Introduction

Although the advent of immune checkpoint blockade has changed the treatment landscape across multiple tumor types, about 30% of patients are classified as non-responders, or develop progression after initial response.1 Immunotherapy resistance arises as a consequence of various attributes within the tumor microenvironment (TME). These attributes encompass the composition of tumor-infiltrating lymphocytes (TILs), particularly cluster of differentiation 8 (CD8)+ T cells, which have been linked to treatment response.2 Additionally, the presence of tumor-associated macrophages (TAMs) and the activation state of specific regulators like phosphoinositide-3-kinase gamma or Paired Box 4 also contribute to immunotherapy resistance, with higher occurrences observed in non-responding individuals. Moreover, a lower proportion of programmed death ligand 1 (PD-L1) expressing cells, or a lower tumor mutational burden has been associated with reduced treatment responsiveness. Understanding these intricate interactions between the TME and immunotherapy response is crucial for devising effective therapeutic strategies.3 4 Sustaining a consistently strong and specific immune response in the human TME using the innate and adaptive branches of the immune system has continued to be a major hurdle in achieving or maintaining durable responses. This area of research requires the employment of strategic combinatory molecules and augmented techniques.1

In resting antigen-presenting cells and other myeloid cells, nucleotide-binding domain and leucine-rich repeat-containing protein 3 (NLRP3) is expressed at a low level and exists as an inactive cytoplasmic protein. Natural activators include ATP released from dying cancer cells that binds to cell membrane P2X purinoceptor 7 resulting in a change of cytoplasmic potassium concentrations.5 Chronic, low-grade inflammatory response mediated by NLRP3-dependent cytokines such as IL-1b is known to be associated with tumor-promoting effects. However, several lines of evidence also suggest that the acute activation of NLRP3 inflammasome results in “hyperactive” state of dendritic cells (DCs) which in turns leads to the robust antitumor immune response.6 This leads to brief conformational changes of NLRP3 allowing association with apoptosis-associated speck-like protein containing C-terminal caspase recruitment domain (ASC) and pro-caspase 1 to form the active NLRP3 inflammasome.7 Preclinical studies targeting this pathway with BAY-117082, a selective NLRP3 inflammasome inhibitor demonstrated tumor suppression and abscopal effects in an in vivo orthotopic model of oral squamous cell carcinoma.8 The novel NLRP3 agonist BMS-986299, a small molecule which targets the NLRP3, can improve the presentation of antigens through caspase 1-induced pyroptosis and stimulate the activation of T cells by releasing interleukin (IL)-1β and IL-18.9 IL-1β promotes the formation of T-cell memory response, while IL-18 activates natural killer (NK) cells.10 These effects, combined with its ability to polarize macrophages towards an inflammatory M1 type, may facilitate the immune control of immunologically “cold” cancers that lack TILs.11 12 During this process, pro-caspase 1 undergoes self-cleavage, attaining enzymatic activity to cleave pro-IL-1β and pro-IL-18 into their active, mature forms.13 Since NLRP3 expression is highly expressed in myeloid cells, and pro-IL-1β and pro-IL-18 are regulated by nuclear factor-kappa B, the activity of BMS-986299 is restricted to sites of inflammation and contact with antigens.14 15 Preclinical studies have demonstrated good tolerance to intravenous administration of BMS-986299 in tumor-free animals, while intratumoral administration of BMS-986299 induced systemic antitumor T cell response and effective immunological memory. Furthermore, in combination with systemic anti-PD1 or anti-CTLA4 therapies intratumorally administered BMS-986299 led to enhanced antitumor efficacy in both injected (I) and non-injected (NI) tumors as well as improved survival in several syngeneic mouse tumor models.9

The clinical safety, preliminary antitumor efficacy, pharmacodynamic (PD) and pharmacokinetic (PK) profile of BMS-986299 were evaluated in this first-in-human (FIH) study, as a monotherapy and in combination with nivolumab and ipilimumab in advanced solid tumors.

Patients and methodsPatient population

Patients with age ≥18 years and Eastern Cooperative Oncology Group (ECOG) performance status ≤1 with histologically proven, advanced, or metastatic solid tumors (excluding brain tumors) following at least one line of prior systemic therapy were eligible. Prior treatment with programmed cell death protein 1, PD-L1, or CTLA4 inhibitors was permitted if the best response to therapy was either progression or stable disease lasting less than 6 months. Additionally, specific types of cancer, including microsatellite stable colorectal cancer (CRC), pancreatic adenocarcinomas, castrate-resistant prostate cancer (CRPC), and hormone-refractory breast cancer (BC), could be included under specific conditions such as without need for documented checkpoint inhibitor therapy refractory state and without need for previous therapy with checkpoint inhibitors. Participants were required to have at least two accessible tumor sites for biopsies, with measurable target lesions per Response Evaluation Criteria in Solid Tumors (RECIST v1.1).16 Key exclusions were patients with active primary or secondary central nervous system tumor involvement, a history of symptomatic cryopyrin-associated periodic syndrome, conditions requiring systemic treatment with corticosteroids (>10 mg daily prednisone equivalent per day) or other immunosuppressive medications within 14 days of start of study treatment and unstable cardiovascular function.

Study design

This was an FIH, phase 1, open-label, multicenter study in which BMS-986299 (NCT03444753) was administered intratumorally as a monotherapy and in combination with intravenous nivolumab and ipilimumab to participants with advanced solid tumors and clinical characteristics indicating an inadequate cancer-directed T cell response. Here, we report the results of the study conducted at the University of Texas, MD Anderson Cancer Center until trial termination. The study involved dose escalation of BMS-986299 as a monotherapy, and once the safety of at least two monotherapy dose levels (DL) was confirmed, the combination with intravenous nivolumab and ipilimumab was explored to determine the maximum tolerated dose (MTD), maximum administered dose, and one or more recommended phase 2 doses for both monotherapy and combination treatment. Based on preclinical studies, increase in blood IL-1β levels was expected to occur early (around 4 hours) and be of short duration (resolved at 6 hours). To address this, a sentinel participant approach was employed, with overnight observation after the first two doses, and a minimum of 6 hours of observation at the treatment center after all other administrations of BMS-986299 for both sentinel and non-sentinel participants.

Throughout the study, PD and PK analyses were performed. The PD endpoints, which could influence dose selection, focused on assays that provided insights into whether BMS-986299 treatment resulted in immune-permissive changes in tumor tissue. The assessment of serum cytokines, in particular IL-1b, IL-18 and IL-1b-induced granulocyte colony-stimulating factor (G-CSF) and IL-6, was deployed to confirm target engagement on cycle 1 day 1 (C1D1) (pre-dose/2 hours/4 hours/6 hours post-dose); C1D2; C1D8 (pre-dose/2 hours post-dose/4 hours post-dose/6 hours post-dose); C1D9; C1D23 and subsequent cycles until end of treatment. Peripheral blood studies included immunophenotyping or characterization of the immune cell subsets in the periphery, including, but not limited to, T cells, monocytes, and NK cells and assessment of T cell receptor (TCR) rearrangements. These assessments required baseline and on-treatment biopsies for fresh tumor biopsy and archival formalin-fixed paraffin-embedded (FFPE) specimens to compare the expression of putative biomarkers to expression in the fresh tumor biopsy and could be used to assess NLRP3, CD3, CD8, and PD-L1 if the available archived material permits. The PD effects of BMS-986299, and BMS-986299 in combination with nivolumab and ipilimumab on changes in the level of IL-1β, IL-18, and NLRP3 messenger ribonucleic acid (mRNA) expression in tumor samples was assessed by real time-polymerase chain reaction (RT-PCR). The number, frequency, and composition of immune infiltrates within the tumor was also assessed before and after exposure to BMS-986299 and BMS-986299 in combination with nivolumab and ipilimumab using immunohistochemistry (IHC) of the FFPE tumor biopsy specimen. IHC assays were performed using, but not limited to, the following markers: CD8, CD3, CD4, FOXP3, PD-L1, NLRP3. Pretreatment biopsy was collected at least 3 days prior to administration of study treatment. On-treatment biopsy was conducted on cycle 1 day 23 (maximum of 3 days after day 22 dosing). The study had primary objectives to characterize the safety, tolerability, dose limiting toxicities (DLTs), and MTD of intratumorally administered BMS-986299 alone and in combination with intravenous nivolumab and ipilimumab. The primary endpoints included the incidence of DLTs, AEs (adverse events), SAEs (serious adverse events), AEs leading to discontinuation, deaths, and incidence of clinical laboratory abnormalities. Secondary objectives included assessing whether BMS-986299 alone and in combination with intravenous nivolumab and ipilimumab increased infiltration of the I and NI tumors with CD8+ lymphocytes. Other secondary objectives included to characterizing the systemic PK profile of BMS-986299, both as a monotherapy and in combination with nivolumab and ipilimumab, if feasible based on systemic BMS-986299 levels. Secondary endpoints included measure of change from baseline in CD8+ T cell infiltration as per cent of nucleated cells by IHC staining and as expression levels of CD8 T-cell specific genes, measures of BMS-986299 PK parameters in plasma and urine, such as but not limited to, Cmax, Tmax, area under the curve from 0 to 24 hours (AUC (0–24)), AUC (0–T), from concentration-time data during BMS-986299 monotherapy, and BMS-986299 concentrations during combination treatment with nivolumab and ipilimumab.

The lesions must have a long axis of at least 2 cm. The tumor lesion chosen for BMS-986299 injection must either be located in the skin, subcutaneous tissues, muscle, lymph nodes, or oral cavity. The lesion chosen for biopsy to investigate for a distant immunological effect at a non-injected site (abscopal effect) must not be located in the brain or bone but may be located in any other organ that can be biopsied safely. Both tumors may not involve a major blood vessel. BMS-986299 was administered via a 2 mL injection of 5% dextrose solution directly into the tumor, using either a single push or multiple injections within the same tumor site as clinically required. For all DLs, re-injection at the same site was contingent on the tumor being at least 5 mm in size, precisely localizable, and safely accessible via palpation or imaging techniques, without signs of abscess or hematoma, and with any prior injection site reactions diminished to grade 1 or lower. If the tumor shrank below 5 mm or became unsafe to inject, and if all BMS-986299-related AEs were reduced to grade 1, an alternative tumor fitting the initial injection criteria could be chosen for injection. This alternative tumor is not selected for biopsy of a non-injected lesion unless the required day 23 biopsy had been completed. If no suitable tumor sites were available, injections were halted. Additionally, given BMS-986299’s absorption of ultraviolet-A (UV-A) light (315–400 nm), patients receiving injections in skin-adjacent tumors were advised to minimize sun exposure to the treated area for 1 week post-injection, using protective clothing or sunscreen as necessary.

BMS-986299 as monotherapy

The study commenced with participant accrual to part 1A of BMS-986299 monotherapy, involving escalating DLs. Each new DL initially involved the treatment of three participants. Cohort tolerability assessment and dose recommendations were made based on the Bayesian logistic regression model (BLRM) after at least two evaluable participants completed the DLT period during the first 21 days of BMS-986299 monotherapy.17 The DLT period was arbitrarily defined. BMS-986299 has a short half-life, and we hypothesized that the 21-day timeframe would be appropriate to interrogate DLTs on repetitive weekly administration of BMS-986299. These boundaries were similar to the toxicity boundaries used by a rule-based design (ie, 3+3 design) in that a minimum is set at 16% (~1 in 6) DLT rate and a maximum at 33% (~2 in 6) DLT rate. Based on the Dose-DLT plot, for BMS-986299 doses 600 µg, 1200 µg, and 2000 µg, the median dose-DLT curve lay between the pre-specified toxicity interval limits of 16% and 33%.

BMS-986299 in combination with nivolumab and ipilimumab

Accrual to the combination arm (part 1B) was initiated once the safety of at least two DLs of BMS-986299 in monotherapy (part 1A) was established. Part 1B of the study was opened when part 1A was still underway. BMS-986299 dose selection for part 1B was driven by safety data (no DLTs) and evidence of TE/PD (greater than or equal to twofold increase in greater than or equal to two peripheral biomarkers (G-CSF, IL-1b, calprotectin)). The objective was to select three tolerable, pharmacodynamically diverse DLs to further identify optimal BMS-986299 dose for combination with nivolumab and ipilimumab. The dose escalation was determined using the BLRM-Copula method. The selection of dose and schedule for combination therapy was based on the safety, PK, and PD results observed in the monotherapy phase. The doses of BMS-986299 used in combination therapy were set to be at or below the highest tolerable dose identified in monotherapy, with a minimum evaluation period of DLT in at least two participants during the first 28 days of BMS-986299 with fixed-dose nivolumab and ipilimumab.

Study schedule

In part 1A, a total of seven dose escalation cohorts were established, ranging from 75 μg to 4000 μg for two cycles, following BLRM to determinine the optimal dose. During cycle 1, patients received IT BMS-986299 on days 1, 8, 15, 22, and 29, and in C2, the treatment was administered on days 1 and 29, with an interval of 8 weeks followed by active surveillance.

In part 1B, the study continued with three additional dose escalation cohorts for BMS-986299, ranging from 75 μg to 2000 μg for two cycles with IT therapy in C1 on days 1, 8, 15, 22, and 29 and in C2 on days 1 and 29. In combination with BMS-986299, patients also received fixed intravenous dosing of nivolumab at 480 mg every 4 weeks and ipilimumab at 1 mg/kg every 8 weeks, starting from day 1 of each 56-day cycle. Nivolumab was administered before ipilimumab on days when both were given. The combination immune checkpoint inhibitor (ICI) therapy was continued for up to 24 months or until disease progression.

Cross-over (CS) was allowed for patients who participated in part 1A to transition into part 1B of the study, providing an opportunity for patients to receive the combination therapy with nivolumab and ipilimumab only after monotherapy completion with BMS-986299. Patients receiving only BMS-986299 who do not show at least a partial response according to RECIST v1.1 criteria by week 16 or who see their disease worsen could switch to a combination therapy of nivolumab and ipilimumab (without additional BMS-986299). Transition to combination therapy required that patients have either completed a minimum of 21 days on BMS-986299 without experiencing significant adverse effects or any previous DLTs have resolved to a toxicity grade of 1 or less. Additionally, the final dose of BMS-986299 must have been administered at least 7 days before starting the combination therapy, and the patient must have continued to meet the trial’s eligibility criteria. Participants who switched to nivolumab and ipilimumab continued this regimen for up to 2 years and remained under study observation for the same duration from the commencement of the combination treatment.

Assessments

Safety analyses in the study were conducted based on reported AEs and measurements of vital signs, ECGs, physical examinations, and clinical laboratory tests. Both AEs and laboratory test results were graded using the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI CTCAE) v5.0. DLTs were grade 4 elevations in liver enzymes or bilirubin, grade 4 myelosuppression, grade 3 thrombocytopenia with bleeding or platelet transfusion requirement, grade 3 neutropenia with fever, grade 3 hemolysis, grade 4 rash, grade 3 rash with no improvement after an infusion delay, severe drug-related eye issues, drug-related pneumonitis, bronchospasm, or neurological toxicity, severe hypersensitivity reactions, and other grade ≥3 study drug-related toxicities.

PK assessments (online supplemental table 1A,B) involved collecting serial plasma samples from participants receiving monotherapy and combination therapy to determine the levels of BMS-986299 by using a validated bioanalytical method. PK parameters such as AUC (0–24), maximum concentration (Cmax), and time to maximum concentration (Tmax) were calculated using non-compartmental analysis (Phoenix WinNonline 8.3).

PD biomarker assessments (online supplemental table 2) aimed to measure target engagement, tumor infiltration by immune effector cells, and potential predictive markers. Baseline and on-treatment biomarker measures were obtained, including biopsies of I and NI tumors. Serial blood draws were taken to assess serum cytokine production over the course of BMS-986299 treatment (IL-1β/HMPCore2 Luminex panel, IL-6, IL-18/HMPCORE1 Luminex panel, G-CSF/HMP9 Luminex panel, Rules Based Medicine, Austin, TX). Additionally, tumor biopsies were analyzed by IHC (single plex PD-L1, Clone 28-8; single plex myeloperoxidase (MPO), clone 59A5; single plex CD163, clone 10D6; multiplex CD3/CD8/granzyme B-CD3, mouse polyclonal; CD8, clone C8/144B; granzyme B, clone 11F1; Cell Carta, Irvine, CA) to evaluate immune responses and cell types. Paired t-tests were used for IHC statistical analysis.

Injectable lesion selection for intratumoral delivery was performed in a multidisciplinary manner. Considerations included safety, accessibility from a percutaneous approach, and lesion composition (ie, necrosis, vascularity, etc). Intratumoral delivery was performed under image guidance. To standardize the delivery rate, the investigational agent was administered via a syringe pump (Medfusion 4000) at 1 cc/min. Furthermore, given prior experience with the improvements in delivery through various needle designs injections were performed using a multi-side hole needle (ProFusion, Cook Medical).18 Furthermore, to investigate the biophysical characteristics of the lesion, a digital in-line piezoelectric pressure transducer (CompassCT, Cook Medical) was used to measure interstitial pressure at baseline, during the injection, and immediately following the injection. Pressure measurements were read directly from the digital screen of this device at three-time points: immediately prior to injection, halfway through the injection, and at the conclusion of the injection.

Preliminary antitumor efficacy assessments used imaging techniques such as computed tomography (CT) and magnetic resonance imaging (MRI). Tumor assessments occurred every 8 weeks starting from the date of the first dose ±7 days for the first year, then every 12 weeks ±7 days until participant treatment was discontinued. Tumor responses were determined using RECIST v1.1,16 and for prostate cancer participants, Prostate Cancer Working Group 3 criteria were also used.19 Tumor measurements and responses were reported on the case report form, with a specific focus on injected tumor lesions and overall response assessment which included both injected lesions and non-injected lesions.

Study oversight

To ensure comprehensive safety monitoring, collaboration among study site investigators, the Bristol-Myers Squibb (BMS) study team, and the Safety Management Team. This collaborative process served as the safety monitoring plan for the study. Consensus support from all members of the meeting was sought, and investigators provided written agreement before proceeding with dose escalation.

Statistical analysis

Baseline characteristics, treatment-related adverse events (TRAEs), and preliminary antitumor efficacy assessments were summarized using descriptive statistics. Progression-free survival (PFS) in months was obtained from cycle 1 start date to the date of disease progression or death (if the patient died without disease progression), or the last evaluation date. Patients who were alive and did not experience progression at the last follow-up date were censored. Overall survival (OS) was computed from the cycle 1 start date to the last known contact if consent was provided for follow-up. Patients alive at the last follow-up date were censored. Paired t-tests were used for analysis of tumor IHC data. The Kaplan-Meier method was used to estimate PFS and OS.20 All statistical analyses were performed using SAS V.9.4 for Windows (Copyright 2002–2012 by SAS Institute, Cary, NC) and R 4.3.1.

ResultsPatient population

36 patients were enrolled in the study between April 2018 and April 2021 at the University of Texas MD Anderson Cancer Center. Demographic and clinical characteristics of all patients enrolled are summarized in table 1. Participants gave informed consent to participate in the study before taking part. In part 1A, 27 patients enrolled where the median age of the patients was 50.5 years (range 35–74 years), 67% were female and all patients had an ECOG performance status of 1. In part 1B, nine patients enrolled where the median age of the patients was 57.5 years (range 46–71 years), 67% were female and all patients had an ECOG performance status of 1. The ethnic distribution of patients across the study showed that the majority were Caucasian (72%), followed by African-American (14%), Asian (6%), and Hispanic (8%). The median injected tumor size varied depending on the DL, ranging from 1.3 to 7.4 cm. The trial enrolled patients with various primary tumor types, and the most common tumor types enrolled were: 31% had BC (either triple-negative breast cancer (TNBC) at 19% or hormone receptor positive (HR+) human epidermal growth factor receptor 2 (HER2) negative BC at 11%); 17% CRC; and 14% head and neck squamous cell cancer. In the cohort of 36 patients, liver metastasis was observed in 21 (58.3%), followed by lymph nodes in 25 (69.4%), lungs and bones in 16 each (44.4%), peritoneum in 5 (13.9%), adrenal glands in 4 (11.1%), and soft tissues in 3 (8.3%). Less frequent occurrences were noted in breasts, kidneys, vagina, face, ovaries, and muscles. Most patients (36%) had received 2–3 prior lines of therapy with 28% of patients having received >5 lines of therapy. 21 patients (58%) had received prior immunotherapy treatments. Nine patients crossed over from part 1A monotherapy to part 1B combination at the time of progression. Background information on the cancer under study and the representativeness of the study population is depicted in online supplemental table 3.

Table 1

Baseline clinical characteristics and demographics of patients treated with BMS-986299 administered intratumorally as monotherapy and in combination with intravenous nivolumab and ipilimumab

Safety and tolerability

At the time of study termination, all 36 patients (100%) had discontinued study treatment with the majority at 69% (25 patients) discontinuing for progressive disease while one patient completed 2 years of therapy. Four patients (11%) withdrew consent while two patients (6%) came off study due to treatment related toxicities (one each of grade 4 interstitial nephritis and grade 3 hepatotoxicity). One patient came off study after developing grade 3 stress-related cardiomyopathy that was unrelated to trial therapy while another patient developed grade 5 myocardial infarction resulting in death, considered unrelated to trial therapy. Finally, one patient was lost to follow-up while another patient was discontinued from trial due to non-compliance. The median number of treatment cycles with BMS-986299 monotherapy and combination therapy was 1 (range, 1–13). For patients with clinical benefit (partial remission (PR) or stable disease (SD) >6 months), the median number of treatment cycles was 1 (range, 1–13).

All patients were evaluable for safety who received any dose of intratumoral BMS-986299. In total, 33 patients (91.7%) had at least one TRAE. Summaries of all TRAEs are presented in tables 2 and 3. Across the study, the most common grade (G)1-G2 TRAEs were fever (n=24/33; 73%), neutrophilia (n=12/33; 36%), and leukocytosis (n=11/33; 33%). In part 1A of the study, 24 patients experienced TRAEs. The most common G1-G2 TRAEs were fever (n=19/24; 79%); leukocytosis (n=10/24; 42%); and neutrophilia (n=9/24; 38%). The most common G3-G4 TRAEs were injection site pain (n=1/24; 4%) in DL2 and G4 interstitial nephritis requiring dialysis and eventual recovery of kidney function (n=1/24; 4%) in DL3. During post-therapy follow-up, the patient reported normalization of kidney function and no longer required dialysis. Hence, additional clinical details or laboratory data regarding renal recovery were not collected or further investigated at that time. In part 1B of the study, nine patients experienced TRAEs. The most common G1-G2 TRAEs were fever (n=5/9; 56%); neutrophilia (n=3/9; 33%), fatigue (n=2/9; 22%), and nausea (n=2/9; 22%). In DL6, one patient experienced G3 colitis and G3 aspartate aminotransferase/alaine aminotransferase (AST/ALT) elevation with resolution after tapering steroids. Four patients experienced G1-G2 cytokine release syndrome (CRS). None of the participants who experienced CRS required anti-IL-6 or other immunosuppressive therapy.

Table 2

Summary of treatment-related adverse events (TRAEs) in patients treated with BMS-986299 administered intratumorally as monotherapy

Table 3

Summary of treatment-related adverse events (TRAEs) in patients treated with BMS-986299 administered intratumorally in combination with intravenous nivolumab and ipilimumab

Efficacy

Among the 36 treated patients, 30 (83%) were evaluable for efficacy due to having RECIST v1.1 measurable disease and having completed post-baseline tumor assessment by radiographic imaging studies or physician-determined clinical progression. Six patients were not evaluable for efficacy secondary to withdrawal of consent (n=2); non-compliance (n=1); clinical progression (n=2); and death from myocardial infarction, which was unrelated to study (n=1). Waterfall and swimmer plots of overall and injected tumor responses are depicted in figure 1A,B. A summary of the best overall tumor response and injected tumor response per RECIST v1.1 is shown in online supplemental table 1A,B. The objective response rate (ORR), defined as the proportion of patients with either a complete response (CR) or PR was 10% with three PRs seen in TNBC, HR+ BC, and cutaneous squamous cell carcinoma (cSCC). No patients experienced CR on this study. Concordant responses were also seen in the injected lesions.

Figure 1

(A and B) Antitumor activity and response duration on therapy among patients treated with BMS-986299 administered intratumorally as monotherapy and in combination with systemic nivolumab and ipilimumab. In (A), waterfall plot depicting antitumor activity of patient’s overall response and injected tumor response per RECIST v1.1 treated with BMS-986299 administered intratumorally as monotherapy and in combination with systemic nivolumab and ipilimumab. In (B), swimmer plot depicted response duration on therapy among patients treated with BMS-986299 administered intratumorally as monotherapy and in combination with systemic nivolumab and ipilimumab. Combo, combination; Mono, monotherapy; PD, progressive disease; PR, partial response; SD, stable disease; TNBC, triple-negative breast cancer.

In part 1A, across all DLs (DL1 to DL7), a total of 21/30 (70%) patients were evaluable for efficacy. One patient with TNBC, who had received one line of prior ICI therapy, experienced concordant PR (−67%) in DL6 (2000 µg IT) with a duration of response (DOR) of 22.9 months after crossing over to part 1B combination arm after 2 months on monotherapy due to progression at first restaging to complete 2 years of therapy.

In part 1B, across three DLs (using DL1, DL3, and DL6 monotherapy dose of BMS-986299 in combination with intravenous nivolumab and ipilimumab), a total of nine patients were evaluated (five patients at DL6 and two patients each at DL1 and DL3) where the ORR was 33% (n=3/9) for both the injected lesion and overall best response in an ICI-refractory cSCC, an ICI-naive HR+ BC patient and the TNBC patient who crossed over from part 1A monotherapy cohort, respectively. In DL6, across five patients enrolled, the ORR was 40% (n=2). In terms of the best response in the injected lesion, one patient with ICI-refractory TNBC in DL3 achieved a PR (−52%), one ICI refractory TNBC patient who crossed over from part 1A achieved a PR (−67%) and one patient with ICI-naïve HR+ BC in DL6 achieved a PR (−67%). However, in terms of the overall best response, the TNBC patient in DL3 experienced an unconfirmed PR (−39%) while the three patients at DL6 achieved confirmed PRs (one patient with ICI-naive HR+ BC had PR (−86%) with DOR of 4.5 months coming off study for G3 hepatotoxicity; one patient with cSCC had PR (−43%) with DOR of 3.9 months and one crossover patient with PR (−67%) with DOR of 22.9 months).

The median overall survival was 5.2 months (95% CI 2.5, 9.0). The OS rates at 6 and 12 months were 46.1% and 23.0%, respectively. The median PFS was 1.7 months (95% CI 1.0, 2.0). Kaplan-Meir curves depicting OS and PFS among all trial participants are shown in online supplemental figure 1A,B.

Pharmacodynamics

While not driven in dose-dependent fashion, part 1A BMS-986299 intratumoral injected cohort resulted in increasing trends for cytotoxic T-lymphocytes (CTLs) (8/12, 64% increase, median fold change (MFC): 1.3), CD8−/CD3+ tumor infiltrating lymphocytes (TILs) via IHC (7/11, 64% increase, MFC: 2.8) and in subset of patient for CD8−/CD3−/granzyme B+ (indirectly NK cells) via IHC (5/12, 42%, MFC 1.5) in figure 2A. NI tumors demonstrated similar increasing trends for immune cell content as reflected by influx of CTLs (6/9, 67%, MFC: 1), CD4+ TILs (5/9, 56%, MFC: 1.8) and NK cells (5/9, 56%, MFC: 0.8). These increasing trends in tumor immune cell content was further characterized by expression of PD-L1 on immune cells, and I and NI lesions demonstrated an influx of PD-L1+ immune cells (I: 5/10, 50%, MFC: 1.5, NI: 5/6, 83%, MFC: 5).

Figure 2

(A–C) Comprehensive pharmacodynamic analysis of patients treated with BMS-986299 administered intratumorally as monotherapy and in combination with systemic nivolumab and ipilimumab. In (A), tumor immune cell content was measured in injected (Inj.) and non-injected (NI) tumor lesions following three weekly doses of BMS-986299 monotherapy treatment. Pretreatment and post-treatment (C1D28) biopsies were collected and characterized by immunohistochemistry for CTLs (CD3+/CD8+/Gzb+), CD4+ T cells (CD3+/CD8−), NK cells (CD3−/CD8−/GzB+) and PDL1+ immune cells. In (B), tumor myeloid cell content was measured by immunohistochemistry for MPO+ TANs and CD163+ TAMs. In (C), serum cytokine analysis was measured in the first 24 hours following initial BMS-986299 monotherapy (part 1A) or combination with nivolumab and ipilumimab (part 1B, N+I) for IL-1β, IL-18, G-CSF, and IL-6. Paired t-tests were used for pretreatment and post-treatment immunohistochemistry analysis. CD, cluster of differentiation; CTL, cytotoxic T-lymphocyte; G-CSF, granulocyte colony-stimulating factor; IL, interleukin; MPO, myeloperoxidase; N+I, nivolumab and ipilimumab; NK, natural killer; PDL1, programmed death-ligand 1; TAMs, tumor-associated macrophages; TAN, tumor-associated neutrophils.

Tumor myeloid content was also assessed for TAMs and tumor-associated neutrophils (TANs) by CD163 and MPO IHC assays, respectively in figure 2B. In the part 1A, treated patients, a subset of patients had an increase in TAMs (5/12, 42%, MFC: 1.54) and TANs (5/11, 45%, MFC: 2.2) in I lesions. In NI lesions, CD163+ TAMs had an increasing trend (6/11, 54%, MFC: 1.1) and little change was observed in TANs (3/11, 27%, MFC: 2.2).

Peripheral BMS-986299 analysis was performed through characterization for key cytokines associated with NLRP3 inflammasome activity (IL-1β, IL-18) and downstream mediators (IL-6 and G-CSF). Part 1A and 1B patients were assessed for cytokine changes in the first 24 hours following BMS-986299 treatment. While serum IL-18 did not increase following treatment, serum IL-1β was denoted to have at least a twofold-change from baseline in 1200 (2.9-fold increase), 2000 (6.6-fold increase), and 4000 µg (2.6-fold increase) BMS-986299 treated patients in figure 2C. Downstream cytokines G-CSF and IL-6 had at least twofold change from baseline in BMS-986299 1200 (G-CSF:4.8-fold increase, IL-6: 9.3-fold increase) and 2000 µg (G-CSF: 21-fold increase, IL-6: 9.1-fold increase) cohorts. Similarly, the part 1B 2000 µg combination cohort had at least a twofold increase in G-CSF (4.7-fold increase) and IL-6 (3.4-fold increase).

The metastatic TNBC patient with PR in part 1A (figure 3A) (2000 µg BMS-986299 monotherapy cross-over) demonstrated a ~57% decrease in skin lesions that was accompanied by robust peripheral and tumor PD changes. Following two weekly doses of BMS-986299, serum IL-18 increased by 2.5-fold, serum IL-1B by 8.5-fold, serum IL-6 by 15-fold, and serum G-CSF by 31-fold change in figure 3B. While no noticeable changes in CD8+ TILs were observed post-treatment, the injected lesions demonstrated an increase in TAMs (1.7-fold change) and both I and NI lesions had approximately fivefold increase in PD-L1+ immune cells in figure 3B.

Figure 3

(A and B) Translational analysis of a TNBC patient with PR (−67%) originally assigned to BMS-986299 administered intratumorally as monotherapy who later crossed over to therapy with ipilimumab and nivolumab. In (A), patient with metastatic TNBC is featured with clinical timeline and therapy with clinical and radiological response. In (B), serum IL-1β, IL-18, G-CSF, and IL-6 have robust increases following two BMS-986299 (2000mcg) doses. Treatment also resulted in increases in PDL1+ on immune cells in both injected and non-injected tumors, along with increased CD163+ TAMs in the injected lesions. C1D1, cycle 1 day 1; Carbo, carboplatin; CR, complete response; G3, grade 3; G-CSF, granulocyte colony-stimulating factor; IHC, immunohistochemistry; IL, interleukin; LN, lymph node; Mono, monotherapy; N+I, nivolumab+ipilimumab; PD-1, programmed cell death protein 1; PDL1, programmed death-Ligand 1; Rt, right; TNBC, triple-negative breast cancer.

Since systemic levels of BMS-986299 were observed after intratumoral administration and notable systemic inflammatory cytokines increased after treatment, a potential causal relationship between the maximal systemic BMS-986299 levels (Cmax) and increased systemic cytokines were evaluated. Systemic BMS-986299 Cmax was positively associated with the highest percentage change from baseline for systemic IL-6 on day 1 (first dose) and day 8 (second dose) and the relationships were statistically significant, suggesting causal inference for BMS-986299 and IL-6 (figure 4B). Systemic BMS-986299 Cmax was positively associated with the highest percentage change from baseline for systemic G-CSF on day 1 and day 8 and demonstrated statistical significance of the relationship on day 8. Systemic levels of the highest percentage change from baseline for IL-18 and IL-1B were not positively associated with BMS-986229 Cmax on day 1 but showed some positive association on day 8 with IL-18 reaching statistical significance (figure 4B). Taken together, this suggests BMS-986229-dependent effects on elevated proinflammatory cytokines is consistent with the mechanism of drug action.

Figure 4

(A and B) Comprehensive pharmacokinetic analysis of patients treated with BMS-986299 administered intratumorally as monotherapy and in combination with systemic nivolumab and ipilimumab. In (A), mean concentration–time profiles of plasma BMS-986299 following BMS-986299 intratumorally dosing. Lines with colors represent the mean concentration–time profiles obtained with different doses on cycle 1 day 1 and day 8. Solid line and dotted line represent the mean concentration–time profiles as monotherapy (P1A) and in combination with nivolumab and ipilimumab (P1B). Dashed gray lines are reference lines representing LLOQ. Values for individual participants that were below LLOQ (0.1 ng/mL) were set as missing. Mean values below LLOQ are omitted from the graph. LLOQ, lower limit of quantification. In (B), cytokine level change is defined as highest percentage change from baseline (%) in relation to BMS-986299 Cmax for each subject following BMS-986299 intratumorally dosing. Symbols with colors represent each individual subject administered different BMS-986299 dose on cycle 1 day 1 and day 8. Solid cycles and triangles represent each individual subject administered BMS-986299 as monotherapy (P1A) and in combination with nivolumab and ipilimumab (P1B). Gray lines are linear regression lines and gray shaded area around linear regression lines represent SE. R2 and p value represent the results of the linear regression between cytokine level change and BMS-986299 Cmax. Cmax, maximum plasma concentration; G-CSF, granulocyte colony-stimulating factor; IL-1 beta, interleukin 1 beta Simoa; IL-18, interleukin 18; IL-6, interleukin 6.

For tumor pharmacodynamic analysis, tumor biopsies from part 1A and part 1B patients were collected from the I and NI tumor lesions prior to treatment and at C1D28. Matched pair tumor biopsies were analyzed by IHC for changes in immune cell content and stromal cell types. Part 1A matched pair analysis was pursued, while part 1B matched pair analysis was confounded by lack of matched tumor biopsies.

Tumor pressure assessments during intratumoral injections

Intratumoral pressure values varied significantly depending on the location of the injectable lesion (online supplemental figure 2A,B). Deep lesions as well as lesions in solid organs demonstrated significantly greater interstitial pressure relative to superficial lesions as well as lesions within lymph nodes. On injection of the investigational agent, there was an increase in interstitial pressure for all lesions, though the increase was greatest for lesions with the highest baseline interstitial pressure. Following injection, interstitial pressures decreased and approached baseline values.

Pharmacokinetics

BMS-986299 was rapidly absorbed into the systemic circulation following intratumor administration with a median Tmax ≤1–2 hours across DLs. Plasma exposure increased with increasing dose after single- and multiple-dose administration. The PK of BMS-986299 is similar between day 1 and day 8, with no accumulation after two weekly doses observed in participants across the dose groups evaluated. While subject numbers were limited for part 1B, systemic BMS-986299 exposures tended to be similar in the combination as compared with monotherapy (P1A), suggesting no drug–drug interaction. Figure 4A (up to 24 hours) and online supplemental table 5 represent the PK concentration-time profiles and PK parameters following BMS-986299 dosing on cycle 1 day 1 and day 8 up to 168 hours post-dose. Inter-subject variability was moderate with 7.2%–88% CV across DLs for AUC (0–24 hours).

Discussion

This study demonstrated that intratumoral administration of BMS-986299, both as a monotherapy and in combination with nivolumab and ipilimumab, was generally well-tolerated. The adverse events observed were mostly manageable and consistent with the known safety profiles of ICIs. The use of a sentinel participant approach allowed for the early detection and appropriate management of potential IL-1β-mediated toxicities.

The monotherapy arm of the study revealed encouraging preliminary efficacy signals. One patient with immunotherapy-refractory TNBC had a PR with an ORR of 5%. The combination of BMS-986299 with nivolumab and ipilimumab demonstrated enhanced antitumor efficacy compared with monotherapy especially in cSCC, HR+ BC, and TNBC. When comparing monotherapy with combination therapy, it is evident that the addition of ipilimumab and nivolumab to BMS-986299 increased the ORR from 0% (n=0/21) to 33% (n=3/9). However, it should be noted that the combination therapy was evaluated in a smaller patient population (n=9) compared with the monotherapy (n=21), which could impact the interpretation of these results.21

Intratumoral levels of BMS-986299 were not measured. However, there were positive increases in systemic BMS-986299 levels with dose, providing a surrogate for tumor exposure. Systemic BMS-986299 levels increased with increasing dose with moderate to high inter-subject exposure variability (generally >40% CV) across DLs. BMS-986299 exposures (Cmax and AUC 0–24 hours) were maintained after the first (day 1) and second doses (day 8) at each DL, suggesting the tumor had consistent BMS-986299 exposure after each dose for at least 24 hours after dosing. The optimal dosing frequency of intratumorally administered innate immune agonists is poorly understood, and syngeneic mouse tumor models have provided limited value in informing dosing schedules for further clinical development. We hypothesized that a more frequent dosing might be required initially to induce NLRP3-mediated T cell priming. Later on, a less frequent dosing might be preferable to prevent tachyphylaxis. This was reflected in the study design: during C1 (weeks 1–8), study participants received weekly BMS-986299 dosing during weeks 1–4, and then switched to Q4W dosing during C1 W5-8 and following C2. Additionally, because tumor responses were not anticipated after single agent treatment with BMS-986299, the study protocol allowed participants of part 1A who did not achieve PR or had insufficiently controlled disease to cross over to therapy with nivolumab and ipilimumab. A less frequent administration of BMS-986299 during C2 allowed better alignment with nivolumab and ipilimumab dosing schedule and helped reduce clinical site visits.

The enhanced infiltration of CD8+ T cells into the TME suggests that BMS-986299 induces a proinflammatory state. This conversion is crucial for the successful application of ICIs, as TILs are required for their effectiveness. This conversion into an inflamed TME was evident with BMS-986299 monotherapy administration, as it resulted in increases in CD8+, CD4+ TILs, and NK cells in the injected lesions. This influx of immune cells was not limited to the localized injection lesions, as the NI (“abscopal”) lesions also demonstrated increases in TILs, indicating that immune cells were potentially being recruited by inflammatory factors from the injected lesions. Increases in TAMs and PDL1+ expressing immune cells may reflect recruitment by IL-1β and could reflect recruitment of pro-inflammatory TAMs.22 Thus, NLRP3 inflammasome agonism can drive recruitment of immune cells to the injected lesion and promote tumor inflammation in distal lesions.

While not directly measured in the patient tumors, increases in serum cytokines may likely reflect changes in the TME. NLRP3 agonism is expected to induce IL-1β and IL-18 production, resulting in recruitment of macrophages, dendritic cells and NKs cells, that could be source for additional pro-inflammatory cytokines.23 At BMS-986299 monotherapy doses above 1200 µg, twofold increases in serum IL-1B were observed. The increases in serum IL-6 and G-CSF may also reflect the activation of these cell types, and the promotion of an inflamed TME. Increases in systemic BMS-986299 Cmax with increases in systemic inflammatory cytokines (G-CSF, IL-6, and IL-18) is consistent with the mechanism of action and suggests a causal inference of drug action.

Quantification of secreted cytokines in tumor biopsy samples remains technically challenging, particularly in FFPE tissues, where protein crosslinking and degradation can limit assay sensitivity. Despite efforts to develop LC-MS-based assays for detecting cleaved IL-1β and IL-18 to provide direct evidence of NLRP3 inflammasome activation, these attempts were unsuccessful due to technical limitations. While alternative approaches, such as multiplex IHC or spatial transcriptomics, could potentially address these challenges, we relied on systemic biomarkers for evidence of target engagement. The observed dose-proportional increases in serum IL-1β and IL-18, along with IL-1β-induced cytokines such as G-CSF and IL-6, provide strong indirect evidence of inflammasome activation, as their secretion is critically dependent on NLRP3 activity. Although we did not assess activated caspases or gasdermins, these serum-based findings offer robust support for inflammasome engagement and suggest a practical, clinically relevant strategy for monitoring therapeutic effects in future studies.

One of the significant