You are here

  • Home
  • Archive
  • Volume 12, Issue 9
  • A multicenter, randomized, non-comparative, phase II study of nivolumab ± ipilimumab for patients with metastatic sarcoma (Alliance A091401): expansion cohorts and correlative analyses

Article Text

Article menu
Immunotherapy biomarkers
Original research
A multicenter, randomized, non-comparative, phase II study of nivolumab ± ipilimumab for patients with metastatic sarcoma (Alliance A091401): expansion cohorts and correlative analyses
  1. Nathan D Seligson1,
  2. James L Chen2,
  3. Austin C Goodrich3,
  4. Brian A Van Tine4,
  5. Jordan D Campbell3,
  6. Allison L Richards5,
  7. Cristina R Antonescu5,6,
  8. David A Liebner2,
  9. Mohammed M Milhem7,
  10. Howard Streicher8,
  11. William D Tap5,6,
  12. Gary K Schwartz9,
  13. Suzanne George10 and
  14. Sandra P D'Angelo5,6
  1. 1University of Florida, Jacksonville, Florida, USA
  2. 2Ohio State University, Columbus, Ohio, USA
  3. 3Mayo Clinic, Rochester, Minnesota, USA
  4. 4Washington University in Saint Louis, Saint Louis, Missouri, USA
  5. 5Memorial Sloan Kettering Cancer Center, New York, New York, USA
  6. 6Weill Cornell Medical College, New York, New York, USA
  7. 7University of Iowa, Iowa City, Iowa, USA
  8. 8National Cancer Institute, Bethesda, Maryland, USA
  9. 9Case Western Reserve University, Cleveland, Ohio, USA
  10. 10Dana Farber Cancer Institute, Boston, Massachusetts, USA
  1. Correspondence to Dr Sandra P D'Angelo; dangelos{at}mskcc.org

  • NDS and JLC are joint first authors.

Abstract

Background In this open-label, randomized, non-comparative, multicenter phase II study (Alliance A091401) we report on three expansion cohorts treated with nivolumab (N) with and without ipilimumab (N+I) and provide a multi-omic correlative analysis of actionable biomarkers.

Methods Patients were randomized (non-comparative) to receive either N or N+I. The primary endpoint was a 6-month confirmed response rate (CRR) defined by Response Evaluation Criteria in Solid Tumors version 1.1. Secondary endpoints included treatment-related adverse events (TRAEs), progression-free survival, and overall survival. Multi-omic correlative analyses were conducted using samples from both the primary and expansion cohorts.

Results A total of 66 patients were evaluated for the primary endpoint with disease including gastrointestinal stromal tumor (GIST, n=18), undifferentiated pleomorphic sarcoma (UPS, n=24), and dedifferentiated liposarcoma (DDLPS, n=24). Neither N nor N+I achieved a complete or partial response in the GIST expansion cohort. In DDLPS and UPS, the primary response endpoint of CRR was met with N+I (both 16.6%, 2/12) but not with N alone (both 8.3%, 1/12). In the GIST cohort, TRAE was higher with N+I treatment, halting enrollment as required per protocol. In a correlative analysis of patients for the expansion cohort and the original cohort (n=86), traditional biomarkers of immunotherapy response were not correlated with response in any histological subtype. Markers of genomic instability including the presence of gene fusions and increased subclonal mutations correlated with improved clinical outcomes.

Conclusions This expansion cohort reaffirms the outcomes of A091401. There remains a pressing need to determine the role of and predictive biomarkers for immunotherapy in sarcoma.

Trial registration number NCT02500797.

  • Immunotherapy
  • Biomarker

Data availability statement

Data are available in a public, open access repository. Data are available, via controlled access, from the NCBI dbGaP archive (database accession number phs003785.v1).

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/jitc-2024-009472

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Recent studies in soft-tissue sarcomas (STS) have identified a potential role for immunotherapy in a minor, but meaningful portion of these diseases. Biomarkers for response to this therapy are unknown.

    WHAT THIS STUDY ADDS

    • We confirmed clinical response to immunotherapy in STS, specifically in undifferentiated pleomorphic sarcoma and dedifferentiated liposarcoma subtypes. Furthermore, we provide preliminary evidence for the potential of markers of genomic instability, including the presence of gene fusions and increased subclonal mutations, as correlated with response to immunotherapy in STS.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • This study confirms the initial findings of A091401 and presents novel biomarkers, which could be used to guide immunotherapy in STS.

    Background

    Sarcomas represent a rare, histologically and molecularly diverse set of mesenchymal malignancies.1 Despite several new drug approvals for use in sarcoma, response rates to therapy remain poor for most histological subtypes.2–5 There remains a significant need to improve therapeutic approaches for patients with sarcoma.6 Over the past decade, checkpoint blockade has resulted in treatment paradigm shifts in many cancer types; however, this has not been the case in sarcoma.7–9 In Alliance for Clinical Trials in Oncology (Alliance) trial A091401, the immune checkpoint inhibitors (ICIs) nivolumab (N) and nivolumab with ipilimumab (N+I) demonstrated a confirmed response rate (CRR) of 5% and 16%, respectively, in patients with advanced sarcoma. In the monotherapy cohort, responses occurred in alveolar soft part sarcoma (ASPS) and non-uterine leiomyosarcoma (LMS). In the combination cohort, responses occurred in undifferentiated pleomorphic sarcoma (UPS), myxofibrosarcoma, LMS, sarcoma not otherwise specified, and angiosarcoma. Corroborating these findings, SARC028, a multicenter phase II study of pembrolizumab in advanced soft-tissue sarcoma (STS) and bone sarcoma, underscored that responses were highest in certain histological subtypes such as UPS, and less so in LMS and bone sarcomas. Additional clinical trials have confirmed dramatic responses in ASPS and the ICI atezolizumab is now FDA-approved for this disease (NCT03141684).10 11 These prior efforts underscore the importance of histology-driven approaches for drug development in sarcoma.

    Recent studies have shed light on the immune microenvironment of sarcomas. These include the presence of tertiary lymphoid structures (TLS), immune infiltrates such as CD8+T cells, and PD-1 expression.12 13 TLS may predict benefit to monotherapy immune checkpoint blockade in STS; however, whether that biomarker will be sustained across other combinatorial strategies remains uncertain. Tumor mutational burden (TMB), a key indicator of ICI response in most tumors, is generally low across sarcoma subtypes. TMB is not typically associated with response to ICI efficacy with the exception of rare, TMB-high sarcomas including pleomorphic dermal sarcoma and cutaneous angiosarcoma.13 Additional markers of genomic instability, such as gene fusions14 and subclonal mutations,15 have been associated with response to ICI but have not been fully explored in STS. Taken together, there is a need to identify biomarkers that are subtype driven to select patients for ICI therapy.

    To better understand the clinical benefit of ICI in sarcoma and to explore additional biomarkers, an expansion to Alliance A091401 was performed. Given prior signals of activity, these expansion cohorts included UPS and dedifferentiated liposarcoma (DDPLS). Gastrointestinal stromal tumors (GIST) were also included given their complex microenvironments.16 Here, we report on the results of this expansion cohort and the correlative results from the primary and exploratory biomarkers. Correlative results include data from immunohistochemistry (IHC) of both tumor and immune microenvironment, DNA sequencing, RNA expression data, and molecular pathway analysis.

    Methods

    Study design and patients

    Patient characteristics included eligible patients aged ≥18 years with histologically confirmed GIST, UPS, or DDLPS that were locally advanced, unresectable, or metastatic. Patients needed to have measurable tumor lesions per Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST version 1.1), an Eastern Cooperative Oncology Group performance status of 0 or 1, and an estimated life expectancy of >3 months. Patients received at least one previous systemic therapy with a minimum washout period of 28 days before study drug initiation. Previous therapy with any anti-PD-1, anti-PD-L1, or anti-CTLA-4 blocking antibodies was not permitted. Exclusion criteria included active brain metastases or a history of autoimmune diseases. Patients also needed adequate kidney, liver, and bone marrow function. Full eligibility criteria including other exclusion criteria and the required initial laboratory values are as previously published.17 Patients were recruited from multiple clinical sites across the USA (online supplemental table S1). Patients were assigned to treatment in an unblinded manner, as this trial was conducted as three independent, non-comparative phase II trials. Patients were assigned treatment in a 1:1 ratio using a dynamic allocation algorithm automatically generated by the Alliance Statistics and Stat Center and immediately transferred to the clinical site. Site of randomization was the only stratification factor used.

    Supplemental material

    Each patient was randomly assigned one of the two treatments (1:1), nivolumab 3 mg/kg as an intravenous infusion every 2 weeks until disease progression (N) or nivolumab 3 mg/kg and ipilimumab 1 mg/kg every 3 weeks for four doses followed by nivolumab 3 mg/kg every 2 weeks until disease progression (N+I). Patients were allowed to cross over due to disease progression from the nivolumab monotherapy group to the combination group provided there was radiographical or clinical confirmation of disease progression, a minimum of 10 weeks of single agent nivolumab treatment, and resolution of any toxic effects of previous therapy except for fatigue and alopecia. Each treatment arm and each tumor type were considered independent for a total of six trials.

    Evaluations

    Tumor assessments were conducted by radiological evaluation at each institution at baseline, every 6 weeks for the first 12 weeks, and every 8 weeks thereafter. Responses detected at these assessments required confirmation 4 weeks after the initial response. Activity analyses were done per protocol, including all patients who started treatment. Patients who had progressive disease within the first 12 weeks of treatment were permitted to continue treatment if the following prespecified criteria were met: evidence of clinical benefit, tolerance of study treatment, ≤4 new lesions, and less than a 40% increase in tumor burden. If progressive disease was confirmed at the subsequent 4-week assessment, the date of the initial progressive disease was used for analyses and the patient stopped study treatment. If progressive disease was not confirmed, the patient could remain on protocol treatment and the prior scan where progressive disease was suspected was used, going forward, as the baseline for determining tumor response.

    Adverse events (AEs) were graded in accordance with the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0 during treatment and up to 30 days after treatment discontinuation. Review boards at each of the participating institutions approved the study. The study was done in accordance with the Declaration of Helsinki and the Guidelines for Good Clinical Practice. Each patient signed an institutional review board-approved, protocol-specific, informed consent form in accordance with federal and institutional guidelines. This study was monitored at least twice per year by the Alliance Data and Safety Monitoring Board, a standing committee consisting of individuals outside of the Alliance Clinical Trials in Oncology Group for accrual, safety, and the primary endpoint evaluation.

    Outcomes

    The primary endpoint was confirmed objective response rate (CRR), defined as complete response (CR) or partial response (PR) by RECIST version 1.1 on two consecutive evaluations at least 4 weeks apart during the protocol-directed treatment. Secondary endpoints included duration of response, clinical benefit rate (CBR), progression-free survival (PFS), and overall survival (OS). CBR was defined as the proportion of patients with a best objective status of CR, PR, or stable disease (SD) at least once after study entry at a minimum interval of 6 weeks as defined by RECIST version 1.1 while receiving protocol treatment. Duration of response was defined as the time from first CR or PR to date of progressive disease or death. PFS was defined as the time from randomization to date of progressive disease or death. Patients who discontinued treatment for reasons other than progressive disease (such as AEs) were censored at the date of their most recent disease evaluation prior to receiving any future systemic treatment regimens. OS was calculated as the time between randomization and date of death. Patients who were lost to follow-up were censored for survival at the date they were last known to be alive and with progressive disease at the date of their most recent disease assessment.

    Statistical analysis

    The original study design was a two-stage Simon phase II clinical trial to assess whether the confirmed response rate was at most 5% against the alternative that the confirmed response rate was at least 20%. The rapid registration led to the study design being converted (protocol update 1, dated November 15, 2015) to a single-stage design with the same hypotheses, significance level, and power following consensus from the study team, the Alliance Data and Safety Monitoring Board, the Alliance Experimental Therapeutics and Rare Tumors Committee leadership, and the National Cancer Institute Cancer Therapy Evaluation Program. The expansion cohort was based on these data and as previously described.17 All enrolled patients who began study treatment were considered evaluable for the analysis of AEs, including data up to the date of ineligibility for any patients found ineligible after initiating study treatment. Categorical data analyses and summary statistics were used to report AEs. Cox-proportional hazard regression and Kaplan-Meier methods were used to estimate the distributions of all time-to-event endpoints. Data collection and statistical analyses were conducted by the Alliance Statistics and Data Center. All statistical analyses were done with SAS V.9.4 and R V.4.1.0, and were based on the study data lock on November 01, 2022.

    Correlative analysis

    Molecular correlative analysis was conducted for all tumors with samples and data available. Due to concerns about multiple comparisons and the low response rate of the cohort, we selected a cut-off of confirmed 12-week PFS (PFS<12 weeks: poor response; PFS≥12 weeks: good response) as the screening correlative endpoint.18 For molecular markers that correlated with 12 weeks PFS, we then tested PFS and OS by that marker. We conducted this correlative analysis as an exploratory aim. IHC, DNA, and RNA analyses were performed for available samples.

    Immunohistochemistry

    Multiplex immunofluorescence IHC was conducted at CellCarta using the Mosaic platform. Tested markers included PD-L1, FOXP3, CD68, CD8, CD4, and CD3. Data were measured as the percentage of cells positive for each marker. Three separate images were generated for each patient sample. Data from these images were averaged to create a single measure per patient for each marker.

    Next-generation sequencing

    Next-generation sequencing was conducted for available samples at Memorial Sloan Kettering Cancer Center. Formalin-fixed paraffin-embedded slides were scraped or microdissected based on pathological review of the H&E slide. Normal and tumor sample DNA isolation was followed by sequencing using Illumina Hiseq to coverage depths of 97× and 164×, respectively. The resulting FASTQs were aligned using Burroughs-Wheeler Aligner-MEM to the GRCh37 reference genome.19 Somatic genomic variants were called using the union of Mutect220 and Strelka221 with filtering using RepeatMasker and GNOMAD.22 TMB was calculated as the number of non-synonymous mutations in coding regions covered by baits. Microsatellite status was determined using MSISensor.23

    Transcriptome sequencing libraries were made following quantification and quality control as described above. Samples were barcoded and sequenced on a HiSeq 4000 in a PE100 run, using the HiSeq 3000/4000 SBS Kit (Illumina). On average, 88 million paired reads were generated per sample and 21% of the data mapped to the transcriptome. FASTQs were processed using an in-house RNA sequencing pipeline including STAR V.2.7.024 to Ensembl V.7525 for alignment and quality control by Picard. Gene fusions were called using high-confidence calls from either FusionCatcher or Arriba.26 Expression was quantified using HTSeq.27 Further sequencing methodology is available in the online supplemental methods.

    Mutation clonality

    Clonality was called by first calculating the cancer cell fraction of each mutation using ccf-annotate-maf in facets-suite28 (https://github.com/mskcc/facets-suite). A mutation was considered clonal if cancer cell fraction was greater than 0.8 or the cancer cell fraction was at least 0.7 and the upper bound of the cancer cell fraction was greater than 0.9. Otherwise, the mutation was considered subclonal. Only pathogenic genomic variants were included in the mutational clonality analysis.

    Additional statistical considerations

    Continuous molecular variables were compared using the Wilcoxon-Mann-Whitney test and categorical by χ2 test. Differential expression was conducted using DESeq229 for each unique gene with a mean expression of at least 50 reads for each histological subtype. Fusion analysis was performed by examining the presence or absence of fusions in the studies in aggregate and in individual tumor types. Gene pathway expression was evaluated with GSEA30 using the Hallmark gene sets31 with a significance filtering of absolute normalized enrichment scores (NES) ≥1.5 and q<0.1. Immune deconvolution was computed using CIBERSORT algorithm32 33 correcting for batch differences between the RNA-seq data in this study and the microarray-derived LM22 signature.32 P values ≤0.05 were considered statistically significant. For all molecular analysis, the false discovery rate was controlled with the Benjamini-Hochberg method using a threshold of q<0.05.34

    Results

    Expansion cohort demographics

    A total of 108 patients underwent central pathology review for eligibility between October 19, 2017 and July 26, 2018. Preregistration by histological category involved a potential 24 GIST, 37 DDLPS, and 47 UPS patients. Of the pre-registered 108 patients, 22 did not meet eligibility/inclusion criteria, 3 died or medically were unable to enroll, 2 were withdrawn by investigator decision, and 2 withdrew before enrollment. 79 patients proceeded to registration: 21 GIST, 29 DDLPS, and 29 UPS. Patient demographics and disease characteristics at registration are summarized in table 1 for the two treatment groups for each tumor type (online supplemental table S2). Following registration, two patients with UPS came off study before initiating therapy, both due to rapid disease progression requiring immediate initiation of alternative therapy. These two patients with UPS were thus not evaluable, due to coming off protocol before receiving treatment. Of the 21 GIST patients, 10 were randomized to receive N, and 11 were randomized to receive N+I. For the 29 DDLPS patients, 15 were randomized to receive N, and 14 were randomized to receive N+I. Finally, for the 27 UPS patients, 13 were randomized to N and 14 to N+I. Consolidated Standards of Reporting Trials diagrams are available for each histology included in online supplemental figure S1. Median follow-up was 271 days.

    View this table:
    Table 1

    Cohort demographics at registration

    Confirmed response rate

    The primary endpoint for this study was confirmed response over the first 6 months of treatment. Each arm and subtype were considered separately (three subtypes, two arms) for a total of six independent tests for confirmed response (figure 1, table 2).

    Figure 1
    Figure 1

    Response to immunotherapy. Swimmers plot for patients in the expansion cohort who received either nivolumab alone or nivolumab in combination with ipilimumab by histological subtype of sarcoma. Vertical dashed lines represent the 6-week and 12-week restaging. Squares, empty or filled in, represent suspected and confirmed progressive disease, respectively. At the time the data were frozen for this analysis, all patients were no longer actively receiving study treatment. Diamonds represent objective responses. DDLPS, dedifferentiated liposarcoma; GIST, gastrointestinal stromal tumor; UPS, undifferentiated pleomorphic sarcoma.

    View this table:
    Table 2

    Response endpoints

    For the DDLPS cohort, the first 12 evaluable patients per treatment arm were included in the primary analysis. The CRR for DDPLS was 8.3% (1/12) for N and 16.6% (2/12) for N+I. Duration of response for the one responder in the N arm was 32.1 months and 8.3 and 33.1 months in the N+I arm. There was an over-accrual of 3 patients in the N arm and 2 patients in the N+I arm. This led to an overall CRR of 1/15 (6.7%) in the N arm and 2/14 (14.3%) in the N+I arm for patients with DDPLS.

    Similarly, for the UPS cohort, the first 12 evaluable patients per treatment arm were included in the primary endpoint analysis. The CRR for UPS was 8.3% (1/12) for N and 16.6% (2/12) for N+I. Duration of response for the one responder in the N arm was 38.5 months and 7.6 and 31.6 months in the N+I arm. There was also an over-accrual with one additional patient in the N arm and two patients in the N+I arm. None of the over accrued patients experienced a confirmed response, leading to an overall CRR of 1/13 (7.7%) in the N arm and 2/14 (16.6%) in the N+I arm. Notably, there were two patients with UPS in the N+I arm that were reported to have an unconfirmed response due to a lack of confirmation scans.

    For the GIST subtype, enrollment was ended early due to failure to see a response within the first nine evaluable patients enrolled in either the N or N+I arm. There was also an over-accrual with one additional patient in the N arm and two patients in the N+I arm. Across all treated patients with GIST, there were no confirmed responses.

    For this expansion cohort, it was hypothesized that at least two confirmed responses out of the first 12 evaluable patients, for each arm and tumor subtype, would be considered evidence of promising activity. For both DDLPS and UPS, the N+I arm met the threshold to be considered promising, each achieving two confirmed responses by 6 months. The N arm treatment for DDLPS and UPS did not meet the threshold, with only 1 of 12 patients achieving a confirmed response within the first 6 months of treatment. In patients with GIST, both treatment arms demonstrated a CRR of 0%. In the total enrolled and treated cohort (n=77), the CRR for GIST was 0% (0/10) treated with N and 0% (0/11) treated with N+I, for UPS was 7.7% (1/13) treated with N and 14.3% (2/14) treated with N+I, and for DDLPS was 6.6% (1/15) treated with N and 14.3% (2/14) treated with N+I. While this cohort included heavily pretreated patients, we were not powered to identify a correlation between CRR and prior lines of therapy. While not statistically significant, none of the patients who achieved a PR on either N or N+I had received more than three prior lines of systemic therapy (p=0.24). Finally, 9 patients crossed over from the N to N+I treatment arms; however, there were no responses post-crossover.

    Secondary response endpoints

    For secondary response endpoints, all patients receiving treatment, including over-accrued patients, were included in the outcome measures. Therefore, a total of 21 GIST, 29 DDLPS, and 27 UPS patients were evaluable for secondary endpoints including CBR, PFS, and OS. In patients with GIST, CBR was 40% and 54.5% for N and N+I, respectively. Median PFS for N was 1.5 months (95% CI 1.3–Inf) versus 2.9 months (95% CI 1.4–Inf) for N+I. Median OS for N was 9.1 months (95% CI 4.9–Inf) versus 12.2 months (95% CI 6.0–Inf) for N+I. In patients with DDLPS, the CBR was 66.6% and 78.6% for N and N+I, respectively. Median PFS for patients receiving N was 4.6 months (95% CI 3.2–Inf) and 5.5 months (95% CI 2.8–Inf) for patients receiving N+I. Median OS for patients receiving N was 8.1 months (95% CI 6.8–Inf) and 14.6 months (95% CI 9.1–Inf) for patients receiving N+I. In patients with UPS receiving N alone CBR was 28.6% versus 53.3% for N+I. Median PFS for patients receiving N was 1.4 months (95% CI 1.4–14.2) and 2.7 months (95% CI 1.5–Inf) for patients receiving N+I. Median OS for patients receiving N was 6.6 months (95% CI 2.4–Inf) and 15.2 months (95% CI 5.1–Inf) for patients receiving N+I.

    Treatment safety

    For patients with DDLPS receiving N (n=15), 12 patients experienced a grade 1/2 treatment-related adverse event (TRAE, 80%), and 4 experienced a grade 3/4 TRAE (26.6%). One grade 5 event, death not otherwise specified, was reported but not deemed treatment related. For patients with DDLPS receiving N+I (n=14), 13 patients experienced a grade 1/2 TRAE (92.9%), and 2 patients experienced a grade 3/4 TRAE (14.3%). No grade 5 TRAEs were reported.

    For patients with UPS receiving N (n=13), nine patients experienced a grade 1/2 TRAE (69.2%), and two patients experienced a grade 3/4 TRAE (15.4%). Three grade 5 event including sepsis, sudden death not otherwise specified, and neoplasms (benign, malignant, and unspecified), were reported but not deemed treatment related. For patients with UPS receiving N+I (n=14), 12 patients experienced a grade 1/2 TRAE (85.7%), and 3 patients experienced a grade 3/4 TRAE (21.4%). One grade 5 event, infections and infestations, was reported but not deemed treatment related.

    For patients with GIST receiving N (n=10), seven patients experienced a grade 1/2 TRAE (70%), and two patients experienced a grade 3/4 TRAE (20%). One grade 5 event, neoplasms (benign, malignant, and unspecified), was reported but not deemed treatment related. For patients with GIST receiving N+I (n=11), nine patients experienced a grade 1/2 TRAE (81.8%), and five patients experienced a grade 3/4 TRAE (45.5%). No grade 5 AEs were reported. TRAE was highest in the GIST cohort for patients treated with N+I, holding enrollment as required per protocol.

    Of the 77 patients in this cohort initiating treatment, 76 patients experienced an AE of grade 1 or greater. Grade 3/4 TRAEs were rare and varied in frequency between treatment regimens as well as histologies (table 3). No trend was observed in the frequency of grade 3/4 AEs between N and N+I regimens. Within each treatment arm for each histology type, there were no grade 3/4 events at least possibly related to treatment in more than a single patient (online supplemental tables S2–S4). No grade 5 AEs were deemed at least possibly related to treatment for any cohort.

    View this table:
    Table 3

    Summary of adverse events by histology and treatment regimen

    Correlative analysis cohort

    As an exploratory endpoint, we tested biomarkers of response in archived pretreatment tissue from 86 patients enrolled in either the N or N+I treatment arm that were collected from either the original trial cohort17 or this expansion cohort (online supplemental table S6). Samples were prioritized for testing by IHC, DNA sequencing, and RNA sequencing, respectively. All samples were evaluated by each method separately, resulting in sample size differences per correlative analysis (online supplemental figure S2). Four of the 86 patients were excluded from correlative analysis due to the lack of a reported PFS (n=2) or censoring of PFS before 12 weeks of treatment (n=2). Patients represented DDLPS (n=28), UPS (n=27), GIST (n=13), LMS (n=10), and four other histologies. Patients included in this analysis were evenly distributed between N (n=41) and N+I (n=42) treatment arms. Full demographic data for the correlative analysis is available in online supplemental table S7. To identify potential biomarkers for response to immunotherapy, we correlated molecular findings with 12-week PFS status regardless of treatment arm. All correlative findings presented here should be considered exploratory in nature.

    Standard tumor immune biomarkers were not correlated with response

    Studies of predictive biomarkers of response to immunotherapies across cancers have identified biomarkers such as PD-L1 expression, immune infiltration, and TMB as highly associated with clinical response. To test these markers in sarcoma, we first applied multiplex IHC to identify the expression of PD-L1 as well as immune infiltration. PD-L1 expression was not highly prevalent across the dataset (median 0.4%, IQR: 0.1%–1.8%; online supplemental figure S3A). High expression of PD-L1 was rare (PD-L1≥50%=1, ≥5%=15, ≥1%=28). PD-L1 expression was not associated with clinical response in any sarcoma histology (online supplemental table S8). To assess immune cell infiltration of these tumors, immune markers and TLS, including FOXP3, CD68, CD8, CD4, and CD3, were measured by IHC. Increased prevalence of CD4-positive cells in DDLPS was associated with a PFS ≥12 weeks on therapy (p=0.05), but was not associated with a statistically significant change in PFS or OS (PFS: p=0.18; OS: p=0.06). No other immune cell markers were correlated with response in any sarcoma histological subtype (online supplemental table S9). As an additional method to measure immune infiltration, we used the CIBERSORTx deconvolution algorithm to infer immune cell infiltration in 29 tumors with RNA sequencing and clinical response data available for analysis. No immune cell types identified by deconvolution were associated with response to immunotherapy (online supplemental table S10). Finally, we assessed both TMB and microsatellite status across tumors. TMB was low (median 1.7 Mut/Mb, IQR: 1.1–2.1 Mut/Mb; online supplemental figure S3B) and tumors of high TMB were rare (>10 Mut/Mb, n=1) and was not associated with response to therapy (online supplemental table S11). No tumors were identified as microsatellite instable and MSI scores were not associated with response to therapy (online supplemental table S12).

    Presence of fusions correlated with improved clinical response across STS histologies

    Within the sarcoma cohort with RNA-seq data analyzed (n=30), we identified 17 unique tumors that had a median of 1 fusion per tumor (range 1–20 fusions; online supplemental table S13). Only one fusion was identified in more than a single tumor (XRCC5::KLHL41). We assessed clinical response by the presence or absence of any gene fusion in each patient sample. Across histologies, the presence of any gene fusion was associated with a PFS ≥12 weeks on therapy (p=0.01, figure 2A). Additional exploratory analyses indicated that fusion presence was associated with a longer PFS (HR 0.20, 95% CI 0.08–0.51, p<0.001, figure 2B) and OS (HR 0.29, 95% CI 0.12–0.71, p=0.006, figure 2C) when all histologies were treated as a single cohort. Limited cohort size prevented concluding too much from this analysis at the subtype level. Nevertheless, a similar trend emerged favoring the presence of fusions as a positive predictive biomarker in all STS histologies included (online supplemental figure S4).

    Figure 2
    Figure 2

    Fusion-positive soft-tissue sarcomas were associated with improved survival outcomes across histologies. (A) Fusions were more prevalent across all tumors demonstrating PFS ≥12 weeks. (B,C) Across all histologies, the presence of a fusion (n=17) was associated with an improved PFS (HR 0.20, 95% CI 0.08–0.51, p=0.0008, (B) and OS (HR 0.29, 95% CI 0.12–0.71, p=0.006, (C) compared with patients’ tumors that did not demonstrate a fusion (n=13). DDLPS, dedifferentiated liposarcoma; GIST, gastrointestinal stromal tumor; OS, overall survival; PFS, progression-free survival; UPS, undifferentiated pleomorphic sarcoma.

    Tumor mutation subclonality predicts clinical outcomes

    Another method to measure the existence and complexity of the subclonal structure of an individual cancer is to measure the portion of subclonal mutations present in a bulk DNA-sequencing sample.15 First, we assessed the number and prevalence of clonal, subclonal, and indeterminate mutations across histologies by clinical response. While clonal and indeterminate and total mutations were not associated with PFS at 12 weeks (p>0.05, online supplemental figure S5A,B), an increased number and percentage of subclonal mutations were associated with a PFS ≥12 weeks (online supplemental figure S5C–F). As a continuous biomarker (z-score), the percentage of subclonal mutations trended toward improved PFS (HR 0.75, 95% CI 0.55–1.02, p=0.07) but demonstrated no difference in OS (HR 0.86, 95% CI 0.61–1.21, p=0.38).

    Given the potential predictive value of subclonal mutations across STS histologies, we further assessed correlations within individual histologies. The strongest correlation between subclonal mutations and PFS at 12 weeks was identified in DDLPS (figure 3A, online supplemental figure S5G,H); therefore, we focused further analysis within this STS histology. While subclonal mutations were higher in DDLPS tumors demonstrating a PFS ≥12 weeks, total mutation count did not differ by response status (figure 3B). Using a median cut-off for the portion of subclonal mutations identified an improved PFS for tumors exhibiting a high proportion of subclonal mutations (HR 0.27, 95% CI 0.09–0.83, p=0.003, figure 3C). In a multivariate model including study treatment arm, portion of subclonal mutations, total reads, and total mutations, PFS was only associated with the proportion of subclonal mutations (figure 3D). Using a median cut-off for the portion of subclonal mutations demonstrated an improved OS for tumors exhibiting a high proportion of subclonal mutations (HR 0.26, 95% CI 0.08–0.85, p=0.01, figure 3E). In a multivariate model for OS, treatment with N only (p=0.02) and total reads (p=0.05) were associated with poor OS while the proportion of subclonal mutations was associated with improved OS (p=0.007, figure 3F). No single gene was associated with subclonality across histologies or within DDLPS.

    Figure 3
    Figure 3

    Prevalence of subclonal mutations confer survival advantage in dedifferentiated liposarcoma (DDLPS). (A) The proportion of genomic variants predicted to be subclonal in DDLPS was higher in responsive compared with non-responsive tumors. (B) Total mutation count was similar in DDLPS despite tumor response. (C) Progression-free survival (PFS) was improved in patients whose tumors demonstrated a greater than median proportion of subclonal mutations (HR 0.27, (95% CI 0.089–0.83, p=0.003). (D) Multivariate analysis for PFS of subclonal mutation proportion, study arm, and total mutations. (C) Overall survival (OS) was improved in patients whose tumors demonstrated a greater than median proportion of subclonal mutations (HR 0.26, 95% CI 0.08–0.85, p=0.01). (D) Multivariate analysis for OS of subclonal mutation proportion, study arm, and total mutations. *p<0.05, ns: not significant.

    Inflammatory signaling associated with tumor response

    To identify potential RNA markers of response to immunotherapy, 30 patient tumors underwent RNA sequencing prior to treatment with study treatment and had matching response data available. Tumors included in RNA analysis represented three histologies, GIST (n=8), DDLPS (n=12), and UPS (n=10). We then assessed the correlation between RNA expression and clinical outcomes in this cohort. Few single RNA biomarkers were associated with response (online supplemental figure S6A–C). Gene Set Enrichment Analysis enrichment in immune and inflammatory related pathways in tumors from patients who responded to immunotherapy (absolute NES ≥1.5, q<0.1, online supplemental figure S6D–F), similar to prior finding of response to immunotherapy in cancer.35–37

    Discussion

    This expansion cohort corroborates the findings in the original A091401 analysis and confirms that response seems to be higher with N+I (figure 4). The highest confirmed response rate occurred in UPS and DDLPS with combination therapy (N+I 16.6% each, N 8.3% each). Notably, there were two patients with UPS in the N+I arm that demonstrated an unconfirmed response. GIST failed to meet the threshold for response, with no tumors in either treatment arm demonstrating confirmed response. Given that the design of this study required all therapeutic agents to be held at least 28 days prior to beginning treatment on study, it is possible that the absence of consistent tyrosine kinase inhibitor use common in GIST affected the response of these tumors to immunotherapy. GIST remains heterogeneous and generally cold tumors with limited immune infiltrates. While it is not possible to compare the findings of this trial directly to those of others, it is notable that the response rates and PFS found here were less than those of prior studies.38–40 Further study is necessary to identify methods to harness the immune system against this malignancy.

    Figure 4
    Figure 4

    Summary of findings across the cohort. N+I outperformed N for PFS (A) and trended towards improved OS (B). OS, overall survival; PFS, progression-free survival.

    To develop potential biomarkers of ICI response in STS, we incorporated samples from both the original and expansion cohorts. Classical biomarkers of immunotherapy response (immune infiltrates, TLS, PD-L1, TMB), did not correlate with response.41 Given the sample size of the cohort, the low response rate to therapy, and the heterogeneity of STS, it is possible that these classical biomarkers maintain utility in STS; however, our study design did not allow enough power to identify these correlations. Two less well-described biomarkers of genomic instability, gene fusions, and subclonal mutations, correlated with 12-week PFS. While some sarcomas are molecularly classified by the presence of a disease-defining gene fusion,42–44 which are often genomically quiet, the STS histologies included in this RNA correlative analysis are not considered to be driven by gene fusion events. Furthermore, the lack of the same recurrent gene fusion suggests that our finding may be related more to the genomic instability rather than the molecular signals driven by the present fusions. Others have noted that the neoantigens formed by gene fusions are considered more immunogenic than neoantigens caused by single-nucleotide variants and insertions or deletions.45 We also identified an association between the subclonal mutations and response in this study, driven by a strong correlation in DDLPS. DDLPS are characterized by DNA copy number amplifications of the 12q chromosomal locus and a low TMB compared with other cancers.46 47 Propagation of cancer cells gives rise to genomic heterogeneity between primary and subclonal tumor cell populations.47–49 Previous studies have demonstrated subclonal mutations as a marker of response to immune checkpoint inhibition15 48–50 as well as response to traditional chemotherapy.51–53 Further studies are needed to assess the potential of these genomic instability markers, and their evolution across treatment time, as predictors of response to immunotherapy.

    In this expansion cohort, we confirm the findings from the parent study A091401.17 One limitation of this study is that it was not designed to make direct comparisons between N and N+I. Our data may suggest increased activity of N+I in comparison to N alone; however, there can be no statistical comparison. These results also demonstrate the histology-dependent nature of immunotherapy for sarcomas. Future studies should continue to be designed to identify promising signals across histologies and allow for expansion studies in the most responsive disease subtypes. The molecular correlative analysis was intended to be exploratory in nature. A limitation of our analysis was the low response rate across cohorts. Further study is required to see if these findings apply to larger cohorts. Nevertheless, we confirm that standard ICI biomarkers such as PD-L1/PD-1 IHC status and TMB may have more limited applicability in sarcoma.10 12 54 We also find that genomic instability biomarkers may merit additional exploration. In this study, we conducted whole-exome sequencing. Given the findings of genomic instability biomarkers here, future studies should consider the use of whole genome sequencing. Additional studies are clearly necessary to determine the optimal patient and tumor factors for immunotherapy in STS.

    Conclusion

    This expansion cohort recapitulates the previous data generated from the parent study A091401. In DDLS and UPS, the primary response endpoint was met with combination therapy but not with monotherapy. However, neither treatment arm leads to confirmed responses in GIST. Subtype heterogeneity was notable and reduced overall analytical power hindering corollary analyses. Standard biomarkers were unrevealing regarding standard IHC biomarkers and singular molecular changes. Nevertheless, novel markers of genomic instability were correlated with improved clinical outcomes. Additional studies will be needed to corroborate these predictive biomarkers for immunotherapy findings in STS.

    Data availability statement

    Data are available in a public, open access repository. Data are available, via controlled access, from the NCBI dbGaP archive (database accession number phs003785.v1).

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and the review boards at each of the participating institutions approved the study. The study was done in accordance with the Declaration of Helsinki and the Guidelines for Good Clinical Practice. Each patient signed an institutional review board-approved, protocol-specific, informed consent form in accordance with federal and institutional guidelines. This study was monitored at least two times per year by the Alliance Data and Safety Monitoring Board, a standing committee consisting of individuals outside of the Alliance Clinical Trials in Oncology Group for accrual, safety, and the primary endpoint evaluation. Participants gave informed consent to participate in the study before taking part.

    Acknowledgments

    Data quality was ensured by review of data by the Alliance Statistics and Data Center and by the study chairperson following Alliance policies.

    References

    1. ↵
      1. Siegel RL,
      2. Miller KD,
      3. Jemal A
      . Cancer Statistics, 2017. CA Cancer J Clin 2017;67:7–30. doi:10.3322/caac.21387
      OpenUrlCrossRefPubMed
    2. ↵
      1. Demetri GD,
      2. Chawla SP,
      3. von Mehren M, et al
      . Efficacy and Safety of Trabectedin in Patients With Advanced or Metastatic Liposarcoma or Leiomyosarcoma After Failure of Prior Anthracyclines and Ifosfamide: Results of a Randomized Phase II Study of Two Different Schedules. JCO 2009;27:4188–96. doi:10.1200/JCO.2008.21.0088
      OpenUrlAbstract/FREE Full Text
    3. ↵
      1. Schöffski P,
      2. Chawla S,
      3. Maki RG, et al
      . Eribulin versus dacarbazine in previously treated patients with advanced liposarcoma or leiomyosarcoma: a randomised, open-label, multicentre, phase 3 trial. Lancet 2016;387:1629–37. doi:10.1016/S0140-6736(15)01283-0
      OpenUrlCrossRefPubMed
    4. ↵
      1. van der Graaf WTA,
      2. Blay J-Y,
      3. Chawla SP, et al
      . Pazopanib for metastatic soft-tissue sarcoma (PALETTE): a randomised, double-blind, placebo-controlled phase 3 trial. Lancet 2012;379:1879–86. doi:10.1016/S0140-6736(12)60651-5
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Seddon B,
      2. Strauss SJ,
      3. Whelan J, et al
      . Gemcitabine and docetaxel versus doxorubicin as first-line treatment in previously untreated advanced unresectable or metastatic soft-tissue sarcomas (GeDDiS): a randomised controlled phase 3 trial. Lancet Oncol 2017;18:1397–410. doi:10.1016/S1470-2045(17)30622-8
      OpenUrlCrossRefPubMed
    6. ↵
      1. Seligson ND,
      2. Warner JL,
      3. Dalton WS, et al
      . Recommendations for patient similarity classes: results of the AMIA 2019 workshop on defining patient similarity. J Am Med Inform Assoc 2020;27:1808–12. doi:10.1093/jamia/ocaa159
      OpenUrl
    7. ↵
      1. Chalabi M,
      2. Fanchi LF,
      3. Dijkstra KK, et al
      . Neoadjuvant immunotherapy leads to pathological responses in MMR-proficient and MMR-deficient early-stage colon cancers. Nat Med 2020;26:566–76. doi:10.1038/s41591-020-0805-8
      OpenUrlCrossRefPubMed
    8. ↵
      1. Schild SE,
      2. Wang X,
      3. Bestvina CM, et al
      . Alliance A082002 -a randomized phase II/III trial of modern immunotherapy-based systemic therapy with or without SBRT for PD-L1-negative, advanced non-small cell lung cancer. Clin Lung Cancer 2022;23:e317–20. doi:10.1016/j.cllc.2022.04.004
      OpenUrl
    9. ↵
      1. Vos JL,
      2. Elbers JBW,
      3. Krijgsman O, et al
      . Neoadjuvant immunotherapy with nivolumab and ipilimumab induces major pathological responses in patients with head and neck squamous cell carcinoma. Nat Commun 2021;12:7348. doi:10.1038/s41467-021-26472-9
    10. ↵
      1. Chen AP,
      2. Sharon E,
      3. O’Sullivan-Coyne G, et al
      . Atezolizumab for Advanced Alveolar Soft Part Sarcoma. N Engl J Med 2023;389:911–21. doi:10.1056/NEJMoa2303383
      OpenUrl
    11. ↵
      1. Bergsma EJ,
      2. Elgawly M,
      3. Mancuso D, et al
      . Atezolizumab as the First Systemic Therapy Approved for Alveolar Soft Part Sarcoma. Ann Pharmacother 2024;58:407–15. doi:10.1177/10600280231187421
      OpenUrl
    12. ↵
      1. Petitprez F,
      2. de Reyniès A,
      3. Keung EZ, et al
      . B cells are associated with survival and immunotherapy response in sarcoma. Nature New Biol 2020;577:556–60. doi:10.1038/s41586-019-1906-8
      OpenUrl
    13. ↵
      1. D’Angelo SP,
      2. Richards AL,
      3. Conley AP, et al
      . Pilot study of bempegaldesleukin in combination with nivolumab in patients with metastatic sarcoma. Nat Commun 2022;13:3477. doi:10.1038/s41467-022-30874-8
    14. ↵
      1. Wang Y,
      2. Shi T,
      3. Song X, et al
      . Gene fusion neoantigens: Emerging targets for cancer immunotherapy. Cancer Lett 2021;506:45–54. doi:10.1016/j.canlet.2021.02.023
      OpenUrlCrossRefPubMed
    15. ↵
      1. Keenan TE,
      2. Guerriero JL,
      3. Barroso-Sousa R, et al
      . Molecular correlates of response to eribulin and pembrolizumab in hormone receptor-positive metastatic breast cancer. Nat Commun 2021;12:5563. doi:10.1038/s41467-021-25769-z
    16. ↵
      1. Li B,
      2. Chen H,
      3. Yang S, et al
      . Advances in immunology and immunotherapy for mesenchymal gastrointestinal cancers. Mol Cancer 2023;22:71. doi:10.1186/s12943-023-01770-6
    17. ↵
      1. D’Angelo SP,
      2. Mahoney MR,
      3. Van Tine BA, et al
      . Nivolumab with or without ipilimumab treatment for metastatic sarcoma (Alliance A091401): two open-label, non-comparative, randomised, phase 2 trials. Lancet Oncol 2018;19:416–26. doi:10.1016/S1470-2045(18)30006-8
      OpenUrlCrossRefPubMed
    18. ↵
      1. Seymour L,
      2. Bogaerts J,
      3. Perrone A, et al
      . iRECIST: guidelines for response criteria for use in trials testing immunotherapeutics. Lancet Oncol 2017;18:e143–52. doi:10.1016/S1470-2045(17)30074-8
      OpenUrlCrossRefPubMed
    19. ↵
      1. McKenna A,
      2. Hanna M,
      3. Banks E, et al
      . The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 2010;20:1297–303. doi:10.1101/gr.107524.110
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Benjamin D,
      2. Sato T,
      3. Cibulskis K, et al
      . Calling somatic snvs and indels with Mutect2. bioRxiv [Preprint]. doi:10.1101/861054
    21. ↵
      1. Kim S,
      2. Scheffler K,
      3. Halpern AL, et al
      . Strelka2: fast and accurate calling of germline and somatic variants. Nat Methods 2018;15:591–4. doi:10.1038/s41592-018-0051-x
      OpenUrlCrossRefPubMed
    22. ↵
      1. Karczewski KJ,
      2. Francioli LC,
      3. Tiao G, et al
      . The mutational constraint spectrum quantified from variation in 141,456 humans. Nature New Biol 2020;581:434–43. doi:10.1038/s41586-020-2308-7
      OpenUrl
    23. ↵
      1. Niu B,
      2. Ye K,
      3. Zhang Q, et al
      . MSIsensor: microsatellite instability detection using paired tumor-normal sequence data. Bioinformatics 2014;30:1015–6. doi:10.1093/bioinformatics/btt755
      OpenUrlCrossRefPubMed
    24. ↵
      1. Dobin A,
      2. Davis CA,
      3. Schlesinger F, et al
      . STAR: ultrafast universal RNA-seq aligner. Bioinformatics 2013;29:15–21. doi:10.1093/bioinformatics/bts635
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Yates AD,
      2. Achuthan P,
      3. Akanni W, et al
      . Ensembl 2020. Nucleic Acids Res 2020;48:D682–8. doi:10.1093/nar/gkz966
      OpenUrlCrossRefPubMed
    26. ↵
      1. Uhrig S,
      2. Ellermann J,
      3. Walther T, et al
      . Accurate and efficient detection of gene fusions from RNA sequencing data. Genome Res 2021;31:448–60. doi:10.1101/gr.257246.119
      OpenUrlAbstract/FREE Full Text
    27. ↵
      1. Anders S,
      2. Pyl PT,
      3. Huber W
      . HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics 2015;31:166–9. doi:10.1093/bioinformatics/btu638
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Shen R,
      2. Seshan VE
      . FACETS: allele-specific copy number and clonal heterogeneity analysis tool for high-throughput DNA sequencing. Nucleic Acids Res 2016;44:e131. doi:10.1093/nar/gkw520
    29. ↵
      1. Love MI,
      2. Huber W,
      3. Anders S
      . Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. doi:10.1186/s13059-014-0550-8
    30. ↵
      1. Subramanian A,
      2. Tamayo P,
      3. Mootha VK, et al
      . Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545–50. doi:10.1073/pnas.0506580102
      OpenUrlAbstract/FREE Full Text
    31. ↵
      1. Liberzon A,
      2. Birger C,
      3. Thorvaldsdóttir H, et al
      . The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 2015;1:417–25. doi:10.1016/j.cels.2015.12.004
      OpenUrlCrossRef
    32. ↵
      1. Newman AM,
      2. Liu CL,
      3. Green MR, et al
      . Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 2015;12:453–7. doi:10.1038/nmeth.3337
      OpenUrlCrossRefPubMed
    33. ↵
      1. Newman AM,
      2. Steen CB,
      3. Liu CL, et al
      . Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol 2019;37:773–82. doi:10.1038/s41587-019-0114-2
      OpenUrlPubMed
    34. ↵
      1. Benjamini Y,
      2. Hochberg Y
      . Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Stat Soc Series B Stat Methodol 1995;57:289–300. doi:10.1111/j.2517-6161.1995.tb02031.x
      OpenUrl
    35. ↵
      1. De Ridder K,
      2. Locy H,
      3. Piccioni E, et al
      . TNF-α-Secreting Lung Tumor-Infiltrated Monocytes Play a Pivotal Role During Anti-PD-L1 Immunotherapy. Front Immunol 2022;13:811867. doi:10.3389/fimmu.2022.811867
    36. ↵
      1. Ayers M,
      2. Lunceford J,
      3. Nebozhyn M, et al
      . IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade. J Clin Invest 2017;127:2930–40. doi:10.1172/JCI91190
      OpenUrlCrossRefPubMed
    37. ↵
      1. Färkkilä A,
      2. Gulhan DC,
      3. Casado J, et al
      . Immunogenomic profiling determines responses to combined PARP and PD-1 inhibition in ovarian cancer. Nat Commun 2020;11:1459. doi:10.1038/s41467-020-15315-8
    38. ↵
      1. Tawbi HA,
      2. Burgess M,
      3. Bolejack V, et al
      . Pembrolizumab in advanced soft-tissue sarcoma and bone sarcoma (SARC028): a multicentre, two-cohort, single-arm, open-label, phase 2 trial. Lancet Oncol 2017;18:1493–501. doi:10.1016/S1470-2045(17)30624-1
      OpenUrlCrossRefPubMed
    39. ↵
      1. Haddox CL,
      2. Nathenson MJ,
      3. Mazzola E, et al
      . Phase II Study of Eribulin plus Pembrolizumab in Metastatic Soft-tissue Sarcomas: Clinical Outcomes and Biological Correlates. Clin Cancer Res 2024;30:1281–92. doi:10.1158/1078-0432.CCR-23-2250
      OpenUrl
    40. ↵
      1. Cho HJ,
      2. Yun K-H,
      3. Shin S-J, et al
      . Durvalumab plus pazopanib combination in patients with advanced soft tissue sarcomas: a phase II trial. Nat Commun 2024;15:685. doi:10.1038/s41467-024-44875-2
    41. ↵
      1. Bai R,
      2. Lv Z,
      3. Xu D, et al
      . Predictive biomarkers for cancer immunotherapy with immune checkpoint inhibitors. Biomark Res 2020;8:34. doi:10.1186/s40364-020-00209-0
    42. ↵
      1. Seligson ND,
      2. Awasthi A,
      3. Millis SZ, et al
      . Common Secondary Genomic Variants Associated With Advanced Epithelioid Hemangioendothelioma. JAMA Netw Open 2019;2:e1912416. doi:10.1001/jamanetworkopen.2019.12416
    43. ↵
      1. Walker V,
      2. Jin DX,
      3. Millis SZ, et al
      . Gene partners of the EWSR1 fusion may represent molecularly distinct entities. Transl Oncol 2023;38:101795. doi:10.1016/j.tranon.2023.101795
    44. ↵
      1. Seligson ND,
      2. Maradiaga RD,
      3. Stets CM, et al
      . Multiscale-omic assessment of EWSR1-NFATc2 fusion positive sarcomas identifies the mTOR pathway as a potential therapeutic target. NPJ Precis Oncol 2021;5:43. doi:10.1038/s41698-021-00177-0
    45. ↵
      1. Wei Z,
      2. Zhou C,
      3. Zhang Z, et al
      . The Landscape of Tumor Fusion Neoantigens: A Pan-Cancer Analysis. iScience 2019;21:249–60. doi:10.1016/j.isci.2019.10.028
      OpenUrl
    46. ↵
      1. Bill KLJ,
      2. Seligson ND,
      3. Hays JL, et al
      . Degree of MDM2 Amplification Affects Clinical Outcomes in Dedifferentiated Liposarcoma. Oncologist 2019;24:989–96. doi:10.1634/theoncologist.2019-0047
      OpenUrlAbstract/FREE Full Text
    47. ↵
      1. Dentro SC,
      2. Leshchiner I,
      3. Haase K, et al
      . Characterizing genetic intra-tumor heterogeneity across 2,658 human cancer genomes. Cell 2021;184:2239–54. doi:10.1016/j.cell.2021.03.009
      OpenUrlCrossRefPubMed
    48. ↵
      1. McGranahan N,
      2. Furness AJS,
      3. Rosenthal R, et al
      . Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. Science 2016;351:1463–9. doi:10.1126/science.aaf1490
      OpenUrlAbstract/FREE Full Text
    49. ↵
      1. McGranahan N,
      2. Swanton C
      . Clonal Heterogeneity and Tumor Evolution: Past, Present, and the Future. Cell 2017;168:613–28. doi:10.1016/j.cell.2017.01.018
      OpenUrlCrossRefPubMed
    50. ↵
      1. Litchfield K,
      2. Reading JL,
      3. Puttick C, et al
      . Meta-analysis of tumor- and T cell-intrinsic mechanisms of sensitization to checkpoint inhibition. Cell 2021;184:596–614. doi:10.1016/j.cell.2021.01.002
      OpenUrlCrossRefPubMed
    51. ↵
      1. Craig DJ,
      2. Bailey MM,
      3. Noe OB, et al
      . Subclonal landscape of cancer drives resistance to immune therapy. Cancer Treat Res Commun 2022;30:100507. doi:10.1016/j.ctarc.2021.100507
      OpenUrl
    52. ↵
      1. Morris LGT,
      2. Riaz N,
      3. Desrichard A, et al
      . Pan-cancer analysis of intratumor heterogeneity as a prognostic determinant of survival. Oncotarget 2016;7:10051–63. doi:10.18632/oncotarget.7067
      OpenUrlPubMed
    53. ↵
      1. Seligson ND,
      2. Tang J,
      3. Jin DX, et al
      . Drivers of genomic loss of heterozygosity in leiomyosarcoma are distinct from carcinomas. NPJ Precis Oncol 2022;6:29. doi:10.1038/s41698-022-00271-x
    54. ↵
      1. Wagner MJ,
      2. Othus M,
      3. Patel SP, et al
      . Multicenter phase II trial (SWOG S1609, cohort 51) of ipilimumab and nivolumab in metastatic or unresectable angiosarcoma: a substudy of dual anti-CTLA-4 and anti-PD-1 blockade in rare tumors (DART). J Immunother Cancer 2021;9:e002990. doi:10.1136/jitc-2021-002990

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors All authors contributed meaningfully to this work. Conceptualization: SDA; methodology: SDA; statistical analysis: ACG and JDC; investigation: all authors; data contribution: all authors, writing–original draft: NDS, JLC, and SDA; writing–review and editing: all authors; visualization: NDS and JLC; supervision: SDA. Data quality was ensured by review of data by the Alliance Statistics and Data Center and by the study chairperson following Alliance policies.The corresponding author (SDA) is the guarantor of the study.

    • Funding Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health (U10CA180821, U10CA180882, U24CA196171 (to the Alliance for Clinical Trials in Oncology), UG1CA233180, UG1CA233290, UG1CA233331, UG1CA233339, and P30CA08748 (to Memorial Sloan Kettering Cancer Center)); Cycle for Survival (#n/a); and the Marie-Josée and Henry R Kravis Center for Molecular Oncology (#n/a). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

    • Competing interests JLC: Foundation Medicine (personal fees), Tempus (employment); BVT: Daiichi Sankyo Inc (personal fees), Deciphera Pharmaceuticals (personal fees), Advenchen (personal fees), Putnam (personal fees), Boxer Capital LLC (personal fees), Acuta Capital Partners, LLC (personal fees), Aadi (personal fees), Race Oncology (personal fees), Hinge Bio, Inc (personal fees), Kronos Bio, Inc, (personal fees) Sonata Therapeutics, Inc. (personal fees), Total Health Conference (personal fees), Adaptimmune (other support), Boehringer Ingelheim (consulting/licensing), Accuronix Therapeutics (patent), Polaris (non-paid member of Scientific/Safety Board); DAL: AADI Bioscience (consulting), MatchTx, LLC (patents); all other authors declare no competing interests.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.