Preclinical evaluation of an anchored immunotherapy strategy with aluminum hydroxide-tethered interleukin-12 in dogs with advanced malignant melanoma

Abstract

Melanoma is an aggressive cancer in dogs involving skin and mucosa similar to people. Anchored immunotherapeutics offer a novel approach to increase intratumoral retention of therapeutic payloads while decreasing systemic exposure, and this strategy can be critically evaluated through a comparative oncology approach. JEN-101 is an anchored canine interleukin-12 (IL-12) tethered to aluminum hydroxide administered by local injection. A Phase I study was conducted to determine the tolerability, activity, and immune responses of JEN-101 in dogs with advanced melanoma.

A 3+3 dose escalation design was used to evaluate intratumoral injection of JEN-101 at 1, 3, 10, or 20 μg/kg every three weeks for four cycles. A second course was allowable in the absence of disease progression or toxicity. Peripheral blood, serum, and tumor biopsies were collected at baseline and at pre-specified timepoints for pharmacokinetic and immune analyses, which included serum cytokines, immunohistochemistry, and gene expression assessment.

JEN-101 was well tolerated with adverse events being fever, lethargy, and isolated elevated liver enzymes. Five dogs experienced grade 3 events and no grade 4 events were observed. Pharmacokinetic analysis showed a trend towards dose-related C_max within 8 hours of injection. Responding dogs demonstrated increased systemic interferon-γ and IL-10 AUC levels and local recruitment of CD3⁺ T cells. Increased pro-inflammatory and antigen processing gene expressions were identified in responding lesions.

JEN-101 was well tolerated with evidence of biologic and therapeutic activities. Anchored IL-12 immunotherapy merits further investigation in dogs with melanoma and our approach represents an immune competent model to inform human clinical trials.

INTRODUCTION

Interleukin-12 (IL-12) is a potent pro-inflammatory cytokine with remarkable anticancer activities in preclinical models yet has remained disappointingly unrealized for treating human cancer patients. In clinical studies, the efficacy of IL-12 at tolerable doses has been minimal, and dose escalation has led to unacceptable toxicities (1). Optimistically, with innovations in drug delivery technologies renewed efforts to advance IL-12 towards clinical translation are being explored utilizing improved localized therapeutic strategies including drug fusion conjugates, gene transfection approaches, and controlled release biomaterials (2).

In tandem with advancing drug delivery approaches, there is scientific consensus that predictive animal tumor models would accelerate the advancement of therapeutics towards successful translation in humans (3). As a preclinical model, pet dogs offer unique advantages for evaluating novel cancer treatments (4), and in particular the assessment of immunotherapeutic strategies (5). While biology, genetics, and clinical behaviors of several tumor histologies are shared between dogs and people (6–8), it would be predicted that the evaluation of novel immunotherapeutics would be most informative when tested in immunocompetent animals harboring solid tumors (e.g. melanoma) that have proven targetable with immune modulating strategies.

Melanoma is a common tumor in dogs and accounts for 7% of all malignant canine cancers (9). While involvement of cutaneous tissues is common and accounts for approximately 30% of diagnosed melanomas, the most biologically aggressive and metastatic forms of melanoma classically arise from mucosal tissues, particularly the oral cavity (10). Mounting scientific momentum supports the translational value of canine malignant melanoma for improving treatment of advanced melanoma in humans given the shared clinical behavior, genetics, and biology between species (11–13). Powerfully, the success of immune checkpoint blockade that has transformed the therapeutic landscape for human melanoma has also proven to be a conserved and druggable target in pet dogs with malignant melanoma (14–17). Collectively, these shared similarities in biology and therapeutic responsiveness of malignant melanoma in dogs and humans underscore the potential of a comparative oncology approach for evaluating novel immunotherapies (9).

Anchored immunotherapy is a promising platform that uses inert scaffolding to bind bioactive agents for delivery to the tumor microenvironment (TME), and we previously reported that murine IL-12 could be successfully linked to aluminum hydroxide (18,19). Such anchoring resulted in prolonged retention in the TME with up to 30% bioactive drug available 21 days following injection. The anchored kinetics allow for slow systemic absorptions, induction of local immune activation, and limited on-target, off-tumor toxicity. With this technology, we previously described potent monotherapy and combinatorial responses in a variety of murine tumor models, including B16 melanoma (18,19). Mechanistically in mice, treatment was associated with recruitment of CD8⁺ T and NK cells to the TME, M1 macrophage polarization, and increased gene expressions related to immune activation, interferon signaling, and antigen processing and presentation (18,19). Additionally, mice and non-human primates receiving anchored IL-12 demonstrated minimal toxicity with only isolated non-dose-dependent laboratory abnormalities and local injection site reactions observed (20).

In this study, a canine version of anchored IL-12, composed of the p35 and p40 subunits linked to Alhydrogel^®, was generated and designated as JEN-101 (formerly cANK-101). To explore the therapeutic potential of JEN-101 and identify toxicity and pharmacokinetic parameters in a relevant animal model, we conducted an exploratory, phase I clinical trial in dogs with malignant melanoma. This report represents a preliminary analysis, providing insights into the safety and efficacy of JEN-101 in the context of advanced melanoma in a naturally occurring immunocompetent tumor model.

METHODS

Design and Production of JEN-101

The cloning of canine IL-12-ABP (cIL-12-ABP) was carried out as previously described for human and mouse constructs (18,19). Briefly, single-chain canine IL-12 p40/p35 constructs were designed using a (G₄S)₃ linker with fusion to an Alum binding peptide (ABP) at the C-terminus, followed by a 6X His tag. Plasmids encoding cIL-12-ABP and human Fam20C kinase were transiently transfected in suspension HEK-293 cells (RRID:CVCL_0063). The details of cIL-12-ABP sequence were published, International Publication Number, WO 2022/235755 A2, sequence ID 103 (21) and methods as described previously (18). Supernatants were harvested and IL-12 fusion polypeptides purified by affinity chromatography on NiSepharose Excel (Cytiva 17–3712-02). Eluted fusion polypeptides were formulated in Tris-buffered saline (TBS), pH 7.4 and purity determined on a Perkin Elmer GXII capillary electrophoresis system. Polypeptide aggregation was assessed by HPLC-SEC with a 300Ȧ pore size. When needed, proteins were further purified by FPLC-SEC on a HiLoad 16/600 or 26/600 Superdex 200 pg column (Cytiva 28–9893-36) and monomeric peak fractions were pooled. IL-12-ABP polypeptides were further polished by anion exchange chromatography to enrich for highly phosphorylated species.

Malachite Green Assay

Phosphorylation of purified species-specific IL-12-ABP proteins were measured using the Pierce Phosphoprotein Phosphate Estimation Assay Kit (Thermo Fisher) per manufacturer’s instructions.

Alum retention assay

Binding and retention of cIL-12-ABP on Alhydrogel^® particles was assessed in vitro. Proteins were complexed with a 10-fold mass excess of Alhydrogel^® by mixing in non-phosphate buffer and incubating for 30 minutes at room temperature. Complexes were then diluted in elution buffer containing a final concentration of 1 mM phosphate and 40% human serum as a control and incubated at 37°C with rotation. At various times, samples were centrifuged to pellet the Alhydrogel^® particles and free IL-12 or IL-12-ABP in the supernatant was quantified by an IL-12p70 enzyme-linked immunosorbent assay (ELISA).

IL-12 reporter assay

In vitro IL-12 signaling activity was assessed using the IL-12 Bioassay (Promega, Catalog# JA2601) according to manufacturer’s instructions. The Promega IL-12 Bioassay consists of human cells engineered to express the IL-12 receptor complex (IL-12R) and a luciferase reporter driven by an IL-12 induced response element. When IL-12 binds to IL-12R, it transduces intracellular signaling resulting in luminescence that can be quantified with a Bio-Glo^™ Luciferase assay system. Since canine IL-12 cross-reacts with the human IL-12 receptors, this cell line can be used to assess potency of canine IL-12 derived constructs.

The canine IL-12-ABP proteins in TBS were mixed with 10X excess of Alhydrogel^®, incubated at RT for 30 min and the complex referred as JEN-101 (formerly cANK-101). Unanchored cIL-12-ABP was used as a control. Titrations were prepared in assay media with 10 μg/mL top concentration and 3-fold dilutions. Frozen cells were thawed and resuspended in complete media, plated in a 96-well plate and incubated with sample titration for 6 hours at 37°C. After incubation, the plates were removed from the incubator, equilibrated to RT for 10–15 mins and incubated for another 10 min after the addition of Bio-Glo reagent. Luminescence was measured on Envision plate reader (PerkinElmer).

Primary Canine PBMC IFN-γ Assay

Frozen peripheral blood mononuclear cells (PBMCs) from healthy canines (BioIVT, Catalog# CAN00PBMC) were thawed, plated at 5×10⁵ cells and stimulated with staphylococcus enterotoxin B (SEB) (500 ng/mL) in the presence or absence of cIL-12, cIL-12-ABP or JEN-101 (8-point, 3-fold dilution starting at 100 ng/mL). After a 3-day incubation, cell culture supernatants were collected and analyzed for IFN-γ release by ELISA (Abcam; Catalog#: ab193684). Unstimulated and SEB stimulated PBMCs treated with vehicle (TBS) were used as controls.

Patient Population

The study population were dogs weighing ≥15 kg with histologic diagnosis of melanoma and a primary tumor measuring ≥1 cm in longest dimension and accessible for direct injection. All dogs were required to have a good performance status (modified ECOG criteria 0–1) and had not received prior systemic therapy within four weeks of trial initiation. All dog owners were required to provide written, informed consent and agree to necropsy if the dog died while on study. Exclusion criteria included dogs with significant co-morbid illnesses, serum creatinine >3.0 mg/dL, total bilirubin >2.0 mg/dL, hematocrit <25%, platelets <120,000 cell/μL, ALT or AST >1.5-times the upper reference range, or any hematologic or biochemical abnormality grade 2 or greater in accordance with the Veterinary Cooperative Oncology Group’s Common Terminology Criteria for Adverse Events (VCOG-CTCAE v2) (22).

Study Design and Objectives

This phase 1 study was approved by the IACUC at the University of Illinois at Urbana-Champaign. The study utilized a standard 3+3 dose-escalation design to identify a minimum effective dose (MED), and evaluated the tolerability, immune-activating properties, and cytoreductive effects of JEN-101 in dogs with malignant melanoma. Dogs were enrolled into four cohorts of 1, 3,10, and 20 μg/kg administered by local injection every three weeks for four cycles. Dogs had baseline computed tomography (CT) scans and blood work within 7 days of starting treatment. After four cycles, repeat CT imaging and assessment were performed. For dogs without disease progression or clinically significant toxicity, another four cycles of JEN-101 every three weeks were allowable. Dogs were followed for up to one year following completion of treatment. Baseline tumor biopsy was required in all dogs and post-treatment biopsies were collected when clinically feasibility or at the time of death. Peripheral blood was collected for pharmacokinetic (PK) and pharmacodynamic (PD) analyses as described below. Clinical response was assessed by CT imaging or caliper measurements using RECIST v1.1 criteria.

Study Treatment

JEN-101 was prepared by mixing canine IL-12-ABP with Alhydrogel^® at a 1:10 ratio 30 minutes prior to injection. The starting dose was 1 μg/kg, which was one log order below the effective dose identified in murine tumor models. Additional dosing cohorts included 3, 10, and 20 μg/kg and JEN-101 was given by direct injection into accessible tumors with volume of injection corresponding to the maximum diameter of the tumor as listed in Supplementary Table S1.

Study Assessments and Endpoints

The goals of the study were to assess safety, cytoreductive activity, immune modulation, and PK/PD endpoints after treatment with JEN-101. Safety was assessed using VCOG-CTCAE v2 (22). Adverse Events (AE) were defined as grade 1, 2 or 3 toxicities, whereas a Serious Adverse Event (SAE) was any grade 4 or 5 toxicity. All adverse events were graded for severity and relation to JEN-101 treatment (unrelated, unlikely, possible, probable, or definitely). Dose-limiting toxicities (DLTs) were defined as unexpected or related grade 3 or greater events. Safety was determined using symptoms reported by owners, identified on physical examination and/or blood tests, and reported from baseline up to 30 days after the last exposure to JEN-101. Clinical responses were determined by RECISTv1.1 criteria (23) and investigator assessment. Kaplan Meier survival and swimmer plots were generated for exploratory analysis only.

Pharmacokinetic Analysis

Canine serum samples were collected for pharmacokinetic (PK) analysis of serum cIL-12-ABP. An 8-point PK analysis per dose treatment 1 (day 01), 2 (day 21), and 4 (day 63) included pre-treatment serum samples, and samples at 1h, 2h, 4h, 8h, 24h, 48h, and 72h following JEN-101 administration. Serum samples were also collected during visits or prior administration of JEN-101 at day 7, day 42, day 84, and day 105. Serum samples were stored at −80°until analysis. The PK assay protocol was developed by using canine IL-12/IL-23 p40 DuoSet ELISA Kit (R&D Systems, #DY1969) and adapted to measure cIL-12-ABP. ANK-P104-K-2 pool (cIL-12-ABP) at various dilutions was used as a standard curve. For the assay, ELISA plates were coated with IL-12 capture antibody (R&D Systems, #AF2118 at 2 μg/mL in PBS, pH 7.4), and incubated at 4°C overnight. Samples were diluted in TBS with canine serum (Beagle Serum Gender Pooled; BioIVT, #CAN00SRM-0107409) and incubated for 2 hours followed by the addition of the detection antibody anti-His (Abcam, #ab1187) for 2 hours at RT. Following incubation, TMB substrate (BD Biosciences, Catalog #555214) was added to the plate for 45 minutes at 20°C. The reaction was stopped by addition of an acid solution, and the plates were read using a spectrophotometer set at 450 nm and 570 nm. A non-compartmental PK analysis was performed using Phoenix WinNonlin software.

Anti-Drug Antibody (ADA) Detection

Serum was collected for immunogenicity (anti-drug antibody; ADA) evaluation, and analyses were performed in batches. ADA formation against the cIL-12-ABP component of the JEN-101 complex was assessed by a bridging ELISA assay at pre-dose and on days 21 and 63. For the ADA assay, 96 well ELISA plates were coated with cIL-12-ABP at 0.2 μg/mL in 0.2M Bicarbonate buffer, pH 9 and incubated overnight at 4°C. The plates were washed with 1X PBST and blocked with 300 μl of casein for 1 hour at 37°C. Test samples, negative (beagle serum gender pooled; BioIVT, #CAN00SRM-0107409) and positive (R&D Systems, #AF2118) controls were diluted in the sample diluent and loaded onto the assay plate in duplicate and incubated at RT for 1 hour. After the plate wash, 100 μl per well of primary detection reagent (biotinylated cIL12-ABP at 0.1 μg/mL) was added to the assay plate and incubated at RT for 1 hour. Followed by the wash, 100 μL per well of secondary detection reagent (HRP-Avidin; BioLegend, #405103 at 0.2 μg/mL). After the plate wash, TMB substrate (SurModics, #TMBW-1000–01) was added and color was allowed to develop for approximately 20 minutes at RT. The reaction was quenched by adding 0.1 M sulfuric acid and the plate was read on SpectraMax plate reader at 450 nm. Prior to treatment, all dogs were ADA-negative. Using a cut-off value of 2-fold above background, the ADA positivity of the samples was determined.

Multiplex cytokine assay and ELISA

Serum samples obtained from patients following treatment were assessed for levels of cytokine and chemokine analytes. These included GM-CSF, IFN-γ, IL-2, IL-6, IL-7, IL-8/CXCL8, IL-10, IL-15, IL-18, IP-10/CXCL10, KC-like, MCP-1/CCL2, and TNFα (Canine MILLIPLEX Magnetic Bead Panel, Millipore Sigma). The concentrations of individual analytes were determined based on standard curves specific to each cytokine or chemokine within the panel according to the manufacturer instructions. Response of patients following treatment was analyzed over time by calculating the log10 fold-change in analyte concentrations relative to the initial pre-treatment measurements. IFN-γ and IL-10 serum levels post-treatment with JEN-101 were measured using the Canine IFN-γ Quantikine ELISA kit (R&D) and the Canine IL-10 Quantikine ELISA kit (R&D) as per the manufacturer’s instructions.

RNA isolation and Nanostring analysis

RNA was extracted from 10 μm tissue thickness FFPE samples obtained from canine primary melanoma tumor or metastatic tumor lesions using the RNEasy FFPE Kit and deparaffinization solution (Qiagen). Subsequently, isolated RNAs underwent assessment of fragment size by Bioanalyzer (Agilent), and samples were processed according to the manufacturer instructions using the Canine IO nCounter Panel (NanoString Technologies) through the Tumor Engineering and Phenotyping Share Resource, Cancer Center at Illinois in batches. Canine gene counts were normalized using nSolver software (NanoString Technologies) with background thresholding based on the mean of 8 negative control probes. Normalized gene counts were processed using the nSolver Advanced Analysis module for cell type profiling and pathway score analysis.

Immunohistochemistry

Canine tumors biopsies were collected at specific indicated timepoints. Tumor tissues were fixed in 10% formalin and embedded in paraffin and immunohistochemistry was used to characterize immune cell infiltration. Sections were stained for CD3 (T lymphocyte; 1:50 Biocare, #CP215A) and IBA-1 (macrophage; 1:100 - Biocare, #CP290A). Staining was carried out using Leica BOND RX fully automated research stainer and Bond Polymer Refine Red Detection Kit (Leica, #DS9390). Stained slides were imaged with Olympus VS120 (Olympus, Center Valley, PA) and analyzed using OlyVIA software. Samples were histologically evaluated and classified by a board-certified pathologist.

Statistical Analysis

The clinical trial was designed as an exploratory 3+3 dose-escalation phase I study to identify a minimum effective dose and all statistical analyses were descriptive. There was no power calculation of sample size. Continuous variables were described, whereas categorical variables were summarized with frequency counts or percentages, as appropriate.

Safety evaluation included all subjects that received at least one dose of JEN-101. Dose limiting toxicities (DLT) followed adverse event reporting guidelines in the clinical protocol with three additional subjects being enrolled in the event of one DLT, and two DLTs at any dose would prohibit further expansion at that dose. The response population included all subjects that had at least one post-baseline tumor response assessment determined by investigator evaluation, serial CT scan and/or caliper measurements and followed RECISTv1.1 guidelines (23). Dogs withdrawn prior to the first assessment were considered to have disease progression. Assessment of PK/PD endpoints utilized serum concentration-time data and descriptions with non-compartmental methods. Gene expression was analyzed as described above. Statistical analyses were conducted using SAS, version 9.0 or later and using Prism v10 (GraphPad).

DATA AVAILABILITY

All data relevant to the study are included in the article or available as supplementary information (Supplementary Data S1). NanoString expression data for canine tumor expression in dogs in response to JEN-101 treatment has been made publicly available in Gene Expression Omnibus (RRID:SCR_005012) at GSE278391.

RESULTS

Characterization of Anchored Canine IL-12 (JEN-101)

JEN-101 consists of a single-chain canine IL-12 protein fused to a c-terminal phosphorylated alum-binding peptide (IL-12-ABP) complexed with aluminum hydroxide (Alhydrogel^®) (Figure 1A). Gene constructs were generated to encode canine IL-12β (p40) and IL-12α (p35), connected through a (G₄S)₃ linker, with fusion to an alum-binding peptide (ABP) containing 8 repeats of the sequence motif (SEEGGGG), followed by a 6X His-tag to facilitate protein purification using affinity chromatography. The construct was co-expressed in HEK cells with a vector encoding the Fam20C kinase, which phosphorylates the amino acid motif SEE at multiple sites within the ABP motif. Canine IL-12-ABP (cIL-12-ABP) proteins were consistently phosphorylated with an average of 7.25 phosphates, as determined by malachite green assay, and closely resembled that of human and mouse versions of the protein, which contained 7 and 5.25 phosphates, respectively (Figure 1B).

Figure 1. In vitro characterization and biological activity of cIL-12-ABP and JEN-101.

(A) Ribbon diagram and schematic illustration of JEN-101 components. (B) Recombinant murine IL-12 (mIL-12; control), mouse (mIL-12-ABP), human (hIL-12-ABP), and canine (cIL-12-ABP) proteins are comparably phosphorylated at multiple serines following co-expression with Fam20C kinase as measured by malachite green assay (mean ± SD, n=2 replicates). (C) cIL-12-ABP, mIL-12-ABP and hIL-12-ABP were complexed with Alhydrogel^® and incubated in elution buffer containing 1 mM phosphate and 40% serum. Free IL-12-ABP proteins in the supernatant were measured at various times by ELISA. >90% of cIL-12-ABP protein was bound to Alhydrogel^® at 24 hours (mean ± SD, n=2 replicates), comparable to human and mouse IL-12-ABP constructs. (D) Reporter assay measuring IL-12 activity of unanchored cIL-12-ABP and anchored JEN-101 (mean ± SD, n=2 replicates). (E) IFN-γ production from canine PBMCs stimulated for 3 days with SEB (500 ng/ml) and test agents. The functional potency of free cIL-12-ABP and the JEN-101 complex were similar in both assay systems (mean ± SD, n=2 replicates). Schematic illustration of JEN-101 adapted using BioRender.com.

When mixed with a 10-fold mass excess of Alhydrogel^® in Tris-buffered saline (TBS), the cIL-12-ABP proteins bind through ligand exchange reactions between the phosphate groups on the cIL-12-ABP and the hydroxide groups on the surface of Alhydrogel^® (the cIL-12-ABP-Alhydrogel^® complex is referred to as JEN-101). Following exposure to phosphate and serum to model physiological conditions, the cIL-12-ABP proteins remained tightly bound within JEN-101, with more than 95% still complexed at 24 hours and comparable to mouse and human versions of the protein (Figure 1C).

To assess whether complexation with Alhydrogel^® impacts biological activity, we titrated JEN-101 into an IL-12 signaling reporter assay and compared it to unanchored cIL-12-ABP. Both cIL-12-ABP and JEN-101 induced potent IL-12 signaling, with EC50 values of 1.3 ng/mL and 4.2 ng/mL, respectively (Figure 1D). To confirm activity in a species-specific and intended cellular-targeting context, we used primary canine immune cells and conducted titrations of wildtype canine IL-12 (cIL-12), cIL-12-ABP, and JEN-101 in an IFN-γ release assay. All three molecules exhibited comparable ability to stimulate IFN-γ release from staphylococcus enterotoxin B (SEB)-activated canine PBMCs in a concentration-dependent manner (Figure 1E). These results indicate that cIL-12-ABP retains its biological activity when complexed with Alhydrogel^®.

Study Population and Demographics

At time of data analysis, 18 dogs had been enrolled across four dose-escalation cohorts with 3 dogs enrolled at the 1 and 3 μg/kg cohorts, 4 dogs enrolled at the 10 μg/kg cohort, and 8 dogs enrolled at the 20 μg/kg cohort (Table 1). One dog had metastatic cutaneous melanoma and 17 dogs had primary oral (mucosal) malignant melanoma. Eleven dogs had World Health Organization (WHO) Stage II, III, or IV disease at baseline. Five dogs required a single dose of local radiotherapy (8 Gy) while on study as a result of either progressive disease or large tumors involving anatomic areas that compromised airflow and swallowing. Fifteen dogs completed at least 4 doses of JEN-101, including a subset of 7 dogs that completed the full 8 doses, and 3 dogs received fewer than 4 doses. The clinical study design is depicted in Figure 2A.

Table 1.

Patient demographics and baseline characteristics.

	Cohort 1 1 μg/kg (n=3)	Cohort 2 3 μg/kg (n=3)	Cohort 3 10 μg/kg (n=4)	Cohort 4 20 μg/kg (n=8)	Total (n=18)
Breed [n (%)]
Purebred	3 (100%)	1 (33%)	2 (50%)	5 (57%)	11 (61%)
Beagle	0	0	0	1	1
Golden Retriever	1	0	0	0	1
Labrador Retriever	2	1	1	1	5
Cocker Spaniel	0	0	1	0	1
Border Collie	0	0	0	2	2
German Shepherd	0	0	0	1	1
Crossbred	0 (0%)	2 (67%)	2 (50%)	3 (43%)	7 (39%)
Golden Retriever/Standard Poodle	0	0	1	0	1
Pit Bull/Chinese Shar Pei	0	1	0	0	1
Pit Bull/Beagle	0	0	1	0	1
Pit Bull/Wire Fox Terrier	0	0	0	1	1
German Shepherd/Beagle	0	0	0	1	1
Labrador Retriever/Cocker Spaniel	0	0	0	1	1
Shetland Sheepdog/German Shepherd	0	1	0	0	1
Primary site of malignant melanoma [n (%)]
Cutaneous	1 (33%)	0	0	0	1 (5.5%)
Lip	0	1 (33%)	1 (25%)	3 (37.5%)	5 (28%)
Mandible	1 (33%)	0	1 (25%)	1 (12.5%)	3 (17%)
Maxilla	0	1 (33%)	0	0	1 (5.5%)
Hard Palate	0	1 (33%)	0	0	1 (5.5%)
Soft Palate	1 (33%)	0	1 (25%)	2 (25%)	4 (22%)
Sublingual	0	0	0	1 (12.5%)	1 (5.5%)
Tongue	0	0	1 (25%)	1 (12.5%)	2 (11%)
Age (years)
Median	9	9	9	8	9
Min, Max	9, 12	5, 10	4, 11	7, 16	4, 16
Gender [n (%)]
Male	1 (33%)	1 (33%)	1 (25%)	5 (62%)	8 (44%)
Female	2 (67%)	2 (67%)	3 (75%)	3 (38%)	10 (56%)
Baseline ECOG [n (%)]
0	3 (100%)	3 (100%)	4 (100%)	8 (100%)	18 (100%)
1	0	0	0	0	0
Baseline weight (kg)
Median	33.6	25.4	33.3	25.6	26.8
Min, Max	26.5, 40.2	22.1, 37.2	16.1, 36.8	12.7, 37.2	12.7, 40.2
Baseline WHO Stage[1] [n (%)]
I	0	1 (33%)	2 (50%)	3 (37%)	6 (33%)
II	0	1 (33%)	0	1 (13%)	2 (11%)
III	2 (67%)	0	2 (50%)	0	4 (22%)
IV	0	1 (33%)	0	4 (50%)	5 (28%)
NA	1 (33%)	0	0	0	1 (6%)

^[1]WHO stage calculated retrospectively based on entered baseline EDC data.

Figure 2. JEN-101 clinical trial study design and assessment of clinical responses.

(A) Schematic of the clinical trial design. Eligible dogs received JEN-101 IT every 21 days for 4 cycles. Three dogs were treated at cohorts of 1, 3, 10, and 20 μg/kg. Eligible dogs also received a second course of JEN-101 IT every 21 days for 4 cycles. Dogs were assessed for clinical responses, pharmacokinetic, pharmacodynamic, immunologic, and CT imaging assessments as described in Methods. (B) Kaplan-Meier survival curve for JEN-101 treated dogs across all dosing cohorts with a median survival time of 164 days (green dotted line). Comparator historical studies for untreated (red dotted line) or palliative radiation (brown dotted line) treated dogs with oral malignant melanoma with a median survival times of 65 days or 147 days, respectively. (C) Swimmer’s plot of individual subject response and durability (arrow denotes alive, without disease). Schematic clinical trial design created with BioRender.com.

Safety Analysis

Eighteen dogs were evaluable for safety and the most common adverse events (AEs) included anemia (n=3; 17%), hypoalbuminemia (n=3; 17%), transient elevated alanine transaminase (n=5; 28%), elevated serum alkaline phosphatase (n=4; 22%), lethargy (n=4; 22%), local edema (n=4; 22%), and thrombocytopenia (n=3; 17%). AEs that occurred in two or more dogs are listed in Supplementary Table S2. There were two Grade 3 JEN-101-related AEs seen at the 10 μg/kg dose (pyrexia and lethargy) and three grade 3 JEN-101-related AEs at 20 μg/kg dose (pyrexia, lethargy, and elevated serum alkaline phosphatase). All other adverse events were ≤ Grade 2.

Elevated liver transaminases (ALT and/or ALP) were seen in six dogs and all were ≤ Grade 2. There was one dog that experienced Grade 1 neutropenia at the 20 μg/kg. There was a general trend for more AEs reported in the 10 μg/kg cohort (Supplementary Fig. S1). Serious adverse events (SAEs) were seen in two dogs. One dog (10 μg/kg cohort) died from sepsis secondary to soft tissue infection that was not caused by JEN-101, and the SAE was ascribed to clinical trial procedures. Specifically, this patient developed lymph node abscessation and subsequent rupture, with aerobic and anaerobic culture supporting translocation of skin bacterial flora introduced during intratumoral injections. The other SAE occurred in a dog (20 μg/kg cohort) with lethargy and fever 13 days after the first JEN-101 dose and was considered as possibly related to study treatment.

Antitumor Activity

At the time of data analysis, 15 dogs had completed treatment with a minimum of 4 JEN-101 injections and were evaluable for response. Of the remaining 3 dogs receiving fewer than 4 doses of JEN-101, 2 were categorized as progressive disease and 1 was lost to follow up. Based on investigator assessment, serial CT scan or caliper measurements, the best response achieved was five dogs with an objective response and three dogs with stable disease, resulting in 47% (8/17) disease control rate (Supplementary Table S3). Objective responses to JEN-101 alone were seen at all doses ranging from 3–20 μg/kg. The median survival time of our study patients treated with JEN-101 is 164 days, compared to reported median survival time of 65 days for untreated dogs with oral melanoma (24), and median survival time of 147 days for dogs with oral melanoma treated with palliative coarse fractionated radiotherapy (9 Gy x 4 fractions) (25) (Figure 2B). In addition, eight dogs are alive and have maintained responses 8–18 months after treatment (Figure 2C).

Dog 001 had six metastatic lesions from a primary cutaneous melanoma and had four of the lesions injected. After completing all planned 8 doses of JEN-101, the dog had stable disease and underwent surgery of all residual lesions achieving a surgical complete response. This dog has been free of melanoma for over 18 months and demonstrated site specific and durable immune activation elicited by JEN-101 treatment (Supplementary Fig. S2B). Dog 006 presented with a locally advanced hard palate mucosal melanoma and was treated with 3 μg/kg JEN-101. The tumor underwent significant regression and the dog achieved a partial response (Figure 3A). However, by day 252 post-treatment, a small lesion reemerged, and biopsy confirmed recurrent melanoma; this lesion slowly progressed in size until day 524, and was then re-treated with JEN-101 as a single agent, and then in combination with anti-CTLA-4 antibody and achieved a second partial response (Supplementary Fig. 3A). Dog 010 presented with a lip mucosal melanoma with regional lymph node metastases and was injected in both the primary tumor and regional effaced lymph node. After 6 doses of JEN-101 there was regression of the primary tumor (Figures 3B–E). Additionally, this dog also had CT imaging on day 84 when an enlarging pulmonary metastatic lesion (12 mm) was noted. A repeat CT scan on day 105 demonstrated shrinkage of the lesion to 8 mm, equating to a 70% reduction in metastatic lesion volume and indicative of an abscopal response. Additionally, a smaller pulmonary metastatic lesion measuring 5 mm identified on day 84 was minimally detectable by CT on day 105 (Figure 3D). Corroborating these radiologic changes, serial primary tumor biopsy and gene expression analysis (Figure 3C and Supplementary Fig. S2E) showed global upregulation of key genes representing a tumor inflammation signature (26). Moreover, gene expression analysis of the regressing metastatic lung nodule (12 mm to 8 mm) also indicated upregulation of key inflammatory genes (Figure 3E) relative to the primary tumor at both pre- and post-treatment timepoints, and suggested that intratumoral JEN-101 administered to the primary tumor was sufficient to generate anticancer immune activities at an uninjected distant site. To further characterize potential abscopal mechanisms, circulating cIL-12-ABP levels following initial dosing were detectable at all assessed timepoints (Supplementary Fig. S4A–C), indicating that metastatic lung nodule regression could be attributed to slow, but systemic, leakage of cIL-12-ABP from the injected primary tumor site. Dog 017 presented with a rostral mandibular melanoma affecting both sides of the incisor teeth, and was treated with JEN-101 at a dose of 20 μg/kg. Following two doses, the patient exhibited pseudoprogression of the primary lesion that started to regress following the third dose of JEN-101, and the dog achieved a partial response (Figure 3F). Corroborating the reduction in tumor size, serial tumor biopsies revealed a global increase in key genes associated with a tumor inflammation signature (26), particularly on day 42 when the mass began to regress (Figure 3G). Further serial biopsies could not be conducted due to the decrease in tumor size.

Figure 3. Case studies of patients with JEN-101 treatment.

(A) Dog 006 treated with JEN-101 at 3 μg/kg over 8 cycles showing partial response. (B) Dog 010 treated with 10 μg/kg JEN-101. A primary lip melanoma demonstrated substantive tumor regression based upon visual inspection and caliper measurements and confirmed quantitatively by serial computed tomography (day 1 versus day 84; red shaded overlay indicating tumor mass). (C) Dog 010 serial NanoString analysis of primary tumor biopsies following repeated JEN-101 treatment underscores robust immune activating properties of intratumoral anchored IL-12 (D) Dog 010 also demonstrated an abscopal response with partial regression of uninjected pulmonary metastases. (E) NanoString analysis of regressing pulmonary metastatic lesion from Dog 010 indicated a strong tumor inflammation signature at an uninjected distant site. (F) Dog 017, which presented with rostral mandible involvement, was treated with JEN-101 at a dose of 20 μg/kg. Partial regression was observed. (G) Serial tumor biopsies from dog 017 presenting with pseudoprogression which demonstrated upregulation of key inflammatory genes immediately before tumor regression was observed.

Pharmacokinetics and Anti-Drug Antibody Responses

To detect systemic absorption of the cIL-12-ABP component of JEN-101 and characterize its PK profile, serum was collected after treatment and concentration of cIL-12-ABP quantified using an ELISA-based assay. The PK profiles obtained after the first cycle of JEN-101 are presented for each treatment cohort in Figure 4A, and the calculated PK parameters are listed in Supplementary Table S4. Overall, the levels of cIL-12-ABP were very low, ranging from picograms to a maximum of 1.68 ng, and in four dogs (001, 002, 009, and 016), levels fell below the range of detection at all timepoints for dogs 002 and 009, and at cycle 2 for dogs 001 and 016. Among dogs with measurable profiles, the median maximum serum concentration (C_max) value of cIL-12-ABP after the first dose of JEN-101 was 419 pg/mL (range 105 pg/mL to 1.68 ng/mL). Both C_max and the total systemic exposure (area under the time-concentration curve from the first to the last measured value [AUC_last]) exhibited a trend toward dose-dependency but failed to reach statistical significance, as shown by the median cIL-12-ABP levels by dose cohort (Figure 4B and Supplementary Table S4). The median time to reach maximum serum cIL-12-ABP concentration (T_max) was 1 hour, with a range of 1–72 hours.

Figure 4. Pharmacokinetics, pharmacodynamics, and immunological responses of JEN-101.

(A) Serial serum samples were obtained at timepoints shown according to the study protocol outlined in the Methods and circulating IL-12 levels were measured by ELISA. Data is shown by treatment cohorts following first dosing and (B) the IL-12 median for all cohorts following the first dosing is shown in the right panel. A transient and consistent increase in serum cIL-12-ABP levels was detected in all dogs following JEN-101 treatment (mean ± SD, n=2 replicates). (C) Serial serum samples collected at specified timepoints, as detailed in the Methods, and ELISA analysis for circulating IFN-γ is shown by treatment cohort following first dosing and (D) the IFN-γ median for all cohorts following first dosing is shown in right panel. Increased serum IFN-γ was observed, peaking around 8 hours after treatment (mean ± SD, n=2 replicates). (E) Composite dose cohort maps of circulating cytokine and chemokine profiling from all patients determined using MILLIPLEX. JEN-101 treatment led to an increase in multiple cytokines and chemokines associated with IL-12 modulation, including IFN-γ, IL-10, IL-8, IL-18, IP-10, and MCP-1. (mean ± SD, n=2 replicates). (F) 72-hour IFN-γ and IL-10 median levels and area under the curve (AUC) of IFN-γ and IL-10 following dose 1 in dogs that achieved disease control (green) and dogs without response (red). Statistics: IL-12-ABP and IFN-γ levels were compared at each time point (hours) across groups using Kruskal Wallis test. Results were not significant.

Serum was collected for anti-drug antibody (ADA) titers at baseline and on days 21 and 63 (Supplementary Table S5). Using a bridging ELISA, a fold change >2 over baseline values was considered positive for ADA. Overall, 10/18 dogs tested positive for ADAs: 2/3 at 1 μg/kg, 1/3 at 3 μg/kg, 3/4 at 10 μg/kg, and 4/8 at 20 μg/kg. There was a trend towards stronger ADA responses in the 20 μg/kg cohort, but the magnitude of ADA fold changes across dosing cohorts following 63 days of JEN-101 exposure were not significant (Supplementary Fig. S5A). However, the development of ADA titers could be a contributing factor to the higher rate of undetectable drug levels and overall lower AUC_last values after the second cycle of JEN-101 treatment. The potential neutralizing effects of ADA on the biologic activity of JEN-101 was not identified in this study, as there is no significant inverse and negative correlation between peak IFN-γ levels and ADA titers at day 63 of treatment across different dosing cohorts (Supplementary Fig. S5B).

Immune Responses

IL-12 directly stimulates the production of IFN-γ, and circulating IFN-γ serves as a key pharmacodynamic marker of IL-12’s in vivo bioactivity (27). Thus, the kinetics of serum IFN-γ were measured using an ELISA assay at similar timepoints as the cIL-12-ABP PK assessments. Most dogs (17/18) displayed detectable increases in IFN-γ with peak concentrations occurring between 1- and 8-hours following the first JEN-101 administration (Figure 4C). Generally, levels of IFN-γ returned to or near baseline within 48 hours. While the magnitude of IFN-γ release varied considerably within and across dose cohorts, higher levels of IFN-γ were consistently observed in cohorts 2 (3 μg/kg) and 4 (20 μg/kg) (Figure 4D).

IL-10 is an anti-inflammatory cytokine induced by IL-12 as a counter regulatory response, and attenuates IFN-γ production upon repeated dosing of IL-12 (27). To assess IL-10 release in response to JEN-101, serum IL-10 levels were measured by ELISA. Overall, IL-10 levels were very low across all dose cohorts and in those patients with detectable IL-10 release, peak levels generally occurred later (≥ 8 hours) than was observed with IFN-γ, with higher levels of IL-10 being consistently observed in cohort 4 (20 μg/kg) (Supplementary Fig. S6).

To explore a broader panel of markers and assess cytokine responses following repeated dosing, patient sera were analyzed kinetically after each treatment cycle using Luminex, and a composite representing fold-change in cytokine values in each dose cohort is shown in Figure 4E. Consistent with the ELISA data, increases in IFN-γ were observed in all four cohorts with a peak at 8 hours after each cycle of treatment followed by a gradual decline. In some instances, IFN-γ remained above the pre-treatment baseline for up to 72 hours. Similarly and congruent with ELISA data, increased IL-10 levels were also observed, most noticeably at the highest dose of 20 μg/kg, which may indicate stronger counter regulation at this dose level. The other measured cytokines and chemokines (IL-2, IL-6, IL-8, IL-15, KC-like, IL-18, MCP-1, and TNFα) showed either no change from baseline or minimal change with no discernible kinetic pattern (Figure 4E).

To explore a potential association between serum markers and clinical responses, we compared the IFN-γ and IL-10 profiles of dogs with disease control (SD, PR, CR) to dogs with PD. Dogs with disease control showed trends towards higher levels of IFN-γ and concurrent lower IL-10 AUCs than dogs with PD, indicating a potential inverse relationship between JEN-101 mediating IFN-γ and IL-10 release and response to treatment (Figure 4F).

Immune changes in the TME were assessed by IHC for CD3⁺ T cells and IBA1⁺ macrophages on biopsy specimens obtained pre-treatment and after JEN-101 injection. We routinely observed an increase in CD3⁺ T cells in the tumors after two injections. Interestingly, we saw a similar influx of CD3⁺ T cells even in dogs with disease progression (Supplementary Fig. S7), suggesting that the presence and magnitude of lymphocyte influx was not a sole determinant of therapeutic benefit. Changes in the number of IBA1⁺ macrophages within the TME were more variable, and their M1 versus M2 phenotype could not be definitively differentiated using commercially available reagents validated in dogs.

Given some limitations associated with immunohistochemistry, NanoString^™ was employed to obtain transcriptional profiles from patients that had multiple serial biopsies collected. Even at the lowest dose cohort (1 μg/kg), intratumoral JEN-101 induced specific and durable immune activation, as exemplified in biopsies taken from dog 001 (1 μg/kg) on study day 196 (approximately 8 weeks after final JEN-101 injection). Both non-injected tumors (2 lesions) and injected tumors (4 lesions) were collected for comparative analysis (Supplementary Fig. S2B). Immune cell gene signatures in the 2 non-injected tumors revealed an immunologically ‘cold’ phenotype characterized by very low levels of CD8⁺ T cells and macrophages, and very high levels of neutrophils. In contrast, the 4 tumor lesions injected with JEN-101 exhibited an immunologically ‘hot’ transcriptional profile characterized by elevated levels of CD8⁺ T cells and macrophages, as well as enrichment of other immune cell signatures including B cells, cytotoxic cells, Th1 cells, and CD56^dim NK cells.

Transcriptional profiling was also conducted on serial tumor biopsies collected before and after treatment from several other patients across different dose cohorts. When biopsies were compared using immune-related pathway score analysis, post-treatment samples exhibited a strong enrichment in multiple pathways indicative of an inflammatory state (Supplementary Fig. S2C–F). The most highly enriched pathways included interferon signaling, chemokines, cytokines, cytokine & chemokine signaling, interleukins, JAK-STAT signaling, and the myeloid compartment. Additionally, upregulation of key inflammatory genes at primary treated lesions (Figures 3C,G), as well as at a non-injected distant pulmonary lesion (Figure 3E) was observed following JEN-101 treatment.

Relationship of PK/PD Markers and Therapeutic Response

The local PK of intratumoral (IT) therapies is impacted by unique dosing parameters, including the ratio of dose volume (DV) to tumor volume (TV) (27). Given the utilization of standard weight-based dosing in the present study, it was expected that these parameters would exhibit considerable variability. Consequently, we retrospectively explored trends with systemic PK, pharmacodynamics (IFN-γ levels), and/or clinical activity. Table 2 provides a summary of dose, DV, TV, DV/TV ratio, and dose/TV ratio, together with C_max levels of cIL-12-ABP, peak IFN-γ levels, and best response for all dogs.

Table 2.

Relationship between dosing parameters, PK/PD, and response in evaluable cohorts.

Dose cohort	Patient ID	Tumor Dose (μg)	Dose volume (mL)	Tumor volume (cc)	DV/TV ratio	Dose/TV ratio (μg/cc)	Cmax (pg/mL)	Peak IFNγ (pg/mL)	Best response
1 μg/kg (n=3)	001	19	2	34	0.06	0.6	104.6	223	CR*
		10	1	2	0.5	3.9
		3	0.3	1	0.3	5.7
		2	2	0.2	10	11.1
	002	40	4	103	0.04	0.4	NR	15.5	PD
	003	15	2.2	25	0.09	0.6	230.5	66.1	SD
		7	1	8	0.13	0.9
		5	0.8	5	0.16	1.2
3 μg/kg (n=3)	004	110	1	3	0.33	32.1	266.7	441.4	PD
	005	60	4	125	0.03	0.5	104.7	288	PD
		15	1	2	0.5	7.4
	006	67	2	10	0.2	6.8	1057	367.1	PR
10 μg/kg (n=4)	007	320	4	187	0.02	1.7	661	93	PD
	008	337	0.5	1	0.5	674.3	224	40.3	CR*
	009	327	1	2	0.5	194	NR	952.8	SD
	010	109	2	14	0.14	8	406.7	172.4	PR
		54	1	6	0.17	8.9
20 μg/kg (n=8)	011	323.2	4	83	0.05	3.9	836.8	1042.3	NE
		80.8	1	14	0.07	6
	012	439	4	70	0.06	6.2	1682.3	969.1	PD
		110	1	3	0.29	32
	013	265	2.5	43	0.06	6.2	1593	493.9	PD
		159	1.5	10	0.15	16.1
		106	1	4	0.25	26.5
	014	744	1	2	0.49	363.3	37.4	145.2	PD
	015	412	0.5	0.9	0.58	476.9	1032	12.7	SD
	016	572	2	27	0.07	20.8	193.4	447.2	PD
	017	258	0.5	1.4	0.36	188	536.1	230.7	PR
		258	0.5	0.9	0.58	298.6
	018	260	1	5.3	0.19	48.8	432	92.4	PD

DV, dose volume; TV, tumor volume; NE, not evaluable; NR, not reportable;

^*CR by surgery

As anticipated, weight-based dosing resulted in extensive variability across both dosing parameters. The DV/TV ratio demonstrated no discernible trends in relation with PK/PD or therapeutic response. Additionally, the dose/TV ratio did not exhibit a trend with either IFN-γ release (Table 2) or therapeutic response (Supplementary Fig. S8). Amongst dogs with PD, 3 out of 9 were treated at very low dose/TV ratios (<2) and exhibited weak to moderate IFN-γ responses (<290 pg/mL). Dogs with PD with dose/TV ratios (>2) exhibited IFN-γ responses that ranged from 92.4 – 493.9 pg/mL. Conversely, among dogs that experienced disease control, 5 out of 8 had dose/TV ratios ranging from 6.8 to 674.3. Dog 008, which achieved strong response with JEN-101 alone prior to achieving a surgical CR, had the highest dose/TV ratio (674.3) by a wide margin. Furthermore, total dose injected per patient did not predict disease response (Supplementary Fig. S8), and C_max values displayed wide variability and did not demonstrate a consistent trend with the dosing parameters, IFN-γ levels, or best response. These observations align with the anticipated decoupling of systemic PK from pharmacodynamics and efficacy, which is characteristic of IT therapies.

DISCUSSION

This is the first report describing the safety and immune-activating properties of an engineered IL-12 cytokine with tight binding to aluminum hydroxide as an anchored immunotherapeutic strategy evaluated in a spontaneous cancer model. JEN-101 (formerly cANK-101) is composed of canine IL-12 stably linked to Alhydrogel^®, allowing for the facile and reproducible localized retention of immune-modulatory payloads when given by intratumoral injection. The inclusion of pet dogs to advance novel immunotherapies towards human translation has been successfully leveraged previously (29–31), and we sought to utilize pet dogs with malignant melanoma to serve as an immune competent tumor model to critically evaluate the safety of our localized cytokine strategy, JEN-101. While cutaneous malignant melanoma develops spontaneously in dogs, primary mucosal involvement in the form of oral malignant melanoma is most common (10), and recent cross-species comparative genomic studies have suggested a close relationship between canine melanoma and human mucosal melanoma (12,13,32). Collectively, these investigations underscore the potential value of a comparative oncology approach, and our study supports the potential use of dogs as a surrogate non-human model for early immunotherapy drug development for mucosal melanoma.

Through protein engineering strategies, JEN-101 was designed to tightly bind with Alhydrogel^®, an approved vaccine adjuvant possessing reproducible depot effects and augmentation of adaptive immune responses. Corroborating murine studies (18,19), we found that >95% of canine IL-12 binds to Alhydrogel^® and anchored IL-12 retains functional activity, inducing IFN-γ release in both canine reporter assays and in peripheral blood lymphocyte assays (Figure 1). In vivo, we found that JEN-101 was generally well tolerated and induced largely low-grade constitutional symptoms (e.g. pyrexia, lethargy) and local injection site reactions (e.g. pain and edema) as well as isolated, non-dose-dependent, and reversible laboratory abnormalities. Some of the injection site reactions identified might be related to aluminum hydroxide, which is known to induce local inflammation and granuloma formation following subcutaneous injection (33), and supported by some of our patients treated with JEN-101 (Supplementary Fig. S9). We did not observe any definitive dose-limiting toxicities, however dogs treated at the 20 μg/kg dose had a higher incidence of adverse events. There was one death on study in a dog treated at 10 μg/kg that developed sepsis following 6 injections; however, this event was not considered related to JEN-101’s mechanism of action but ascribed to procedural needle injection techniques that resulted in inadvertent cutaneous bacterial flora translocation into an injected metastatic lymph node with consequent abscessation. This adverse event prompted the institution and practice of improved injection site preparation and increased vigilance for maintaining sterility over the course of repeated intratumoral treatments. Overall, JEN-101 was well tolerated and appears to be safe.

While safety was a primary study endpoint, we did document local disease control (CR, PR, or SD) in 47% (Table 2) of the dogs treated with JEN-101. Of the 8 dogs achieving disease control, one dog treated at 3 μg/kg, one dog treated at 10 μg/kg, and one dog treated at 20 μg/kg achieved durable partial responses. Additionally, two dogs treated with 1 μg/kg and 10 μg/kg achieving stable disease were rendered complete responders by adjuvant surgical resection following completion of 8 JEN-101 treatment cycles. Further, we also observed infrequent, yet provocative, abscopal responses including partial to near complete regressions of uninjected lung metastatic lesions (Figure 3E) in a patient achieving partial remission of their injected primary oral melanoma.

An important secondary objective of this study was to demonstrate the immunobiological activity of JEN-101. We observed production of IFN-γ and IL-10 following JEN-101 in most dogs, with a trend towards dose-dependency. Dogs with very low IFN-γ levels generally had progressive disease, whereas dogs with higher IFN-γ experienced more disease stabilization and/or tumor regression (Figure 4F). IL-10 is also induced by IL-12 and serves as a critical counter-regulatory cytokine to modulate immune activation, and IL-10 has been implicated in the loss of sensitivity to repeated IL-12 exposure (34). We found that low levels of IL-10 were consistently detected after JEN-101 administration (Supplementary Fig. S6). Interestingly, we observed higher IFN-γ and concurrent lower IL-10 AUCs in dogs treated with JEN-101 that achieved disease control compared to dogs with progressive disease (Figure 4F). These findings suggest the potential utility of these 2 diametrically opposed cytokines in predicting response to JEN-101 and further characterization of cytokine relationships associated with anchored IL-12 will be an important area of future study.

In addition to circulating factors, tumor biopsies were analyzed pre- and post-treatment when clinically feasible. The lack of consistent biopsy collection is a recognized limitation of the current study, with feasibility of serial biopsies often limited by the size and location of injected lesions, as well as the intent to minimize inadvertent JEN-101 leakage following intralesional injection. In specimens available, nearly all showed a robust increase in CD3⁺ T cell infiltration following JEN-101 injection (Supplementary Fig. S7). We also explored if JEN-101 could alter the myeloid population within the TME given that IL-12 mediates conversion of M2 macrophages to an anti-tumor M1 phenotype in mouse models (35). IBA1 is a glial and macrophage marker purported to aid in the identification of M2 polarized macrophages (36), and used to characterize tumor associated macrophages (TAMs) within canine melanocytic tumors (37). While observational, our findings suggest that JEN-101 might mediate tumor regression, at least partially, through repolarization and/or reduction of M2-like TAMs, a supposition which is also consistent with prior reports of murine ANK-101 that was associated with a transition of M2 to M1 macrophages in a murine model of head and neck cancer (20). Lastly, transcriptional profiling consistently showed immune activation in post-treatment tumor samples (Figures 3C, G and Supplementary Fig. S2B–F) and provides strong corroborative evidence for the pro-inflammatory activities exerted by intralesional JEN-101 locally and also at an uninjected distant site (Figure 3E).

The PK profile observed in dogs treated with JEN-101 suggested a reduction or loss of cIL-12-ABP systemic exposure after repeated dosing (Supplementary Table S4). This attenuating effect may be related to induction of anti-drug antibodies (ADAs), which were observed in nearly all dogs by day 63 (Supplementary Table S5). However, there was no statistically significant correlation between ADA and peak IFN-γ concentrations at day 63 identified in our study (Supplementary Figure S5). These findings would suggest that other counter-regulatory tachyphylactic mechanisms might be operative and contribute to reductions in circulating IFN-γ following repeated treatment with JEN-101.

We observed that peak drug levels and total systemic exposure occurred in a less than dose-proportional manner, as expected. This ‘flip-flop’ PK profile suggests that systemic absorption of cIL-12-ABP is rate-limited following intralesional delivery. Consequently, the local bioactive drug concentration, influenced by dose, dose volume, and tumor volume, is expected to impact the level of systemic exposure. In our dog study, we utilized a weight-based dosing design since this has been used in prior canine cytokine studies (31). This resulted in a wide range of C_max levels and systemic exposure within and across dose cohorts, which, at least in part, can be attributed to wide variation in the dose/tumor volume ratio, a parameter we believe is important to consider for IT therapies. However, we did not observe a trend toward better response in dogs that achieved higher dose/TV ratio (Table 2 and Supplementary Fig. S8). In part, the absence for identifying higher dose/TV ratio to be predictive of response could have been confounded by multiple factors including differences in tumor porosity, heterogenous intratumoral biodistribution of JEN-101, and inaccurate tumor volumetric calculations (e.g. assumption that all tumors adopt spheroid growth).

There were several important limitations of this study. First, we were not able to define MTD for JEN-101 across the 4 dosing cohorts evaluated. However, for molecular therapeutics and immunotherapies, there has been increased awareness that MTD study designs might be suboptimal (e.g. Project Optimus), and dose optimization for drug delivery strategies inclusive of intratumoral immunotherapeutics are better suited for the identification of a minimum effective dose (MED). In our current study, we believe the MED for JEN-101 is less than 20 μg/kg, and most likely within the 3–10 μg/kg dose range, if a weight base dosing strategy is to be clinically adopted. This supposition is supported by several linked findings including objective responses across several dose cohorts (3–20 μg/kg), increasing ADA titers and higher IL-10 levels at the highest dose cohort (20 μg/kg), and near equivalence of IFN-γ levels elicited by JEN-101 at either 3 μg/kg or 20 μg/kg dosages. Second, the study was designed with the intention of using standard RECIST criteria, which are inadequate for evaluation of IT therapeutics (38), and response assessment was further complicated by the anatomic location of many of the treated tumors. Third, the study did not investigate specific cell subsets within the treated tumorous lesions by flow cytometry or lineage specific immunohistochemical staining, which would have provided more definitive characterization of infiltrating immune cells. In lieu of these proteomic methods in which validated canine reagents are limited (e.g. commercial antibodies that recognize NK cell subsets), we utilized RNA NanoString profiling to characterize the presence of different cellular subsets based on gene transcripts. Favorably and supporting our approach, recently published investigations with NanoString profiling have demonstrated good correlation between transcript profiles and immunohistochemical staining results in canine melanoma samples (39). Finally, this was an exploratory study and the number of subjects is small making broad conclusions difficult. Nonetheless, the study demonstrates the feasibility of using exploratory canine clinical trials to further validate potential human immuno-oncology agents.

In summary, anchored immunotherapy with JEN-101 was safe and well tolerated in dogs with advanced melanoma. There was evidence for JEN-101 to induce local and systemic immune activation and exertion of therapeutic activity across all doses tested based upon collective physical, radiological, immunologic, and genomic endpoints. Further studies are warranted to better define the clinical benefit of JEN-101 in dogs with melanoma and other solid tumors, not only as a single agent, but also in combination with therapies such as radiation and/or checkpoint blockade. Canine clinical studies represent an immunologically relevant and clinically feasible model for pre-clinical development of potential human intratumoral immunotherapy agents.