at least one measurable lesion by RECIST 1.1, and had at least one lesion accessible for biopsy. The primary endpoint was objective response rate (ORR) by RECIST 1.1. Secondary endpoints included safety, ORR by immune-related response criteria (irRC), progression-free survival (PFS), overall survival (OS), and correlation of PD-L1 expression with clinical benefit. The STS arm had 10 pts in each of 4 cohorts: undifferentiated pleomorphic sarcoma (UPS), dedifferentiated liposarcoma (DDLPS), synovial sarcoma (SS) and leiomyosarcoma (LMS). The BS arm included 40 pts with osteosarcoma (OGS), Ewing sarcoma (ES) or dedifferentiated chondrosarcoma (CS). All patients were treated with pembrolizumab at 200 mg intravenously every 3 weeks. Imaging was performed at week 8 and every 12 weeks thereafter. Pre- and on-treatment biopsies and blood were required and collected for correlative studies. Patients who received at least one dose of pembrolizumab were included in the safety analysis and patients who progressed or reached at least one scan assessment were included in the efficacy analysis. The study has completed accrual to all BS cohorts, as well as to LMS and SS cohorts, however the UPS and DDLPS cohorts have resumed accrual to a planned 30 patient expansion in each cohort. This trial is registered on ClinicalTrials.gov, NCT02301039.

Findings:

86 pts were enrolled, 80 were evaluable for response. For STS, median follow-up was 19.1 months. The ORR in the STS cohort was 18% and clinical activity was variable by histologic subtype: 40% ORR in UPS (1 CR+3PR/10), 2 PR/10 in LPS, 1PR/10 in SS and 0/10 in LMS. For BS, median follow-up was 17.8 months, ORR 5%, with 1PR/22 in OGS, 1PR/5 in CS and 0/13 in ES. The most common grade 3–4 adverse events included fatigue in 11 patients, anemia in 9 patients, and lymphopenia in 8 patients. Nine patients experienced serious adverse events including 2 with adrenal insufficiency, 2 with pneumonitis, and 1 with interstitial nephritis, all considered immune-related.

Interpretation:

Pembrolizumab has meaningful clinical activity in UPS and LPS, and expansion cohorts in those subtypes are ongoing to confirm and better characterize its efficacy.

The study was partially funded by Merck, Inc., SARC, the Sarcoma Foundation of America, QuadW, and other philanthropic sources.

INTRODUCTION

Sarcomas are broadly categorized into soft tissue sarcoma (STS) and bone sarcomas (BS) and represent a heterogeneous group of mesenchymal malignancies with more than 50 histologic subtypes(1). Studies to better understand sarcomas and to improve therapeutic outcomes are limited by their rarity and diversity. The median overall survival (OS) is around 2 years for advanced leiomyosarcoma but under one year for most other advanced STS and only about 10% of patients are alive at 5 years(2). Treatment options are limited and generally palliative while the expected benefits are tempered by significant side effects. Response to conventional chemotherapy and radiation therapy is dependent on specific histology, as some subtypes are relatively chemotherapy resistant. The last decade has seen novel agents investigated in a collaborative fashion in the treatment of sarcoma with large randomized controlled clinical trials leading to the FDA approvals of several agents including pazopanib, trabectedin, eribulin, and olaratumab (3–6). However, such therapies remain without a substantial cure rate, prompting the need for development of novel agents. Similarly, adult patients with metastatic bone sarcomas have a 5-year overall survival rate of less than 25% and a dearth of therapeutic or curative options(7, 8).

Immunotherapy is already approved in some countries for osteosarcoma, in the form of adjuvant mifamurtide, a non-specific immune stimulator that was demonstrated to improve overall survival in a phase III trial(9). The promise of immunotherapy gained broader appeal as anti-PD1 antibody studies demonstrated the benefit of immune checkpoint inhibition beyond melanoma; for instance, pembrolizumab has shown therapeutic benefit in non-small cell lung cancer, renal cell carcinoma, bladder cancer, Hodgkin’s lymphoma, and Merkel cell carcinoma(10–14). However, beyond mifamurtide, immunotherapy has had limited therapeutic benefit in STS and BS since studies utilizing cytokines or immune adjuvants did not achieve their primary endpoints(9, 15–17). In this open-label multi-centre phase 2 study (SARC028), we sought to determine the safety and efficacy of immune checkpoint blockade with pembrolizumab in patients with advanced STS and BS.

METHODS

Study design and participants

SARC028 was a multicenter, two-cohort, open label phase 2 trial of pembrolizumab (Merck, Kenilworth, NJ, USA) monotherapy performed at 12 US sites that are members of the Sarcoma Alliance for Research through Collaboration (SARC). Patients were required to be age 12 years or older for the BS cohort, and 18 years or older for the STS cohort, and to have histological evidence of metastatic or surgically unresectable locally advanced sarcoma including one of the following histologic subtypes: leiomyosarcoma (LMS), poorly differentiated/de-differentiated liposarcoma (LPS), undifferentiated pleomorphic sarcoma (UPS), synovial sarcoma (SS), Ewing sarcoma (ES), osteosarcoma (OGS), and de-differentiated or mesenchymal chondrosarcoma (CS). The histologic subtypes included were chosen on the basis of prevalence therefore we decided to study the most common STS and BS given the limitations of available funding. Patients had to have measurable disease by CT or MRI (per Response Evaluation Criteria In Solid Tumors [RECIST] v1.1), at least one site of disease that was safely accessible for pre- and on-treatment core biopsies, a life expectancy of more than 12 weeks, and Eastern Cooperative Oncology Group (ECOG) Performance Status of 0 or 1. Patients may have received up to three prior lines of systemic anticancer therapy.

Baseline laboratory assessments to determine eligibility included adequate kidney, liver, and bone marrow function, and were completed within 14 days of pembrolizumab administration. Key exclusion criteria were active brain metastases or any serious or uncontrolled medical disorder that would affect study participation. Patients with an active autoimmune disease or syndrome, or patients requiring chronic steroid or immunosuppressive agent use were not eligible, except those with vitiligo or resolved childhood asthma/atopy. Prior treatment with anti-PD-1 or anti-PD-L1 antibodies was not allowed.

Treatment with chemotherapy, radiotherapy, biologics for cancer, or investigational therapy was not permitted within 28 days of first administration of pembrolizumab, and previous palliative radiotherapy had to have been completed at least 2 weeks (or 4 weeks if wide field) before pembrolizumab administration.

The protocol was approved by the institutional review boards or independent ethics committees at each site and conducted according to Good Clinical Practice guidelines, per the International Conference on Harmonisation. Patients provided written informed consent based on Declaration of Helsinki principles.

Procedures

Patients were enrolled into one of two arms based on histology: STS (Arm A) or BS (Arm B). Within Arm A, patients were enrolled into one of 4 cohorts based on sarcoma sub-type with a pre-planned goal of enrolling 10 patients per cohort. Enrollment of 40 patients was planned for Arm B but the number of patients with ES, OGS or CS was not pre-specified. Patients received pembrolizumab 200 mg by a 30-minutes intravenous infusion every 3 weeks until disease progression or unacceptable toxicity. Population pharmacokinetic studies have shown that a fixed dose of pembrolizumab has a comparable pharmacokinetic and pharmacodynamic profile to weight-based dosing and is currently the FDA approved regimen for non-small cell lung cancer, head and neck cancer, and other indications. Disease assessments were performed using CT or MRI imaging at baseline, at 8 weeks, then every 12 weeks until disease progression. Response was determined by investigators using RECIST v1.1 and no central radiology review was performed. Treatment beyond RECIST-defined progression was permitted if pembrolizumab was tolerated, and clinical benefit was noted based on the investigator’s assessment. No dose modifications were allowed but dose delays were permitted for adverse events. Patients requiring treatment discontinuation due to adverse events were followed until disease progression or initiation of subsequent therapy and at 30 days after the last dose of pembrolizumab. Safety assessments including laboratory monitoring were performed during screening and on day 1 of each cycle of therapy. Adverse events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE v4.0) during treatment and up to 30 days after treatment discontinuation.

Pre- and on-treatment biopsies were required and obtained during screening [prior to first study drug administration] and at week 8 of therapy; biopsies were optional at the time of disease progression. Blood for correlative analysis was obtained during screening, at the time of subsequent disease assessments (e.g., at week 8 and then every 12 weeks) and at progression.

Immunohistochemistry (IHC) and Analysis of PD-L1

Hematoxylin & eosin (H&E) slides from all tissue specimens were reviewed by 3 pathologists to evaluate the presence of malignant cells and to select the best representative tumor block from each case. The slides were stained in a Leica Bond Max stainer (Leica Biosystems, Nussloch, GmbH). The primary antibody was PD-L1, clone 22C3 (Dako, Santa Clara, CA, catalog # M365329-1, dilution 1:50). PD-L1 staining was detected using diaminobenzidine (DAB) counterstained with hematoxylin (Leica Bond Polymer Refine kit, Leica Biosystems, Nussloch, GmbH). The slides were digitalized in an Aperio AT2 scanner (Leica Biosystems, Nussloch, GmbH). PD-L1 evaluation was performed by 3 pathologists and the IHC score was expressed as percentage of tumor cells positive for PD-L1 showing a distinct cell membranous accentuation. A tumor was considered PD-L1 positive if >1% of tumor cells showed membranous staining. The final IHC score was reviewed by a pathologist expert in soft tissue tumors (AJL).

Outcomes

The primary endpoint was objective response by RECIST v1.1. Objective response rate (ORR) was defined as the proportion of patients in each cohort with best overall response; complete or partial response assessed by the investigator. Responses were confirmed with a second scan at least 4 weeks after criteria for objective response were met. The duration of overall response was measured from the time criteria were met for CR or PR until the first date that recurrent or progressive disease was objectively documented. Secondary endpoints included objective response rate by immune-related response criteria (irRC), incidence of adverse events (AEs) and immune-related adverse events (irAEs), progression-free survival (PFS), and OS. Exploratory objectives included the correlation of PD-L1 expression with clinical outcome, as well as other biomarker analyses including immune monitoring in peripheral blood and tumor tissues for which data is still being analyzed.

Statistical analysis

We hypothesized that pembrolizumab would lead to clinical benefit in patients with advanced soft tissue and bone sarcomas. A patient was classified as a treatment success if he/she achieved a PR or better by RECIST 1.1. For each arm, an ORR of 25% was considered clinically meaningful and a response rate less than 10% was considered ineffective. In this study with a single stage phase 2 design, the treatment was to be considered a success if 8 or more of 40 enrolled patients in each arm had a PR or better by RECIST 1.1. This design has a one-sided type I error of 4.2% and a power of 82% to detect a difference in objective response rates between 10% and 25%. In advanced STS, a regimen that improves the 12-weeks PFS rate (PFR) from a historical rate of 20% to >40% is considered indicative of an active therapy. We estimated that 40 patients will provide 87% power to detect an improvement in the 12-week PFR from 20% to 40% with a one-sided type I error of 4%(18).

Objective response rate was summarized by considering the best response observed while on study treatment. Progression-free survival was calculated from the first date of study treatment to the earlier date of progression or death, and patients who did not progress or die were censored at their date of last contact. The Kaplan-Meier method was used to estimate OS and PFS(19), with two-sided confidence intervals (CIs) for the medians calculated using the method of Brookmeyer and Crowley(20), and for the point estimates using standard methods. Confidence intervals for response and PFS rates were two sided and calculated using the exact method. All statistical analyses were performed using SAS version 9.4.

Patients who received at least one dose of pembrolizumab were included in the safety analysis. Patients who received at least one dose of treatment and either progressed or underwent at least one disease assessment were considered evaluable and included in the primary efficacy endpoint of ORR. No interim analyses were planned or performed. Analyses by histological subtype, as well as correlative analyses were exploratory. All 95% confidence intervals were 2-sided.

This trial is registered on ClinicalTrials.gov, NCT02301039.

Role of the funding source

The study was investigator-initiated and received study drug and partial funding from Merck. Additional funding was obtained from philanthropic sources. The investigators designed the study, collected, analyzed, and interpreted the data independently. All drafts of this report were prepared by the corresponding author with input from all co-authors. Merck, Inc. reviewed the manuscript prior to submission. Raw data were made accessible to all authors. All authors made the decision to submit the report for publication. This study was not supported fully or in part by the National Institutes of Health (NIH) and none of the authors are employed by the NIH.

RESULTS

Between March 13, 2015 and February 18, 2016, 86 patients were enrolled and 84 patients were treated with pembrolizumab (Table 1). The last data cutoff was March 1, 2017, which provided a median follow-up of 18.7 months (IQR, 12.3–19.3 months). Demographic and safety data are presented for all 84 treated patients (2 patients were deemed ineligible prior to initiating study treatment and were never dosed with pembrolizumab).

Table 1.

Demographic and Clinical Characteristics (Among 84 Eligible, Treated Patients)

	Treatment Arm
	Bone (N= 42)	Soft Tissue (N= 42)
Age (years)
Median (Range) IQR	33 (16–70) (22–48)	53 (18–81) (45–63)
Sex
Female	16 (38%)	15 (36%)
Male	26 (62%)	27 (64%)
Prior Therapies
1	8 (19%)	8 (19%)
2	16 (38%)	17 (40.5%)
3	18 (43%)	17 (40.5%)
Prior Treatment in Metastatic Setting
No	16 (38%)	20 (48%)
Yes	26 (62%)	22 (52%)

Demographic and clinical characteristics are presented in Table 1 and specific histologic subtypes included are presented in Table 3. A total of 86 patients were enrolled, 84 were treated with at least one dose of pembrolizumab, 42 on either arm. Patients with STS had median age of 53 (Range, 18–81); patients with BS were younger with median age of 33 (Range, 16–70). Thirty one (37%) of 84 patients were female and the proportion was similar in both arms. In STS patients, 27 (64%) out of 42 received prior adjuvant or neo-adjuvant therapy, but 20 (48%) out of 42 patients had not received prior systemic therapy in the metastatic setting. Of all 84 patients treated with pembrolizumab, 35 (42%) had received 3 prior systemic therapies. All BS patients had received prior therapy in the neo-adjuvant or adjuvant setting, while 16 (38%) out of 42 had no prior therapy in the metastatic setting.

Table 3.

Best Response among patients evaluable for response

Sarcoma Subtype	Best Response
Soft Tissue Sarcomas	CR	PR	SD	PD	Total
Leiomyosarcoma	0	0	6	4	10
Undifferentiated Pleomorphic Sarcoma	1 (10%)	3 (30%)	3	3	10
Liposarcoma	0	2 (20%)	4	4	10
Synovial sarcoma	0	1 (10%)	2	7	10
STS Total	1 (2.5%)	6 (15%)	15	18	40
Bone Sarcomas	CR	PR	SD	PD	Total
Chondrosarcoma	0	1 (20%)	1	3	5
Ewing’s	0	0 (0%)	2	11	13
Osteosarcoma	0	1 (5%)	6	15	22
BS Total	0	2 (5%)	9	29	40

At the time of this analysis, only 2 patients remain on study treatment. One patient with osteosarcoma had a PR and requested to discontinue study therapy but is receiving pembrolizumab on a compassionate access basis, and remains on therapy at the time of this report. The most common reason for discontinuation was disease progression or death (35 on STS cohort and 33 on BS cohort). Pembrolizumab was well tolerated in this population but 9 patients discontinued secondary to toxicity (3 on STS arm and 6 on BS arm). While most patients experienced AEs of any grade (Webappendix Toxicity Data-All AEs, page 5), only 9 patients (11%, 5 BS, 4 STS) out of 84 treated patients had treatment-related serious adverse events (SAEs), of which only pneumonitis was seen in both arms. Other treatment-related SAEs included interstitial nephritis (1), infectious pneumonia (1), bone pain (1), pleural effusion (1) and hypoxia (1) in the BS arm, while adrenal insufficiency (2) and pulmonary embolism (1) were reported in the STS arm. None of these treatment-related SAEs resulted in a fatal outcome (Table 2).

Table 2.

Serious Adverse Events Related to Treatment

	BONE (N = 42)			SOFT TISSUE (N = 42)
	Grade1			Grade1
	2	3	4	2	3	4
Adverse Event Description
Overall	1 (2%)	3 (7%)	1 (2%)	1 (2%)	3 (7%)	0 (0%)
Cardiovascular disorders					1 (2%)
PULMONARY EMBOLISM					1 (2%)
Endocrine disorders				1 (2%)	1 (2%)
ADRENAL INSUFFICIENCY				1 (2%)	1 (2%)
Immune system disorders		1 (2%)
INTERSTITIAL NEPHRITIS		1 (2%)
Infections and infestations		1 (2%)
INFECTIOUS PNEUMONIA		1 (2%)
Musculoskeletal and connective tissue disorders		1 (2%)
BONE PAIN		1 (2%)
Respiratory, thoracic and mediastinal disorders	1 (2%)	1 (2%)	1 (2%)		1 (2%)
HYPOXIA			1 (2%)
PLEURAL EFFUSION		1 (2%)
PNEUMONITIS	1 (2%)				1 (2%)

¹No Grade 1 or Grade 5 serious adverse events related to treatment reported during the course of the trial

Efficacy data are presented for all 80 patients evaluable for the primary endpoint, six patients were considered not evaluable for the primary efficacy endpoint (2 ineligible and 4 withdrew consent prior to the first disease assessment). In the STS arm, confirmed objective responses by RECIST were observed in patients with UPS and LPS. One patient with UPS achieved a confirmed CR, a 50 year-old female with primarily pulmonary target lesions with a duration of CR exceeding 13 months. Three patients experienced durable confirmed PRs out of 10 evaluable patients for an ORR in UPS of 40% (Table 3). Similarly, 2 (20%) patients with LPS had confirmed PRs out of 10 evaluable patients. One patient with SS experienced a short-lived PR, while another remains on treatment (>15 months) and continues to derive clinical benefit. This SS patient experienced initial disease progression in one lung lesion but remained on treatment as she was deemed to be deriving clinical benefit by the treating investigator. Indeed the patient (classified as irPR using irRC) eventually had the progressing lesion resected and remains on treatment with continued disease control of the remaining evaluable lesions. ORR was therefore observed in 7 (18%, 95% CI, 7%−33%) out of 40 evaluable patients (Figure 1a, and Table 3). Responses were generally durable with a median duration of 33 weeks (IQR, 23–49 weeks) with some responses still ongoing at the time of this analysis (Figure 1b). Response assessment with irRC was generally concordant with RECIST in the STS cohort; two patients with RECIST SD being classified as irPR and vice versa (Webappendix Tables S2a, page 4). The overall response rate remained the same, and since the protocol allowed for continued therapy for patients deriving clinical benefit, no patient discontinued therapy secondary to progression by RECIST that was later determined to be responding by irRC.

Figures 1a:

Waterfall plot of best percent change from baseline in size of Target Lesions (sum of diameters)- STS

Figure 1b:

Spider plots of % of baseline tumor size (sum of target lesion diameters) at each assessment-STS

In the overall STS population, 37 out of 40 evaluable patients have progressed and the median PFS was 18 weeks [95% CI, 8–21] (Webappendix Figures S1a, page 3), and the 12-weeks PFR was 55% [95% CI, 40–70] which is statistically significantly higher than the historical threshold of 40% expected from an active regimen in STS. In patients with UPS the median PFS was 30 weeks [95% CI, 8–68], and 12-week PFR 70% [95% CI, 42–98], and in LPS the median PFS was 25 weeks [95% CI, 8–42], and 12-week PFR 60% [95% CI, 30–90]. The median OS was 49 weeks [95% CI, 34–73] in the overall STS population where 25 deaths have occurred, all were related to disease progression, and in UPS the median OS has not been reached at the time of this analysis (Webappendix Figure S1b, page 3).

In the BS arm, one confirmed PR was observed out of 22 OGS patients (<5%), and one confirmed PR was observed in a patient with CS (Webappendix Table S2b, page 2; Figure 1c). The median duration of response was 43 weeks (IQR, 25–61) indicating durable responses. Thirty-eight patients have progressed, one patient was censored, and the median PFS was 8 weeks [95% CI, 7–9] indicating that most patients with BS had progressed at the time of first disease assessment (Webappendix Figure S2a, page 4). Twenty-five deaths have occurred in the overall BS population, all were related to disease progression, and the median OS was 52 weeks [95% CI, 40–72] and the median OS was not reached in CS patients as would be expected for this population. Response assessment with irRC compared with RECIST in the BS cohort indicated that only one ES patient with RECIST SD would be re-classified as irPR (Webappendix Table S2b, page 4).

Figure 1c:

Waterfall plot of best percent change from baseline in size of Target Lesions (sum of diameters)- BS

Tumor biopsies were safely obtained on 78 patients pre-treatment (>90%) and on 68 patients on-treatment (72%, all of which had matching pre-treatment samples). Of 78 total samples obtained pre-treatment, 70 passed quality control for presence of tumor tissue and were successfully analyzed for pre-treatment PD-L1 expression levels. PD-L1 was positive at the 1% threshold only in 3 out of 70 pre-treatment tissue, all three had UPS. Of those 3, only 2 patients were evaluable for response, 1 had a CR and the other a PR.

DISCUSSION

To our knowledge, SARC028 is the first prospective multi-centre open-label phase 2 study of immune checkpoint blockade in patients with advanced soft tissue and bone sarcomas, pembrolizumab monotherapy was associated with clinically meaningful and sustained objective responses in 18% in STS, while the response rate was only 5% in BS. Despite the response rate in STS being short of reaching the primary endpoint (which required 8 responses out of 40), the 12-weeks PFR in STS was 55% (95% CI, 40–70) and was significantly higher (p=0.001) than the 40% 12-weeks PFR required for an active regimen, suggesting meaningful clinical activity in this population. Importantly, the objective responses were limited to patients with UPS and LPS, and so was the improvement in 12-weeks PFR, which was 70% in patients with UPS. The median duration of responses in UPS (30 weeks) and median OS (median not reached) are highly encouraging and are suggestive of similar clinical activity for immune checkpoint blockade observed in other malignancies.

The UPS cohort is being expanded to confirm and further characterize the clinical activity. UPS was interestingly noted to have PD-L1 expression correlating with T cell infiltration. This indicates that UPS may fit the “inflamed tumor” model and could explain the activity of single agent PD-1 anitbodies(21). PD-L1 expression was observed in only 2 (4%) of samples tested in which tumor response was evaluable, but both tumors were UPS and responded to therapy. This is consistent with previous reports where PD-L1 expression was found only in 5% of STS, 2 of 36 non-GIST tumors, both of which were UPS(22). More importantly, it is clear that responses were seen even in the absence of PD-L1 expression, consistent with other tumor types including melanoma where PD-L1 negative tumors may still respond to checkpoint blockade. The role of the PD-L1 expression in STS remains unclear(23–25), however, ongoing analyses with multi-color IHC being performed on samples collected from our study will help examine this question and will be reported separately.

The activity observed in LPS is similarly encouraging with clinical activity observed in the absence of PD-L1 expression. Also, the LPS cohort included multiple high grade LPS subtypes, however, due to lack of responses in the myxoid/round cell LPS, this cohort will be expanded only with DDLPS.

The results in LMS are consistent with a phase 2 evaluation of nivolumab that was stopped early for futility, confirming that single agent anti-PD1 therapy may not be able to elicit an immune response in LMS(26). The mechanisms of resistance to immunotherapy in LMS remain unclear although a recent report implicated PTEN loss as a potential mechanism of resistance. Such investigations could yield targetable pathways, in this case the PI3K-AKT pathway, that could be pursued in combination with checkpoint blockade(27).

SS has a high expression of cancer testis antigens (CTAs) and was anticipated to be responsive given prior successes in TCR-modified TILs therapy, however the vast majority of patients on this study experienced rapid progression(28). A Phase II pilot study trial evaluating ipilimumab in SS expressing the NY-ESO-1 antigen enrolled 6 patients and was terminated early because of lack of activity and immune responses(29). In our study, only one patient with SS had a short-lived PR and another had prolonged stable disease and is still deriving clinical benefit while on therapy for almost 2 years. The latter patient’s response was indeed classified as a partial response (irPR) by immune-related Response Criteria (irRC) confirming that this pattern of response can correlate with clinical benefit.

The activity of pembrolizumab in bone sarcomas was limited, although the two responses observed; one each in osteosarcoma and chondrosarcoma were deep (>50% tumor shrinkage) and durable (>6 months).

Although immunotherapy had shown promise in the adjuvant therapy of osteosarcoma, only one patient with metastatic OS had a response to pembrolizumab. Osteosarcoma is reported to have variable PD-L1 expression and frequent loss of MHC I potentially facilitating immune evasion (30–32). Mifarmutide has been shown to increase immune cell infiltration into osteosarcoma metastases, a critical step to improve the efficacy of anti-PD-1 antibodies(33).

Pembrolizumab had no activity in the 13 patients with ES, which could be related to the highly suppressive immune microenvironment in the tumor. Only one ES patient had an irPR suggesting that a small subset of ES patients could benefit from single agent therapy. Nevertheless, it is clear that combination approaches will be needed to assess activity for this rapidly growing aggressive malignancy. In a retrospective study of ES tumor tissues, high concentrations of regulatory T cells (Tregs) were seen in patients that presented with metastatic ES potentially contributing to inhibition of cytotoxic CD8+ T lymphocytes and promoting tumor escape (34, 35).

Only 5 patients with chondrosarcoma were enrolled onto this study, one patient achieved a PR. This is consistent with one retrospective series in which one patient with dedifferentiated chondrosarcoma responded to nivolumab(36). This is highly significant given the lack of therapeutic options for chondrosarcoma that has traditionally been refractory to most if not all anti-cancer therapies(37). In a retrospective report, 11 of 21 (52%) dedifferentiated chondrosarcomas examined had PD-L1 expression associated with infiltrating TILs and HLA-I expression(38). Taken together, further evaluation of immune checkpoint blockade in CS is warranted.

The safety profile of pembrolizumab in this population was consistent with what has been observed for other approved indications; specifically there was no apparent increase in the incidence of pneumonitis despite the fact that the vast majority of advanced sarcomas patients have lung metastases.

Sarcoma is generally considered a “non-immunogenic” tumor; however, our study corroborates once more that sarcomas are highly variable in their biology and that each histologic subtype should be considered a separate therapeutic challenge that requires a distinct understanding of its immune and molecular biology. The role of PD-L1 and other potential biomarkers will be explored as we continue to evaluate the serial blood and tumor that were collected on this study to help further our understanding of the immunophenotype underlying the observed clinical differences of individual sarcoma subtypes.

The findings of our study are limited by its non-randomized design and the small sample size but are highly encouraging and, if the clinical activity is confirmed in larger studies, could be practice changing given that UPS and LPS are two of the three most common STS and together represent over 30% of all STS. The results of planned correlative analyses will hopefully improve our understanding of the immune response and therapies in sarcoma and help to prioritize combination strategies which could include chemotherapy, radiation therapy(39), targeted agents, and/or other immune checkpoint inhibitors with or without additional immune approaches.

Our study was a testament to the collaborative efforts of SARC investigators, pharmaceutical companies, and philanthropic organizations that allowed for rapid accrual of patients and the collection of high quality data and biospecimens that will lay the foundation for the future of immunotherapy in sarcoma.

Figure 1d:

Spider plots of % of baseline tumor size at each assessment- Bone Sarcomas

* Study day for 1 patient was truncated at 250 days to better show the earlier part of the graph.

RESEARCH IN CONTEXT.

Evidence before this study

We searched PubMed with the terms “metastatic sarcoma”, “clinical trials”, “PD-1”, “PD-L1”, “immune checkpoint blockade”, “immune response”, “pembrolizumab”, “nivolumab” for articles published in English until May 5^th, 2017 (date of our final search). The findings of this search suggested that patients with metastatic soft tissue or bone sarcomas have a poor prognosis with limited survival rates. First line single agent treatment of STS with chemotherapy remains a de facto standard of care that imparts response rates of 10–15%. Combination therapies have led to limited or no impact on overall survival, while consistently increasing the rates of adverse events. The addition of olaratumab to doxorubicin improved overall survival in a randomized Phase II setting and is being confirmed in an ongoing Phase III study. Second or third line targeted therapy with pazopanib provides benefit for a small proportion of patients although the improvements in outcome are limited to progression-free survival. Treatment of bone sarcomas with combination chemotherapy in the neo-adjuvant and adjuvant setting can be curative despite the significant toxicity, however in advanced bone sarcomas such therapies as associated with limited survival advantage. Sarcoma has not been traditionally considered an immunogenic tumor, however, several reports indicate that PD-L1 expression could be found in up to 30–40% of certain sarcoma subtypes. Also, the presence of infiltrating CD8+ T cells correlated with improved outcomes suggesting that the immune response plays a role in the natural history of the disease. Anecdotal reports as well as retrospective series have described responses to immune checkpoint blockade in select sarcomas. One phase 2 study of nivolumab in leiomyosarcoma was stopped early for futility but no prospective studies have been published to date in other soft tissue sarcoma subtypes or bone sarcomas. Treatment with antibodies targeting the PD-1/PD-L1 axis has resulted in durable responses in multiple cancers. Preclinical and clinical evidence provided the rationale to evaluate the role of immune checkpoint blockade in metastatic sarcomas.

Added Value of this study

To our knowledge, this study represents the first multi-centre prospective phase 2 study examining the role of PD-1 antibody therapy in soft tissue and bone sarcomas. The study was designed to take into account the heterogeneity of sarcomas and was limited to specific histologic subtypes that were intentionally balanced to allow for better efficacy signal identification. Among patients with soft tissue sarcomas, we observed an objective response rate of 18% in the overall population. Importantly, the benefit was limited to two of the three most common sarcoma histologic subtypes: 4/10 undifferentiated pleomorphic sarcoma had partial responses (PR, median overall survival was not reached), and 2/10 dedifferentiated liposarcomas (LPS) had a PR, prompting planned expansion cohorts for formal phase 2 evaluations in each of those subtypes. The quality of the responses was consistent with the durable benefit observed with checkpoint blockade in other cancers. We observed limited to no benefit in synovial sarcoma and leiomyosarcoma. Similarly in bone sarcomas, we observed infrequent responses suggesting the need for combination approaches in this population. PD-L1 expression was only observed in 3/70 tumors, two evaluable patients whose tumors expressed PD-L1 responded suggesting a correlation between PD-L1 expression and response, although 6 additional responses were observed in patients with PD-L1-negative tumor.

Implications of all the available evidence

These findings, taken together, suggest that pembrolizumab is an active agent in the specific soft tissues sarcoma subtypes UPS and DDLPS. Our findings in UPS and DDLPS, which represent more than 30% of all soft tissue sarcomas, are clinically meaningful and confirmation in larger cohorts is currently underway. These results provide a promising therapeutic option for patients with soft tissues sarcomas that can alter the natural history of the disease. Further investigation is required to determine the utility of predictive biomarkers for response and to understand the mechanisms of resistance in the other STS and BS subtypes where rational combination therapies could be considered.