Phase II Trial of IL-12 Plasmid Transfection and PD-1 Blockade in Immunologically Quiescent Melanoma

This article is featured in Highlights of This Issue, p. 2769

Interleukin-12 (IL-12), a cytokine produced by activated innate immune cells such as macrophages and DCs (1), coordinates many facets of a productive immune response. It triggers T cell and NK cell activation (2), Th1 polarization of CD4⁺ cells (3), reduces regulatory T cell (Treg) activity (4), and, by a positive feed-forward loop with IFN-γ, establishes an inflammatory environment (5–7). Recently, we have shown that IL-12 production by intratumoral DCs in response to IFN-γ in the tumor microenvironment is required for successful anti-PD-1 immunotherapy (8). While systemic IL-12 has limited clinical efficacy and significant toxicity in patients (9, 10), preclinical (11), and clinical data (12, 13) show that intratumoral IL-12 (tavokinogene telseplasmid; “tavo”) leads to sustained local expression of IL-12 and can produce objective responses in both injected and uninjected tumors. This is observed even at distant sites, suggesting that intratumoral tavo electroporation (i.t.-tavo-EP) can induce productive immune responses systemically (12).

On the basis of these data, we hypothesized that i.t.-tavo-EP could increase immune infiltration in patients with nonimmune-infiltrated, or “cold,” melanoma, making these tumors more responsive to the anti-PD-1 antibody pembrolizumab and yielding more frequent and sustained clinical responses. To this end, we conducted a phase II clinical trial of i.t.-tavo-EP and the anti-PD-1 antibody pembrolizumab in patients who are unlikely to respond to anti-PD-1 antibody monotherapy, as defined by an absence of intratumoral “exhausted” or checkpoint-positive cytotoxic lymphocytes (cpCTL) in the tumor microenvironment (14). In addition to clinical response data, we also interrogated clinical samples from study patients to investigate associated immunologic changes in the tumor microenvironment and periphery.

This multi-center, phase II, open label, single-arm trial in individuals with a low frequency of tumor-infiltrating lymphocyte (TIL) melanoma was registered at cancer.gov as NCT02493361. The trial was conducted in accordance with International Conference on Harmonization guidelines for Good Clinical Practice and the Code of Federal and guided by the ethical principles of the Belmont Report. The protocol was approved by each local Institutional Review Board. All patients provided written, informed consent at the time of screening.

Anti-PD1 antibody experienced and naïve adults (≥18 years old) with metastatic or unresectable melanoma with accessible lesions were eligible. We have previously described how the proportion of CD8⁺ tumor-infiltrating lymphocytes (TIL) coexpressing the exhaustion markers CTLA-4^hi and PD-1^hi, termed partially exhausted CTLs (peCTL) or checkpoint-positive CTLs (cpCTLs), accurately predicts response to anti-PD-1 monotherapy in melanoma (14, 15). Melanoma tumors with CD8⁺ TILs consisting of less than 20% cpCTL are unlikely to respond (5.9%; 1 of 17), whereas patients with greater than 30% cpCTL are highly likely to respond (77.8%; 14 of 18). A fresh tissue biopsy was collected at screening and the frequency of CD8⁺CD45⁺ TIL that were PD-1^hiCTLA-4^hi was determined using flow cytometric analysis as described previously (15). Predicted PD-1 nonresponders were identified on the basis of a low proportion (<25%) of checkpoint-positive cytotoxic lymphocytes (cpCTL) or an absence of TIL infiltration at baseline. Individuals on immunosuppressive medication and patients with uveal melanoma, active central nervous system metastases, carcinomatous meningitis, or active autoimmune disease requiring immunosuppressive agents were excluded.

I.t.-tavo-EP (pIL12; Tavokinogene telseplasmid, tavo) was injected into tumors at ¼ tumor volume at a concentration of 0.5 mg/mL followed by in vivo electroporation of six pulses at field strengths (E+) of 1,500 V/cm and pulse width of 100 μs at 300-millisecond intervals (i.t.-tavo-EP). A minimum of 0.1 mL tavo was injected per lesion for lesions < 0.1 cm³. Pembrolizumab was administered as a flat dose of 200 mg i.v.

Individuals initiated the first cycle of i.t.-tavo-EP concurrently with the first injection of pembrolizumab, receiving both on day 1 of the first cycle. Pembrolizumab was administered on day 1 of each 3-week cycle. I.t.-tavo-EP was administered on days 1, 5, and 8 of every odd cycle (every 6 weeks; cycles 1, 3, 5, 7, etc.).

Treatment was continued for up to 2 years if individuals had stable disease or better and continued to benefit from treatment. Patients with unconfirmed disease progression were permitted to continue on treatment until disease progression was confirmed. Treatment with i.t.-tavo-EP was discontinued if there was regression of all superficial, accessible lesions, in which case patients continued with pembrolizumab monotherapy for up to 2 years.

Best overall response rate (BORR) was evaluated using RECIST v1.1 and clinical assessment by investigator evaluation at each restaging assessment performed every 12 weeks (Fig. 1, rBOR and Fig. 2 RECIST). Duration of response (DOR), progression-free survival, overall survival, and overall objective response rate evaluated by RECIST was also assessed. In addition to scoring responses by RECIST v1.1, we also scored responses by investigator clinical assessment, which was defined as ≥50% decrease in the sum of bidimensional measurements of all measured lesions, and latent responses compared to baseline (after RECIST progression or pseudoprogression) were included (Fig. 1, cBOR and Fig. 2, clinical assessment).

Safety was assessed during the study by documentation of adverse events (AE) according to CTCAE 4.0.

After the end of treatment, individuals were followed for 30 days for adverse event monitoring (90 days for serious adverse events). For subjects who discontinued for reasons other than progressive disease, every attempt was made to continue monitoring their disease status using radiologic imaging. Monitoring was continued until the start of a new anticancer treatment, documented disease progression, or death, whichever occurred first. Radiographic imaging in follow-up was performed as clinically indicated or per local standard of care. Individuals were followed for overall survival until death, withdrawal of consent, or the end of the study, whichever occurred first.

Tumor tissue procured by core-needle or punch biopsy was fixed in formalin-filled containers for 24–48 hours before being transferred into PBS. One biopsy per tissue container was stored at 4°C until shipped to Cureline Inc. for further processing. Specimens were embedded in paraffin and 8 × 10 μm tissue curls were cut by Cureline Inc. and shipped to OncoSec Medical. Upon delivery, tissue curls were stored at −80°C until RNA extraction.

Paraffin was removed from FFPE/formalin-fixed paraffin-embedded (FFPE) tissue curls prior to RNA isolation using the protocol in RecoverAll Total Nucleic Acid Kit (catalog no. AM1975). Tissue curls were treated with xylene and heated at 50°C for (3) minutes. After xylene incubation, an equal part of ethanol was added then spun down. Pellet was then resuspended in ethanol twice to wash. After centrifugation, ethanol was allowed to evaporate at 37°C. Following paraffin removal, the RNA was isolated following the RecoverAll protocol. Briefly, proteinase k was incubated with the resuspended pellet for 1 hour at 55°C. After incubation, the solution was transferred into a PureLink spin column with Isolation Additive and flow through was used in further steps for RNA isolation. Samples were washed twice and then DNase was added to the column for 15–30 minutes. An additional three washes were performed prior to addition of elution buffer. RNA was collected from the spin column following a 1-minute incubation and the elution step was repeated with fresh buffer. Samples were measured on a NanoDrop 2000c spectrophotometer to test for purity and concentration. RNA was stored at −80°C until use.

Approximately 60 mL of venous blood was collected in sodium heparin vacutainer tubes (BD Biosciences, catalog no. 367874) and shipped overnight from the clinical site to OncoSec Medical at ambient temperature. Heparin tubes were immediately processed for peripheral blood mononuclear cells (PBMC) using standard Ficoll-Pacque (GE) gradient extraction as described previously (1). PBMCs were frozen in CryoStor 10 (BioLife Solutions) and stored in liquid nitrogen until use.

Approximately 60 mL of venous blood was collected in BD Vacutainer CPT tubes (BD Biosciences, catalog no. 367874) and spun at 1,500 to 1,800 RCF for at least 15 minutes within 2 hours of blood collection. Spun tubes were then shipped overnight from the clinical site to OncoSec Medical at ambient temperature. The top portion of the tubes was removed and washed twice in DPBS (Gibco) and PBMCs frozen in CryoStor 10 (BioLife Solutions) and stored in liquid nitrogen until use.

Approximately 8.5 mL of veinous blood was collected in PAXgene DNA tubes (PreAnalytix) and shipped overnight from the clinical site to OncoSec Medical at ambient temperature. Samples were then allowed to cool to 4°C before storage at −80°C.

Total RNA was used with the nCounter system, according to the manufacturer's protocol (NanoString Technologies). In brief, 100 ng of total RNA was hybridized at 96°C overnight with the nCounter (Human Immunology v2 Gene Expression Panels). This panel profiles approximately 600 immunology-related human genes as well as internal controls. Hybridized samples were then digitally analyzed for frequency of each RNA species using the nCounter SPRINT profiler. Raw mRNA abundance frequencies were quantified using the nSolver analysis software 3.0 pack. Data was normalized to control genes. Data were excluded if binding density, positive controls, or normalization factors were outside of the acceptable ranges set by NanoString.

Normalized counts of RNA transcripts from patient biopsies were determined and expression was reported as relative to the average pre-treatment counts. These values were graphed in GraphPad Prism 7.03 and statistical analysis was determined as detailed in the statistical section below.

FFPE Tissue DNA Isolation: Paraffin was removed from PFPE tissue curls prior to DNA isolation. Tissue curls were treated twice with xylene. After last xylene incubation, equal parts of ethanol were added then spun down. The pellet was then resuspended in ethanol to wash. After centrifugation, ethanol was allowed to evaporate for 5 minutes at 37°C. Following paraffin removal, the protocol in Qiagen PAXgene Tissue AllPrep DNA/RNA/miRNA Kit (Venlo, catalog no. 80224) was followed. Briefly, proteinase k was incubated with the resuspended pellet, shaking, for 2 hours at 45°C. After incubation, the solution was transferred into a PAXgene DNA spin column and washed once. An additional proteinase K incubation occurred on the spin column, followed by a wash and drying of the column membrane. Resuspension buffer containing DNA was collected from the spin column following a 5-minute incubation and measured on a NanoDrop 2000c spectrophotometer to test for purity and concentration. DNA was stored at −80°C until use.

Genomic DNA (gDNA) was isolated using the spin protocol in Qiagen's DNeasy Blood and Tissue Kit (Venlo, catalog no. 69504) using no more than 5 × 10⁶ cells per column. Briefly, previously separated and frozen PBMCs were thawed and incubated with proteinase K for 10 minutes at 56°C, after which ethanol was added. The entire sample was transferred to a DNeasy spin column and washed twice. Resuspension buffer was run through the column twice before resuspended DNA was measured on a NanoDrop 2000c spectrophotometer to test for purity and concentration. Samples were stored at −80°C until use.

Data was made available through the immunoSEQ Analyzer (Adaptive Biotechnologies). Differential abundance was calculated with 99% confidence through immunoSEQ to determine the number of clones differentially expressed between patient samples at screening and posttreatment.

PBMCs isolated as previously described from blood samples obtained from patients were analyzed for immune cell subsets by flow cytometry. Frozen PBMCs were thawed quickly by the addition of warm (37°C) X-Vivo 15 culture media (Lonza, catalog no. 04-418Q) containing IL-2 at 1,000 CU/mL (6,000 IU/mL). Cells were rested overnight at 37°C in humid conditions and 5% CO₂ as needed. Cells were stained for viability using BD HorizonTM Fixable Viability Stain 780 for 30 minutes at 4°C. Cells were stained for surface markers by incubating in the presence of fluorescence-conjugated antibody (Supplementary Table S1) below) for 30 minutes at 4°C. Intracellular staining was carried out by first incubating cells in 1× transcription factor fixation/permeabilization solution (BD Biosciences, catalog no. 562574) according to the manufacturer's protocol to fix and permeabilize the cells. Antibodies for intracellular staining were then added to cells. After a 45-minute incubation at 4°C, cells were washed, resuspended in FACS buffer and acquired on an LSRFortessa X-20 flow cytometer (BD Biosciences).

Tissue sections were cut at 4 μm from FFPE blocks. All the sections were deparaffinized and subjected to heat-induced epitope retrieval in citrate buffer pH 9.0 (Biogenex). Six-plex panel IHC was performed for each tissue slide using the following antibodies: anti-FoxP3 (clone 236A/E7, dilution 1:100, Abcam), anti-PD-L1 (clone E1L3N, dilution 1:250, Cell Signaling Technology), anti-CD8 (clone SP16, dilution 1:50, Spring Bioscience), anti-CD3 (clone SP7, dilution 1:50, Spring Bioscience), anti-CD163 (clone MRQ26, Ventana), anti-Cytokeratin (clone AE1/AE3, dilution 1:100, DAKO). Antigen–antibody binding was visualized with TSA-Cy5 (PerkinElmer), TSA-Cy3 (PerkinElmer), TSA-FITC (PerkinElmer), TSA-Alexa594, TSA-Cy5.5 (PerkinElmer), and TSA-Coumarin (PerkinElmer), respectively.

Digital images were captured with PerkinElmer Vectra platform. Tumor areas with the highest immune cell (CD3⁺CD8⁺) infiltrates were scanned at 20X and selected for analysis in a blinded fashion. Three images of 0.36-mm² each were analyzed per sample with InForm Software (PerkinElmer). Hematoxylin and eosin staining was performed on for each sample and reviewed by a pathologist to ensure a representative tissue sample.

Punch or core biopsies were taken from treated lesions up to two weeks before treatment and at different time points on therapy. Untreated biopsies were also taken after treatment. Biopsies were FFPE, and cut into 5-μm sections. These FFPE of PFPE sections were immunostained for CD8, clone 144B, (Dako, catalog no. M7103) using methods established by PhenoPath for use in clinical diagnostics and for PD-L1, clone 22C3, (Dako, catalog no. SK006) per kit instructions. Whole slide digital images were captured using Aperio imaging software. All slides were evaluated by a board-certified pathologist and blindly scored for PD-L1 on a scale of 0–5 per the melanoma scoring guidelines listed below. Hematoxylin and eosin (H&E) staining was also performed for morphologic assessment and tumor location. A score of 0 = no staining; 1 = 0% to < 1%; staining; 2 = 1% to < 10% staining; 3 = 10% to < 33% staining; 4 = 33% to < 66% staining; 5 = ≥ 66%.

The primary endpoint was best overall response rate (BORR), CR + PR as determined by RECIST v1.1. Progression-free survival (PFS) was defined as the duration between the date of the first date of treatment on study to the first date of either disease progression or death; Overall survival was defined as was defined as the duration between the date of first treatment to the time of death. Individuals with measurable disease at baseline, which was defined separately under investigator evaluation, who received at least one dose of i.t.-tavo-EP and pembrolizumab study treatment, were included in the full analysis set population (FAS).

Analysis was performed using the R language. Progression-free survival (PFS) and overall survival (OS) were estimated using the method of Kaplan–Meier. To obviate the need to make distributional assumptions, statistical significance was assessed using rank-based methods. In particular, the Mann–Whitney test was used to for unpaired analysis and the Wilcoxon signed rank test was used for paired analysis between pretreatment and posttreatment. Exact tests using the Wilcoxon. exact function in the package exact RankTests were performed due to small sample sizes. Tests with P values less than 0.05 were considered significantly significant. For the gene expression analysis depicted in Fig. 4, raw gene scores (NanoString Inc) were analyzed using a negative binomial generalized linear model with individual level random effect adjustment for pairing. The raw gene level P values were adjusted for multiple comparisons using the Benjamini–Hochberg method (Fig. 4A) or the Wilcoxon matched-pairs signed rank test (Fig. 4B–D).

This study was a single-arm combination study with a historical comparator, thus the sample size depended on the presumed “baseline pembrolizumab ORR” in the biomarker-selected population, as defined by the low TIL status (a cutoff of 25% cpCTLS was chosen in the study) on the screening flow cytometric assay measures. The predicted response rate of patients in this population was estimated at 12.5% (based on RECIST1.1) based on analysis of melanoma patients at University of California, San Francisco treated with pembrolizumab as a monotherapy who had fewer than 25% cpCTLs prior to the initiation of treatment. We posited that ORR of >30% in patients treated with pIL-12 EP combined with pembrolizumab would be clinically meaningful. Therefore, inclusion of 42 individuals would meet the following statistical parameters: H₀ ORR = 12.5%; H_alt ORR = 30%; alpha of 0.025 (actual; one-sided test); power = 85%.

Interim safety and efficacy assessments were performed after 23 evaluable were enrolled. The interim results showed the ORR rate, 41% by RECIST (and 50% by clinical assessment) with 36% complete responses, far exceeded the projected response rate in the hypothesis. The investigators stopped accrual and a larger confirmatory trial was initiated.

A total of 41 patients were screened for this trial (study schema, Fig. 1A and study design Fig. 1B), and 23 patients, including 22 patients with RECIST measurable disease at baseline, were eligible to participate based on a CD8⁺ cpCTL (or peCTL) percentage of less than 25% in tumor biopsy specimens taken at screening (Fig. 1A). We also examined the trial population with previously described PD-1 predictive biomarker assays including PD-L1 IHC using the 22C3 mAb (16) and IFNγ gene expression signatures (17). Tumors evaluated by either method had a low likelihood of response (Fig. 2A; Supplementary Fig. S1). Baseline patient characteristics are described in Fig. 1C. Of note, 10 patients had prior exposure to anti-PD-1 antibodies and 7 patients had prior exposure to anti-CTLA4 antibodies.

The ORR was 48% (11/23) by clinical assessment (including 1 patient with RECIST nonmeasurable disease at baseline and 1 patient with a delayed response after progression) and 41% (9/22) by RECIST version 1.1 (Fig. 2A and B). Patients responses by RECIST and clinical criteria are shown in Fig. 2A. Responses were rapid and sustained (Fig. 2E). The median progression free survival by RECIST was 5.6 months (95% CI undefined). The median overall survival was not reached at a median follow up of 19.6 months. (Fig. 2C and D). These results are in contrast to those expected for predicted PD-1 nonresponders by the cpCTL assay (14). Objective responses were observed in patients with elevated LDH levels at baseline, prior exposure to anti-PD-1 antibodies, and in patients with stage IIIB to stage IVC disease (Fig. 1C). Clinical responses were observed in patients regardless of baseline TIL levels (Fig. 2A) or baseline PD-L1 protein levels by IHC (Fig. 3B). The median baseline interferon-γ expression signature of enrolled patients (Supplementary Fig. S1), 0.81, was > 1 SD below the mean reported for responders by Ayers and colleagues (17). Overall, 13 patients had untreated lesions measured at baseline and 10 of these patients had untreated RECIST target lesions at baseline. On a per lesion basis including both target and nontarget lesions (Supplementary Table S2), 29.2% of untreated lesions, including liver and lung metastases, decreased by > 30% in maximum diameter. The diameters of treated and untreated lesions were similar (mean = 2.6 ± 2.1 cm vs. 3.4 +/2.5 cm) as were the diameters of responding and progressing lesions (mean 3.4 ± 1.9 cm vs. 2.2 ± 2.1 cm). A patient with > 50 in-transit lesions at baseline and prior disease progression on ipilimumab, pembrolizumab, and nivolumab had a partial response after RECIST confirmed progression (Fig. 2F–I). In this patient, biopsied lesions posttreatment showed a siderotic scar with no evidence of residual melanoma by histopathology (Fig. 2I). In contrast to systemic IL-12 administration, the combination of tavo and pembrolizumab was well tolerated (Supplementary Table S3) and grade 3 or higher treatment associated toxicity was limited to treatment associated pain (n = 1), cellulitis (n = 1), chills (n = 1), and sweats (n = 1).

The presence of tumor-infiltrating CD8⁺ lymphocytes and high PD-L1 levels in the tumor microenvironment have been identified as dynamic predictors of response to PD-1 antibodies in patients with metastatic melanoma (18–20). The combination of i.t. tavo EP and pembrolizumab drove an increased density of CD8⁺ TILs (Fig. 3A, P = 0.027) as well as a significant increase in the number of unique intratumoral T cell clones (Fig. 5B, P = 0.041). In addition, significant increases in intratumoral PD-L1 expression by IHC (Fig. 3B, P = 0.016), transcriptomic analysis (Fig. 3C, P = 0.026), and systemic proliferating PD-1⁺ T cells (Fig. 5D, P = 0.012) were noted.

While multispectral IHC highlighted that i.t.-tavo-EP combination therapy can elicit an increase in adaptive tumoral and stromal PD-L1 expression in both responding and nonresponding patients (Fig. 3D), these groups differed in the relative balance of infiltrating immune sub-populations. In quantitative spatial analysis, nonresponding patients had more Tregs in close proximity to CD8⁺ T cells than responding patients (Fig. 3F, P = 0.018). This balance of effector to suppressive immune subsets could also be seen in the positive ratio of CD8⁺ T cells to M2 macrophages (PD-L1⁺/CD163⁺) in responding patients (Fig. 3G, P = 0.011).

A critical hypothesis of this study was that IL-12 licenses the tumor immune environment by multiple mechanisms and increases the likelihood of effective anti-PD-1 therapy. Increased intratumoral transcription of an immune-based gene set following treatment was observed with i.t.-tavo-EP plus pembrolizumab (Fig. 4A). Effective anti-PD-1 therapy has also been shown to require the presence of cross-presenting cDC1 dendritic cells (20). In a pooled analysis of all patients at cycle 2 (week 3) of concurrent i.t.-tavo-EP + pembrolizumab, we found that cDC1-associated antigen presentation genes were also significantly increased intratumorally following combined treatment (Fig. 4B). In addition to antigen cross-presentation, expression of genes associated with T cell activation and adaptive resistance (Fig. 4C) as well as T cell trafficking and costimulation (Fig. 4D) was also significantly increased, similar to what Garris and colleagues recently reported (8). An increase in natural killer (NK) cell–associated genes was also noted (Fig. 4B). Transcriptome analysis highlighted treatment-related changes in relevant cell type scores, which were particularly evident when separated by clinical response (Supplementary Fig. S2).

IFN response factor-1 (IRF-1) is a transcription factor produced in response to IFNγ, linking stimuli to immune evasion via PD-L1 expression (21). Because the transcriptional analysis demonstrated that responding patients had a treatment-related increase in the expression of PD-L1 (Fig. 3B and C), we examined intratumoral mRNA levels of IRF-1 and PD-L1 in i.t.-tavo-EP + pembrolizumab treated patients. As shown in Fig. 5E the correlation between IRF-1 and PD-L1 became significant with treatment, suggesting that the “capability” of tumor/immune cells to respond to IFNγ via IL-12 may factor into clinical response with this combination.

To understand how an intratumoral therapy can drive systemic immune responses, the architecture and context of immune infiltration in both electroporated and nonelectroporated tumors were examined by multispectral IHC (mIHC). As shown in a sample of nonresponding and responding patients (Fig. 5A), intense infiltration was seen at cycle 2 (week 3) in all interrogated lesions. This was the case in patients with progressive disease as well as in patients with objective responses. Regardless of electroporation, the composition of the infiltrate was a key determinate of clinical response as seen in the ratio of CD8⁺ cells:M2 macrophages (Fig. 5A); similarly observed with the entire cohort (Fig. 3G). Additional support for combination therapy initiating systemic immunity beyond regression of nonelectroporated lesions was evident in the treatment-related increase of activated immune cells measured in the periphery. In particular, combination of i.t.-tavo-EP and pembrolizumab resulted in a significant increase in proliferating T_emra cells in peripheral blood (Fig. 5C, P = 0.008).

IL-12 is a pivotal type 1 immune regulatory cytokine (22, 23), which induces a positive feedback loop inducing T cells and NK cells to produce IFNγ that in turn primes DCs for IL-12 production (7). Despite its importance, effective strategies for increasing IL-12 in immunologically quiescent tumors have been lacking. Previously, we have demonstrated that immunologically relevant temporal intratumoral expression of IL-12 using gene electroporation can convert non-inflamed to inflamed Th1-polarized tumors in animal models (24, 25).

Given these data, we administered i.t.-tavo-EP and pembrolizumab concurrently to patients determined to be unlikely to respond to anti-PD1 monotherapy based on a low frequency of cpCTL in the tumor microenvironment (26–31). Clinically, a RECIST objective response rate of 41% was observed with 36% complete responses in this poorly inflamed, “cold” melanoma population with a predicted anti-PD-1 antibody response rate of < 12.5% based on the cpCTL assay and a low probability of response based on PDL-1 IHC (16) and IFNγ scores (17). Notably, 42% of patients had previously experienced disease progression on anti-PD-1 antibodies and the consistency between these clinical and immunologic negative predictors of response to anti-PD-1 antibodies suggests that the overall population was unlikely to respond. Our earlier work suggested that the duration of clinical responses to i.t.-tavo-EP monotherapy might be limited by the induction of adaptive resistance (13). In this study, analysis of serial biopsy specimens suggests substantial mechanistic interplay whereby pembrolizumab limits i.t.-tavo-EP–induced adaptive resistance, supporting the maintenance of the IL-12/IFNγ feedforward loop, which is fundamental for DC maturation and the resulting adaptive cellular responses. Reshaping of immunologically inert tumors was evident after 1 cycle of treatment with a global increase in immune-associated gene expression, including key transcripts associated with innate and adaptive immune responses. Specifically, we noted increases in antigen presenting cell (APC) and NK-cell–associated transcripts, inflammatory cytokines and chemokines, and genes related to adaptive cellular immunity including T cell activation, costimulation, and adaptive resistance. Marked ingress of CD8⁺ T cells, measured by mIHC and emerging TCR clones, were all consistent with productive immune responses. This treatment-related change in the tumor microenvironment could be seen in both i.t.-tavo–EP injected and in uninjected tumors. When combined with the supportive peripheral NK and T cell phenotypic changes, these data collectively demonstrate promotion of a systemic anti-tumor immune response. The sustained, systemic effects of i.t.-tavo-EP in combination with pembrolizumab observed here are in contrast to previous studies of systemic IL-12, which induces objective responses in only 3–5% of patients and which is associated with high rates of severe side effects (9).

While the safety profile, clinical responses, and concordant translational data obtained in patients treated with i.t.-tavo-EP and pembrolizumab are highly encouraging in responding patients, further studies will be useful to understand mechanisms of resistance in nonresponding patients. In particular, the impact of differences in HLA variability, tumor antigenicity (32, 33), disregulation of WNT–β-catenin pathway (34, 35), PTEN loss (36), p53 loss (37), and deletional mutations in the JAK1/2–STAT (38) signaling pathway may influence responses to i.t.-tavo-EP. In addition, this study is limited to patients with superficial, accessible disease. While this current applicator can only treat tumors to a depth of 1.5 cm, a novel lesion applicator able to deliver plasmid to deep visceral organs is under development and could broaden the spectrum of treated tumors.

In summary, we show that i.t.-tavo-EP, combined with pembrolizumab in cpCTL poor melanoma can result in deep, durable, and frequent clinical responses, including responses in patients who have failed previous anti-PD-1 therapy. This approach may help to induce remissions in patients who are clinically refractory to anti-PD-1 therapies. A confirmatory trial to establish the therapeutic value of combining i.t.-tavo-EP and anti-PD-1 antibody therapy in anti-PD-1 relapsed/refractory patients with stage III/IV metastatic melanoma is currently underway with encouraging preliminary data to date.

A.P. Algazi is a paid consultant for OncoSec Medical, Inc., Array, Regeneron, Valitor Biosciences, reports receiving other commercial research support from Acerta, BMS, Merck, AstraZeneca, Incyte, Tessa, OncoSec, Dynavax, Genentech, Sensei, Amgen, and Medimmune, holds ownership interest (including patents) in OncoSec Medical, Inc. and Valitor Biosciences, and is an advisory board member/unpaid consultant for Sensei and Merck. C.G. Twitty is an employee/paid consultant for and holds ownership interest (including patents) in Oncosec, K.K. Tsai reports receiving other commercial research support from Oncosec and Regeneron. M.H. Le is an employee/paid consultant for Oncosec Medical, Inc. R.H. Pierce is an employee/paid consultant for OncoSec Medical. E. Browning is an employee/paid consultant for and holds ownership interest (including patents) in OncoSec Medical, Inc. R. Hermiz, D.A. Canton and D. Bannovong are employees/paid consultants for OncoSec. C.B. Bifulco holds ownership interest (including patents) in PrimeVax, is an advisory board member/unpaid consultant for PrimeVax, HalioDx, and BMS, and is listed as an inventor on a United States patent application for an image processing system and method for displaying multiple images of a biological specimen. S. Bhatia is an employee/paid consultant for BMS, EMD Serono and Sanofi-Genzyme, and reports receiving commercial research grants from BMS, Merck, Oncosec, Nantkwest, Exicure, Novartis, Immune Design and EMD Serono. R.H. Andtbacka is an employee/paid consultant for Merck, Novartis, Amgen, Pfizer and Seven and Eight Biopharmaceuticals Inc., and holds ownership interest (including patents) in Seven and Eight Biopharmaceuticals Inc. B.A. Fox is an employee/paid consultant for Definiens/AstraZeneca, UltiVue and PerkinElmer, reports receiving commercial research grants from Akoya/PerkinElmer and OncoSec, and is an advisory board member/unpaid consultant for Akoya. M. Rosenblum is an employee/paid consultant for TRex Bio and Sitryx Bio. A.I. Daud reports commercial research funding from Merck, BMS, Roche, Pfizer, Incyte, and Novartis, ownership interest (including patents) in SQZ bio, Trex Bio, and OncoSec, and advisory/consultant relationships with Xencor, Checkmate, Incyte, Array, and Genentech. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.P. Algazi, C.G. Twitty, M.H. Le, R.H. Pierce, S. Bhatia, R.H.I. Andtbacka, A.I. Daud

Development of methodology: A.P. Algazi, C.G. Twitty, R.H. Pierce, C. Ballesteros-Merino, S. Bhatia, B.A. Fox, A.I. Daud

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.P. Algazi, C.G. Twitty, K.K. Tsai, M.H. Le, R.H. Pierce, E. Browning, R. Hermiz, A. Oglesby, M. Francisco, C.B. Bifulco, C. Ballesteros-Merino, S. Bhatia, S. Gargosky, R.H.I. Andtbacka, B.A. Fox, M. Rosenblum, A.I. Daud

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.P. Algazi, C.G. Twitty, R.H. Pierce, E. Browning, R. Hermiz, D.A. Canton, C.B. Bifulco, S. Bhatia, S. Gargosky, R.H.I. Andtbacka, B.A. Fox, M. Rosenblum, L. Fong, M.J. Pittet, S.P. Arlauckas, C. Garris, L. Levine, A.I. Daud

Writing, review, and/or revision of the manuscript: A.P. Algazi, C.G. Twitty, K.K. Tsai, E. Browning, R. Hermiz, D.A. Canton, D. Bannavong, C.B. Bifulco, S. Bhatia, S. Gargosky, R.H.I. Andtbacka, B.A. Fox, L. Fong, M.J. Pittet, S.P. Arlauckas, C. Garris, L. Levine, A.I. Daud

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.P. Algazi, C.G. Twitty, K.K. Tsai, D. Bannavong, A. Oglesby, S. Gargosky, M. Rosenblum, A.I. Daud

Study supervision: A.P. Algazi, M.H. Le, S. Gargosky, R.H.I. Andtbacka, A.I. Daud

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.