A Phase II Study of Epacadostat and Pembrolizumab in Patients with Advanced Sarcoma

This article is featured in Highlights of This Issue, p. 2013

Sarcomas are heterogeneous, rare bone, and soft tissue malignancies of mesenchymal origin with poor prognoses (1). Treatments that are more effective are urgently needed for patients with advanced disease.

Over the last decade, immune checkpoint inhibitors targeting cytotoxic T-lymphocyte associated protein 4 (CTLA4), programmed cell death protein 1 (PDCD1), and CD274 (PD-L1) have been approved to treat many cancer types (2–4). Although sarcoma was among the first tumor models for which immunotherapy was proposed (5), immune checkpoint inhibition (ICI) has demonstrated activity in only limited sarcoma subtypes. Modest activity in liposarcoma (LPS; 10%, n = 4/39) to moderate activity in undifferentiated pleomorphic sarcoma (UPS; 23%, n = 9/40) and alveolar soft part sarcoma (37.2%, n = 16/43) has been observed with ICI and no predictive biomarkers of response have been identified (6–9).

Tumor cells use multiple mechanisms to evade immune surveillance; combination therapeutic strategies targeting different aspects of the immune system may restore immune function and improve response to ICI. Known immunomodulating pathways with therapeutic potential include the PD-1/PD-L1 pathway and the kynurenine pathway incorporating indole 2,3 dioxygenase 1 (IDO1), an intracellular enzyme expressed by tumor, endothelial, and dendritic cells and macrophages within the tumor microenvironment (TME; refs. 10, 11). IDO1 upregulation is associated with poor prognosis in patients with advanced cancers (12, 13). IDO1 catalyzes the first and rate-limiting step in the degradation of tryptophan to kynurenine, which helps the cell evade immune surveillance (14, 15). IDO1-mediated depletion of tryptophan and downstream metabolites may result in cell-cycle arrest, apoptosis of effector T cells, and activation of immunosuppressive cells (14, 16, 17), whereas IDO1 inhibition shifts the TME from an immunosuppressive state to an immune-stimulated one.

A retrospective study showed that IDO1 expression varies by histologic sarcoma subtype. Pathology specimens from 328 patients who underwent surgical resection for localized, primary sarcomas with genomic complexity were examined to assess the expression and prognostic value of PD-L1, IDO1, and kynurenine (18). Most patients had high-grade UPS, myxofibrosarcoma (MFS), leiomyosarcoma (LMS), or dedifferentiated liposarcoma (DDLPS); PD-L1 (≥1%), IDO1 (≥5%), and kynurenine (≥1%) were detected by IHC in 19.5%, 42%, and 59% of sarcoma samples, respectively. PD-L1 expression was detected in <10% of tumor cells in 97% of PD-L1-positive cases. IDO1 and kynurenine expression were generally scored as one or two on a three-point scoring system. The study highlighted that IDO1 expression varied by histologic sarcoma subtype and was significantly more common in UPS (48%) than LMS (30%). PD-L1 and IDO1 expression were associated with CD8⁺ effector T-cell infiltration into the tumor (18).

A phase II study of pembrolizumab and metronomic cyclophosphamide demonstrated limited activity among 50 evaluable patients with soft tissue sarcoma (STS), with only 3 patients experiencing tumor shrinkage and 1 of these 3 meeting the definition of a partial response (PR) by RECIST v.1.1. Correlative analyses revealed PD-L1 tumor expression and low levels of tumor-infiltrating CD-8 effector T cells. However, the tumors were predominantly infiltrated by M2 macrophages expressing IDO1, and patents with STS had significantly increased kynurenine-to-tryptophan plasma ratio during treatment (19). These findings highlighted the possible role of the IDO1/kynurenine pathway in primary resistance to PD-1/PD-L1 immunotherapy in STS.

Epacadostat is a highly selective oral inhibitor of the IDO1 enzyme. Phase I and II studies demonstrated the safety and tolerability of epacadostat monotherapy among patients with advanced cancer, with doses of ≥100 mg twice daily achieved optimal inhibition of IDO1 (20, 21).

In preclinical models, the combination of epacadostat and anti-PD-1/PD-L1 or anti-CTLA4 antibodies suppressed tumor growth more effectively than either drug alone through reactivation of anticancer immunity (22). A phase I/II study of epacadostat in combination with pembrolizumab demonstrated promising activity in patients with advanced treatment-naive melanoma, with an objective response rate of 56% (23). However, in the subsequent phase III placebo-controlled trial (ECHO-301), the combination failed to improve survival in patients with advanced melanoma (24). A retrospective pooled analysis of epacadostat clinical studies determined that maximal blockade of IDO1 activity in the context of anti-PD-1 treatment requires significantly higher doses of epacadostat than was examined in the ECHO-301 trial (25).

On the basis of emerging data regarding the importance of IDO1 as a mechanism of immune escape in sarcomas and prior to the emergence of the ECHO-301 data, we designed and completed a study that examined IDO1 inhibition in combination with anti-PD-1 therapy for patients with advanced STS.

This single-center phase II trial evaluated the efficacy of the IDO1 inhibitor epacadostat in combination with the anti-PD-1 mAb pembrolizumab in patients with advanced sarcoma. The study was designed to enroll 30 patients into four cohorts stratified by histology: (i) UPS, LPS, MFS (n = 10); (ii) LMS (n = 5); (iii) vascular sarcoma, including angiosarcoma and epithelioid hemangioendothelioma (EHE; n = 5); and (iv) other (n = 10). The key inclusion criteria included age ≥18 years; histologically confirmed locally advanced or metastatic sarcoma; measurable disease by RECIST v1.1; ≥1 line of prior standard systemic therapy (where a proven standard of care systemic option is available); Eastern Cooperative Oncology Group performance status 0 or 1; and adequate organ function. Key exclusion criteria included known active central nervous system metastases; symptomatic autoimmune disease; clinically significant immunosuppression; known active tuberculosis, HIV or acute/chronic HBV or HCV; receipt of a monoamine oxidase inhibitor (MAOI) within 21 days of starting study therapy or a history of serotonin syndrome; QTc interval ≥480 milliseconds; and history of another concurrent, active malignancy. Study representation of underserved communities is available in Supplementary Data S1.

The study protocol was approved by the FDA and institutional review board at Memorial Sloan Kettering Cancer Center. The study was conducted in accordance with the Declaration of Helsinki and the Guidelines for Good Clinical Practice and was registered at ClincalTrials.gov (NCT03414229). Each participant provided protocol-specific, written informed consent in accordance with institutional guidelines.

Patients received both study drugs beginning on day 1 of each 21-day cycle. Epacadostat 100 mg was administered orally, twice daily, continuously for 21 days. Pembrolizumab 200 mg was administered once every 3 weeks by intravenous infusion. Treatment was continued until patients had complete response (CR), progressive disease (PR), or unacceptable toxicity or after 12 months of therapy. Patients experiencing disease progression were permitted to continue treatment beyond progression if deriving clinical benefit.

Participants underwent radiographic tumor assessments at baseline, every 8 weeks until week 56, and every 12 weeks thereafter until disease progression. Primary response assessment was by RECIST v1.1 (26) and secondary response assessment by immune-related (ir)RECIST (27). Safety was assessed every 3 weeks, and adverse events (AE) evaluated by the NCI Common Terminology Criteria for Adverse Events (CTCAE v4.03). Participants underwent mandatory tumor biopsies, when feasible, at baseline and week 8. Blood for research purposes was also obtained at screening and selected time points on study.

The primary outcome of this study was best objective response rate (ORR), defined as CR or PR by RECIST v1.1, at 24 weeks. Secondary outcomes included safety, best ORR by irRECIST, progression-free survival (PFS), and overall survival (OS).

The study used a one-stage design based on the exact binomial test. A 10% ORR was considered not promising; a 30% ORR was considered promising. The planned population was 30 patients. The design had a type I error rate of 0.07 and a type II error rate of 0.08. Efficacy analyses included patients who received study therapy and underwent evaluation for response. Safety analyses included patients who received at least one dose of treatment. Statistical tests were two-sided, and categorical data analyses and summary statistics were used to report AEs. The Kaplan–Meier method was used to estimate the distributions of time-to-event points.

IHC analysis of tumor biopsy material was performed by Qualtek molecular laboratory and the following biomarkers were examined: PD-L1, IDO1, kynurenine, CD3, CD8, and FOXP3. PD-L1 (22C3 antibody) reactivity (tumor expression) was defined as membranous PD-L1 expression at any intensity in ≥1% of tumor cells or tumor-infiltrating mononuclear inflammatory cells, which demonstrate membrane staining and/or the presence of a distinctive PD-L1 staining pattern at the tumor/stroma interface. For each clinical sample, the modified percent score (MPS) for PD-L1 was calculated; MPS is defined as the overall percentage of tumor and any tumor-infiltrating mononuclear inflammatory cells with membrane staining at low (1+) intensity or greater.

IDO1 (SP260 antibody) and kynurenine (3D4-F2 antibody) expression were interpreted by evaluating the estimated percentage of tumor cells that stain in the cytoplasm at a corresponding differential intensity. IDO1 reactivity was defined as expression at any intensity in ≥5% of tumor cells. IDO1 and kynurenine expression were evaluated separately using H-scores, the sum of each percentage score (0%–100%) multiplied by its corresponding intensity score (0, 1+, 2+, 3+), with scores ranging from 0 to 300.

Bioanalysis of prospectively collected (at baseline and prior to weeks 4, 7, 13, and 22) serum samples for kynurenine and tryptophan was performed by Worldwide Laboratory using LC/MS-MS. Summary statistics and paired t tests were used to analyze biomarker levels and their changes over time. Biomarkers were transformed on the log scale where appropriate. Landmark analyses along with log-rank tests and Cox proportional hazard models were used to assess the associations between biomarkers at on-study timepoints and PFS when sufficient data were available.

After RiboGreen quantification and quality control by Agilent BioAnalyzer, 107 to 500 ng of total RNA with RIN values of 5.4 to 10 underwent polyA selection and TruSeq library preparation according to instructions provided by Illumina (TruSeq Stranded mRNA LT Kit, catalog no. RS-122–2102), with eight cycles of PCR. Samples were barcoded and run on a HiSeq 4000 in a PE100 run (HiSeq 3000/4000 SBS Kit; Illumina). An average of 43 million paired reads were generated per sample. Ribosomal reads represented 1.9% to 19% of the total reads generated and the percent of mRNA bases averaged 69%.

Expression was quantified using Kallisto (28) and summarized at the gene level using Enembl v75. Normalized transcripts per million (TPM) were calculated using Sleuth (sleuth_to_martix). IDO pathway genes were manually curated. Hierarchical clustering of samples was performed using Manhattan distance and Wald D clustering.

To identify genes differentially expressed between patients with long versus short PFS (median of 7.64 weeks as cutoff, excluding 5 patients that were removed due to toxicity), the following model was used in Sleuth (29).

TrialCohort is one of four cohorts in the trial and PFS is “yes” or “no” according to whether the patient had PFS longer than 7.64 weeks. A Wald test was used to identify differentially expressed genes. A gene was considered differentially expressed if q < 0.05.

The data generated in this study are available upon request from the corresponding author, Ciara Kelly ([email protected]).

A total of 30 patients enrolled and received at least one dose of combination therapy between May 22, 2018 and August 22, 2019. The data cut-off date was November 17, 2020. Baseline demographics and tumor characteristics of the patients enrolled are shown in Table 1. Notably, 6 patients (20%) who entered the study were treatment-naive and 2 patients (7%) had received prior immunotherapy. Represented histological subtypes included LMS (16.7%), UPS (16.7%), MFS (6.7%), LPS (10%), EHE (10%), angiosarcoma (6.7%), and “other” SAR subtype (33%). Patients had received were refractory to 0 (20%), 1 (40%), 2 (23.3%), and ≥3 (16.7%) prior lines of therapy.

All 30 patients were evaluable for efficacy. The best ORR at 24 weeks by RECIST v1.1 was 3.3% (PR, n = 1; two-sided 95% CI, 0.1%–17.2%). This PR was observed at week 16 in a patient with treatment-naive LMS who experienced disease progression 9 months later. The disease control rate [CR + PR + stable disease (SD)] was 47% by RECIST v1.1.

The median follow-up among survivors was 29.7 months and the median PFS was 7.6 weeks (two-sided 95% CI, 6.9%–26.7%; Fig. 1). The 3-month, 6-month, and 12-month PFS rates were 46.7% (two-sided 95% CI, 31.8%–68.4%), 30.0% (two-sided 95% CI, 17.4%–51.8%), and 22.9% (two-sided 95% CI, 11.7%–44.5%), respectively. The median OS was 16.9 months (two-sided 95% CI, 9.4–not estimable: Fig. 2). The 12- and 24-month OS rates were 55.6% (two-sided 95% CI, 40.1%–76.9%) and 34.7% (two-sided 95% CI, 21.1%–57.2%), respectively.

Treatment with epacadostat and pembrolizumab was well tolerated. The most frequently observed treatment-related adverse events (TRAE) occurring at any grade in ≥15% of patients included fatigue (33.3%), rash (30%), and elevated aspartate aminotransferase (AST), increased serum amylase, and anemia (20% each). Grade 3 TRAEs were reported in 7 patients and included increased AST, anemia, arthritis, hypophosphatemia, increase in serum amylase, and increase in lipase (Table 2).

Three patients discontinued study treatment due to grade 3 or recurrent grade 2 elevations in transaminases. There were no deaths related to study treatment.

A total of 46 samples (24 baseline; 22 post-treatment) from 27 patients were available for IHC analysis. At baseline, 20 of 24 samples contained adequate tumor for analysis of PD-L1 expression; 13 were positive and 7 negative for PD-L1. The majority of PD-L1-positive tumors had PD-L1 expression within immune cells, and 3 of the 13 had PD-L1 expression ≥1% in the tumor cell component of the sample tested.

Of 24 patient samples, 19 contained adequate tumor for analysis of tumor IDO1 expression at baseline; 18 of 22 post-treatment samples had adequate tumor. All 19 baseline samples had IDO1 expression (≥5% tumor cells at intensity ≥1). In tumor samples at baseline and post-treatment, the median IDO1 H-scores were 200 [interquartile range (IQR), 142–255] and 220 (IQR, 200–254), respectively, whereas the median kynurenine H-scores were 230 (IQR, 185, 280) and 202 (IQR, 184, 258), respectively. There were nonsignificant differences in IDO1 H-scores (P = 0.09) and kynurenine H-scores (P = 0.7) between baseline samples and 7-week post-treatment samples (analyzed on the log scale). Week 7 landmark analyses showed no significant differences in PFS among those who had increased versus decreased IDO1 H-scores (logrank P = 0.12) or increased versus decreased kynurenine H-scores (log-rank P = 0.3), although the number of patients available at this landmark timepoint was limited (n = 8).

Analysis of immune cells within the tumor showed that the relative reactivity of CD3, CD8, and FOXP3 often stayed the same post-treatment, either increasing or remaining unchanged in >83% of pairs.

In the setting of adequate IDO1 inhibition, the serum tryptophan level should increase, and serum kynurenine level should normalize. Serum tryptophan and kynurenine were analyzed in 29 participants at baseline, week 4, week 7, and week 13 when adequate data were available. There was no evidence of significant change in log serum tryptophan levels observed between baseline and week 4, week 7, or week 13 (P = 0.3, 0.4, >0.9, respectively, paired t test; see Supplemenry Data S2).

The median serum kynurenine level at baseline was 1,806 nmol/L, which is lower than that observed in participants with several other cancer types who were enrolled in other clinical trials evaluating epacadostat. In our study, serum kynurenine values decreased between baseline and week 4 (median 1,595 nmol/L) but not significantly (mean difference on the log scale: 0.09; 95% CI, −0.01–0.19; P = 0.07, paired test). The levels did not significantly change between baseline and week 7 (P = 0.4) or week 13 (P = 0.7), whereas median levels at later time points remained slightly above the median healthy control value of 1,500 nmol/L: 1,595 nmol/L at week 4, 1,657 nmol/L at week 7, and 1,527 nmol/L at week 13 (see Supplementary Data S3). Landmark analyses showed no significant associations between percent change in serum kynurenine levels and PFS at week 4 (n = 29; percent change HR, 1.00; 95% CI, 0.98–1.01; P = 0.7) or week 7 (n = 18; percent change HR, 1.00; 95% CI, 0.99–1.01; P > 0.9). Serum tryptophan and kynurenine results were interpreted by stratifying patients based on the best objective response achieved per RECIST. All patients achieving a best response of SD (n = 13) had results available for the first three time points (baseline to week 6); 38.5% had increased serum tryptophan, whereas 46% had decreased serum kynurenine. Among those with progressive disease (n = 16), 15 patients had results available for the first three time points; 33.3% had increased serum tryptophan, whereas 66.6% had decreased serum kynurenine. The patient with a PR had decreased serum tryptophan and stable serum kynurenine levels over the initial 22 weeks on study therapy.

RNA sequencing was performed on 38 samples (17 paired samples [baseline and on-treatment week 6] and 4 unpaired samples) from 21 patients to compare the expression of genes in the IDO pathway. Pairs of samples from the same patient clustered together, suggesting little change in IDO expression following treatment (Fig. 3). By comparing baseline gene expression between patients with longer versus shorter PFS (using the median PFS of 7.64 weeks as cutoff), we identified 14 genes that were differentially expressed. Although no pathway enrichments were identified among these genes, the top two differentially expressed genes, PTGDS and CRIP1 (Fig. 4), were previously associated with longer OS in endometrial cancer and osteosarcoma, respectively (30, 31). Length of PFS was not associated with tumor mutational burden or number of expressed neoantigens at either time point (t test P value > 0.05, t test).

This phase II trial of epacadostat (100 mg twice daily) plus pembrolizumab did not result in clinically meaningful activity in patients with advanced sarcoma. The best ORR per RECIST v1.1 observed at 24 weeks was 3% and the median PFS was 7.6 weeks. The combination of epacadostat 100 mg twice daily plus pembrolizumab was well tolerated. No new safety signals were identified for either study drug. TRAEs were primarily grade 1 or 2 and manageable with dose modification or supportive therapy. Three patients discontinued study therapy due to hepatotoxicity.

Intratumor PD-L1 and IDO1 expression levels are known to vary depending on the sarcoma histologic subtype (32). PD-L1 and IDO1 expression in tumor cells is also associated with CD8⁺ effector T-cell tumor infiltration (18), although responses to immunotherapy have been observed in sarcoma subtypes regardless of PD-L1 expression (9, 33). This study did not include PD-L1 or IDO1 as part of the eligibility criteria. Tumor PD-L1 expression was detected in 65% of evaluable patient samples (n = 13/20) at baseline per the Qualtek definition of PD-L1 positivity assigned at analysis; however, further interrogation showed that only three of these cases had PD-L1 expression ≥1% in the tumor cell component of the sample evaluated, which aligns with the degree of tumor PD-L1 expression reported in the literature for sarcoma. In a retrospective analysis of 50 patients with sarcoma, PD-L1 expression in tumor cells was found in 12% of cases (32). The SARC028 phase II study of pembrolizumab monotherapy defined positive tumor PD-L1 expression as ≥1% antigen staining in malignant tumor cells and showed that 5% of patients (n = 2/40) demonstrated PD-L1-positive tumor expression and responded to therapy. The majority of patients in the study reported here had sarcomas expressing IDO1 at baseline but this did not translate into clinical activity of the combination regimen.

The open-label phase I/II ECHO-202/KEYNOTE-037 trial evaluated epacadostat plus pembrolizumab in patients with advanced solid tumors. The combination was well tolerated, and the maximum tolerated dose was not reached. Epacadostat 100 mg twice daily plus pembrolizumab 200 mg every 3 weeks were selected as the recommended phase II dose on the basis of slightly better tolerability and similar projected time-averaged IDO1 inhibition with epacadostat 100 or 300 mg twice daily. Promising antitumor activity was reported in patients with multiple tumors types, especially melanoma.

The phase III KEYNOTE-252 study of epacadostat or placebo plus pembrolizumab in patients with advanced melanoma failed to reach its survival primary endpoints despite the promising antitumor activity of epacadostat 100 to 300 mg twice daily plus pembrolizumab or nivolumab observed in patients with melanoma in two early-phase clinical trials (23, 34). When the results of KEYNOTE-252 were first reported, our phase II study of epacadostat plus pembrolizumab in advanced sarcoma was almost fully accrued. The limited activity of this combination therapy in sarcoma is unsurprising given its results in melanoma, an immunogenic tumor. However, the discordant results of the phase I/II and phase III trials in advanced melanoma raise the possibility that the epacadostat dose-level chosen did not achieve target inhibition.

Our correlative analyses in a sarcoma population suggest that adequate IDO1 target inhibition was not achieved. The IDO1 and kynurenine tumor expression measured by IHC did not change significantly following treatment. There was no evidence of significant change in serum tryptophan or kynurenine levels observed between baseline and later timepoints among participants. The serum kynurenine values decreased between baseline and week 4 but not significantly (paired test, P = 0.07) and thereafter did not normalize or fall below the median healthy control value (1500 nmol/L). The change observed in serum kynurenine levels between baseline and later timepoints was not associated with a significant difference in PFS. RNA sequencing of paired pre- and on-treatment tumor samples showed that expression of genes in the IDO pathway was similar in the baseline and on-treatment samples from the same individual, suggesting that IDO1 expression did not change following treatment exposure.

The KEYNOTE-252 study was a negative trial however, one of the biggest lessons drawn from this study was the importance of incorporating a robust correlative analysis plan into a study design. To improve future trial outcomes with IDO1 inhibition, a correlative analysis plan designed to inform patient selection and identification of predictive biomarkers and guide optimal dose selection in the context of the combination strategy under investigation should be emphasized. IDO1 induces an immunosuppressive effect on effector T cells and hyperactivates Tregs through the depletion of tryptophan and the accumulation of kynurenine in the tumor microenvironment. On target IDO1 inhibition can be examined by measuring kynurenine levels in pre- and on-treatment tumor tissue and blood samples taken from patients exposed to IDO1 inhibition. This information should help to inform optimal dose selection.

IDO1 inhibition alone is not expected to cause direct cytotoxicity or initiate a de novo antitumor immune response. In clinical studies, epacadostat alone did not show anticancer effect in most patients (23). Combinatorial strategies with other agents are necessary to cause tumor regression. When designing clinical trials incorporating IDO1 inhibition it is imperative to understand the effect of the combinatorial partner on the tryptophan–kynurenine pathway at baseline ideally in the study population of interest. A retrospective pooled analysis of kynurenine, as a pharmacodynamic marker of epacadostat activity, and epacadostat concentrations in prospectively acquired samples from participants in epacaostat trials showed that pembrolizumab monotherapy increased the production of plasma kynurenine. This study also showed that when combined with an anti-PD1 agent that epacadostat doses <600 mg twice daily were unable to maintain suppression of plasma kynurenine production (25).

Whether epacadostat at a higher dose and target coverage, particularly in less immunogenic tumor types such as sarcoma, would result in clinical activity remains unknown. The POD1UM-102 study is evaluating an anti-PD-1 mAb (retifanlimab) plus epacadostat at doses up to 900 mg twice daily. Preliminary results suggest that the maximally tolerated epacadostat dose is 600 mg twice daily, and that this dose maintains suppression of kynurenine to levels found in healthy control subjects (25).

Similar to KEYNOTE-252, this study did not prospectively screen patient's tumors for IDO1 expression. This was retrospectively examined in both studies. In this sarcoma study, most patient's tumors did express IDO1 at baseline. In KEYNOTE-252, 60% of participants had IDO1 positive tumors. It would be helpful if future trials incorporating IDO1 inhibition examined IDO1 expression in pre- and on-treatment tumor samples and correlated the intensity of tumor IDO1 expression with clinical outcomes to determine if this biomarker should be used to select patients most likely to benefit from this treatment.

Exploration of a broader range of relevant biomarkers may provide meaningful information to guide further development of IDO1 inhibition as a therapeutic approach. Two other enzymes, tryptophan 2,3-dioxygenase (TDO) and IDO2, are involved in the tryptophan–kynurenine pathway resulting in kynurenine production and have the potential to contribute to tumor resistance to IDO1 inhibition. Compensatory overexpression of TDO or IDO2 has been proposed as a mechanism of resistance to IDO1 inhibition in the KEYNOTE-252 study (35). Examining TDO and IDO2 expression in pre- and on-treatment tumor samples in the context of IDO1 inhibition may inform the success or failure of future combinatorial strategies such as concurrent TDO and IDO inhibition.

In summary, epacadostat 100 mg twice daily and pembrolizumab was safe and well tolerated in patients with advanced sarcoma. However, limited clinical activity was observed in this population and the study did not meet its primary endpoint, with the best ORR at 24 weeks of 3%, and median PFS was 7.6 weeks. One patient with treatment-naive LMS achieved a PR. The correlative analyses of paired pre- and post-treatment tumor samples did not show an association between treatment and PD-L1 or IDO1 tumor expression by IHC, and did not show IDO pathway–related gene expression by RNA sequencing. Similarly, there was no relationship between treatment response and the serum levels of tryptophan and kynurenine observed during the initial 6 weeks of treatment. The results of the correlative analyses suggest that on-target IDO1 inhibition was not achieved by the epacadostat dose used in this study. However, we do not know if the failure of this combination therapy in sarcoma is due to insufficient epacadostat dose alone or whether inate features of the sarcoma immune microenvironment may also limit the ability to effectively inhibit the IDO1 pathway.

C.M. Kelly reports other support from Incyte and Merck during the conduct of the study; other support from Amgen, Server, Kartos Therapeutics, Exicure, and Xencor outside the submitted work; and also reports employment with ChemoCentryx, Kartos Therapeutics, Immunicum, and Exicure. A.L. Richards reports grants and other support from Incyte and other support from Merck during the conduct of the study. V. Avutu reports grants and nonfinancial support from Incyte and nonfinancial support from Merck during the conduct of the study. J.E. Chan reports research support from ONO Pharmaceuticals relating to lung adenocarcinoma, which is not relevant to this work. P. Chi reports grants and personal fees from Deciphera and Ningbo NewBay; and grants from Pfizer outside the submitted work. M.A. Dickson reports grants and nonfinancial support from Incyte and nonfinancial support from Merck during the conduct of the study; grants from Eli-Lilly, Sumitomo, and AADi outside the submitted work. M.M. Gounder reports other support from Incyte and Merck during the conduct of the study; personal fees from Bayer and Regeneron; personal fees and other support from Boehringer Ingelheim, Ayala, Epizyme, Karyopharm, and Rain Oncology; other support from Springwork Therapeutics, Agios, and Glaxo SmithKline outside the submitted work. S. Movva reports grants and nonfinancial support from Incyte and nonfinancial support from Merck during the conduct of the study; grants from Ascentage Pharma, Hutchinson Medipharma, and Trillium and nonfinancial support from Clovis and Merck outside the submitted work. B.A. Nacev reports other support from Incyte and Merck, and grants from NCI during the conduct of the study; grants from NCI, CTOS, and QuadW Foundation/AACR outside the submitted work. E. Rosenbaum reports grants and other support from Incyte Corporation and Merck during the conduct of the study. T.E. Adamson reports non-financial support and other support from Incyte and Merck during the conduct of the study; other support from Bayer, Pfizer, Johnson & Johnson, GlaskoSmithKline, Takeda, and Catalyst Pharmaceuticals outside the submitted work. E.K. Bartlett reports grants from SkylineDx and personal fees from Excite International outside the submitted work. A.M. Crago reports nonfinancial support from Merck and grants and nonfinancial support from Incyte during the conduct of the study; personal fees from Springworks Therapeutics outside the submitted work. S. Hwang reports grants and nonfinancial support from Incyte and nonfinancial support from Merck during the conduct of the study. J.P. Erinjeri reports personal fees from AstraZeneca outside the submitted work. W.D. Tap reports personal fees from Eli Lilly, EMD Serono, Mundipharma, C4 Therapeutics, Daiichi Sankyo, Servier Pharmacueticals, Deciphera, Adcendo, Ayala, Kowa, Bayer, Epizyme, Cogent, Medpacto, Amgen, Foghorn, AmMaxBio, Boehringer Ingelheim, Bioatla, and Inhibrx during the conduct of the study; in addition, W.D. Tap has a patent for Companion Diagnostic for CDK4 inhibitors - 14/854,329 pending to MSK/SKI and a patent for Enigma and CDH18 as companion Diagnostics for CDK4 inhibition – SKI2016-021-03 pending to MSK/SKI; and reports employment with Certis Oncology Solutions; is a co-founder of Atropos Therapeutics and reports ownership of Atropos Therapeutics stock; reports ownership of Innova Therapeutics stock and reports employment with Innova Therapeutics. S.P. D'Angelo reports personal fees from Aadi bioscience, Adaptimmune, GI innovation, GSK, Pfizer, and Servier outside the submitted work. No disclosures were reported by the other authors.

C.M. Kelly: Conceptualization, data curation, software, formal analysis, validation, investigation, writing–original draft, project administration, writing–review and editing. L-X. Qin: Resources, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. K.A. Whiting: Data curation, software, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. A.L. Richards: Resources, data curation, software, formal analysis, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. V. Avutu: Resources, formal analysis, investigation, writing–original draft, writing–review and editing. J.E. Chan: Formal analysis, validation, investigation, writing–original draft, writing–review and editing. P. Chi: Resources, funding acquisition, validation, investigation, writing–original draft, writing–review and editing. M.A. Dickson: Resources, formal analysis, validation, investigation, writing–original draft, writing–review and editing. M.M. Gounder: Resources, formal analysis, validation, investigation, writing–original draft, writing–review and editing. M.L. Keohan: Resources, formal analysis, investigation, visualization, writing–original draft, writing–review and editing. S. Movva: Resources, formal analysis, investigation, writing–original draft, writing–review and editing. B.A. Nacev: Resources, validation, investigation, visualization, writing–original draft, writing–review and editing. E. Rosenbaum: Resources, formal analysis, validation, investigation, writing–original draft, writing–review and editing. T. Adamson: Resources, data curation, software, formal analysis, validation, investigation, methodology, project administration, writing–review and editing. S. Singer: Resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. E.K. Bartlett: Resources, formal analysis, investigation, writing–original draft, writing–review and editing. A.M. Crago: Resources, formal analysis, validation, investigation, writing–original draft, writing–review and editing. S.S. Yoon: Resources, data curation, formal analysis, investigation, writing–original draft, writing–review and editing. S. Hwang: Resources, validation, investigation, visualization, writing–original draft, writing–review and editing. J.P. Erinjeri: Resources, software, formal analysis, supervision, validation, investigation, visualization, writing–original draft, writing–review and editing. C.R. Antonescu: Conceptualization, resources, software, formal analysis, funding acquisition, validation, visualization, writing–original draft, writing–review and editing. W.D. Tap: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S.P. D'Angelo: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

We thank the patients who participated in this study and the clinical trial personnel at Memorial Sloan Kettering Cancer Center who were involved in this study. Editorial assistance at Memorial Sloan Kettering Cancer Center was provided by Hannah Rice, BA, ELS and Jessica Massler, MSW. This trial was funded in part by Incyte, Merck, and a Cycle for Survival Grant. At Memorial Sloan Kettering Cancer Center, support was also provided by the NIH/NCI Cancer Center Support Grant P30 CA008748 and NIH/NCI Grant P50 CA217694. The funders of the study did not contribute to the study design, data collection, data analysis, data interpretation, or writing of the report. The funders did review the manuscript before submission. C.M. Kelly, T. Adamson, KH, S.P. D'Angelo, and L-X. Qin had full access to all the data in the study. The corresponding author (to C.M. Kelly) had final responsibility for the decision to submit for publication.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.