Clinical and immunological responses in metastatic melanoma patients vaccinated with a high-dose poly-epitope vaccine

Abstract

Background

Safety and cellular immunogenicity of rising doses and varying regimens of a poly-epitope vaccine were evaluated in advanced metastatic melanoma. The vaccine comprised plasmid DNA and recombinant modified vaccinia virus Ankara (MVA) both expressing a string (Mel3) of seven HLA.A2/A1 epitopes from five melanoma antigens.

Methods

Forty-one HLA-A2 positive patients with stage III/IV melanoma were enrolled. Patient groups received one or two doses of DNA.Mel3 followed by escalating doses of MVA.Mel3. Immunisations then continued eight weekly in the absence of disease progression. Epitope-specific CD8+ T cell responses were evaluated using ex-vivo tetramer and IFN-γ ELISPOT assays. Safety and clinical responses were monitored.

Results

Prime-boost DNA/MVA induced Melan-A-specific CD8+ T cell responses in 22/31 (71%) patients detected by tetramer assay. ELISPOT detected a response to at least one epitope in 10/31 (32%) patients. T cell responder rates were <50% with low-dose DNA/MVA, or MVA alone, rising to 91% with high-dose DNA/MVA. Among eight patients showing evidence of clinical benefit—one PR (24 months+), five SD (5 months+) and two mixed responses—seven had associated immune responses. Melan-A-tetramer+ immunity was associated with a median 8-week increase in time-to-progression (P = 0.037) and 71 week increase in survival (P = 0.0002) compared to non-immunity. High-dose vaccine was well tolerated. The only significant toxicities were flu-like symptoms and injection-site reactions.

Conclusions

DNA.Mel3 and MVA.Mel3 in a prime-boost protocol generated high rates of immune response to melanoma antigen epitopes. The treatment was well tolerated and the correlation of immune responses with patient outcomes encourages further investigation.

Introduction

In metastatic melanoma spontaneous regressions are occasionally observed and have been associated with tumour infiltration by immune cells [1]. Amongst these immune cells, peptide-specific CD8+ T lymphocytes with the ability to lyse melanoma cells and secrete cytokines are seen [2, 3]. Melanoma antigens recognised by CD8+ T cells have been identified and are targeted in therapeutic vaccine approaches using full-length antigens or specific epitopes [4, 5] and a variety of antigen delivery systems. However, by applying standard criteria to evaluate tumour response, the evidence for efficacy of therapeutic vaccination has been limited [6].

Plasmid DNA vaccines and recombinant viral vectors are particularly suited to inducing CD8+ T cell responses as they introduce antigens into the MHC class I antigen processing and presentation pathway. Clinical studies with plasmid DNA vaccines in cancer patients have shown only modest immune responses and little clinical efficacy [7, 8]. However, viral vectors activate strong innate immune responses facilitating recruitment of antigen-specific T cells through toll-like receptor (TLR) dependent and TLR-independent pathways [9].

In this study, plasmid DNA (pSG2.Mel3) and a recombinant poxvirus vector based on modified vaccinia virus Ankara (MVA.Mel3) were evaluated. Both vectors express a polyepitope string, Mel3, targeting seven CD8+ T cell epitopes derived from five melanoma antigens: tyrosinase, Melan-A/MART-1, MAGE-A1, MAGE-A3 and NY-ESO-1 [10]. The epitopes are restricted mainly by HLA-A2 but in the case of MAGE-A1 by HLA-A1 only. Using both plasmid and viral vector in a prime-boost strategy may optimise the induction of melanoma-specific CD8+ T cells.

A phase I study of this approach in melanoma patients in an adjuvant setting indicated good tolerability but inadequate immune responses [10]. Two doses each of 1 mg DNA.Mel3 and 5 × 10⁷ plaque-forming units (pfu) MVA.Mel3 were administered at 2-week intervals in this earlier study. Results from preventive vaccine studies suggest that increasing doses of DNA and particularly MVA improves T cell responses further [11]. We hypothesised that increasing doses of DNA.Mel3 and MVA.Mel3 would result in stronger melanoma-specific CD8+ T cell responses and an increased frequency of vaccine responders, and so doses were increased with DNA.Mel3 doubled and MVA.Mel3 escalated 20-fold during the trial. To extend the dosing period and thus hopefully the duration of immune response the interval between initial vaccinations was increased to 3 weeks and subsequently additional, less frequent MVA.Mel3 boosts were used to enhance or maintain T cell responses.

This multi-centre, non-randomised, open-label, dose-escalation study set out to determine the tolerability of, and cellular immune response to, rising doses and varying regimens of DNA.Mel3 and MVA.Mel3 vaccines in patients with unresectable Stage III or IV melanoma. Tumour responses and survival data were recorded to correlate with immunological data.

Patients and methods

Investigational agents

DNA.Mel3 and MVA.Mel3 were manufactured by Qiagen GmbH/DSM Biologics and Impfstoffwerk Dessau-Tornau GmbH (IDT) (Dessau, Germany), respectively, according to cGMP guidelines. The Mel3-encoded epitopes are tyrosinase1–9, melan-A26–35 analogue (substitution Ala to Leu at residue 2725), tyrosinase369–377,26 MAGE-3168–176, MAGE-3271–279, MAGE-1161–169, NY-ESO-1155–167. A murine H2-Db restricted influenza virus (flu) nucleoprotein epitope 27 was included in the construct for preclinical analyses (see [12] for full insert sequence). The tumour antigen epitopes are restricted by HLA-A* 0201, except MAGE-3168–176 and MAGE-1161–169, which are HLA-A*01-restricted.

Patients

Patients were human leukocyte antigen (HLA) A2 positive, had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, life expectancy greater than 3 months, adequate bone marrow (including normal lymphocyte count), renal and hepatic function.

Patients were not eligible if they had brain metastases; surgery, chemotherapy, or radiotherapy within 4 weeks of trial entry; immunotherapy or any other investigational drug within 6 weeks; prior treatment with gene therapy at any time; or treatment with systemic corticosteroids or other immuno-suppressants within 6 weeks. The study protocol was approved by the relevant national regulatory authorities and local ethics committees. All patients provided written informed consent.

Study design

The primary objectives were to assess the tolerability of, and immune responses generated by, rising doses and varying regimens of DNA.Mel3 and MVA.Mel3. Tumour response assessment was a secondary objective. Patients were assigned sequentially to one of seven treatment groups to assess three different aspects of this prime-boost regimen (Table 1). After considering evidence from an earlier malaria challenge study [11], a dose of DNA.Mel3 up to 4 mg and dose of MVA.Mel3 up to 1 × 10⁹ pfu were selected. To evaluate the potential benefit of a rapidly induced immune response, a shortened 6-week regimen with one 4-mg DNA.Mel3 priming dose was compared to a 9-week regimen with two 2-mg DNA.Mel3 doses. The need for DNA priming and effect of increasing the MVA.Mel3 boosting doses (5 × 10⁷ – 1 × 10⁹ pfu) were assessed by including a control group that received MVA alone at the highest-tolerated dose, and by increasing MVA.Mel three doses between groups.

Table 1.

Treatment regimens

Regimen	Number of Patients	DNA ‘Prime’ (DNA.Mel3) (mg)	MVA ‘Boost’ (MVA.Mel3) (pfu doses)
Group 1	8	Two 2	Two 5 × 107
Group 2	5	One 4	Two 5 × 107
Group 3	5	Two 2	Two 2 × 108
Group 4	7	Two 2	Two 5 × 108
Group 5	6	Two 2	Two 1 × 109
Group 6	5	None	Four 1 × 109
Group 7	5	One 4	Two 1 × 109

Vaccine-related dose-limiting toxicities were defined as all grade IV, or any grade III toxicity with the exception of skin rash/desquamation, fever, haematological toxicity or hepatotoxicity reverting to baseline within 2 weeks. Dose limiting toxicities were reviewed by a safety committee, consisting of the investigators and the sponsor, prior to escalating the dose of MVA.Mel3. Patients received DNA.Mel3 via the intramuscular (im) route and MVA.Mel3 via the intradermal (id) route in the same arm. In the absence of disease progression, additional MVA.Mel3 immunisations were given at weeks 16 and 24. Two patients elected to receive additional off-protocol boosters at 12 week intervals on a named-patient basis.

Evaluations

Baseline investigations included medical history, physical examination and computed tomography (CT) of head, chest, abdomen and pelvis. Blood samples were taken for haematology, biochemistry and immunological assays at baseline, just prior to and 1 week after each vaccination.

Tumour assessments by examination and CT scanning were performed every 8 weeks until disease progression. Response was calculated using Response Evaluation Criteria in Solid Tumours (RECIST) [13]. All patients were followed for survival. Adverse events were graded according to the National Cancer Institute Common Toxicity Criteria (version 2.0).

Immunological assessments

Blood (50 mL) samples were collected at screening, just prior to and 1 week after each vaccination. Peripheral blood mononuclear cells (PBMC) were isolated and cryopreserved using standard laboratory methods. Frequencies of Melan-A/A2 analogue-specific CD8+ T cells were determined by a tetramer assay [10] using an MHC class I tetramer–PE conjugate (Proimmune, UK). Tetramer responders were classified as those patients for whom the mean percentage of tetramer+/CD8+ T cells exceeded the baseline mean by >2 standard deviations (sd) at one or more time points. Frequencies of Mel3 epitope-specific IFN-γ-secreting T cells were measured by ex vivo enzyme-linked immunospot (ELISPOT) assay [10]. ELISPOT responders were defined as those patients for whom the mean spot forming cells (SFC) per million exceeded the mean of the negative control responses by >3 sd SFC per million. Sample time points from each patient were analysed in the same experiment to minimise inter-assay variability, and all analyses were carried out at the same central laboratory. PBMC sample quality was assessed and samples with viability less than 80% were not included in the analysis. Anti-dsDNA and anti-MVA antibodies were determined before treatment and at week 16.

End Points and Statistical Analyses

The primary endpoints were tolerability and immunogenicity (induction of melanoma-specific CD8+ T cell responses) of the vaccine. Secondary endpoints included tumour response, time-to-progression (TTP) and survival. A sample size of five subjects per treatment group was chosen as sufficient to assess tolerability and cellular immune responses.

The primary analysis for melanoma-specific CD8+ T cell response was conducted by comparing the proportion of responders in each group using the Fisher’s exact test. Individual patient data profiles were plotted over the trial period. TTP and overall survival were analysed using 3-year Kaplan–Meier estimated survival curves for each group; differences between groups was assessed using the log-rank test. Cellular immune and tumour response rates were tested using the Fisher’s exact test.

Results

Patient population

Eligible patients had histologically confirmed unresectable stage III or stage IV melanoma measurable by clinical examination or CT scan. Between March 2003 and December 2004, 41 patients were enrolled. The intention-to-treat analysis included all patients; two were excluded from the per protocol analysis due to non-compliance. Thirty-seven patients (90%) had stage IV disease (Table 2); a larger proportion of patients with more advanced M1c disease were recruited to the higher-dose treatment Groups 4–7 (12/23;52%) compared to Groups 1–3 (1/18; 6%). Twenty-six patients (63%) had received at least one prior drug therapy and 18 of these were allocated to Groups 4–7. Thirty-three patients (80%) completed the initial course of treatment and 12 patients received at least one additional MVA.Mel3 boost immunisation.

Table 2.

Baseline characteristics of patients

Characteristic	N	Percentage (%)
Age
Median	64.0
Range	(27–86)
Gender
Male	20	49
Female	21	51
HLA status
HLA-A2	34	83
HLA-A2 + A1	7	17
ECOG
0	41	100
1	0	0
AJCC staginga
M0	4	10
M1A	7	17
M1B	17	41
M1C	13	32
Number of previous drug therapiesb
1	9	22
2	10	24
≥3	7	17
LDH
Normal	30	73
>1 to <2× ULN	11	27

^aAJCC, American Joint Committee on Cancer

^bDrug therapy includes, chemotherapy, biochemotherapy, immunotherapy, and gene therapy

Safety

Overall, 52 doses of DNA.Mel3 and 95 doses of MVA.Mel3 were administered. Adverse events (AE) considered at least possibly related to treatment occurred in 37 patients (90%), most commonly injection site reactions and flu-like symptoms (Table 3). Nine subjects suffered grade 3 AEs, including 6 of 16 subjects who received the highest dose of MVA.Mel3 (10⁹ pfu), mainly consisting of injection site reactions. Only one grade 4 treatment-related event was reported in a subject experiencing a brief episode of syncope approximately 12 h after a second MVA.Mel3 dose of 1 × 10⁹ pfu associated with a suspected allergic-type reaction to the study treatment. None of these reactions led to early withdrawal of the subject from study or were considered dose-limiting. Ten per cent of patients discontinued treatment due to disease progression. One death occurred (cerebral haemorrhage) which was attributed to development of brain metastases. Locally, intramuscular immunisation with DNA.Mel3 was better tolerated than intradermal MVA.Mel3. There was a trend towards increased toxicity with rising MVA.Mel3 dose, but subsequent doses were better tolerated. No autoimmune events and no change in dsDNA antibodies between baseline and week 16 were noted.

Table 3.

Adverse events

Adverse event	N	Percentage (%)
N (ITT)	41	100
Grade 4
Syncope	1	2
Grade 3
Injection site reactions	8	20
Fatigue	1	2
Bone pain	1	2
Petechiae	1	2
Any Grade (>15%)
Injection site reactions	32	78
Headache	13	32
Pyrexia	12	29
Nausea	12	29
Cough	11	27
Rigors	8	20
Fatigue	8	20
Lethargy	7	17
Limb pain	7	17
Vomiting	6	15

Immune monitoring

Thirty-six patients were evaluable for cellular immune response based on the number and quality of PBMC isolated.

Tetramer Assay

At one or more time points during the 16-week treatment period, 24/36 (67%) patients exhibited a tetramer-detected response (Table 4). The proportion of tetramer responders in the higher-dose groups (Groups 4, 5, and 7; MVA ≥ 5 × 10⁸ pfu) was significantly higher than in the lower-dose groups (88 vs. 53%, P = 0.043). Priming doses of DNA.Mel3 generated transient, low-frequency Melan-A analogue tetramer responses in 9/36 (25%) patients; however, booster immunizations with MVA.Mel3 elicited more substantial, longer-lasting epitope-specific CD8+ T cell responses (Fig. 1, data not shown for lower-dose groups 1–4). Heterologous immunisation at the highest MVA dose elicited Melan-A/A2 tetramer responses in 10/11 (91%) patients (Table 4), whilst homologous immunisation with MVA.Mel3 elicited short-lived and low-magnitude responses in only 2/5 (40%).

Table 4.

Cellular and humoral immune responses according to MVA dose

Vaccine dose	2 or 4 mg DNA + MVA dose shown below				MVA alone	Totals
	5 × 107	2 × 108	5 × 108	1 × 109	1 × 109
Treatment group	1 and 2	3	4	5 and 7	6
N (evaluable)	10	5	5	11	5	36
Tetramer
Melan-A	5 (50%)	3 (60%)	4 (80%)	10 (91%)	2 (40%)	24 (67%)
IFN-γ ELISPOT
Melan-A/A2 only	3 (30%)	0	1 (20%)	3 (27%)	1 (20%)	8 (22%)
NY-ESO-1 only	0	1 (20%)	1 (20%)	0	0	2 (6%)
Melan-A/A2 + NY-ESO-1 + MAGE-3/A1	0	0	1 (20%)	0	0	1 (3%)
Anti-MVA IgG
Seroconversion	5 (50%)	4 (80%)	4 (80%)	11 (100%)	5 (100%)	28 (78%)

Fig. 1.

Melan-A tetramer assay profiles for individual patients treated with higher dose regimens. a Group 5: two 2 mg DNA.Mel3 + two 1 × 10⁹ pfu MVA.Mel3, b Group 7: one 4 mg DNA.Mel3 + two 1 × 10⁹ pfu MVA.Mel3, and c Group 6: four 1 × 10⁹ pfu MVA.Mel3. Immunisations at 3-week intervals. Dotted line signifies mean + 2 standard deviations (SD) of baseline response (above line positive response). open triangles administration of DNA.Mel3, closed triangles MVA.Mel3

IFN-γ ELISPOT assay

According to this assay 11/36 (31%) patients showed a positive response to at least one epitope. Responses were observed against the Melan-A/A2 epitope in 9/11 patients and to NY-ESO-1/A2 in 2/11 patients. One patient simultaneously responded to Melan-A/A2, MAGE3/A1 and NY-ESO-1 epitopes (Fig. 2, patient 047). This patient exhibited a substantial response to Melan-A/A2 at baseline (612 ± 122 SFC million), further expanded by the administration of the MVA.Mel3 booster immunizations (1315 ± 157 SFC million). Responses against MAGE3/A1 and NY-ESO-1 epitopes were evident at weeks 10 and 13, after administration of the second MVA.Mel3 booster. Although inconsistencies were noted between results obtained from ELISPOT and tetramer measurements, this is not unusual. Tetramer staining is a structural assay measuring T cells of a given specificity, whilst ELISPOT is a functional assay measuring only those T cells able to secrete IFN-γ following stimulation [14].

Fig. 2.

T cell immune responses elicited against multiple melanoma epitopes in Patient #47. a Mean Melan-A specific response detected by tetramer assay (circles) and ELISPOT assay (squares) ± 1 SD, b Response against epitopes other than Melan-A detected by ELISPOT assay. Dotted line mean +2 SD of baseline response in the tetramer assay, dashed line mean +3 SD of the negative control responses in the ELISPOT assay, open triangles priming immunisations with 2 mg pSG2.Mel3, closed triangles boosting immunisations with 5 × 10⁸ pfu MVA.Mel3

As expected, anti-vector responses developed; by week 16 a total of 28/36 (78%) patients had seroconverted to MVA, 100% in patients that received the highest-dose (1 × 10⁹ pfu) of MVA.Mel3. Despite this, anti-MVA antibody titres did not prevent boosting of the melanoma-specific immune response, 5/7 evaluated patients given additional booster immunisations after week 16 demonstrated a tetramer response against the Melan-A/A2 epitope,and two of these patients also responded in the ELISPOT assay. The Melan A/A2-specific response in Patient #33 (Fig. 3), who had a prolonged clinical response, continued to expand with the administration of repeated MVA.Mel3 boosts reaching a peak of 0.655 ± 0.03% in the tetramer assay and 457 ± 71 SFC/million in the ELISPOT assay.

Fig. 3.

Enhanced Melan-A specific T cell responses elicited in Patient #33 after extended boosting with MVA.Mel3. Mean Melan-A specific response detected by tetramer assay (circles) and ELISPOT assay (squares) ±1 SD. T cell immune responses were not detected against other epitopes (data not shown). Initial prime with two 2 mg pSG.Mel3 (open triangles) and boost with two 1 × 10⁷ pfu MVA.Mel3 (solid triangles). Two additional boosters of a higher dose 2 × 10⁸ pfu MVA.Mel3 on weeks 18 and 25 (large solid triangles)

Clinical response

At 16 weeks 39 patients were evaluable for tumour response; one was excluded as no scan was performed and another because the solitary target lesion was resected. 6/39 patients (15%) achieved disease stabilisation durable for greater than five months (Table 5) and two others a mixed response with regression of target lesions but development of an isolated new lesion. One was removed from study, but the other had the new subcutaneous lesion resected and received additional boosting vaccinations with further disease regression until week 91 when new lesions developed. One initially stable patient exhibited a partial response beyond 16 weeks with regression continuing over a 25-month period following ongoing boosting. Clinical benefit was only observed in patients treated with the heterologous prime-boost regimen (DNA/MVA).

Table 5.

Clinical responses

Dosea (pfu)	Patient no.	Metastases TNM (site)	Prior therapy	Baseline LDH	Tetramer response	ELISPOT response	Tumour response	TTP (weeks)	Survival (weeks)
5 × 107	1	M1b (LN, Lu)	−	<ULN	NE	NE	SD	34	86
	6	M1a (Cu)	−	<ULN	+	−	SD (MxR)	15	137+
	31	M1b (LN, Lu)	DTIC	<ULN	+	+	SD	27	69
	33	M0 (LN)	−	<ULN	+	+	PR	107	144+
5 × 108	16	M1c (Li, Lu, Ki)	IL-2	<ULN	+	−	SD	49	129+
1 × 109	21	M1b (Lu)	−	<ULN	+	−	SD	33+	117+
	71	M1c (LN)	DTIC, IFN	<ULN	+	−	SD(MxR)	95	110+
	75	M1c (Li, Lu)	GBN, 3S	<ULN	+	+	SD	24	103+

LN lymph node, Lu lung, Cu cutaneous, Li liver, Ki kidney, DTIC dacarbazine, IL-2 interleukin-2, IFN interferon, GBN gemcitabine, 3S treosulfan, ULN upper limit of normal range, NE not evaluable, PR partial response, SD stable disease, MxR mixed response: appearance of new lesions with target lesion regression

^aTwo 2 mg DNA.Mel3 + MVA.Mel3 at stated dose

The median time-to-progression (TTP) across all the treatment groups was 9 weeks (95% CI: 8–16) with a median survival of 51 weeks (95% CI: 39–86) and did not indicate any trend according to dose. Median survival for patients receiving DNA/MVA was 60 weeks (95% CI: 42–89), more than double that for patients who received MVA alone (26 weeks; 95% CI: 19–∞).

A retrospective analysis was performed using a random cohort of subjects from two centres that were screened, but not enrolled into the study due to their HLA phenotype, and thus received standard care. The median survival of these 23 patients was 42 weeks (95% CI: 25–91) compared to 49 weeks (95% CI: 37–82) for a comparable sub-group of 21 HLA.A2-positive trial patients treated with DNA/MVA at the same trial centres.

Correlation of clinical and immune response

Seven of the eight patients (88%) with a partial response, mixed response, or stable disease had a Melan-A-specific tetramer response (Table 5) as compared with 17/29 (59%) of those with disease progression. The estimated median TTP of tetramer responders treated with prime-boost DNA/MVA was 16 weeks (95% CI: 16–24) compared to 8 weeks (95% CI: 8–9) for non-responders (Log rank test, P = 0.037) (Fig. 4a). Patient #33 demonstrated Melan-A-specific immune responses temporally associated with tumour regression (PR). This patient along with Patient #71 (mixed response) continued to be treated with boosts at 12-week intervals until progression at 31 months and 22 months, respectively.

Fig. 4.

Kaplan-Meier survival curves showing Melan-A tetramer responders compared to non-responders. a Per protocol analysis of time-to-progression (TTP) for stage III and IV patients (29 subjects) receiving prime-boost regime (Groups 1–5 and 7), b Intention to treat analysis of survival for stage III and IV patients (31 subjects) receiving prime-boost regime (Groups 1–5 and 7). Open circles censored patients

Subjects demonstrating Melan-A-specific T cell responses tended to have prolonged survival. Median survival in stage IV Melan-A-specific tetramer responders was estimated to be 108 weeks (95% CI: 51–137) and in non-responders 37 weeks (95% CI: 35–44; Log rank test, P = 0.0002) (Fig. 4b). If stage III patients were included the result was similar (median 100 vs. 37 weeks, log rank P < 0.001). Analysis conducted on the basis of IFNγ ELISPOT response to Melan-A, NY-ESO-1 or MAGE-3, demonstrated the median survival of responders to be 108 weeks (95% CI: 69–∞), against 45 weeks (95%CI: 37–86) for non-responders (Log rank test, P = 0.265).

Discussion

Acknowledging the relatively low patient numbers the most intriguing finding in this study was a median survival of 27 months in the 19 stage IV melanoma patients demonstrating significant Melan-A/A2-specific tetramer responses, compared with 9.25 months in nine non-responders. The result for the non-responders was similar to the 10.5 months median survival of an HLA-A2 negative cohort of patients, excluded from the study, that received standard of care at the trial sites, and also to that from other recently published trials reporting median survival of patients with stage IV melanoma of 7–9 months from first treatment [15–17]. The data compare favourably to trials of other vaccines such as Melacine (Corixa, Seattle) [18]. The difference in survival was highly statistically significant but should be considered cautiously in a non-randomised study of this size, and the question remains as to whether this difference in survival is related to the vaccine or to other pre-existing immunological factors. The results seemed to indicate that a significant minority of patients benefited from the vaccine either due to the nature of their immune response or the biology of the tumour. Eight patients showed disease stabilisation for over 5 months, or a mixed response. Given the low toxicity of treatment it could be considered that in future, patients showing evidence of response continue with vaccination, even if there is isolated disease progression in single, resectable lesions. One patient whose subcutaneous lesion was resected at 16 weeks elected to receive additional boosters and subsequently developed a partial response at week 48 and remained stable for a further 43 weeks. Disease in a second patient remained stable until week 24 when they developed a partial response, 8 weeks after a higher-dose MVA booster vaccination and coincident with a rise in a tetramer-detected immune response. It could be postulated that those patients with disease indolent enough to allow time for immunity to develop may benefit, whilst for those with more rapidly progressive disease immunotherapy is less suited. This pattern of extended disease stabilisation even after apparent localised progression has been seen with anti-CTLA4 antibody therapy [19], another modifier of the immune response, and may indicate that agents such as these require modified evaluation criteria to fully assess their potential benefits [20].

Toxicity of the vaccine was low. The maximum-tolerated dose of the plasmid component was not reached and although the high-dose MVA vaccine induced more significant local reactions on first exposure, subsequent doses were better tolerated possibly due to anti-vector immunity dampening inflammatory responses. The selection of the intradermal route for MVA.Mel3 was based on early preclinical studies that suggested this route as the most immunogenic [21]. However, later work suggests less reactogenicity and similar immunogenicity for the intramuscular route [22]; use of this route is likely to allow further dose escalation without concerns over more severe injection site reactions and should be considered for future trials.

Although weak, the measured immune responses to DNA.Mel3 were superior to that seen by Triozzi et al. [8] who used a plasmid DNA vaccine encoding the full-length Melan-A/MART-1 antigen. This might be explained by the higher DNA doses in this study, and use of a polyepitope string. Increasing dose of MVA.Mel3 administered following the DNA priming was associated with increased frequencies of melanoma-specific T cell responses. At the highest dose of 10⁹ pfu, 90% of patients mounted a detectable T cell response; with lower doses (5 × 10⁷ pfu) there was a reduced magnitude and frequency of response consistent with that recorded in an earlier phase I study [10]. This dose–response is despite the fact that in the presence of metastatic disease, as in this trial, some T cell priming might have been expected. This dose-dependence is echoed in other clinical studies that have tested plasmid DNA and recombinant MVA-based malaria and HIV vaccines [11]. Also of note was the broadening of T cell response with one or two booster immunisations of MVA.Mel3 inducing T cell responses to NY-ESO and MAGE-3 in some patients [12]. This broadening of response could increase the chances of clinical benefit with more epitopes being recognised and a reduced chance of tumour escape. A concern of poly-epitope approaches has been the dominant immunogenicity of certain epitopes, and the dominance of the Melan-A/A2 epitope over the other HLA-A2 epitopes in the Mel3 epitope string was previously observed in HLA-A2-transgenic mice. Incidentally although HLA-A2 positivity was an entry criterion to the trial some patients were also HLA-A1 positive. The patient who responded to the MAGE-3/A1 epitope was HLA-A1 negative but HLA-B44 positive; binding of MAGE-3/A1 peptide to HLA-B44 has been described previously [23].

The successful targeting of Melan-A may be of particular relevance. Expansion of Melan-A/A2-specific CD8+ T cells was observed 1 week after the first or second MVA immunisation but generally contracted within 2–3 weeks. Even a temporary expansion of these T cells during the 16 weeks of treatment was associated with prolonged survival. As noted in other studies [22, 24] Melan-A/A2-specific CD8+ T cell responses can be maintained in DNA-primed patients by repeated boosting with MVA.Mel3 despite the presence of anti-MVA antibodies. Associations between immune response to Melan-A/A2 and prolonged survival have been seen in peptide vaccine trials [25, 26]; and studies of adoptive transfer of PBMC or TIL-derived Melan-A-specific CD8+ T cells have demonstrated clinical responses in melanoma patients [27, 28].

The trial design should also be reviewed. In this case it was similar to that of a standard chemotherapy phase II trial, which may not be appropriate in trials of biological therapy as a maximum-tolerated dose is less likely to be achieved. Here it resulted in relatively small dose-banded groups, and in hindsight it may have been more useful to simply compare larger low-dose and high-dose groups with an MVA-alone group. Trials of biological therapy on the whole tend to demonstrate more subtle responses than the tumour responses seen with chemotherapy. They include data from biological assays as utilised here, and clinical effects such as disease stabilisation or delayed responses which may be more difficult to identify in small standard-design trials. Immunisation with plasmid DNA and recombinant MVA triggers a range of innate and adaptive immune responses. In this study, the immune monitoring was limited to the detection of melanoma epitope-specific CD8+ T cells in peripheral blood using structural (tetramer) and functional (ELISPOT) assays. In future other parameters such as central versus effector memory markers, polyfunctionality, avidity of responses, T regulatory cells and innate immune cells should be monitored.

In summary, the vaccine used in this trial has several attractions. It has been well tolerated to the highest dose tested and demonstrated good immunological efficacy at this dose, apparently associated with clinical benefit. Compared to autologous vaccines, it is relatively easy to manufacture and can be administered to all HLA-A2 patients. The clinical benefit of the vaccine for the patient population as a whole in this study was small although immune responders may have improved survival. Further work should incorporate additional dose-escalation, and the prime-boost strategy might be improved by employing electroporation to improve immunogenicity of the DNA vaccine prime [29]. This vaccine also has potential to stimulate proliferation of transferred T cells in an adoptive immunotherapy setting, and combinations of this vaccine with new protocols such as lymphodepletion, with chemotherapy, or other immune modulators such as CTLA4 antibody should be explored [30, 31]. Future trials should look to involve larger cohorts of patients and employ a more appropriate design for biological agents with consideration given to continuation of therapy beyond localised progression or the addition of other treatment modalities.