A randomised controlled trial of long NY-ESO-1 peptide-pulsed autologous dendritic cells with or without alpha-galactosylceramide in high-risk melanoma

Aim We have previously reported that polyfunctional T cell responses can be induced to the cancer testis antigen NY-ESO-1 in melanoma patients injected with mature autologous monocyte-derived dendritic cells (DCs) loaded with long NY-ESO-1-derived peptides together with α-galactosylceramide (α-GalCer), an agonist for type 1 Natural Killer T (NKT) cells. Objective To assess whether inclusion of α-GalCer in autologous NY-ESO-1 long peptide-pulsed DC vaccines (DCV + α-GalCer) improves T cell responses when compared to peptide-pulsed DC vaccines without α-GalCer (DCV). Design, setting and participants Single-centre blinded randomised controlled trial in patients ≥ 18 years old with histologically confirmed, fully resected stage II–IV malignant cutaneous melanoma, conducted between July 2015 and June 2018 at the Wellington Blood and Cancer Centre of the Capital and Coast District Health Board. Interventions Stage I. Patients were randomised to two cycles of DCV or DCV + α-GalCer (intravenous dose of 10 × 106 cells, interval of 28 days). Stage II. Patients assigned to DCV + α-GalCer were randomised to two further cycles of DCV + α-GalCer or observation, while patients initially assigned to DCV crossed over to two cycles of DCV + α-GalCer. Outcome measures Primary: Area under the curve (AUC) of mean NY-ESO-1-specific T cell count detected by ex vivo IFN-γ ELISpot in pre- and post-treatment blood samples, compared between treatment arms at Stage I. Secondary: Proportion of responders in each arm at Stage I; NKT cell count in each arm at Stage I; serum cytokine levels at Stage I; adverse events Stage I; T cell count for DCV + α-GalCer versus observation at Stage II, T cell count before versus after cross-over. Results Thirty-eight patients gave written informed consent; 5 were excluded before randomisation due to progressive disease or incomplete leukapheresis, 17 were assigned to DCV, and 16 to DCV + α-GalCer. The vaccines were well tolerated and associated with increases in mean total T cell count, predominantly CD4+ T cells, but the difference between the treatment arms was not statistically significant (difference − 6.85, 95% confidence interval, − 21.65 to 7.92; P = 0.36). No significant improvements in T cell response were associated with DCV + α-GalCer with increased dosing, or in the cross-over. However, the NKT cell response to α-GalCer-loaded vaccines was limited compared to previous studies, with mean circulating NKT cell levels not significantly increased in the DCV + α-GalCer arm and no significant differences in cytokine response between the treatment arms. Conclusions A high population coverage of NY-ESO-1-specific T cell responses was achieved with a good safety profile, but we failed to demonstrate that loading with α-GalCer provided an additional advantage to the T cell response with this cellular vaccine design. Clinical trial registration: ACTRN12612001101875. Funded by the Health Research Council of New Zealand. Supplementary Information The online version contains supplementary material available at 10.1007/s00262-023-03400-y.


Antigenic peptides and α-GalCer
The DCs were loaded with two long peptides from NY-ESO-1 (NY- ESO-179-116, GARGPESRLLEFYLAMPFATPMEAELARRSLAQDAPPL and NY-ESO-1118-143, VPGVLLKEFTVSGNILTIRLTAADHR). As an approach to increase likelihood of measuring T cell responses to evaluate the impact of α-GalCer, HLA class I-binding peptides from influenza proteins were loaded separately onto half of the cells of the vaccine with or without α-GalCer; these peptides were from influenza polymerase basic protein 1 (PB-1489-497, TFEFTSFFY), influenza virus matrix 1(M158-66, GILGFVFTL), and influenza virus nucleoprotein (NP265-273, ILRGSVAHK). However, as none of the participants in the earlier phase I study had detectable MHC class I-restricted CD8 + T cell memory responses to these peptides, and the vaccine did not prime detectable de novo CD8 + T cell responses [12], it was anticipated that this readout would be uninformative in the randomised study, and evaluation of responses to these peptides was not included in the statistical plan for the formal endpoints of this study. Good manufacturing practice (GMP) processes were used to generate the peptides (University of Auckland, Auckland, New Zealand) and α-GalCer (GlycoSyn, Lower Hutt, New Zealand).

Vaccine production
The generation of MoDCs and antigen pulsing was conducted under GMP conditions (Malaghan Institute of Medical Research). A Lymphoprep density gradient (Axis Shield, Oslo, Norway) was used to enrich PBMCs from leukapheresis product, with a 20 % sucrose gradient (Calbiochem, Billerica, MA) then used to remove platelets. Monocytes were enriched by adherence to plastic in complete medium consisting of RPMI 1640 medium (Gibco, Life Technologies, Carlsbad, CA) supplemented with 2 % autologous plasma. The adherent fraction was incubated in complete medium overnight, and then maintained in complete medium supplemented with 1,000 U/ml recombinant human granulocyte maturation colony stimulating factor (rhGM-CSF; Genzyme, Lynnwood, Australia) and 1,000 U/ml rh interleukin (IL)-4 (Gibco CTS, Life Technologies) to promote differentiation to monocyte-derived DCs, with cytokines replenished on day 3. On day 5, samples of immature DCs were collected and stored at -20 °C for later in vitro studies. The remaining DCs were pulsed with peptides and α-GalCer in complete medium supplemented with a cytokine cocktail of 1,000 U/mL IL-1β (CellGenix, Freiberg, Germany), 1,000 U/mL IL-6 (Gibco CTS), 1,000 U/mL TNF (Gibco CTS) and 1 μg/mL PGE2 (Cayman Pharma, Neratovice, Czech Republic). To avoid immunodominance of influenza-specific responses over those for the tumour antigen, one half of the cells were incubated with 10 µM of each of the NY-ESO-1 peptides and the other half with 10 µM of each of the influenza peptides. The peptide-supplemented cells were split again at a ratio of 2:1, with the larger fraction supplemented with 100 ng/mL a-GalCer (thereby providing the larger number of a-GalCerpulsed DCs required for the dosing regimen). On day 6, the separate cultures were washed to remove excess antigens, with the influenza and NY-ESO-1 peptide-pulsed cells then combined at a ratio of 1:1 to give the DCV, and the influenza and NY-ESO-1 peptide-pulsed cells with α-GalCer combined at a ratio of 1:1 to give DCV+α-GalCer. Products were released if the final overnight culture medium was negative for bacterial growth in BacT/Alert FN Plus and BacT/Alert FA Plus cultures (bioMérieux, Marcy-l'Étoile, France) and endotoxin levels were <0.5 EU/mL (Kinetic-chromogenic Limulus amebocyte lysate test; Charles River, Melbourne, Australia). The vaccines were cryopreserved in 90 % autologous plasma and 10 % DMSO (OriGen Biomedical, Austin, TX) in 2 mL CellSeal closed-system cryogenic vials (Cook General BioTechnology, Indianapolis, IN) using a controlled rate freezer (Thermo Fisher Scientific). Released products contained >70 % CD83 + HLA-DR + cells and were >70 % viable as determined by flow cytometry on a thawed sample.

Vaccine administration
Cryopreserved vaccines were transported in a dry shipper containing liquid N2 and thawed immediately prior to injection using a drybath at 37 °C at the bedside. Intradermal test doses consisting of 1 x 10 5 cells of the assigned vaccine were administered to ensure no evidence of an immediate antigen-related wheal and flare reaction over 15 min, with injection of the autologous cryopreservation medium alone as control. The vaccine dose consisting of 10 x 10 6 cells was then administered intravenously via a cannula over 1 min, with the patients kept in the ward for 6 h to be monitored.

IFN-γ ELISpot assay for peptide-specific T cells and NKT cells
For the detection of interferon (IFN)-γ-producing T cells and NKT cells in blood, ELISpot plates (Millipore) were pre-coated in-house with 1 μg/mL of anti-IFN-γ antibody (mAb 1-DK1, Mabtech, Nacka Strand, Sweden) in 100 µL of PBS per well at 4 °C overnight. The next day the plates were washed four times with 200 μL sterile PBS per well. Cryopreserved PBMCs were thawed and washed three times in RPMI 1640 without supplements and then 1-2 x 10 5 live cells were resuspended in fresh AIM-V medium (Gibco) and cultured overnight at 37 °C, 5 % CO2 in the presence of either 10 μM of the individual peptides and 0.5 ng/mL rhIL-7 to quantify antigen-specific T cells by , or with 100 ng/mL α-GalCer to quantify NKT cells by IFN-γ ELISpot, in each case to a final volume of 150 µL per well. Analyses were conducted in triplicate, with five medium-only negative controls. As positive controls, additional samples at each timepoint were stimulated with 5 μg/mL phytohemagglutinin (PHA; ThermoFisher Scientific). After incubation, the plates were washed four times with 200 μL PBS per well, and then 100 μL of 1 μg/mL of anti-human IFN-γ biotinylated antibody (mAb 7-B6-1) and plates were incubated for an hour at 37 °C, 5 % CO2. Plates were then washed again four times and 100 μL of 1:1000 diluted streptavidin-ALP was added to each well and incubated at room temperature for 45 min away from light.
Plates were then washed six times with PBS, and 100 μL of BCIP/NBT-plus substrate (Mabtech) added to each well for spot development. The reaction was stopped by washing wells six times, using 250 μL H2O per well. Numbers of IFN-γ-producing spots determined using an AID reader (Autoimmun Diagnostika GmbH, Strassberg, Germany). For data visualisation of ELISpot data using heatmaps, mean log10-transformed T cell count from technical replicates were plotted to each peptide over time for all treated patients. To avoid zero values, log10 (x+a) was used instead of log10 (x), where "a" was 0.7 -the minimum required after considering the entire data set.

Intracellular cytokine staining (ICS) and detection
Thawed PBMCs from each timepoint were washed as above and incubated in triplicate with comprising PBS supplemented with 5 % fetal calf serum and 0.6 mg/mL human normal immunoglobulin (Intragam P, CSL Behring Pty Ltd, Australia) and then stained in 50 μL of an antibody mixture containing antibodies for CD3, CD4, CD8, CCR7, and CD45 RA diluted in flow buffer. Staining was performed for 10 min at room temperature in the dark. The cells were fixed and permeabilized using Cytofix/Cytoperm (BD Biosciences) and stained with antibodies for IFN-γ, TNF and IL-2 for 20 min in the dark. Antibody fluorophore, identifier, source and dilution are given in Supplementary Table 2. After staining, cells were resuspended in a 1:1 mix of 4 % formalin and PBS and kept on ice in the dark before analysis, which was performed on a BD LSR II flow cytometer (BD Biosciences, San Jose, CA). Machine set up including PMT voltages were performed by the core facility (Hugh Green Cytometry Centre) with data analyzed using FlowJo v9.9.6 software using the gating strategy in Supplementary Fig. 2.

Analysis of NKT cells by flow cytometry
Samples of PBMCs from each timepoint were assessed in triplicate to determine the number of NKT cells within the T cell fraction by flow cytometry using fluorescent α-GalCer-loaded human CD1d tetramers (ProImmune, Oxford, UK) and antibody to CD3 (clone UCHT1; BD Biosciences, San Jose, CA). Antibody fluorophore, identifier, source and dilution are given in Supplementary Table 2. The cells were rested in RPMI medium supplemented with 2 % human AB serum (Sigma) and 1 % Penicillin/Streptomycin (Gibco) before staining was conducted in PBS (Gibco) supplemented with 2 % fetal calf serum and 0.6 mg/mL human normal immunoglobulin (Intragam P, CSL Behring Pty Ltd, Australia). Live/Dead fixable blue stain was used as a viability dye as above. Plates were then washed with flow buffer, and stained with the appropriate antibodies as indicted in Supplementary Table 2 in a final volume of 50 μL. The cells were then washed twice with flow buffer and fixed in 4 % formalin:PBS before analysis was performed on a BD LSR II flow cytometer with data analyzed using FlowJo v9.9.6 software and the gating strategy in Supplementary Fig. 3.

Analysis of cytokines
Serum was tested for increases in the levels of 11 cytokines specified a priori based on earlier studies [19][20][21][22]: IL-4, IL-6, IL-10, TNF, IL-12p70, IFN-γ, MCP-1, MIP-1α, MIP-1β, and IP-10 were analyzed in triplicate by multiplex immunoassays (Bio-Plex Pro Human Cytokine 27-plex) according to the manufacturer's instructions (BioRad, Hercules, CA), and analysis of RANTES was by ELISA (R&D systems, Minneapolis, MN). The assay output was the median fluorescence intensity (MFI) per bead for each bead classifier. All the negative MFI values, where a cytokine's florescence level was lower than the background, were set to zero. Heatmaps were generated using Z-scores, data was separated by patient number and then further stratified based on cytokine before Z-scores were calculated. This was done using the formula Z-score = (x -μ) / σ; where x = an individual MFI replicate value, μ = the mean and σ = the standard deviation, of all the replicates across all the time points for each cytokine. This was performed in an iterative manner for each patient's cytokine data using R software, and finally visualized using Prism software.