Combination of the STING Agonist MIW815 (ADU-S100) and PD-1 Inhibitor Spartalizumab in Advanced/Metastatic Solid Tumors or Lymphomas: An Open-Label, Multicenter, Phase Ib Study

Abstract

Purpose:

The stimulator of IFN genes (STING) is a transmembrane protein that plays a role in the immune response to tumors. Single-agent STING agonist MIW815 (ADU-S100) has demonstrated immune activation but limited antitumor activity. This phase Ib, multicenter, dose-escalation study assessed the safety and tolerability of MIW815 plus spartalizumab (PDR001), a humanized IgG4 antibody against PD-1, in 106 patients with advanced solid tumors or lymphomas.

Patients and Methods:

Patients were treated with weekly intratumoral injections of MIW815 (50–3,200 μg) on a 3-weeks-on/1-week-off schedule or once every 4 weeks, plus a fixed dose of spartalizumab (400 mg) intravenously every 4 weeks.

Results:

Common adverse events were pyrexia (n = 23; 22%), injection site pain (n = 21; 20%), and diarrhea (n = 12; 11%). Overall response rate was 10.4%. The MTD was not reached. Pharmacodynamic biomarker analysis demonstrated on-target activity.

Conclusions:

The combination of MIW815 and spartalizumab was well tolerated in patients with advanced/metastatic cancers, including in patients with anti-PD-1 refractory disease. Minimal antitumor responses were seen.

Introduction

The stimulator of IFN genes (STING; also known as TMEM173 or MITA), a transmembrane protein localized to the endoplasmic reticulum of macrophages, dendritic cells (DC), endothelial cells, and epithelial cells (1, 2), plays a crucial role in mediating the innate and adaptive immune response (3). The activation of the STING pathway via intracellular DNA in antigen-presenting cells leads to increased production of pro-inflammatory cytokines and IFNβ, as well as recruitment and priming of CD8⁺ T cells targeting tumor antigens (3, 4). Increased CD8⁺ T-cell activation and immunologic memory results in regression of established tumors and rejection of metastases (4).

Specifically, STING is activated following the sensing of intracellular DNA by cyclic-GMP-AMP-synthase (cGAS), which catalyzes the production of a cyclic dinucleotide (CDN) called cGAMP. This CDN directly binds and activates STING, leading to a downstream signaling cascade involving phosphorylation and activation of TANK-binding kinase (TBK1), IFN regulatory factor 3 (IRF3) phosphorylation and production of IFNβ, as well as cytokines and chemokines (5, 6).

Resistance to immune checkpoint blockade is linked to a lack of a preexisting antitumor immune response (7). STING agonists therefore have the potential to synergize with checkpoint inhibitors by initiating or augmenting the initial anticancer immune response. MIW815 (ADU-S100) is a novel synthetic CDN, which activates the STING pathway in vitro and in vivo (8). In murine tumor models, intratumoral injection of MIW815 resulted in tumor regression in both injected and noninjected lesions, and demonstrated synergistic antitumor activity and survival in combination with immune checkpoint inhibitors (4, 8). Thus, activation of the STING pathway has the potential to render refractory and relapsed tumors sensitive to immune checkpoint inhibition through antigen-presenting cell activation and IFNβ production (9, 10). The findings of the phase I study suggested that single-agent MIW815 may not be sufficient to overcome immune resistance mechanisms in humans (11). However, evidence of systemic immune activation was observed, supporting the rationale to combine STING agonism with PD-1 inhibition.

Spartalizumab (PDR001) is a high-affinity, ligand-blocking, humanized IgG4 antibody directed against PD-1, which blocks PD-L1 and PD-L2 binding (12, 13). Here we describe the safety, tolerability, pharmacokinetics, pharmacodynamics, and efficacy data from a phase Ib study of MIW815 (ADU-S100) in combination with spartalizumab in patients with advanced/metastatic solid tumors or lymphomas (NCT03172936). The findings from the dose-escalation part of this study are presented here.

Patients and Methods

Study oversight

This study (ClinicalTrials.gov: NCT03172936) was performed in accordance with the Declaration of Helsinki and the principles of Good Clinical Practice. The protocol was approved by the Human Ethics Review Committee at each site, and all patients provided written informed consent.

Study design

This study was a phase Ib, multicenter, open-label dose-escalation study of MIW815 (ADU-S100) in combination with spartalizumab in patients with advanced/metastatic solid tumors or lymphomas. The study consisted of a dose-escalation part to determine safety, tolerability, and the MTD and/or recommended dose for expansion (RDE) of intratumoral injection of MIW815 in combination with spartalizumab, as well as a dose expansion part after reaching MTD or RDE to better characterize safety, tolerability, and preliminary antitumor activity of the combination (Supplementary Fig. S1).

Two schedules of MIW815 intratumorally in cutaneous or subcutaneous-accessible lesions were explored in the dose-escalation part of the study; MIW815 was administered on Days 1, 8, and 15 in Group A, and on Day 1 in Group B with a fixed dose of spartalizumab 400 mg intravenously on Day 1 of each 28-day cycle. The Group B schedule was explored to determine whether comparable biological activity could be achieved with less-frequent MIW815 intratumoral dosing. One lesion was injected repeatedly. The starting dose of MIW815 was 50 μg (based on preclinical safety, tolerability data, and pharmacokinetic/pharmacodynamic data in the phase I study; ref. 11), with prespecified escalating doses of 100, 200, 400, 800, 1,600, and 3,200 μg in cohorts of 3 to 6 people. Group C was designed to explore intratumoral MIW815 to visceral metastatic lesions once MTD/RDE was determined in Group B, however this was not subsequently pursued. Approximately 40 patients were planned to be enrolled in the dose-escalation part of the study.

Patients received MIW815 intratumorally and spartalizumab intravenously until they experienced unacceptable toxicity or disease progression per RECIST v1.1 or immune-related response criteria (irRC) for solid tumors (14–16). Prior to each injection of MIW815, the longest diameter of the injected lesion was measured, and the injection volume was based on the size of that lesion (10–25 mm, 0.5–1.0 mL; >25–50 mm, 1.0–2.0 mL; and >50–<100 mm, 2.0–4.0 mL). In the event the injected subcutaneous or cutaneous lesion regressed to ≤1 cm and/or became too fibrotic for injection, intratumoral MIW815 was no longer administered at that site but other suitable sites could be selected for ongoing intratumoral MIW815 injections.

Outcomes

The primary objective of this study was to characterize the safety and tolerability of MIW815 (ADU-S100) given with spartalizumab, and to identify the recommended dose and schedule for future studies.

Secondary objectives included preliminary evaluation of antitumor activity of MIW815 in combination with spartalizumab including objective response rate (ORR), progression-free survival (PFS), duration of response (DOR), and disease control rate per RECIST v1.1 (14), irRC (15, 16), or Cheson 2014 criteria for lymphomas (17); characterization of the pharmacokinetic properties of MIW815 and spartalizumab; and assessment of the pharmacodynamic effects of MIW815 and spartalizumab combination in injected and uninjected distal lesions. Potential pharmacodynamic biomarkers of interest including modulation of PD-L1 expression, CD8, and cytokines were investigated as an exploratory objective.

Study participants

Eligible patients included adults with advanced/metastatic solid tumors or lymphomas with easily accessible cutaneous, subcutaneous, and/or nodal lesions either clinically or by ultrasound (US) or computed tomography (CT) guidance, who progressed on standard therapy, were intolerant to standard therapy, or for whom reasonably effective standard therapy was not available. All patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1 and were required to have measurable disease per RECIST v1.1 (14) or Cheson 2014 criteria for lymphomas (17) and at least two distinct lesions, both accessible for baseline and on-treatment biopsies, with patient consent. The injected lesion was required to measure ≥10 to <100 mm in longest diameter and be accessible for repeated intratumoral injections.

Patients were excluded if they required local palliative measures such as radiotherapy or surgery within 2 weeks of the study drug, had symptomatic or untreated leptomeningeal disease, had central nervous system (CNS) metastases or CNS metastases that required local CNS-directed therapy or increasing doses of corticosteroids, a history of severe hypersensitivity to any component of study drug, active or a known history of autoimmune disease, known history of human immunodeficiency virus, active Epstein–Barr Virus, active Hepatitis B or C, any active infection requiring systemic antibiotics, a history of drug-induced interstitial lung disease, or pneumonitis grade ≥2. Patients with prior discontinuation of anti-PD-1 therapy due to drug-related toxicity or previous discontinuation on therapy directed against STING were also excluded. Full inclusion and exclusion criteria are detailed in Supplementary Table S1.

Assessments

Tumor response was determined locally according to irRC (15, 16), and RECIST v1.1 (14) for solid tumors, and Cheson 2014 criteria (17) for lymphomas, at screening, and thereafter every 8 weeks until Week 40 and then every 12 weeks until disease progression or patient withdrawal. Response was also assessed at the end of treatment (EOT) if a scan was not conducted within 30 days prior to EOT.

Pre- and postinjection (±2 minutes) blood samples for MIW815 and spartalizumab pharmacokinetic analysis were collected at multiple timepoints on Cycle 1 Days 1, 8, and 15, and Cycle 3, Day 1. Plasma samples were assayed for MIW815 concentration, and serum samples were assayed for spartalizumab concentration, using validated LC/MS-MS assays with a lower limit of quantification (LLOQ) of 50 pg/mL, and LLOQ of 0.25 ng/mL, respectively. Blood samples were also collected for pharmacodynamic effects and tested for plasma cytokines (including GM-CSF, IFNβ, IFNγ, IL10, IL18, IL2, IL3, IL4, IL5, IL6, IL7, IL8, IP-10, MCP-1, MIP1α, MIP1β, TNFα, and TNFβ) and cell-based markers (including flow cytometric profiling of peripheral blood mononuclear cells) on Cycle 1, Day 1 prior to dose, 2 and 6 hours after doses, Day 8 and Day 15, and Cycle 3, Day 1.

Mandatory tumor biopsies from injected and noninjected lesions were collected at screening [or archival sample(s) if new tumor sample could not be safely collected at study entry] and on-treatment during Cycle 2, between Days 18 and 25. Paired baseline and on-treatment tumor biopsies were used for IHC staining of immune markers, including CD8, CD163, and PD-L1. CD8 and CD163 were quantified by % biomarker-positive area % per tumor region. Staining for PD-L1 was carried out using two separate antibodies (clone 22C3, Dako and clone SP142, VENTANA) and quantified by determining the percentage of positive tumor cells [tumor positive score (TPS)] as well as quantifying combined positive tumor cells and tumor-associated immune cells [combined positive score (CPS)]. In addition, in patients with triple-negative breast cancer (TNBC), the algorithm defined for the complementary diagnostic VENTANA PD-L1 (SP142) assay was used, quantifying only PD-L1+ tumor-infiltrating immune cells [p160002s009c.pdf (fda.gov)]. Patients with melanoma were scored using the melanoma (MEL) scale of 0 to 5 by assessing membranous staining in tumor and tumor-associated immune cells (18). Adverse events (AE) were assessed at every visit according to the Common Terminology Criteria for Adverse Events (CTCAE) v4.03. Safety assessments including AEs and laboratory values were closely monitored for all patients to identify any dose-limiting toxicities (DLT). A DLT was defined as a medically significant severe or life-threatening AE or abnormal laboratory value that occurred during the first treatment cycle and was suspected to be treatment related. Full criteria for defining dose-limiting toxicities are detailed in Supplementary Table S2.

Statistical analysis

In the dose-escalation part (groups A and B), a Bayesian Hierarchical Logistic Regression Model (BHLRM) with escalation with overdose control was used to make dose recommendations and estimate the MTD and RDE (19, 20). The BHLRM modeled the dose–toxicity relationship and used all available DLT information accumulated across all dose cohorts and regimens (e.g., groups) during the 28 days of the dose-limiting toxicity evaluation period. For data analysis, statistical analyses were carried out using the most updated SAS, R, WinBUGS, and JAGS versions in the statistical programming and modeling environment. Pharmacokinetic parameters were calculated using noncompartmental methods, using the pharmacokinetic profiles of MIW815 and spartalizumab available in Phoenix WinNonlin version 8.0. As no patients were enrolled in Group C or for dose expansion, no analysis was performed for those parts. No formal statistical power calculations were used to determine sample size for this study.

Data availability

The data generated in this study are available upon request from the corresponding author. Novartis will not provide access to patient-level data, if there is a reasonable likelihood that individual patients could be re-identified. Phase I studies, by their nature, present a high risk of patient re-identification; therefore, patient individual results for phase I studies cannot be shared. In addition, clinical data, in some cases, have been collected subject to contractual or consent provisions that prohibit transfer to third parties. Such restrictions may preclude granting access under these provisions. Where co-development agreements or other legal restrictions prevent companies from sharing particular data, companies will work with qualified requestors to provide summary information where possible.

Results

Patient population, treatment, and disposition

This phase Ib dose-escalation study took place at 12 sites across eight countries (Australia, Canada, Germany, Japan, Spain, Switzerland, Netherlands, and United States of America). At the analysis cutoff date (March 2, 2021), 106 patients were enrolled across seven cohorts (50 μg–3,200 μg) and two treatment schedules [Group A, n = 67 (Day 1, 8, and 15 every 28 days); Group B, n = 39 (Day 1 every 28 days); Supplementary Fig. S1].

Baseline patient demographics and disease characteristics are shown in Table 1. Most patients enrolled had solid tumors (n = 64, 96% in Group A; n = 38, 97% in Group B), with the remainder having lymphomas (n = 3, 3% in Group A; n = 1, 3% in Group B). Most patients with solid tumors had stage IV disease at study entry (n = 49, 77% in Group A; n = 30, 79% in Group B). The most common solid-tumor histologies were melanoma (n = 38, 35.8% of all patients) and TNBC (n = 23, 21.7% of all patients). Most patients had received prior anti-neoplastic therapies (n = 64, 95.5%) in Group A, of which 35 (52.2%) received ≥4 prior regimens; 38 (97.4%) patients in Group B had received prior antineoplastic therapies, of which 21 (53.8%) received ≥3 regimens. Sixty-nine patients (65.1%) received prior treatment with immunotherapies (n = 38, 56.7% in Group A; n = 31, 79.5% in Group B), with a mean duration of treatment of 154.8 days ± Standard Deviation 174.1 days (range 1–1,145 days).

Table 1.

Baseline patient demographics and disease characteristics.

Demographic variable/characteristic	Group A N = 67	Group B N = 39	All patients N = 106
Age (years)
Mean (Standard Deviation)	58.7	61.8	59.8
Median	60.0	64.0	61.0
Range	27–93	27–83	27–93
Sex, n (%)
Male	27 (40.3)	23 (59.0)	50 (47.2)
Female	40 (59.7)	16 (41.0)	56 (52.8)
Race, n (%)
Caucasian	48 (71.6)	35 (89.7)	83 (78.3)
Asian	13 (19.4)	3 (7.7)	16 (15.1)
Black	3 (4.5)	1 (2.6)	4 (3.8)
Other	3 (4.5)	0	3 (2.8)
ECOG performance status, n (%)
0	18 (26.9)	9 (23.1)	27 (25.5)
1	49 (73.1)	30 (76.9)	79 (74.5)
Primary tumor type, n (%)
Melanoma	24 (35.8)	14 (35.9)	38 (35.8)
Cutaneous melanoma	10 (14.9)	7 (17.9)	17 (16)
Melanoma (unknown)	12 (17.9)	5 (12.8)	17 (16)
Uveal melanoma	1 (1.5)	2 (5.1)	3 (2.8)
Mucosal melanoma	1 (1.5)	0	1 (0.9)
Breast cancer	24 (35.8)	4 (10.3)	28 (26.4)
Triple-negative breast cancer	19 (28.4)	4 (10.3)	23 (21.7)
Breast cancer	5 (7.5)	0	5 (4.7)
Othera	19 (28.4)	21 (53.8)	40 (37.7)
Colorectal cancer	2 (3.0)	3 (7.7)	5 (4.7)
Head and neck cancer	2 (3.0)	3 (7.7)	5 (4.7)
Non-Hodgkin lymphoma	2 (3.0)	1 (2.6)	3 (2.8)
Merkel cell carcinoma	0	3 (7.7)	3 (2.8)
Mesothelioma	2 (3.0)	0	2 (1.9)
Ovarian cancer	0	2 (5.1)	2 (1.9)
Prostate cancer	1 (1.5)	1 (2.6)	2 (1.9)
Sarcoma	2 (3)	0	2 (1.9)
Squamous cell carcinoma of skin	1 (1.5)	1 (2.6)	2 (1.9)
Prior immunotherapies, n (%)	38 (56.7)	31 (79.5)	69 (65.1)
Anti-PD-1	33 (49.3)	27 (69.2)	60 (56.6)
Anti-CTLA-4	16 (23.9)	14 (35.9)	30 (28.3)
Anti-CD20	2 (3.0)	1 (2.6)	3 (2.8)
Anti-PD-L1	1 (1.5)	5 (12.8)	6 (5.7)
Interferon	3 (4.5)	2 (5.1)	5 (4.7)
Interleukins	1 (1.5)	2 (5.1)	3 (2.8)
Anti-IDO	2 (3.0)	1 (2.6)	3 (2.8)
Adenosine receptor	0 (0)	1 (2.6)	1 (0.9)
Cancer vaccine	0 (0)	1 (2.6)	1 (0.9)
Otherb	12 (17.9)	4 (10.3)	16 (15.1)
Prior antineoplastic therapies, n (%)
No	3 (4.5)	1 (2.6)	4 (3.8)
Yes	64 (95.5)	38 (97.4)	102 (96.2)
Number of prior antineoplastic therapies, n (%)
1	9 (13.4)	3 (7.7)	12 (11.3)
2	10 (14.9)	14 (35.9)	24 (22.6)
≥3	45 (67.2)	21 (53.8)	66 (62.0)

Abbreviations: CTLA-4, cytotoxic T-lymphocyte–associated. protein 4; ECOG, Eastern Cooperative Oncology Group; IDO, indoleamine 2,3-dioxygenase; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1.

^aAdditional diagnoses occurring in only 1 patient not shown in the table: adenoid cystic carcinoma, cancer of thymus, cholangiocarcinoma, classical sporadic Kaposi’s sarcoma, clear cell sarcoma, esophageal cancer, gastric cancer, gingival cancer, Hodgkin lymphoma, malignant fibrous tumor, mucinous adenocarcinoma, mucoepidermoid carcinoma of submandibular gland, squamous cell carcinoma of the anal canal, superficial urothelial carcinoma.

^bOther drugs included talimogene laherparepvec (TVEC; 8 patients), ibrutinib (1 patient), utomilumab (1 patient), and investigational, non-approved WHO drugs (4 patients).

Median duration of exposure to MIW815 was 6.1 weeks in Group A (range, 0.1–72.3 weeks) and 4.4 weeks in Group B (range, 0.1–50.9 weeks). For spartalizumab, the median duration of exposure was 8.0 weeks for Group A (range, 0.7–119.1 weeks) and 11.9 weeks for Group B (range, 3.3–61.0 weeks).

At the time of data cut-off, all patients had discontinued the study treatment. The most common reason for discontinuation was progressive disease (n = 43, 64.2% in Group A; n = 29, 74.4% in Group B). Other reasons for discontinuation are shown in Supplementary Table S3. Two patients discontinued study treatment due to AEs: 1 patient in Group A after developing grade 3 pneumonitis, and 1 patient in Group B due to grade 2 bullous dermatitis. Two patients discontinued treatment due to death: 1 patient in Group A (death caused by grade 5 failure to thrive) and 1 in Group B (due to underlying cancer).

Safety and tolerability

No DLTs were observed up to 1,600 μg MIW815 + spartalizumab. One of the 6 patients in the 3,200 μg MIW815 + spartalizumab cohort from Group A experienced a DLT (grade 3 injection site reaction) after the third weekly dose in Cycle 1, which resolved. This patient discontinued treatment due to progressive disease. The MTD was not reached for MIW815 in combination with spartalizumab.

Nearly all patients (n = 103; 97.2%) reported at least one AE. The most frequently reported AEs regardless of grade and treatment relationship were pyrexia (n = 30; 28.3%), anemia (n = 29; 27.4%), diarrhea (n = 21; 19.8%), injection site pain (n = 21; 19.8%), fatigue (n = 19; 17.9%), dyspnea (n = 18; 17.0%), and constipation (n = 17; 16.0%; Supplementary Table S4). Grade 3/4 AEs regardless of study drug relationship were reported in 48.1% of patients, with anemia (n = 9; 8.5%) and dyspnea (n = 6; 5.7%) as the most frequently reported (≥5% of patients; Supplementary Table S4).

Approximately two thirds of patients (n = 70, 66.0%) reported at least one AE attributed to be related to study treatment (Table 2). The most common treatment-related AEs were pyrexia (n = 23, 21.7%), injection site pain (n = 21, 19.8%), diarrhea (n = 12, 11.3%), and fatigue (n = 9, 8.5%). Grade 3/4 drug-related AEs were reported in 14 patients (13.2%) including lipase elevation (n = 3; 2.8%), increased alanine aminotransferase (n = 2, 1.9%), diarrhea (n = 2, 1.9%), pyrexia (n = 1, 0.9%), injection site reaction (n = 1, 0.9%), fatigue (n = 1, 0.9%), increased amylase (n = 1, 0.9%), and increased aspartate aminotransferase (n = 1, 0.9%). Most treatment-related AEs were self-limited or manageable with the use of concomitant medication and/or with dose adjustment/interruption

Table 2.

Suspected treatment-related AEs (>2% of patients), for weekly (Group A) and monthly (Group B) treatment schedules.

	Group A N = 67, n (%)		Group B N = 39, n (%)		All patients N = 106, n (%)
N (%)	All grades	Grades ≥3	All grades	Grades ≥3	All grades	Grades ≥3
Total	45 (67.2)	10 (14.9)	25 (64.1)	4 (10.3)	70 (66.0)	14 (13.2)
Pyrexia	17 (25.4)	1 (1.5)	6 (15.4)	0	23 (21.7)	1 (0.9)
Injection site pain	17 (25.4)	0	4 (10.3)	0	21 (19.8)	0
Diarrhea	7 (10.4)	2 (3.0)	5 (12.8)	0	12 (11.3)	2 (1.9)
Fatigue	5 (7.5)	1 (1.5)	4 (10.3)	0	9 (8.5)	1 (0.9)
Rash	5 (7.5)	0	3 (7.7)	0	8 (7.5)	0
Chills	4 (6.0)	0	3 (7.7)	0	7 (6.6)	0
Anemia	6 (9.0)	0	0	0	6 (5.7)	0
Lipase increased	4 (6.0)	2 (3.0)	2 (5.1)	1 (2.6)	6 (5.7)	3 (2.8)
Pruritus	5 (7.5)	0	1 (2.6)	0	6 (5.7)	0
Amylase increased	3 (4.5)	1 (1.5)	2 (5.1)	0	5 (4.7)	1 (0.9)
Alanine aminotransferase increased	3 (4.5)	1 (1.5)	1 (2.6)	1 (2.6)	4 (3.8)	2 (1.9)
Aspartate aminotransferase increased	2 (3.0)	1 (1.5)	2 (5.1)	0	4 (3.8)	1 (0.9)
Injection site reaction	4 (6.0)	1 (1.5)	0	0	4 (3.8)	1 (0.9)
Vomiting	3 (4.5)	0	1 (2.6)	0	4 (3.8)	0
Decreased appetite	3 (4.5)	0	0	0	3 (2.8)	0
Dry mouth	1 (1.5)	0	2 (5.1)	0	3 (2.8)	0
Headache	3 (4.5)	0	0	0	3 (2.8)	0
Nausea	3 (4.5)	0	0	0	3 (2.8)	0

Note: A patient with multiple occurrences of an AE under one treatment was counted only once in the AE category for that treatment.

A patient with multiple AEs was counted only once in the total row.

Serious AEs (SAE) regardless of study drug relationship were reported in 34.9% (n = 37) of patients. The most frequently reported SAEs (>2%) included dyspnea (n = 5, 4.7%), and anemia, diarrhea, fatigue, and pyrexia (all n = 3 patients, 2.8%). Grade 3/4 SAEs were reported in 28.3% of patients (n = 30), with dyspnea (n = 5, 4.7%) and fatigue (n = 3, 2.8%) the most frequent. Treatment-related SAEs were reported in 9 patients (8.5%), and included pyrexia (n = 3), diarrhea (n = 2), partial seizures (n = 1), hyperthyroidism (n = 1), anemia (n = 1), increased amylase (n = 1), increased lipase (n =1), dyspnea (n =1), fatigue (n = 1), and pneumonitis (n = 1).

No deaths were suspected to be related to either MIW815 or spartalizumab treatment. On-treatment grade 5 events (occurring between treatment initiation and within 30 days of treatment discontinuation) were reported in 13 patients (12%); 9 were attributed to disease progression and 4 deaths were caused by pulmonary embolism, respiratory failure, right ventricular dysfunction, or failure to thrive.

Efficacy

The duration of treatment with MIW815 and spartalizumab with individual patient responses are shown in Supplementary Fig. S2. Overall responses for treatment schedules are shown in Table 3. Across all dose levels, overall, an objective response [confirmed complete response (CR) or partial response (PR)] was observed in 11 patients [10.4%; 90% confidence interval (CI), 5.9–16.6] including 9 patients in Group A (ORR, 13.4%; 90% CI, 7.2–22.3) and 2 patients in Group B (ORR, 5.1%; 90% CI, 0.9–15.3). Ten patients experienced a confirmed PR (Group A, 8 patients and Group B, 2 patients), and 1 additional patient in Group A achieved a CR per RECIST v1.1 after five cycles; this patient had TNBC, was naïve to prior anti-PD-1 therapy, and was treated with 200 μg MIW815 and spartalizumab. The ORR was 5/69 (7.2%) for patients pretreated with immunotherapy, versus 6/37 (16.2%) among immunotherapy-naïve patients. Subgroup analysis of ORR and disease control rate by prior immunotherapy exposure is shown in Supplementary Table S5. The estimated median DOR for the combined treatment groups was 11.5 months [95% CI, 4.7–not applicable (NA)]. The disease control rate, defined as the proportion of patients experiencing a response or stable disease, was 29.2% (90% CI, 22.0–37.4).

Table 3.

Best overall response in patients receiving weekly (Group A) and monthly (Group B) treatment schedule.

	Group A N = 67	Group B N = 39	All patients N = 106
Best overall response, n (%)
Complete response	1 (1.5)	0	1 (0.9)
Partial response	8 (11.9)	2 (5.1)	10 (9.4)
Stable disease	11 (16.4)	9 (23.1)	20 (18.9)
Progressive disease	34 (50.7)	22 (56.4)	56 (52.8)
Non-CR/Non-PD	1 (1.5)	0	1 (0.9)
Unknown	12 (17.9)	6 (15.4)	18 (17.0)
Overall response rate (CR or PR), n (%)	9 (13.4)	2 (5.1)	11 (10.4)
90% CI	7.2–22.3	0.9–15.3	5.9–16.6
Disease control rate (CR, PR, or SD), n (%)	20 (29.9)	11 (28.2)	31 (29.2)
90% CI	20.7–40.4	16.7–42.3	22.0–37.4

Abbreviations: CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease.

The best percentage change in tumor measurement from baseline in target lesions is shown in Fig. 1A and B. The best percentage change from baseline in injected versus sum of target lesions is shown in Fig. 1C and D. Lesion size was stable or decreased in most evaluable injected lesions. Best percentage change from baseline in target lesions in patients with TNBC and melanoma specifically on both treatment schedules is shown in Fig. 2A and B.

Figure 1.

Best percentage change from baseline in target lesions in weekly (A) and monthly (B) treatment schedule, and in injected vs. sum of target lesions in patients treated weekly (C) and monthly (D) with MIW815 and spartalizumab per RECIST v1.1 or Cheson 2014 criteria, according to MIW815 dose level. Note: One patient diagnosed with melanoma did not have any target lesions per RECIST v1.1 at baseline, so was not evaluable for best % change, but the patient had a confirmed PR based on nontarget lesion response.

Figure 2.

Best percentage change from baseline in target lesion with corresponding CD8 and PD-L1 expression by MIW815 dose level in patients with TNBC (A) and melanoma (B). Note: PD-L1% positive immune cells (A) were assessed using the SP142 assay; PD-L1 Mel Score (B) was measured using the Dako 22C3 assay. In B, 1 patient diagnosed with melanoma did not have any target lesions per RECIST v1.1 at baseline, so was not evaluable for best % change, but the patient had a confirmed PR based on nontarget lesion response.

Among 23 patients with TNBC, 4 (17.3%) achieved an objective response. All 4 of these patients were naïve to prior anti-PD-1 therapy, and all 4 were PD-L1 positive in the injected lesion (% positive immune cells ≥1% as assessed with SP142; Fig. 2). Among 38 patients with melanoma who had failed prior anti-PD-1 therapy, 3 (7.8%) achieved an objective response. All three patients expressed PD-L1 in the injected lesion at baseline (PD-L1 Mel Scores of 5, 5, and 2).

Median PFS was similar for both treatment groups. For Group A, the median PFS was 1.9 months (90% CI, 1.8–2.0) and the Kaplan–Meier estimated PFS rates at 6 and 12 months were 17.5% (90% CI, 9.2–25.8) and 11.7% (90% CI, 4.5–18.9), respectively. For Group B, the median PFS was 1.9 months (90% CI, 1.8–2.0) and the Kaplan–Meier estimated PFS rates at 6 and 12 months were 23.8% (90% CI, 11.7–35.8) and 8.1% (90% CI, NA–16.8), respectively.

Pharmacokinetics

MIW815 was rapidly absorbed from the tumor injection site into the plasma, with the maximum observed plasma concentration (C_max) coinciding with the end of injection, resulting in a short median time to peak plasma concentrations (T_max) for both treatment schedules (Supplementary Tables S6 and S7). The mean steady state plasma exposure increased dose-proportionally for doses between 50 and 3,200 μg, with no plasma accumulation after weekly dosing observed. Pharmacokinetic variability with both treatment schedules showed a coefficient of variation (CV) >86%, indicating a moderate to high interindividual variability. The concentration of MIW815 declined following a biphasic elimination, with the terminal phase observed after 2 hours postinjection. Plasma concentration–time profiles demonstrated rapid elimination (Fig. 3A).

Figure 3.

Plasma concentration. A, MIW815 plasma concentration–time profile according to MIW815 dose level. B, Change in IFNβ plasma concentration from pre-MIW815 dose to 6-hour post-dose vs. MIW815 AUC_last. Note: Combined 6-hour post-dose change from pre-dose for Cycle 1 Day 1, Cycle 1 Day 8, Cycle 1 Day 15, and Cycle 3 Day 1 are shown. AUC_last, AUC from time zero to time of last measurable concentration; CR, complete response; NCR/NPD, no complete response/no progressive disease; PD, progressive disease; PR, partial response; SD, stable disease; UNK, unknown.

MIW815 pharmacokinetic parameters for Group A (weekly) and Group B (monthly) are summarized in Supplementary Tables S6 and S7. Overall, MIW815 pharmacokinetic profiles were similar to those observed in the phase I dose-escalation study (11). Spartalizumab pharmacokinetic parameters for Group A and Group B are summarized in Supplementary Tables S8 and S9. No impact of MIW815 on spartalizumab pharmacokinetic parameters was observed.

Pharmacodynamics

Consistent correlations were observed between intra-tumoral dosing and serum levels of MIW815. Plasma levels of cytokines, including IFNβ, MIP1β, IL6, and MCP-1, were increased at 6 hours postinjection compared with pre-dose levels, consistent with evidence of immune activation (Supplementary Fig. S3). Plasma levels of IFNβ showed a dose-dependent increase with increasing MIW815 serum concentration, suggesting innate immune activation and STING agonism (Fig. 3B) consistent with biomarker findings observed with single-agent MIW815 (11). However, we cannot rule out the contribution of PD-1 blockade in this response.

Tumor biopsies from injected and noninjected lesions were collected at screening (fresh or archival samples) and on-treatment during Cycle 2, between Days 18 and 25. Supplementary Table S10 shows the numbers of samples scored for PD-L1 by biopsy location, timepoint, and scoring method. At baseline, all 106 patients had both injected and noninjected lesions biopsied. However, not all biopsies yielded sufficient tumor tissue for IHC analysis. At Cycle 2, Day 15, paired biopsies were taken from injected and noninjected lesions for 59 and 61 patients, respectively. At baseline, 49.4% (39/79) of evaluable patients had PD-L1–positive tumor cells in injected lesions (≥1% tumor positive staining by IHC), and 41.9% (31/74) in noninjected lesions. At baseline, 36.3% of evaluable patients (33/91) had CD8-positive cells (percent marker area, PMA) in injected lesions and 36.4% (32/88) in noninjected lesions. There was no evidence of consistent increase in PD-L1 expression on treatment, in injected or noninjected lesions (Supplementary Table S11). However, this is likely confounded by limited tissue material during on-treatment biopsies due to tissue necrosis and observed tumor responses.

Thirty-eight patients had both paired screening and on-treatment biopsies from injected and noninjected lesions with evaluable CD8 IHC. Of the 11 patients who achieved a response (CR/PR), 5 had paired biopsies and all showed an increase in CD8. Figure 2 highlights paired biopsy data in patients with TNBC (n = 2) and melanoma (n = 2; data from the fifth patient with Merkel cell carcinoma not shown). PD-L1 percent positive tumor (using the 22C3 assay) increased in 15 of 41 paired injected lesions and in 10 of 38 paired noninjected lesions. All 11 responding patients demonstrated positive PD-L1 expression in the injected lesion at baseline. Best overall response by PD-L1 in the injected lesion by PD-L1 assay and scoring method can be observed in Table 4.

Table 4.

Best overall response by PD-L1 expression in the injected lesion at baseline by PD-L1 assay and scoring method.

	DAKO CPS			DAKO Mel Score			DAKO TPS			VENTANA SP142
Best overall response	<10	≥10	UNK	<1	≥1	UNK	<1%	≥1%	UNK	<1%	≥1%	UNK
All tumor types	N = 38	N = 30	N = 48	N = 14	N = 22	N = 70	N = 40	N = 39	N = 27	N = 14	N = 10	N = 82
CR	0	0	1 (2.1)	0	0	1 (1.4)	0	0	1 (3.7)	0	1 (10.0)	0
PR	1 (3.6)	6 (20.0)	3 (6.2)	1 (7.1)	3 (13.6)	6 (8.6)	1 (2.5)	8 (20.5)	1 (3.7)	0	3 (30.0)	7 (8.5)
SD	5 (17.9)	6 (20.0)	9 (18.8)	1 (7.1)	8 (36.4)	11 (15.7)	7 (17.5)	8 (20.5)	5 (18.5)	2 (14.3)	0	18 (22.0)
NCR/NPD	0	0	1 (2.1)	0	0	1 (1.4)	0	1 (2.6)	0	0	0	1 (1.2)
PD	15 (53.6)	15 (50.0)	26 (54.2)	11 (78.6)	9 (40.9)	36 (51.4)	21 (52.5)	19 (48.7)	16 (59.3)	7 (50.0)	4 (40.0)	45 (54.9)
UNK	7 (25.0)	3 (10.0)	8 (16.7)	1 (7.1)	2 (9.1)	15 (21.4)	11 (27.5)	3 (7.7)	4 (14.8)	5 (35.7)	2 (20.0)	11 (13.4)
TNBC	N = 10	N = 8	N = 5	N = 0	N = 0	N = 23	N = 11	N = 8	N = 4	N = 10	N = 10	N = 3
CR	0	0	1 (20.0)	0	0	1 (4.3)	0	0	1 (25.0)	0	1 (10.0)	0
PR	0	3 (37.5)	0	0	0	3 (13.0)	1 (9.1)	2 (25.0)	0	0	3 (30.0)	0
SD	1 (10.0)	0	0	0	0	1 (4.3)	0	1 (12.5)	0	1 (10.0)	0	0
PD	6 (60.0)	2 (25.0)	4 (80.0)	0	0	12 (52.2)	5 (45.5)	4 (50.0)	3 (75.0)	5 (50.0)	4 (40.0)	3 (100.0)
UNK	3 (30.0)	3 (37.5)	0	0	0	6 (26.1)	5 (45.5)	1 (12.5)	0	4 (40.0)	2 (20.0)	0
Melanoma	N = 14	N = 19	N = 5	N = 12	N = 21	N = 5	N = 6	N = 14	N = 18	N = 0	N = 0	N = 38
PR	1 (7.1)	2 (10.5)	0	0	3 (14.3)	0	0	3 (21.4)	0	0	0	3 (7.9)
SD	3 (21.4)	6 (31.6)	3 (60.0)	1 (8.3)	8 (38.1)	3 (60.0)	2 (33.3)	6 (42.9)	4 (22.2)	0	0	12 (31.6)
PD	8 (57.1)	11 (57.9)	1 (20.0)	10 (83.3)	9 (42.9)	1 (20.0)	4 (66.7)	5 (35.7)	11 (61.1)	0	0	20 (52.6)
UNK	2 (14.3)	0	1 (20.0)	1 (8.3)	1 (4.8)	1 (20.0)	0	0	3 (16.7)	0	0	3 (7.9)

Note: Data reported as n (%).

Abbreviations: CPS, combined positive score; CR, complete response; Mel, melanoma; NCR/NPD, non-complete response or non-progressive disease; PD, progressive disease; PD-L1, programmed death ligand-1; PR, partial response; SD stable disease; TPS, tumor positive score; UNK, unknown.

During Cycle 2, very little tumor material from injected lesion biopsies could be collected, despite no reported differences in cellularity between injected and noninjected lesions. Remaining stroma and any potential tumor remnants were highly inflamed, with large numbers of CD8-positive cells and high inflammatory and stromal cell PD-L1 expression. An example of a responder is illustrated in Supplementary Fig. S4. Expression levels of CD8 and CD163 in the noninjected lesion were broadly similar to those of the injected lesion, suggesting a systemic immune response. Unlike the injected lesion, limited PD-L1 expression was observed in noninjected lesions, which correlated with macrophage and CD8-positive cell accumulation.

Discussion

There is emerging interest in intratumoral therapy to address local barriers in tumor immunity and trigger adequate local and systemic immunologic responses. Talimogene laherparepvec (T-VEC), the first oncolytic viral immunotherapy, has been approved for the local treatment of unresectable metastatic melanoma (21). Various intratumoral therapies are in development stages for melanoma, sarcoma, and prostate, breast, pancreatic, head and neck, and rectal cancers (22). STING agonists trigger innate immune activation, leading to an enhanced adaptive immune response to control tumor growth alone or in combination with immune checkpoint blockade (23). This is the first published study of an intratumorally administered STING agonist, MIW815 (ADU-S100), in combination with an anti-PD-1 antibody, spartalizumab.

This dose-escalation study demonstrated MIW815 in combination with spartalizumab had a favorable safety profile and was well tolerated in patients with advanced solid tumors and lymphomas. The most common treatment-related AEs included pyrexia, injection site pain, diarrhea, and fatigue, which are broadly consistent with the safety profile of the phase I single-agent MIW815 study as well as with other type 1 IFN-inducing agonists (11, 24, 25). The MTD was not reached and the RDE was not determined. The study was terminated early due to minimal efficacy, without safety concerns. As a result, the dose-expansion part of the study and Group C were not initiated.

Preliminary clinical activity was observed in this study with an overall ORR of 10.4%. The greater number of observed responses in the weekly schedule versus monthly schedule suggests that observed clinical activity is a result of the combinational activity of MIW815 and spartalizumab, although it should be noted that this schedule was also enriched for patients with melanoma and TNBC. Size reduction of noninjected tumor lesions could be attributable to anti-PD-1 treatment, MIW815 abscopal effects, or systemic MIW815 exposure. Despite the heterogenous patient population within the study, most patients were at advanced stages of disease and had been heavily pretreated, including having previously progressed on anti-PD-1–based treatment. The overall response rate of 13.4% with weekly MIW815 intratumoral dosing in combination with spartalizumab, was numerically higher than that observed in the phase I MIW815 monotherapy trial (2.1%; ref. 11).The efficacy was intriguing but insufficient to support advancing to the dose confirmation and dose-expansion parts of the study.

Pharmacokinetic analyses in this study were very similar to those of the phase I study (11), suggesting there was no drug–drug interaction between spartalizumab and MIW815. Furthermore, no differences between Cycles 1 and 3 were observed, indicating there was no MIW815 accumulation. Local drug concentration levels within the injected lesion are expected to be an important factor in mediating a therapeutic effect and are key to driving efficacy. Considering the intratumoral administration, the relevance of systemic plasma concentrations to tumor levels of MIW815 remains unclear, and thus an interpretation of therapeutically relevant drug levels is difficult. In particular, the rapid drop in plasma concentrations, especially in Group B, does not necessarily translate into lack of activity within the tumor.

Increased expression of pro-inflammatory cytokines including IFNβ demonstrated biological activity in response to MIW815 and spartalizumab treatment. This could be partly due to MIW815 increasing IFNβ expression within the tumor (as previously shown by gene expression data; ref. 11) and resulting in increased IFNβ plasma levels; however, increased levels of MIW815 in the plasma could also have induced systemic IFNβ expression. A link between type 1 IFNs and T-cell responses indicates a role for these cytokines in the generation of the adaptive T-cell response (26). Clinical benefit of PD-1/PD-L1 blockade correlates with T-cell activation prior to treatment, with a marked expansion of tumor-infiltrating CD8-positive T cells also observed following anti-PD-1 treatment (27, 28). MIW815 plasma concentration correlated with plasma IFNβ levels, indicating target engagement and is consistent with pharmacokinetic/pharmacodynamic data obtained in the MIW815 single-agent phase I study (11). However, an additional combined pharmacodynamic effect of MIW815 and spartalizumab cannot be excluded. Considering the two potential antitumor modalities of MIW815 (10) and a pharmacodynamic effect across a range of MIW815 doses, the collective biomarker data did not support the selection of one optimal biologically active dose.

Intratumoral CD8 infiltration from biopsies was likely confounded by limited material during second biopsies. We note that in preclinical models, MIW815 demonstrated two potential antitumor modalities (8, 29). At higher concentrations of injected compound, TNFα-dependent direct tumor cell killing was observed (29). This was separate from the stimulation of the adaptive immune response leading to abscopal tumor killing seen at lower doses (7). An open question is whether we observed this direct tumor cell killing in the injected lesions or immune cell–mediated killing, leading to decreased numbers of tumor cells. We also note that CD8 tumor-infiltrating lymphocytes are seen with PD-1 blockade, as previously shown (30). Therefore, the individual contribution of each compound is hard to assess.

In two responders in the TNBC group, CD8 expression levels increased on treatment; however, the small number of patients is insufficient to draw conclusions. Those with CR/PR all had >1% positive PD-L1 expression observed in both injected and noninjected lesions, making it difficult to distinguish the individual contributions from MIW815 versus PD-1 blockade. The time point for on-treatment biopsy was +14 days post-dose, making it difficult to assess an acute pharmacodynamic readout for STING agonism alone. Activation of the STING pathway leads to abiphasic response inwhich early changes in gene expression of cytokines lead to later changes in immune cell activation and potential antitumor effects. However, cytokine increases in blood samples collected soon after MIW815 administration confirm the on-mechanism activity of MIW815.

Several aspects should be considered to further understand the mechanisms of response to STING agonists in combination with checkpoint blockade, including: the impact of the primary tumor microenvironment immunobiology on immunomodulatory and clinical outcomes, optimal dose and schedule to elicit a productive local response and avoid overstimulation and negative regulation, and identification of biomarkers predictive of clinical activity. There is a strong rationale for activation of the STING pathway combined with checkpoint inhibition, and a number of other STING agonists are in various stages of development including intratumorally administered STING agonists in combination with a checkpoint inhibitor (31), systemically delivered STING agonist in combination with an anti-PD-1/PD-L1 agent (32) and an orally bioavailable STING agonist (33). A STING antibody–drug conjugate to improve tumor targeting is also being investigated (34).

Our study has some limitations. Intratumoral administration faces potential barriers to therapeutic efficacy, for example, increased intratumoral pressure, restraints on infusion of injected agents, and challenges with injecting tumors in patients with lesions that are not readily accessible for tumor injections, such as visceral lesions. High interstitial fluid pressure in tumors, which is commonly observed in solid tumors (35) and was reported by some clinicians in this study, can make injection more difficult. The more limited on-treatment tumor material collection from injected sites due to necrosis and responses in injected lesions sometimes meant that PD-L1 and CD8 expression had to be assessed in noninjected lesions, making it difficult to compare the differences between injected and matched noninjected tumor lesions. Furthermore, plasma exposure data only provides information on systemic exposure of MIW815, rather than intratumoral levels. Although unable to measure intratumoral concentration of MIW815, rapid clearance in plasma may reflect a similarly rapid clearance within the injected tumors and a drug formulation that would allow for longer tumoral retention could be considered. Spartalizumab is not approved for the treatment of solid tumors but has been tested in several oncology trials, notably in metastatic nasopharyngeal (36) and anaplastic thyroid cancers (37). Overall, these studies show that spartalizumab has similar activity to other PD-1 blocking antibodies in terms of treatment-related AEs, treatment benefit in the context of PD-L1 expression, and IFNγ pathway activation.

To conclude, we demonstrate that the intratumoral STING agonist, MIW815 (ADU-S100), can be safely combined with anti-PD-1 therapy and results in minimal efficacy in patients with highly refractory cancers. Results of our study are not supportive of further advancement to phase II/III studies. Pharmacologic optimization of intratumoral therapeutics to improve delivery within the tumor microenvironment will be needed to improve systemic efficacy. Determination of the relevant functional pathways in distinct patient populations will be critical for optimization of innate immune agonist combinations with existing immunotherapies (23).

Translational Relevance.

MIW815 is a small molecule agonist of stimulator of IFN genes (STING). Intratumoral administration of this molecule led to tumor regression in murine models, in combination with PD-1 blockade. Clinical testing of this combination had minimal efficacy and highlights the need for stronger immune activation of the tumor immune compartment.