This trial tested a dendritic cell (DC) therapeutic cancer vaccine in which antigen is loaded using a novel non-viral transfection method enabling the uptake of plasmid DNA condensed with a cationic peptide. Proof of principle required the demonstration of diverse T lymphocyte responses following vaccination, including multiple reactivities restricted through both major histocompatibility complex (MHC) class I and II. Patients with advanced melanoma were offered four cycles of vaccination with autologous DC expressing melan A and gp100. Disease response was measured using Response Evaluation Criteria in Solid Tumours. Circulating MHC class I- and II-restricted responses were measured against peptide and whole antigen targets using interferon-γ ELIspot and enzyme-linked immunosorbent assay assays, respectively. Responses were analyzed across the trial population and presented descriptively for some individuals. Twenty-five patients received at least one cycle. Vaccination was well tolerated. Three patients had reduction in disease volume. Across the trial population, vaccination resulted in an expansion of effector responses to both antigens, to the human leukocyte antigen A2-restricted modified epitope, melan A ELAGIGILTV, and to a panel of MHC class I- and II-restricted epitopes. Vaccination with mature DC non-virally transfected with DNA encoding antigen had biological effect causing tumour regression and inducing diverse T lymphocyte responses.
Dendritic cells (DCs) comprise overlapping subsets of cells that have central roles in activating and regulating T and B lymphocyte function in response to microbial challenge.1 Their potential role as adjuvants in vaccine strategies against malignancy has been extensively exploited using diverse methods of antigen loading. Vaccination using whole antigens offers significant advantages over synthetic peptide epitopes because it is independent of major histocompatibility complex (MHC) haplotype and of the previous identification of epitopes. Individual patients might amplify a response through multiple reactivities specific for diverse epitopes and MHC restriction elements. In principle, both MHC class I- and II-restricted cytotoxic T lymphocytes (CTLs) and helper T cells (THC) might be stimulated. Exploiting their evolved nature, transduction using recombinant viruses encoding antigen can result in efficient gene expression. However, there are limitations to the use of viruses in the clinical setting including production, recombination, safety, cost, toxicity and generating immunodominant reactivities against irrelevant targets.2, 3 Non-viral DNA transfection by electroporation results in efficient gene expression and is widely used in varied cell types. Electroporation of vaccine cells using DNA has been rejected in favour of RNA because the latter resulted in better cell viability and whole antigen detection. Furthermore, in T-cell assays, DC electroporated with DNA, but not RNA, induced nonspecific responses masking antigen recognition.4 The use of RNA-electroporated DC has been exploited in several trials.5, 6, 7, 8, 9 Nonetheless, in the setting of cellular vaccination, non-viral DNA transfection retains advantages that DNA is more robust and amenable to standardization and vaccine generation requires fewer steps than for RNA.
CL22 is a 35 amino acid cationic peptide that comprises a lysine rich N-terminus, C-terminal influenza nucleoprotein sequence and a linking cleavage site for endosomal cathepsin. The biophysical properties have been described that result in CL22-condensed DNA efficiently transfecting a wide range of mammalian cells including primary DC.10 Transfection of human DC, differentiated from monocytes, involves transient co-incubation of cells with CL22-condensed DNA with chloroquine as an endosome escape agent. It was reported that functional whole protein was detected in 1–54% of the cell preparations, with most exceeding 10%. This significantly exceeded that achieved using other DNA transfection strategies. In vitro, human DC, transfected with plasmid encoding antigen, stimulated MHC class I-restricted epitope-specific T-cell responses as efficiently as peptide loaded DC or DC transduced with an adenovirus vector. In a murine model, vaccination with DC transfected using CL22 to express the melanocyte differentiation antigen TRP-2 was protective against challenge with the melanoma cell line B16F1.11
Here, we report the first use in a phase I/II trial of CL22-mediated transfection of human DC with plasmid encoding melanocyte differentiation antigens, melan A and gp100, as a therapeutic vaccine for patients with metastatic melanoma. The primary objective was to determine the safety and tolerability of vaccination. The secondary objectives were to demonstrate and characterize the biological effect of vaccination including both clinical tumour regression and circulating T lymphocyte responses. An important aspect of this trial was to demonstrate the diversity of responses, across all patients and within individuals, to show the particular value of vaccinating with DC presenting two full-length antigens expressed from DNA and processed endogenously.
Recruitment took place over a 3-year period between December 2003 and January 2007. A summary of the consenting participants (n=27, 16 men and 11 women) is shown in Table 1. Eleven participants had significantly raised lactate dehydrogenase measurement reflecting bulky metastatic disease. Ten participants had liver metastases. The median age was 56 (range 37–77 years). Trial interventions are shown in Figure 1.
DC yield and phenotype
A total of 107 DC vaccines were prepared during the study. Typical DC morphology was confirmed by light microscopy. Vaccine cell recovery from 150 ml of whole blood was variable (median 6.6 × 106, quartiles 3.3–10.7 × 106). Cell viability was high (median 91%, quartiles 86–95%). Evidence that the tumour necrosis factor-α and interleukin (IL)-1β cocktail upregulated maturation markers in patients' DC is demonstrated in Supplementary Figure 1.
The proportion of the vaccine preparation that contained intact melan A and gp100 proteins following transfection, detected by immune fluorescence and flow cytometry, was highly variable both within and between patients (Figure 2a). Surface presentation of the human leukocyte antigen (HLA) A2-restricted gp100 epitope was sought using surplus cryo-preserved vaccine cells and re-directed T-cell effectors in interferon (IFN)-γ ELIspot assays. Re-directed effectors recognized lymphoblastoid cell line targets loaded with YLE peptide (peptides are listed in Supplementary Table 1) but not with an irrelevant HLA A2-restricted peptide, GLCTLVAML (GLC). Low but detectable recognition of thawed transfected vaccine preparations was observed (Figure 2b). In CL22/DNA-transfected DC, antigen recognition by T cells was observed despite very low detection of whole antigen in additional experiments using a different antigen system (Supplementary Figure 2).
Twenty-five participants received 1 vaccine cycle of whom 22 experienced mild injection site toxicity mostly limited to local erythema. Thirteen patients experienced constitutional symptoms judged possibly or probably attributable to vaccination rather than to disease. For two patients fatigue and ‘flu-like symptoms were grade 2, all others grade 1. Twenty-four patients experienced anaemia, probably due to disease burden, but possibly exacerbated by repeated venesection for DC preparation. The only occurrence of grade 3 anaemia was attributable to the melanoma. Sixteen serious adverse events were recorded for 12 patients. Only two of these events, severe arm pain (patient 2) and atrial fibrillation noted on routine review post vaccination (patient 10) were graded as possibly attributable to vaccination.
Clinical responses are summarized in Table 1. Of the 25 evaluable patients, 18 experienced progressive disease. Stable disease over the vaccine period was achieved for four patients. Partial response across vaccination was observed for three patients (patients 04, 11 and 22). All three patients ultimately progressed 54, 71 and 50 weeks respectively from the first vaccination. In the context of an overall partial response, patient 04 exhibited an increase in volume of mediastinal lympadenoapthy after four cycles that resolved by cycle 8 (not shown). Patient 22 demonstrated a reduction in tumour volume in the liver but an increase in the lungs following four vaccine cycles. Resolution of detectable lung metastases was noted on routine chest X-ray 6 weeks following cycle 4, confirmed by computed tomography scanning 6 weeks later (Figure 3). The patient underwent four more cycles of treatment. In the context of ongoing remission in the liver and lungs, a new skeletal metastasis developed after the eighth vaccine cycle.
Tumour antigen expression
It was not an eligibility criterion to have demonstrated expression of the target antigens in malignant tissue. However, melan A was positive in malignant cells in 9/9 cases and gp100 in 7/7 cases for which immunohistochemistry results were available. Target antigen expression was retained in cutaneous lesions resected post vaccination for patients 02, 16 and for responding patient 04 whereas responding patient 22 developed a new spinal metastasis with loss of expression of melan A and gp100 during regression of disease at other sites (Supplementary Figure 3).
Immune response to whole antigen
A total of 22 patients were evaluable for immune response. Using serial blood samples from vaccinees, the secretion of IFN-γ by T lymphocytes was measured in response to stimulation with autologous B blasts nucleofected with plasmids encoding whole melan A or gp100 antigen (Figure 4a and Supplementary Table S2). For most patients, compared with pre-vaccination samples, IFN-γ release increased in samples harvested after two (n=1), three (n=11) or four (n=7) vaccination cycles. The increase in IFN-γ release measured from the last sample taken after cycle 2 compared with pre-cycle 1 was statistically significant for both melan A (P=0.05) and for gp100 (P=0.03).
It had been noted that vaccine preparations were heterogeneous in the proportion of cells expressing CD83, a measurement of maturation, and in the proportion of cells in which the target antigens could be detected by immune fluorescence. It was plausible that vaccine cell maturation or detection of whole antigen in vaccine cells might relate to immune outcome in vivo. For the 19 patients with data, the difference between pre- and post-vaccination response to each antigen was plotted against the median proportion of DC expressing CD83, melan A and gp100 for the first four vaccine cycles. Figure 4b shows a trend to an association between immune response and vaccine CD83 expression (melan A—R2=0.21, P<0.05; gp100—R2=0.13, P=0.13) whereas there was no indication of an association between vaccine antigen levels and subsequent immune response (melan A—R2<0.01, P=0.74; gp100—R2=0.02, P=0.57). The three clinical responders all made immune responses that fell within the range of other vaccinees.
Immune response to epitopes within melan A and gp100
In ELIspot assays using serial T-cell samples from HLA-A0201-positive patients, the frequency of effectors recognizing a viral peptide pool decreased (P<0.01, n=9) whereas those against target epitopes appeared to increase (ELA P=0.02, n=12; KTW P=0.08, n=7) comparing responses in paired samples from pre-vaccination and after cycle 4 (Figure 5a).
Serial blood samples were available from 17 patients to measure changes in MHC class I-restricted responses across the vaccination course. Patients contributed 1–6 separate reactivities to the analysis depending on MHC type and availability of cells. Taking the mean of multiple reactivities for each patient, the change over time is shown in Figure 5b. An increase in response was observed, comparing measurements before vaccination with after cycle 4 (n=14, P=0.02). Analysis of MHC class II-restricted epitope-specific responses was carried out for 30 reactivities in eight patients. The mean change in response per patient also suggests an increase in response across the vaccination course (Figure 5b). ELIspot data for MHC class I- and II-restricted responses are detailed in Supplementary Table S3.
Individual patients demonstrated evidence of an increase in epitope-specific reactivities in both gp100 and melan A restricted through both MHC class I and II across the vaccination course (Figure 5c). The magnitude of individual peptide-specific reactivities was typically low: the peak frequencies of up to 60 spot forming cells per well corresponded with a T-cell frequency of 150 responders per million peripheral blood mononuclear cell (PBMC). However, collectively, the response to antigen via multiple reactivities was higher.
Immune response in paired injection site biopsies
Melan A and gp100 were tested and not detected. CD83+ cells were observed in 8/48 biopsies. A CD3+ infiltrate was detectable within 84% of injection site biopsies. This was typically perivascular in distribution and CD4+ cells dominated over CD8+ cells. The intensity of T-cell infiltrate increased from cycle 1 to 4 for only four patients including one responder (patient 11).
Immunotherapy exploiting antigen-specific T lymphocytes can be clinically effective against metastatic melanoma.15 However, achieving a high clinical response rate depends on intensive individualized T-cell infusion strategies limited to a few centres.16 Vaccination to amplify anti-tumour responses in vivo might prove valuable as a component of multi-modality treatment strategies. The use of autologous DC as vaccine adjuvants allows the manipulation of antigen presentation and T-cell stimulatory capacity ex vivo. Non-viral transfection of nucleotide sequences encoding full-length tumour antigens enables autologous selection of epitope targets, avoids the generation of competing immune responses against proteins encoded by the vector and does not require precautions to limit dissemination of gene-modified organisms. Previous trials have used DC transfected with RNA encoding antigen.5, 7, 17, 18, 19 This phase IB trial was designed to show proof of principle that vaccination using DC transfected with plasmid DNA encoding antigens is biologically active, while further extending the data set on safety and tolerability for DC vaccination.
In keeping with data from other trials,5, 7, 17, 18, 19, 20 the administration of four or more cycles of autologous DC vaccination was safe and well tolerated. The tumour response rate was 12%, including effect on metastases in the liver, lungs, lymph nodes and cutaneous sites (Figure 3), with a further 16% not progressing over the 12-week vaccination period. This level of activity against established disease is consistent with published data. In 14 trials of vaccination for metastatic melanoma using matured DC, the overall rate of reduction in tumour burden (complete, partial and mixed responses) was 10% (26/262). In all, 13 of 14 trials used exogenous antigen as peptide or antigen, one used autologous RNA transfection.21 Initial tumour expansion before response was observed. A similar phenomenon is now recognized in the context of anti-CTLA4 treatment: in one reported case, a resected lung lesion exhibited necrosis and a T-cell infiltrate without malignant cells. This pattern is now incorporated in the design of immunotherapy trials.22 The development of an aggressive de-differentiated metastasis in patient 22 is consistent with a recognized phenomenon, the selection by an effective immune response of malignant variants with loss of antigen or of antigen-presenting machinery.23, 24, 25 Conversely, long-term survival from melanoma might result from dynamic T-cell recognition responding to sequential immune escape variants.26 The probability of an antigen-loss mutant being selected might be reduced by including more diverse antigens within the vaccine preparation and by combining vaccination with aggressive cytoreduction using other treatment modalities when possible, for example, through metastatectomy.
The results of immune assays illustrated key points of proof of mechanism for this vaccine. Evidence of vaccine immunogenicity was sought by using ELIspot and enzyme-linked immunosorbent assay (ELISA) data across all individuals as continuous variables falsifying the hypothesis that the results after vaccination were drawn from the same data set as those before vaccination. A commonly used approach is, instead, to prospectively define ‘response’ as a discrete variable for individuals and count the proportion of responding patients,27 as shown in the Supplementary data. However, it was not possible to determine individual response with statistical significance because each donated too few independent post-vaccination blood samples. Furthermore, it is not certain what T-cell amplification threshold determines clinical utility whereas very small increases in the frequency of circulating activated T cells might translate to clinical benefit.28 The analytical approach used here, using immune outcome data as continuous variables, enabled proof of principle sufficient to justify further development of this vaccination strategy. Immune responses increased across the vaccination course as shown in IFN-γ ELISA assays using autologous cells nucleofected to express whole melan A and gp100 (Figure 4a), and in ELIspot assays for the HLA A2-restricted substituted melan A epitope, ELA (Figure 5a), and for all MHC class I-restricted responses analyzed together (Figure 5b). This increase occurred against epitopes within the vaccine whereas responses to an irrelevant virus-encoded epitope declined (Figure 5a). Responses to the few known MHC class II-restricted peptide epitopes in gp100 and melan A were demonstrated in ELIspot assays apparently increasing across the vaccine cycles (Figure 5b). CD4+ THC have been shown to have important roles against tumours and to support the generation and maintenance of CD8+ CTL responses. Within individuals, multiple reactivities within both target antigens, restricted through both MHC class I and II, were amplified following vaccination during an overall reduction in disease volume (Figure 5c). This is consistent with a previous report that RNA-transfected DC in vitro can induce T-cell responses against multiple epitopes across multiple MHC restrictions and that co-transfection of two antigens into the same DC, as in this trial, can result in stimulators that activate responses against both antigens.29 In summary, this vaccine strategy was clearly biologically active, being both immunogenic and resulting in reduction in volume and in immune editing at the site of malignancy.
A novel peptide-based method was used to condense DNA for transfection. In 107 separate DC vaccine transfections, we found great variability in the measurement by flow cytometry of the proportion of cells positive for the antigen, labelled with a monoclonal antibody (Figure 2) and no suggestion of association with clinical outcome. It might be argued that such variation in antigen expression, including very low levels, hampers the application of this protocol in clinical trials because standardization is a prerequisite for large-scale application in multicenter studies. However, this protocol did result in both immune and clinical responses. Furthermore, apparently inefficient transfection methods, determined by expression of GFP, have been reported as being immunogenic for MHC class I-restricted epitopes when used for other antigens.11, 30 As immunogenicity depends on protein degradation, it is suggested, but not proven, that using immunofluorescent detection of whole antigen to measure transfection efficiency may not be informative about immunogenicity.2 Antigenic epitopes are not only derived from stable full-length proteins but from newly synthesized polypeptides that are rapidly degraded. In other cell types, about one-third of newly synthesized proteins are degraded within 15 min of expression.31, 32, 33 Most epitopes are derived from proteins expressed and degraded within 2 h. It has been reported that proteasome inhibition triples recovery of cellular proteins, many of which are ubiquitinated, suggesting that two-thirds of cell proteins are degraded shortly after or during synthesis. This recovery is blocked rapidly by inhibiting new protein synthesis.34 In post-trial assays, we showed that T-cell effectors re-directed to recognize gp100 YLE epitope responded to thawed vaccine cells from the trial. Epitope-specificity is inferred because the same effectors recognized peptide-loaded HLA A2+ lymphoblastoid cell line targets only when these were sensitized by the YLE peptide. Recognition did not appear to relate to the level of gp100 detection in the DC (Figure 2b). This dissociation between antigen detection and epitope presentation in CL22/DNA-transfected DC was also observed using another antigen system (Supplementary Figure S2). We have preliminary in vitro data that, following RNA transfection, intact melan A is detected only very briefly whereas MHC class I epitopes continue to be presented in DC that are apparently melan A-negative by immune fluorescence (Rao A, Steele J manuscript in preparation). In summary, for transfected DC vaccine preparations, effective translation and immunogenicity is not usefully measured by detecting whole antigen by immunofluorescence. Standardization of individual vaccine preparations for trials may require demonstration of MHC-peptide presentation through T-cell assays or non-cellular assays using MHC-peptide-specific soluble T-cell receptors or antibodies.
In principle, endogenous antigens can access the MHC class II presentation pathway: membrane or secreted proteins can associate with MHC class II in the endoplasmic reticulum,35 long-lived cytoplasmic proteins access endosomes through autophagy36, 37 and intercellular transfer of antigen can occur.38 Previous reports indicated that RNA-transfected DC can present epitopes through MHC class II via cytosolic and endosomal pathways.39 It has been reported that presentation by DC vaccines to CD4+ effectors is enhanced using antigen tagged with the lysosomal targeting sequence of lysosome-associated membrane protein-117, 40 whereas in this trial, the native antigen sequences were used. In this setting, high levels of whole antigen may be required for lysosomal access to occur in competition with rapid proteosomal processing.41 Thus, the variability between preparations in detection of whole melan A and gp100 might not be optimal for stimulation of CD4+ responders.
Ex vivo DC are more effective vaccines if they are exposed to activating stimuli42 but there is currently no standard process to mature DC vaccines.21 In this study, vaccine preparations did respond to tumour necrosis factor-α and IL-1β by altering the cell surface phenotype (Supplementary Figure 1), but DC functional properties were not tested. Upregulation of CD83 was highly variable between vaccine preparations and we recognize that some vaccine preparations might have been less immunogenic as a result of incomplete maturation. Indeed, we observed that the median proportion of CD83+ DC on cycles 1–4 might account for 21% of the variability in vaccine-specific immune responses to melan A and 13% to gp100 (Figure 4b). Although the P-value was <0.05 for the association of vaccine CD83 expression with melan A response, this should be regarded as an exploratory analysis, albeit an association which is biologically plausible. Interestingly, in the same analysis, there was no trend to an association between antigen detection in DC and immune response, a result that might be predicted following the reasoning discussed above. The maturation cocktail was selected on the evidence at the start of the trial and availability to us of Good Manufacturing Practice grade materials. Prostaglandin E2 has been shown to increase surface expression of CD8343 and of the lymph node homing marker, CCR7, but was omitted because it also inhibits IL-12 production.44, 45 With very low expression of CCR7, the cutaneous route, particularly injection of half the cells subcutaneously, was likely to have been sub-optimal because of inadequate migration to lymphoid tissue. Ligation of Toll-like receptors can result in a DC functional phenotype that includes IL-12 secretion, CCR7 expression and migratory properties46 whereas, in an animal model, DC activated by inflammatory mediators but without direct Toll-like receptor ligation can adopt an activated surface marker phenotype without driving TH1 differentiation.47 Future trials with fully activated DNA-transfected DC, and more consistency in maturation between vaccines, may achieve much higher circulating T-cell responses and more prominent delayed type hypersensitivity at injection sites than observed in this trial.
In conclusion, vaccination of patients with advanced melanoma with autologous DC, transfected by peptide-condensation of a plasmid encoding whole antigens, was safe and well tolerated. There was clear evidence for vaccine immunogenicity in vivo and that this approach could result in effect on malignancy through tumour regression and possibly immune editing. For a DC vaccination strategy to be clinically effective, it will probably require combination with systemic treatments to mitigate the effects of a tolerizing environment in the host.48 The level of activity in our trial does support the use of DC, non-virally transfected with DNA encoding gp100 and melan A, as the vaccine component in an experimental combinatorial strategy in future trials.
Materials and methods
This was a non-randomized open label phase I/II trial of DC vaccination at one dose level in one centre. The primary objective was to determine the safety and tolerability of sub-cutaneous and intra-dermal administration of autologous DC transfected with a plasmid encoding both melan A and gp100 using the CL22 cationic peptide. The secondary objectives were (i) to assess anti-tumour efficacy and (ii) to measure T-cell responses. The study was approved by the Gene Therapy Advisory Committee (study 084; 10 July 2003) and the South Birmingham Research Ethics Committee (203/298; 22 October 2003). Patients gave written informed consent to participate.
Patients had stage IV melanoma for which no conventional therapy was available. Other eligibility criteria were the presence of measurable disease according to Response Evaluation Criteria in Solid Tumours, adequate bone marrow function (white blood cells >4.0 × 109 l–1, haemoglobin >10.0 g dl–1 platelets >100 × 109 l–1), adequate renal function creatinine <150 μmol l–1), adequate liver function (billirubin normal <1.5 × upper limit of normal, albumin >30, serum transaminases and alkaline phosphatases <1.5 × upper limit of normal or 3 × upper limit of normal if patient had liver metastases), age >18 years, performance status of 0–1. Patients were not eligible if there was concurrent systemic corticosteroid treatment, active autoimmune disease, clinically significant ischaemic heart disease or cardiac failure, concurrent infection, co-existing or previous other malignancies (unless in complete remission for not <3 years, and excepting in situ carcinoma of the cervix or basal cell carcinoma of the skin), or if there had been a previous splenectomy or radiotherapy to spleen, surgery (other than excision of skin metastases), chemotherapy, radiotherapy, IFN or experimental therapy within the previous 4 weeks. Hepatitis B and C and human immunodeficiency virus serology were required to be negative. Resection of skin metastases under local anaesthetic was permitted without deferring vaccination. Patients were permitted to continue to receive the full vaccination course despite clinically evident disease progression on early cycles.
Preparation of vaccine
At each treatment cycle, 150 ml of heparinized blood was obtained and PBMCs isolated by density gradient centrifugation over Lymphoprep (Axis-Shield, Oslo, Norway). DC were generated from adherent PBMC (non-adherent cells were cryopreserved for use in immune assays) by culture in RPMI 1640 (Gibco Invitrogen, Paisley, UK) containing 10% fetal calf serum (FCS) (TCS CellGenix, Freiburg, Germany), 2 mM glutamine (Gibco Invitrogen), 500 U ml–1 granulocyte-macrophage colony-stimulating factor (TCS CellGenix) and 500 U ml–1 IL-4 (TCS CellGenix). Plasmid pCV222 is based on Cobra Bio-Manufacturing plc (Keele, UK) proprietary Operation Repressor Titration plasmid maintenance system and does not contain antibiotic resistance genes. The gp100 and melan-A open reading frames are inserted sequentially under independent human Cytomegalovirus and murine Cytomegalovirus promoters, respectively. On day 4, immature DC were harvested and transfected with pCV222 using a non-viral transfection system, CTL901, according to the method described by Irvine et al.11 pCV222 DNA was condensed using a short cationic peptide (CL22) by mixing for 1 h at room temperature in the precise ratio pCV222:CL22 of 1:2.3. Transfection of DC was carried out by the addition of the condensation mixture to cells at a concentration of 1.2 × 106 cells ml–1 and incubation for 1.5 h at 37 °C in the presence of 80 μM chloroquine added as an endosome escape agent. Following transfection, cells were placed in maturation medium comprising RPMI 1640, 10% FCS, 2 mM glutamine, 500 U ml–1 granulocyte-macrophage colony-stimulating factor, 500 U ml–1 IL-4, 20 ng ml–1 tumour necrosis factor-α (TCS CellGenix) and 10 ng ml–1 IL-1β (TCS CellGenix) for a further 48 h. Pre-clinical optimization had been undertaken by the manufacturer. This demonstrated, first, that transfected DC, cultured in FCS and matured using IL-1β and tumour necrosis factor-α, resulted in secretion of IL-12 in preference to IL-10. Second, using enhanced green fluorescent protein as a reporter gene, the optimal transfection conditions were selected. Third, DC transfection resulted in detectable RNA transcripts for gp100 and melan A (CTL-901 Investigators' Brochure—ML Pharmaceuticals plc, now part of Vectura Group plc, Chippenham, UK). Mature DCs were prepared for immunization by washing and resuspending in saline. Up to 5 × 106 DCs were suspended in saline at approximately 107 ml–1. Vaccine quality was monitored using a small number of reserved cells. Morphology was monitored by direct inspection on light microscopy and viability was measured by Trypan blue exclusion. The expression of cell surface antigens on both differentiated and matured DC was measured by flow cytometry (Beckman Coulter Epics XL-MCL, Fullerton, CA, USA) using the phycoerythrin-conjugated monoclonal antibodies anti-human CD14, CD83, CD86, CD25 (Serotec, Oxford, UK), CD1a (BD Biosciences Pharmingen, San Diego, CA, USA), and CCR7 (R & D Systems, Oxford, UK). The presence of intact target antigen was measured by fixing and permeabilizing mature DC (Intraprep Permeabilization Reagent Kit, Beckman Coulter) and staining for the expression of melan A and gp100 using specific primary antibodies followed by a secondary phycoerythrin-labelled goat anti-mouse immunoglobulin G1 (all from Dako, Glostrup, Denmark). The manufacturers' instructions were followed for all reagents and antibodies, and isotype-matched controls were used throughout. Release criteria were that the vaccine preparations should exhibit >90% viability, should be free of infection by visual inspection, and should have completed the process of differentiation, maturation and transfection without deviation from the standard operating procedures. All DC preparations met these criteria. Measurement of cell surface markers of lineage and maturation and target antigen served as markers to be reported rather than release criteria.
Vaccination dose and schedule
According to protocol, four vaccine cycles were offered three times a week (Figure 1). For each cycle, the vaccine preparation was generated from a unique 150 ml venesection and only fresh DCs were used. Up to 5 × 106 DCs were divided between intra-dermal and sub-cutaneous sites close to intact lymph node groups. If clinical benefit was observed, further treatment cycles were undertaken with Gene Therapy Advisory Committee approval. Patients needed to complete one cycle to be evaluable for toxicity and two cycles to be evaluable for immune response.
Patients were monitored clinically for 4 h following the first injection and for 2 h following subsequent injections. Adverse events were documented on each cycle using the Common Toxicity Criteria version 3.1 (National Cancer Institute, Bethesda, MA, USA). Response was determined using Response Evaluation Criteria in Solid Tumours. At screening and following the second and fourth cycle, clinically detectable metastases were mapped and counted and bi-directional calliper measurements of up to six reference skin lesions (>1 cm diameter) or enlarged lymph nodes were recorded. Cross-sectional imaging was undertaken at screening and following the fourth cycle, or when early withdrawal from the trial was mandated by disease progression. Blood samples were harvested using preservative free heparin during screening, after cycle 2 (week 6) and twice after cycle 4 (weeks 12 and 16) for measurement of T-cell responses. PBMC were isolated by density gradient centrifugation and cryopreserved in the vapour phase of nitrogen.
Cell lines and reagents
B-cell blasts were grown from patient PBMC by culture in IMDM (Gibco Invitrogen) containing 10% pooled human serum (HD Supplies, Aylesbury, UK), 5 ng ml–1 IL-4 (R & D Systems), 0.67 μg ml–1 cyclosporin A and 2 mM glutamine. PBMC were seeded at 1.5 × 106 cells ml–1 onto irradiated L cells (transfected with CD40L) in six-well plates (3 ml per well) and passaged every 2/3 days.12 B-cell blasts between passages 4 and 9 were considered suitable for use as antigen-presenting cells in the immune assay and were cryopreserved. The panel of synthetic target epitopes (Alta Bioscience, Birmingham, UK) within melan A and gp100 is listed in Supplementary Table 1. Plasmids encoding full-length melan A and gp100 were manufactured by Plasmid Factory GmbH & Co. KG, Bielefeld, Germany. The T-cell receptor anti-CD3 (TCR-CD3) is a prototype soluble fusion protein with a high affinity for HLA A2-gp100 280–288, generated by TCR phase display and directed evolution13, 14 and was a kind gift from Immunocore Ltd (Oxford, UK).
All samples from the same patient were thawed and analyzed simultaneously without in vitro stimulation. Cells were thawed and recovered overnight in RPMI1640 containing 10% FCS, counted, then seeded in triplicate wells (4 × 105 per well). Synthetic peptides representing known MHC class I- and II-restricted epitopes, derived from melan A and gp100, selected according to HLA type, were added to a final concentration of 5 μg ml–1. IFN-γ release was measured using a commercially available ELIspot kit (Mabtech, Nacka Strand, Sweden) using cytokine capture and detection reagents according to the manufacturers' instructions. Negative control wells contained diluted dimethylsulphoxide. Positive control wells contained 20 μg ml–1 PHA (Sigma-Aldrich, Dorset, UK) and, for HLA-A0201-positive patients, an HLA-A0201-restricted vial peptide pool. Spots were counted using the AID Elispot Reader (Autoimmun Diagnostika GmbH, Strassberg, Germany). An external positive control was included with every run of the assay. This comprised PBMC isolated from the same HLA-A0201-positive buffy coat sample (National Blood Service, Birmingham, UK), cryopreserved in multiple aliquots. At each sample and reactivity, the spot count per 4 × 105 cells was calculated by subtracting the mean spot count of the negative control triplicate from that obtained for the target peptide epitope(s).
Re-directed ELIspot assays were undertaken using a CD8+ HLA B35 restricted Epstein Barr Nuclear Antigen-1-specific T-cell clone. Targets were an HLA A2+B35-ve Epstein Barr virus-transformed B lymphoblastoid cell line, loaded with either gp100 YLE or EBV BMLF1 GLC peptide (5 μM), and cryo-preserved CTL901-transfected patient-derived HLA A2+ DC. Effectors (103) and DC (5 × 103) were added to each ELIspot well with gp100 280–288 TCR-CD3 fusion protein (10−9 M) according to the manufacturer's protocol.
ELISA assays against nucleofected targets
B-cell blasts were nucleofected (Amaxa, Cologne, Germany) with 10 μg of plasmid DNA (encoding melan A, gp100, or empty plasmid) according to the manufacturer's instructions using the Human B-Cell Nucleofector Kit. Matched pairs of PBMC, or non-adherent cell populations from screening, and from subsequent cycles were thawed and rested in RPMI 1640 medium containing 10% FCS overnight. These responders (105 per well) were co-cultured in 96-well plates for 16 h with autologous nucleofected B-cell blasts (105 per well) in triplicates. Transfection conditions had been optimized before these experiments were undertaken. Every run of the assay included internal positive control wells comprising only responder cells and 20 μg ml–1 PHA, and also an external positive control comprising PBMC isolated from a buffy coat sample (cryopreserved in multiple aliquots) cultured with autologous B-cell blasts (batch-produced in advance and cryopreserved in multiple aliquots) nucleofected with plasmid DNA (5 μg) encoding the EBV protein BMLF1 or control plasmid. After overnight incubation, culture supernatants were harvested and IFN-γ was measured by ELISA (Endogen) following the manufacturer's recommended protocol.
In ELIspot assays, the spot count per 4 × 105 cells at screening was compared with that at later time points. This is presented descriptively for individual patients and, for the whole patient population, in spider diagrams in relation to each individual's pre-vaccination measurements. Paired spot counts at screening and after cycle 4 were compared using measurements from all patients with available data across the trial. In the ELISA assays, paired estimates of IFN-γ release at screening and at the latest cycle for which there were available cell samples were compared. In ELIspot and ELISA assays, paired data were tested using the Wilcoxon matched pairs signed-rank test for non-parametric data to falsify the null hypothesis that results at baseline and post vaccination were drawn randomly from the same data set. The relationship between vaccine cell parameters (CD83 and antigen expression) and in vivo immune response using IFNγ ELISA data were analyzed by linear regression. Linearity was confirmed by a runs test for all analyses. All analyses were undertaken using GraphPad Prism 5.0 (GraphPad Software Inc., La Jolla, CA, USA).
Immunohistological examination of tumour samples and vaccine site skin biopsies
On cycles 1 and 4, 0.05 ml of the vaccine was delivered intra-dermally at an additional site, and a punch biopsy was taken at 48–72 h. Paraffin-embedded formalin-fixed 3 micron sections were stained for CD3, CD4, CD8, CD83, melan A and gp100 using conventional methods (antibodies for melan A, HMB45 from Dako UK Ltd (Ely, UK), antibodies to CD3, CD4, CD8, CD83 from Leica Microsystems (UK) Ltd., Milton Keynes, UK) and a Dako Autostainer. Intensity of infiltration with immune cells was graded from 0 to 3 by the histopathologist.
The vaccine materials were generously provided by Innovata Ltd. (part of Vectura Group plc). The trial was supported through the Birmingham Experimental Cancer Medicine Centre and the Cancer Research UK Clinical Trials Unit, Birmingham. The trial was funded through an educational grant from the Moulton Charitable Trust. Clinical activity took place in the Wellcome Trust Clinical Research Facility, University Hospital NHS Foundation Trust, Birmingham, UK. Preparation of vaccine grade cells was undertaken in the National Blood Service, Vincent Drive Birmingham. Immunocore Ltd. generously provided the gp100 280–288 TCR-CD3 fusion protein and technical expertise in setting up the re-directed T-cell assays. We are grateful for being allowed use of this prototype reagent before it has been formally described in the literature.
About this article
Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)
Amino Acids (2012)