Neoadjuvant Intravenous Oncolytic Vaccinia Virus Therapy Promotes Anticancer Immunity in Patients

Patients with advanced solid malignancies can undergo surgical resection of their metastatic disease with curative intent. However, only a minority of these patients remain cancer-free long term, due to either incomplete tumor resection or the presence of micrometastatic disease at the time of surgery. For patients with colorectal liver metastases (CRLM), survival at 5 years following liver resection is approximately 50%, despite the availability of combination perioperative chemotherapy (1). Likewise, approximately 60% of patients with melanoma that has spread to lymph nodes can relapse following surgical resection and adjuvant treatment (2). For these patients, effective novel systemic therapies administered prior to (neoadjuvant therapy), or shortly after surgery (adjuvant therapy), hold the potential to significantly enhance the chances of tumor clearance.

Oncolytic viruses (OV) are principally immunotherapeutic viruses that preferentially replicate in malignant cells, thereby inducing immunogenic cell death. Several engineered viruses have reached randomized studies, with three agents currently licensed for standard care (3). One of the most clinically tested oncolytic viruses, Pexa-Vec (Pexastimogene Devacirepvec; JX-594, TG6006), is an engineered Wyeth-strain vaccinia virus (4) under development by Transgene (5) and SillaJen (6). Pexa-Vec tumor specificity is enhanced by deletion of thymidine kinase, an enzyme of the DNA precursor pathway, which is strictly regulated during the normal cellular cycle, but highly expressed in growing malignant cells (7). Pexa-Vec also expresses GMCSF, which promotes antitumor immunity by inducing proliferation and differentiation of myeloid precursors, alongside the stimulation, recruitment, and maturation of dendritic cells (DC; refs. 8, 9). Clinical and in vitro studies have helped to elucidate the fundamental mechanisms of Pexa-Vec therapy, namely tumor-specific virus replication, expression of GMCSF, and the stimulation of CTL tumor infiltration (10). Other mechanisms of therapy include antibody-mediated, complement-dependent cancer cell cytotoxicity (11) and Pexa-Vec replication in tumor-associated endothelial cells, leading to disruption of tumor blood flow, hypoxia, and necrosis (12).

Pexa-Vec has shown promising clinical signs of efficacy as a single agent, including in a randomized study between low-dose (1 × 10⁸ plaque-forming units; p.f.u) and high-dose (1 × 10⁹ p.f.u) intratumoural injection in patients with hepatocellular carcinoma, where overall survival was significantly longer for patients in the high-dose group (14.1 vs. 6.7 months; ref. 13). Furthermore, intravenous delivery of Pexa-Vec to tumors, a critical feature for the treatment of micrometastatic disease, is achievable using a dose of 1 × 10⁹ p.f.u (14).

Therefore, we sought to clinically develop intravenous Pexa-Vec delivery as a neoadjuvant, multimechanistic therapy for patients with metastatic solid malignancies. Herein, we showed in 9 patients that the administration of a single intravenous infusion of Pexa-Vec ahead of planned surgical resection of advanced CRLM or metastatic melanoma, resulted in acceptable patient safety and pathologic evidence of tumor necrosis. Pexa-Vec associated with the plasma compartment of peripheral blood, resulting in delivery of virus to tumor and promotion of innate anticancer immunity. Furthermore, a single neoadjuvant infusion of Pexa-Vec stimulated long-lived T-cell anticancer immune responses, with repertoire sequencing suggesting that pathologic tumor response associated with a perceived reduction in global T-cell diversity. These results support the continued development of Pexa-Vec and other OVs as perisurgical treatments, to increase the probability of tumor clearance.

All data and materials are available by request from the corresponding author.

The study was approved by the National Research Ethics Service (NRES) Committee London—West London and the Gene Therapy Advisory Committee (GTAC) approval number 101521. All patients granted explicit written consent for clinical trial participation.

The trial, EudraCT (number 2012-000704-15) was open between September 2015 and June 2018 in accordance with Declaration of Helsinki ethical guidelines. This was an open-label, nonrandomized study of Pexa-Vec given as a 1-hour intravenous infusion to patients prior to planned surgical resection of tumor. The trial inclusion and exclusion criteria were as follows:

(i) Patients with histologically proven or radiological findings consistent with locally advanced/poor prognosis or metastatic cancer, planned for surgical resection (curative or palliative) of primary or metastatic disease as part of standard clinical care. Patients with the following diseases were eligible: (a) metastatic melanoma due for lymph node dissection for lymph node macrometastases (stage IIIB/C) or metastasectomy at any other site; (b) muscle-invasive transitional cell bladder cancer suitable for surgical resection, that is, partial or total cystectomy; (c) primary hepatocellular carcinoma, planned for liver resection; (d) colorectal cancer with liver metastases planned for metastasectomy or liver resection; (e) locally advanced/metastatic renal cell cancer planned for palliative nephrectomy. (ii) Patients were willing to have full preoperative work-up prior to planned resection, consistent with standard clinical practice appropriate for the disease site/surgical intervention, (iii) were fit for the planned surgical intervention, (iv) had a life expectancy of at least 3 months, (v) were at least 18 years of age, (vi) had an Eastern Cooperative Oncology Group performance status score 0–2, (vii) hemoglobin ≥ 9 g/dL (correction with transfusion allowed), (viii) platelets ≥ 100 × 109 (without platelet transfusion), (ix) absolute neutrophil count ≥ 1.5 × 10⁹/L, (x) total bilirubin ≤ 1.5 × upper limit of normal (ULN), (xi) aspartate aminotransferase and/or alanine aminotransferase ≤ 2.5 × ULN, (xii) INR ≤ 1.7, and (xiii) serum creatinine ≤ 1.5 × ULN. (xiv) For patients who are sexually active, able, and willing to abstain from sex for 3 weeks and then use a barrier method of contraception for a further 3 weeks after the JX-594 treatment. A negative pregnancy test was required within 24 hours of treatment if patients were a female of child-bearing potential. (xv) Patients were willing and able to provide informed consent and willing and able to comply with scheduled visits, the treatment plan, and laboratory tests.

Patients who: (i) were pregnant or breast-feeding an infant, (ii) were on antiviral therapy, immunotherapy, had known HIV infection or hepatitis B/C, (iii) had clinically significant active infection or uncontrolled medical condition (e.g., pulmonary, neurological, cardiovascular, gastrointestinal, genitourinary), such that the patient was unfit for surgery or results interfered with interpretation of the trial, (iv) had known CNS malignancy (including history of brain metastases completely resected or treated by gamma knife therapy or whole brain radiotherapy), (v) had clinically significant and reaccumulating ascites, pericardial and/or pleural effusions, (vi) had tumor(s) in a location that would potentially result in significant clinical adverse effects if posttreatment tumor swelling were to occur (e.g., tumors causing near total blockage of the common bile duct), and (vii) received anticancer therapy (e.g., chemotherapy, surgery, radiotherapy, investigational agent) within 4 weeks prior to treatment. Patients were also excluded if: (viii) they experienced a severe systemic reaction or side effect as a result of a previous smallpox vaccination (previous smallpox vaccination is not required), (ix) had a history of exfoliative skin condition (e.g., eczema or ectopic dermatitis) requiring systemic therapy, (x) used of IFN/pegylated IFN or ribavirin that could not be discontinued within 14 days prior to any JX-594 dose, (xi) had active cardiovascular disease, including but not limited to significant coronary artery disease (e.g., requiring angioplasty or stenting) or congestive heart failure within the preceding 12 months (the potential risk of hypotension after administration of JX-594 in conjunction with any underlying cardiac disease should be taken into consideration), (xii) inability to suspend treatment with antihypertensive medication if necessary (including but not limited to: diuretics, beta-blockers, angiotensin converting enzyme inhibitors, aldosterone antagonists, etc.) for 48 hours prior to and 48 hours after JX-594 treatment, (xiii) pulse oximetry O₂ saturation less than 90% at rest on room air, and (xiv) previous treatment with JX-594 or other vaccinia vector–based treatment.

A total of 9 patients each received a single dose of 1 × 10⁹ p.f.u. Pexa-Vec via intravenous infusion 10 to 22 days prior to surgery at St. James's University Hospital (SJUH), Leeds, United Kingdom. Three patients had metastatic melanoma and 6 had CRLM. Eight patients completed their planned surgery, and 1 patient (patient 07) had surgery cancelled when an up-to-date CT scan revealed pulmonary metastases. The primary end point of the study was the presence of Pexa-Vec in resected tumor tissue and blood. All patients gave written, informed consent according to good clinical practice guidelines. Protocol, patient information sheet, and consent forms were approved by the United Kingdom Medicines and Healthcare products Regulatory Authority, regional ethics review committee, as well as institutional review at SJUH. The trial management committee met on a monthly basis to discuss study progress, including patient safety and adverse events.

Pexa-Vec (Pexastimogene Devacirepvec; JX-594, TG6006), a replication-competent, transgene-armed therapeutic vaccinia virus, and Pexa-Vec-GFP were provided by Transgene S.A, France. Pexa-Vec is engineered for viral thymidine kinase gene inactivation and expression of GMCSF and β-galactosidase transgenes under the control of the synthetic early-late and p7.5 promoters, respectively. Pexa-Vec was stored at 1 × 10⁹/mL at −80°C for use in in vitro experiments.

Blood and tissue samples (metastatic lesions or lymph nodes, where applicable) were collected, processed, and analyzed using the Translational Cancer Immunotherapy Team quality-assured lab manual, which included SOPs to standardize all processes. Peripheral blood was collected into K₃EDTA and serum clot activator vacuette tubes (both Greiner) and processed within 2 hours of venepuncture. Blood samples were taken on day 1 (pre- and 1-hour postinfusion), day 2, day 3 (optional), day 5 (optional), on the day of surgery, 1-month postsurgery, and 3 months postsurgery. Tumor (and corresponding normal margins, if available) were taken from planned surgical resections.

All blood processing steps were performed at ambient temperature or at a controlled temperature of 20°C during centrifuge spins.

K₃EDTA blood was centrifuged for 10 minutes at 2,000 × g and plasma was harvested from the resulting upper layer. Aliquots were stored at –80°C.

Peripheral blood mononuclear cells (PBMC) were isolated from K₃EDTA blood by density-gradient separation over lymphoprep (Stemcell Technologies) as per manufacturer's instructions. Cells were frozen at 1 × 10⁷/mL in 40% [volume for volume (v/v)] RPMI1640 (Sigma) containing 5 mmol/Ll-glutamine (Sigma), 1 mmol/L sodium pyruvate (Sigma), 50% (v/v) pooled human serum (HS; BioIVT), and 10% (v/v) dimethyl sulphoxide (DMSO; Sigma). Where cultured, PBMCs were placed in RPMI containing 5 mmol/L l-glutamine and supplemented with 10% (v/v) FCS (Invitrogen). PBMCs were stored in liquid nitrogen.

Blood collected in clot activator tubes was left to clot for a minimum of 30 minutes post-venepuncture and then centrifuged at 2,000 × g for 10 minutes. Serum was harvested from the resulting upper layer. Aliquots were stored at −80°C.

Full blood counts were performed, where appropriate, at SJUH as part of standard clinical care. The Patient Pathway Manager and Results Server systems (Leeds Teaching Hospitals Trust in-house electronic health record system for West Yorkshire and Humber region, United Kingdom) were used to acquire total lymphocyte counts (expressed as 10⁹/L) throughout treatment. Normal ranges of lymphocytes were defined by SJUH as 1–4.5 × 10⁹/L.

IHC was performed on formalin-fixed paraffin-embedded tissue obtained from surgical resection of patient tumors or lymph nodes. Tissue for IHC was processed using an automated Bond Max system (Leica Biosystems) as described in ref. 15). IHC for Pexa-Vec was performed by Histalim (France) following company validated protocols using 1:200 dilution of polyclonal rabbit-anti-vaccinia virus antibody (Meridian Life Sciences; RRID: AB_153134). Mouse anti-human CD8 antibody (Dako; RRID: AB_2075537) was used at 1:100 dilution, followed by anti-mouse secondary (Abcam; RRID: AB_10680417) at 1:500. CD8 positivity was detected using the ImmPACT 3,3′-Diaminobenzidine horseradish peroxidase (HRP) Substrate Kit (Vector Labs; RRID: AB_2336520) or ImmPACT Vector Red (Vector Labs; RRID: AB_2336524). Antibodies against tumor-associated antigens (TAA) were used as follows: anti-human carcinoembryonic antigen (CEA; RRID: AB_304463) at 5 μg/mL, anti-human Melan-A (RRID: AB_305836) at 1:50 dilution, both with an anti-mouse secondary antibody at 1:2,000 dilution (all Abcam). Enzymatic detection was performed using ImmPACT VIP for the melanoma (Melan-A) tissue and ImmPACT DAB HRP Substrate Kit (RRID: AB_ 2336520) for the CRLM (CEA) tissue (both Vector Labs). Finally, rabbit anti-human PD-1 (1:200; Abcam; RRID: AB-230_881954) and mouse anti-human Ki67 (1:75; Dako; RRID: AB_2142367) were detected with the ImmPACT VIP HRP or DAB Peroxidase Substrate kits. Control sections were processed as above, without the addition of primary antibody. Hematoxylin and eosin (H&E; both Sigma) staining was performed on all patient tissue sections. Digital images were acquired using an Aperio AT2 Scanscope (Leica Biosytems) at ×20 magnification and quantified using ImageScope software (RRID: SCR_014311).

qPCR was performed using DNA extracted from PBMCs and plasma using a DNeasy Blood and Tissue Kit and a Circulating Nucleic Acid kit, respectively (Qiagen). Primers corresponding to the vaccinia E3 L gene (forward: 3′-TCCGTCGATGTCTACACAGG-5′ and reverse: 5′-ATGTATCCCGCGAAAAATCA-3′) (Integrated DNA Technologies) were used to detect the presence of Pexa-Vec (five replicates), alongside a standard curve of known Pexa-Vec DNA concentration (three replicates) and a control gene (18s; 3 replicates; Qiagen), on an Applied Biosystems QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific) using SYBR green detection method (Thermo Fisher Scientific). 2 μL DNA was included in the reaction mix, which equated to a range of 100–200 ng/μL total DNA. PCR data were analyzed using a QuantStudio three-dimensional AnalysisSuite Cloud (Thermo Fisher Scientific).

The African monkey green kidney cell line (Vero), a melanoma cell line (Mel888) and a colorectal cell line (SW620; all ATCC) were maintained in full growth medium: DMEM (Sigma) containing 5 mmol/L l-glutamine and supplemented with 10% (v/v) FCS for approximately 2 weeks before use. Cells were routinely tested and found to be negative for Mycoplasma infection. Cell lines were authenticated by short tandem repeat profiling.

Neutralizing antibodies (NAb) were detected using a modified serial dilution assay of heat-inactivated (HI) patient serum as described previously (16). Briefly, 3-fold serial dilutions (1:3 to 1:531,441) of HI-serum samples were incubated with Pexa-Vec. After incubation for 3 hours at 37°C, the diluted HI-serum samples were transferred onto monolayers of Vero cells resulting in a final virus dose of 50 p.f.u./cell. Cells alone or with Pexa-Vec dilution only were also cultured as negative and positive controls, respectively. After a further 72 hours of incubation, MTT (5 mg/mL; Sigma) was added to each well and left for 4 hours, before the removal of all medium and addition of 150 μL DMSO (Thermo Fisher Scientific) to each well. Absorbance of samples was then read at 540 nm on a Thermo Multiskan EX plate reader (Thermo Fisher Scientific). NAb titers were calculated as 1/end point, which equates to the last serum dilution at which no antibody neutralization of Pexa-Vec–induced killing was observed (n = 9). Data are shown as mean + SEM.

Bio-Plex Pro Cytokine and Chemokine Assays (21-plex; human group I and 27-plex; human group II or 48-plex; human cytokine; both Bio-Rad) were used to determine concentration of soluble mediators in patient plasma samples throughout treatment, as per manufacturer's instructions. Plasma was diluted 2-fold and 50 μL loaded per well in duplicate. Bio-Plex plates were analyzed using the Bio-Plex 200 system and the Bio-Plex Manager 6.1 for data acquisition and analysis (Bio-Rad). IFNβ was measured in samples using the VeriKine-HS Human Interferon Beta ELISA Kit for plasma (R&D Systems), as per manufacturer's instructions and plates were analyzed using a Thermo Multiskan EX plate reader. Standard curves included within each kit were used to calculate the concentration of each solute in patient plasma based on emission or absorbance spectra, as appropriate to the kit. Data are expressed as mean ± SEM relative fold change in posttreatment samples compared with pretreatment samples. Statistical significance between time points was determined using paired t tests; *, P < 0.05; **, P < 0.01 (n = 9).

PBMCs from specified time points prior to (D1 pre) and following Pexa-Vec infusion were cocultured at a ratio of 1:1 with Mel888 or SW620 tumor cells for patients with melanoma and CRLM, respectively, for 1 hour, and then 1 μL/mL brefeldin (Sigma) was added. Coculture continued for a further 4 hours before PBMCs were harvested, washed, and stained for anti-CD3-PerCP (5 μL; SK7; BD Biosciences), anti-CD56-PE (2 μL; AF12-7H3; Miltenyi Biotec), and anti-CD107a/b-FITC (5 μL; H4A3; BD Biosciences) for 30 minutes at 4°C. PBMCs were washed and fixed in 1% PFA. CD107 positivity was assessed using a CytoFLEX S flow cytometer and analysis was performed using CytExpert software (RRID: SCR_017217; both Beckman Coulter). Data are expressed as mean ± SEM% positive CD107 staining on NK cells (CD3⁻ CD56⁺). Statistical significance was determined by paired t tests (*, P < 0.05; n = 9).

Assessment of natural killer (NK)-cell CD69 expression was performed using flow cytometry either: (i) as described above using anti-CD69-APC (5 μL; Miltenyi Biotec, RRID: AB_2784271) or (ii) using a panel of custom-designed DURAClone tubes (Beckman Coulter) incorporating CD3 (UCHT1; Pacific Blue), CD56 (N901; PC7) and CD69 (TP.55.3; APC). Data were acquired on a CytoFLEX S flow cytometer, and analysis was performed using CytExpert software. Data are expressed as mean ± SEM % positive CD69 staining on NK cells. Statistical significance was determined by paired t tests (*, P < 0.05; n = 9).

PBMC RNA was isolated using an RNeasy Plus Mini kit (Qiagen) following manufacturer's instructions. Analysis was performed using the HTG EdgeSeq Precision Immuno-Oncology Panel by HTG Molecular, Arizona, which contains probes for 1,410 genes. The quantitative nuclease protection assay was used to ensure robust RNA quality to a minimum of 50 bases. All sequencing was performed on the Illumina NextSeq. Differential expression analysis was performed using the DESeq2 package (version 1.14.1; Bioconductor; RRID: SCR_006442), utilizing median data normalization. A read depth of 14 million to 10,000 was used as a threshold. Data are expressed as log₂ (CPM) for the average expression of each probe across all groups after normalization. Statistical significance between time points is shown by an adjusted P value (P_adj); *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (n = 7). Reagents listed in Supplementary Data S1.

Briefly, 1 × 10⁵ patient PBMCs were incubated per well in the presence of either 2 μg/mL CEA or melanoma antigen recognized by T cells-1 (MART-1) overlapping peptide pools, or with 2 μg/mL CEF peptide pool (positive control; all Cambridge Biosciences). Media alone (negative control) and 10 p.f.u/cell Pexa-Vec were also used. IFNγ secretion from activated T cells was detected using a matched paired antibody kit (MabTech) and spot-forming units (SFU) were visualized using BCIP/NBT substrate (MabTech). Images were captured using an AID enzyme-linked immunosorbent spot (ELISpot) reader (AID GmbH). Data are presented as mean SFU per well (in triplicate) + SEM. Statistical significance between time points was determined by fitting a mixed model for one-tailed multiple comparisons; *, P_adj < 0.05 (n = 3–9, dependent on sample availability).

Peripheral blood from healthy donors (HD) was incubated with 2 × 10⁵ p.f.u/mL Pexa-Vec in vitro for 1 hour at room temperature. Blood was processed to isolate PBMCs and plasma before DNA was extracted, and qPCR was performed to quantify Pexa-Vec DNA, as described above. A standard curve of known Pexa-Vec DNA was used to estimate the concentration of Pexa-Vec DNA in PBMCs and plasma. Data are shown as mean + SEM ng DNA per 5 mL initial blood volume (n = 5 replicates) for three donors. Statistical significance was determined by paired t tests; **, P < 0.01; ***, P < 0.001.

DNA was extracted using an AllPrep RNA/DNA mini kit (Qiagen) following manufacturer's instructions. TCRB was amplified from genomic DNA, extracted from patient samples, using a modified version of the BIOMED-2 sequencing protocol (17). Modifications included the generation of amplicons using Phusion High-fidelity Master Mix (Thermo Fisher Scientific) with an annealing temperature of 63°C, with a final magnesium concentration of 3.5 mmol/L. Adapters attached to the PCR primers (Supplementary Data S2) allowed sequences to be indexed using the Nextera XT indexing kit (Illumina) for sample multiplexing. Libraries were sequenced on a single lane of an Illumina MiSeq by the Leeds University sequencing facility using a 2 × 300 bp kit. The quality of demultiplexed reads was assessed by fastqc (18), and reads were trimmed to remove adapters and low-quality bases using trimmomatic (19) and trim galore (20). High-quality reads (Phred>30) were then overlapped with FLASH (21) and aligned to the IMGT database using MiXCR (22), which discarded erroneous and nonproductive rearrangements. Clones were defined by having the same TCRBV/J segment and identical CDR3 sequence, and their abundance was adjusted using the absolute T-cell count to eliminate amplification errors. For generation of inverse Simpson indices, assessment of antigen specificities and generation of Circos (RRID: SCR_011798) plots, the subsequent output was filtered into VDJTools (23) and processed in R (24) using Immunarch (25), circlize (26), and ggplot2 (27). Statistical analysis of the percent change in T-cell diversity at the time of surgery was performed using a one-tailed Mann–Whitney test, *, P < 0.05.

HD PBMCs were isolated from leukocyte apheresis cones supplied by the National Health Service Blood and Transplant unit (United Kingdom) and treated with 0.1 p.f.u./cell Pexa-Vec for 72 hours. Viability of PBMC populations were assessed by staining, as described above, with 5 μL anti-CD3-PerCP (CD3⁺ T cells), 2 μL anti-CD56-APC (CD3⁻CD56⁺ NK cells; REA196, Miltenyi Biotec), 5 μL anti-CD19-FITC (CD19⁺ B cells) (SJ25C1, BD Biosciences), 20 μL anti-CD14-PE (CD14⁺ monocytes; M5E2, BD Biosciences), and Zombie UV fixable viability kit (BioLegend) following manufacturer's instructions. Flow cytometry analysis was performed using a CytoFLEX S and CytExpert software.

Liver metastases and accompanying background liver were obtained from freshly resected liver specimens immediately following surgical resection from 10 patients with CRLM undergoing planned surgical resection of liver metastases but not participating in the trial. Written, informed consent was obtained in accordance with institutional ethics review and approval. Tissue cores were made using a Tru-Cut biopsy needle (CareFusion), resulting in cores of 1 mm diameter and 15 mm length, and cores were subsequently divided into three 5-mm-length cores. Tissue cores were analyzed as soon as possible following resection and treated with 1 × 10⁷ p.f.u. Pexa-Vec-GFP or PBS for 96 hours. Prior to image acquisition, 4′,6-diamidino-2-phenylindole staining was performed following cell penetration with 0.1% Triton X (both Sigma). High-resolution images were taken using a Nikon A1 Confocal Laser Scanning Microscope (Nikon).

HD PBMCs were treated with Pexa-Vec for 24 hours with or without (±) type I IFNα/β blockade or monocyte depletion. For type I IFN neutralization, PBMCs were preincubated for 30 minutes with polyclonal antibodies to IFNα, IFNβ (both anti-sheep), and IFNα/β receptor chain 2 (anti-mouse; all PBL Assay Science) or isotype control (anti-sheep, Sigma; and anti-mouse, R&D Systems) prior to addition of 1 p.f.u./cell Pexa-Vec for 24 hours as described previously (28). For monocyte depletion, CD14⁺ cells were depleted from whole PBMCs using CD14-positive magnetic bead selection on LS MACS columns (Miltenyi Biotec), according to the manufacturer's instructions. The CD14-depleted PBMCs were at least 97% pure of CD14⁺ cells. After virus treatment for 24 hours, NK-cell CD69 expression and CD107 degranulation (as described above) were determined using an Attune NxT Acoustic Focusing Cytometer (Life Technologies), and data were analyzed using Attune Cytometric Software (v2.1.0; Life Technologies). For NK-cell CD107 degranulation, PBMCs (±1 p.f.u. Pexa-Vec) were cocultured with SW620 cells at 10:1 effector:target ratio, following the same protocol as previously described for the patient CD107 assay.

Raw data for mRNA expression analysis were generated at HTG Molecular. Derived data supporting the findings of this study are available from the corresponding author upon request. The TCRB sequencing data generated in this study are publicly available in NCBI; BioProject accession number PRJNA825493.

We recruited 9 patients (3 patients with metastatic melanoma and 6 patients with CRLM) to a phase Ib window-of-opportunity trial. Each patient received a single, 1-hour intravenous infusion of 1 × 10⁹ p.f.u. Pexa-Vec, 14 (±4) days ahead of planned surgical resection of metastatic melanoma or CRLM (Fig. 1A). Pexa-Vec–related grade 3 and 4 adverse events included lymphopenia and neutropenia (Supplementary Table S1). Patient 09 experienced grade 4 intraoperative hypotension that was regarded to be unrelated to Pexa-Vec. Surgery was undertaken in all but 1 patient (patient 07), where plans for surgery were abandoned after finding pulmonary metastases on an up-to-date preoperative CT scan. At the time of writing, after more than 3 years of follow-up post–Pexa-Vec infusion for each patient, 5 of the 9 patients were alive. The 3 patients that remained cancer-free all had CRLM (patients 05, 06, and 08). Pathologic examination of their resected tumor specimens post–Pexa-Vec infusion revealed extensive (patient 05) or complete (patient 08) necrosis of the tumor in 2 of the 3 patients (Fig. 1B; Supplementary Table S1). None of the other resected patient tumors showed any signs of necrosis.

In contrast to our previous clinical trial findings using intravenous reovirus infusions (29, 30), qPCR analysis of peripheral blood compartments revealed that Pexa-Vec associated with plasma, but not PBMCs (Table 1), being detectable at 1 hour postinfusion in 4 of the 9 patients. This was confirmed in Pexa-Vec–pulsed blood, donated from healthy volunteers, where Pexa-Vec could only be detected in the separated plasma fraction (Fig. 1C). However, in patient 05, extensive tumor necrosis was noted following Pexa-Vec infusion (Supplementary Table S1), and qPCR of plasma only revealed the presence of virus immediately presurgery [22 days postinfusion (Table 1)]. This potentially signified ongoing virus production from the tumor at this time point. In common with reovirus infusion (30), Pexa-Vec neutralizing antibodies peaked either at the time of surgery or 1-month post-surgery (Fig. 1D).

IHC analysis of four of five resected CRLMs revealed the presence of Pexa-Vec protein in the tumors from patients 02 and 09, on the periphery of the tumor from patient 08, and its absence from the tumor of patient 06 (Fig. 2A). The available tissue from patient 05′s tumor could not be assessed due to tissue necrosis. To confirm the specificity of Pexa-Vec to CRLM compared with the surrounding liver, we treated ex vivo tissue core biopsy samples from patients undergoing standard CRLM surgical resections outside the trial with Pexa-Vec. At 96 hours posttreatment, large areas of CRLM core biopsies were infected with Pexa-Vec, as determined by immunofluorescence for virus-expressed GFP (Fig. 2B). In contrast, separate core biopsies taken from background livers were resistant to Pexa-Vec infection, with GFP only detected in scattered cells, indicating nonproductive infection (Fig. 2B).

A total of 24 hours after Pexa-Vec infusion, a significant increase in plasma IFNα concentrations (Fig. 3A; Supplementary Table S2), in addition to other proinflammatory, proapoptotic, and DC maturation-associated cytokines such as IL12, TNF-related apoptosis-inducing ligand (TRAIL), and lysosome-associated membrane glycoprotein 3 (LAMP3), respectively (Supplementary Fig. S1A and S1B; Supplementary Table S2; ref. 31). Gene expression of a panel of IFN-stimulated genes (ISG) significantly increased in PBMCs 24 hours post–Pexa-Vec (Fig. 3B). Inflammatory cytokine protein and RNA expression gradually reduced to baseline levels by the time of surgery (Fig. 3A and B; Supplementary Fig. S1A and S1B; Supplementary Table S2).

Given that NK cells are activated by ISGs, including IL12 (32), we measured patient-derived peripheral blood NK-cell activation and cytolytic activity against tumor-relevant cell line targets. We found significantly increased NK-cell activation, as measured by CD69 expression, on days 2 and 3, with a reduction to baseline levels by day 5 [Fig. 3C(i); Supplementary Fig. S2A]. NK-cell cytolytic activity was significantly increased 24 hours following Pexa-Vec infusion, before decreasing to baseline at day 5 [Fig. 3C(ii); Supplementary Fig. S2B]. In common with other oncolytic viruses (29, 33), the critical role of IFNα/β in NK-cell activation following Pexa-Vec stimulation was confirmed by simultaneous blockade of the type I IFN receptor and of soluble IFNα/β within in vitro Pexa-Vec–treated PBMCs. This resulted in a significant reduction in NK-cell activation, as assessed by cell surface CD69 expression and cytolytic activity (Fig. 3D). Previous work indicates that monocytes are the source of both oncolytic reovirus and herpes simplex virus-induced type I IFNs (34, 35). We confirmed a critical role for monocytes in mediating NK activity following Pexa-Vec stimulation, whereby depletion of monocytes from PBMCs significantly reduced NK-cell activation and function (Fig. 3E).

Because IFNs stimulate upregulation of programmed death protein 1 (PD-1) (29), we analyzed expression of this immunosuppressive immune checkpoint in patient 01, comparing the pre–Pexa-Vec tumor biopsy with the resected tumor sample. In keeping with the observed induction of ISGs, we found more concentrated PD-1 staining post-Pexa Vec infusion (Supplementary Fig. S3A). PD-1–expressing cells lacked Ki67 staining, confirming that they were nonreplicating and likely to be immunologically exhausted. Examination of post-Pexa Vec tumor samples from other patients on study similarly revealed areas of PD-1–expressing nonreplicating cells (Supplementary Fig. S3B).

Pexa-Vec infusion was followed by a reduction in the peripheral lymphocyte concentration in all patients at day 2, returning to baseline levels by the time of surgery (Fig. 4A). We confirmed that this was unlikely due to lymphocyte cell death in an in vitro cell viability assay (Fig. 4B). Pexa-Vec also associated with CXCL10 gene expression in PBMCs and a corresponding peak in plasma CXCL10 (also known as IFNγ-induced protein 10; IP-10) protein levels 24 hours following virus infusion (Fig. 4C; Supplementary Table S3). IP-10 is a chemokine known to play an important role in recruiting activated T cells into sites of tissue inflammation and associates with the presence of CD8⁺ T cells in tumors (36, 37). Concordantly, the observed kinetics of peripheral blood lymphopenia closely followed the rise and fall in IP-10 concentrations and CXCL10 gene expression. On examination, all resected tumors harbored infiltrating CD8⁺ T cells (Fig. 4D). CD8⁺ T cells were frequently associated with malignant cells, as exemplified by the resected lymph node from patient 04, where CD8⁺ T cells were found in higher concentrations in association with melanoma cell clusters (Supplementary Fig. S4A) than in areas of the same lymph node with few infiltrating melanoma cells. In patient 01, where a pre–Pexa-Vec tumor biopsy was available, comparison with the resected tumor sample post–Pexa-Vec revealed a shift in the CD8⁺ T-cell population, from a perivascular localization at baseline to a wider infiltrative distribution across the tumor postvirus infusion (Supplementary Fig. S4B). In addition to increased CXCL10, gene expression of CCL2, a chemotactic cytokine that induces directional migration of DCs into infected tissue, also peaked at day 2 and returned to baseline by the time of surgery (Supplementary Fig. S5; ref. 38).

In addition to the observed innate anticancer immune effects, successful anticancer therapy requires a longer-lived T-cell immune response. We therefore measured the effects of Pexa-Vec infusion on functional T-cell responses via PBMC IFNγ secretion using ELISpot assays. Patient-derived PBMCs were stimulated ex vivo using overlapping peptide pools of TAAs, either MART-1 for PBMCs derived from patients with melanoma or CEA for PBMCs isolated from patients with CRLM, which were confirmed to be expressed in the corresponding tumor samples (Fig. 5A). Although very few MART-1 or CEA-specific IFNγ-producing cells were observed at baseline pre-Pexa-Vec, this increased following Pexa-Vec infusion (Fig. 5B). A similar increase was seen in IFNγ-producing cells specific to CEF (cytomegalovirus, Epstein–Barr virus, influenza virus) overlapping peptide pools, which were used as a positive control (Supplementary Fig. S6A). T-cell responses to either CEA and MART-1 for all patients peaked at 1-month postsurgery and remained elevated over pre–Pexa-Vec baseline levels (Fig. 5C; Supplementary Fig. S6B). T-cell responses remained elevated at least 3 months following Pexa-Vec infusion, indicating the induction of long-lived T-cell anticancer immune responses (Fig. 5C). We confirmed these findings for MART-1 and other TAAs by T-cell receptor (TCR) sequencing using DNA derived from trial patient PBMCs at baseline, surgery, and at the end of study. Examination of the relative numbers of individual T-cell clones was performed against published databases of TCR epitope specificity [VDJdb (39), PIRD (40), and McPAS-TCR (41)], where these data were available. Analysis revealed clonal proliferation of MHC class I–restricted (CD8) T cells following Pexa-Vec infusion, targeting MART-1 (Fig. 5D; Supplementary Table S4) and other TAAs (Supplementary Table S4).

Despite the observed NK-cell cytolytic activity and the functional T-cell anticancer immune responses across all the trial patients, tumor necrosis was only observed in the CRLM of patients 05 and 08. We, therefore, characterized changes in peripheral blood T-cell clonal evolution in the 9 trial patients to elucidate whether this could be a determining factor for the observed differences in tumor necrosis. Longitudinal analysis of these data showed changes in the V-J TCR usage, which are displayed as circos plots (Fig. 6A) where the width of each band is proportional to the frequency of usage of that gene segment. Individual gene regions were labeled, and a transient increase in the use of TRBV20-1 was seen in the 2 patients who displayed tumor necrosis (patient 05 and 08), along with a similar increase in TCRBV18 in the patient displaying complete necrosis (patient 08). This was in contrast to a more stable gene family usage profile exhibited by other patients with no visible necrosis, as exemplified by patient 06. The small patient numbers involved limited statistical inference; however, increased usage of TRBV18 and TRBV20-1 draws parallels with a prior study examining tumor-infiltrating lymphocytes within colorectal cancer biopsies and may indicate a TAA-driven T-cell response (42).

Tracking changes in the abundance of T-cell clones following Pexa-Vec infusion revealed that by the time of surgery, there was a reduction in the diversity of the most highly abundant T-cell clones derived from patients 05 and 08 compared with other patients, as estimated by the inverse Simpson index (Fig. 6B). Statistical analysis revealed that this drop in T-cell diversity for patients 05 and 08 was significant (P = 0.036) compared with the other patients that had no tumor necrosis at the time of surgery (Fig. 6C). We investigated whether this could be due to differences in T-cell responses to Pexa-Vec. Comparison of longitudinal PBMC IFNγ production via ELISpot between patients 05 (extensive tumor necrosis) and 06 (no tumor necrosis) revealed an expected increase in responses at the time of surgery, which reduced at the end of study but showed no discernible difference between the 2 patients (Supplementary Fig. S7). Therefore, CRLM necrosis in patients 05 and 08 following Pexa-Vec infusion appeared to be associated with a reduction in the diversity of the most highly proliferated T-cell clones. This observation could, in part, be accounted for by proliferation of select TRBV18 and TRBV20-1 T-cell clonotypes targeting TAAs, in contrast to a more diverse naïve CD8⁺ TCR repertoire in patients without tumor necrosis. This is reminiscent of the same observation in patients with melanoma who experienced durable progression-free survival following treatment with a neoantigen vaccine and PD-1 inhibition (43). In these patients, the presence of increased effector memory CD8⁺ T cells was reflected by a more restricted and less diverse circulating TCR repertoire than in patients who experienced progressive disease.

Our results showed that neoadjuvant intravenous Pexa-Vec therapy is well tolerated in patients with metastatic melanoma and CRLM, has tumor-specific replication, and shows evidence of clinical efficacy, including pathologic tumor response. Pexa-Vec was detected in three of the four available CRLM specimens, confirming previous findings of intravenous delivery of Pexa-Vec to tumors when using the same or higher doses (14). In contrast to our previous findings of the cellular carriage of oncolytic reovirus to tumors following intravenous infusion (29, 30), intravenous Pexa-Vec administration resulted in plasma-based carriage, with no association of virus with PBMCs. The development of neutralizing antibodies to Pexa-Vec at the time of surgery is consistent with the timing of seroconversion previously reported for Pexa-Vec, as well as for other intravenously administered oncolytic viruses (44, 45). In contrast to oncolytic reovirus, which is delivered to tumors by peripheral blood cell carriage (46), and results in the shielding of virus from antibody neutralization (47), the development of Pexa-Vec antibodies will likely hinder intravenous delivery from that time point onward (48). Strategies to reduce antibody neutralization are being tested clinically for other oncolytic poxviruses, including TG6002, an engineered Copenhagen strain vaccinia virus under clinical development by Transgene, which is being delivered by hepatic artery infusion in patients with CRLM (ClinicalTrials.gov Identifier: NCT04194034).

Our study characterizes the innate and adaptive immune response to intravenous Pexa-Vec infusion in patients with cancer. In accordance with results from other oncolytic virus clinical trials, Pexa-Vec administration stimulated IFN expression, which peaked 24 hours after infusion (49, 50). This resulted in a broad inflammatory cytokine and chemokine response, despite the potential for immunosuppressive vaccinia virus gene expression of encoding proteins that act as decoy receptors to block the activity of type I IFNs (51). The expressed cytokines associated with NK-cell activation and the proliferation of CD8⁺ T-cell clones specific for TAAs, despite the presence of exhausted cell populations. The correlation seen between pathological tumor response and reduction in peripheral TCR diversity requires confirmation in larger trials. Although we have not, within the confines of the current study, specified the nature of these T-cell populations, we suggest that pathologic tumor response is driven by a proliferative, low diversity clonal T-cell population targeting TAAs. If confirmed in a larger study, our results would support a strategy to encode selected HLA-matched tumor-specific epitopes within oncolytic viruses, or non–HLA-specific TAAs, instead of the pursuit of broad-spectrum, low-level T-cell stimulation.

Disappointing phase III results using Pexa Vec in combination with sorafenib in advanced hepatocellular carcinoma (NCT02562755) indicate the need to better rationalize combinations and schedules of therapy. Despite the size of the present study and heterogeneity of the tumor cohort, our findings, in conjunction with the excellent safety data from our previous neoadjuvant oncolytic virus studies (29, 30), supports the development of this class of immunotherapies toward standard clinical practice in the neoadjuvant setting. Rational combinations with nonoverlapping toxicities are needed to increase efficacy. The most promising oncolytic virus combination is with the targeting of immune checkpoint proteins, which are frequently upregulated in the presence of both pathogenic and therapeutic viral infections (29, 52–54), and act to dampen CD8⁺ T-cell responses. The sequential combination of an oncolytic virus followed by PD-1 or PD-L1 blockade has produced remarkable results in preclinical models (29, 54) and is currently being tested in numerous early- (55) and late-phase (ClinicalTrials.gov Identifier: NCT02263508) clinical trials for a wide variety of malignancies. Likewise, tumor-conditioning using intratumoral Pexa-Vec followed by PD-1 blockade is currently being tested in patients with advanced hepatocellular carcinoma (ClinicalTrials.gov Identifier: NCT03071094). To aid in the clinical progression of oncolytic vaccinia therapy, further research should seek reliable baseline biomarkers that predict tumor response and clinical prognosis following virus treatment, in conjunction with the associated clinical disease parameters.

A. Samson reports grants from Transgene during the conduct of the study. K.J. Scott reports grants from Transgene during the conduct of the study. K. Bendjama reports personal fees from Transgene SA during the conduct of the study. N. Stojkowitz reports personal fees from Transgene outside the submitted work. J. Bell reports other support from Jennerex Biotherapeutics during the conduct of the study; other support from Jennerex Biotherapeutics outside the submitted work; in addition, J. Bell has a patent for WO2013022764A1d pending. K.J. Harrington reports personal fees from Arch Oncology, BMS, Inzen, Merck-Serono, Pfizer; grants and personal fees from AstraZeneca, Boehringer-Ingelheim, MSD, and Replimune outside the submitted work. C. Ralph reports grants from Jennerex during the conduct of the study; grants from Oncolytics Biotech, non-financial support from Viralytics, personal fees and non-financial support from BMS and Eisai outside the submitted work. A. Anthoney reports grants and personal fees from Transgene during the conduct of the study. F. Collinson reports grants from Transgene during the conduct of the study. No disclosures were reported by the other authors.

A. Samson: Conceptualization, data curation, formal analysis, supervision, funding acquisition, writing–original draft, writing–review and editing. E.J. West: Conceptualization, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. J. Carmichael: Formal analysis, methodology. K.J. Scott: Formal analysis, investigation, methodology, writing–original draft, writing–review and editing. S. Turnbull: Enrolment of patients. B. Kuszlewicz: Formal analysis, methodology. R.V. Dave: Methodology. A. Peckham-Cooper: Methodology. E. Tidswell: Methodology. J. Kingston: Methodology. M. Johnpulle: Methodology. B. da Silva: Methodology. V.A. Jennings: Methodology. K. Bendjama: Conceptualization, methodology. N. Stojkowitz: Conceptualization. M. Lusky: Conceptualization. K.R. Prasad: Patient management. G.J. Toogood: Patient management. R. Auer: Conceptualization. J. Bell: Conceptualization. C.J. Twelves: Conceptualization, management of patients. K.J. Harrington: Writing–original draft. R.G. Vile: Writing–original draft. H. Pandha: Writing–original draft. F. Errington-Mais: Conceptualization, writing–original draft. C. Ralph: Conceptualization. D.J. Newton: Conceptualization, writing–original draft. A. Anthoney: Patient management. A.A. Melcher: Conceptualization, writing–original draft. F. Collinson: Conceptualization.

We are grateful to all the patients that participated in this trial. The research was supported by the National Institute for Health Research (NIHR) infrastructure and the Experimental Cancer Medicine Centre (ECMC) at Leeds. A. Samson was supported by fellowships from Yorkshire Cancer Research (YCR) and Cancer Research UK (CRUK). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.

Funding: Transgene, Strasbourg, Yorkshire Cancer Research, Cancer Research UK and the Institute of Cancer Research/Royal Marsden NIHR Biomedical Research Centre.

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.