Safety and Immunogenicity of DNA/Modified Vaccinia Virus Ankara Malaria Vaccination in African Adults

On average, a child dies every 30 s of malaria in sub-Saharan Africa [1]. An effective malaria vaccine is urgently needed. In several animal models of malaria, cytotoxic T cells are critical for protection against infection [2], and there is indirect evidence that cytotoxic T cells play a similar role in human Plasmodium falciparum malaria [3]. If a so-called priming immunization with a plasmid DNA encoding a preerythrocytic malaria antigen is followed by a boosting immunization with a recombinant, nonreplicating viral vector encoding the same antigen, there results a 10-fold amplification of CD8⁺ T cell immunogenicity, compared with that produced by DNA vaccination alone and, in 2 mouse models of malaria, complete protection [4]. In these experiments, protection correlated with secretion of interferon-γ by splenocytes in an ex vivo ELISPOT assay. Similar results have been published for human immunodeficeincy virus (HIV) [5], tuberculosis (TB) [6], and Ebola virus [7] (DNA/adenovirus) models. A particularly immunogenic viral vector for boosting immunization is modified vaccinia virus Ankara (MVA), a vector suitable for human use. MVA is a highly attenuated vaccinia virus with an exceptionally good safety profile [8]

We have developed 2 novel vaccine candidates, DNA multiple epitope (ME)–thrombospondin related adhesion protein (TRAP) and MVA ME-TRAP. The ME-TRAP construct consists of the ME string [9] of 18 T cell epitopes and 2 B cell epitopes, from 6 preerythrocytic P. falciparum antigens and the T9/96 strain of an entire, well-characterized preerythrocytic antigen, TRAP [10]. Since August 1999, we have investigated, in the first prime-boost trials of DNA-based vaccines in humans, their safety, immunogenicity, and efficacy in phase 1/2a trials in Oxford. In this report, we describe the first African phase 1 trial to assess the potential of this new vaccine technology

Volunteers and methodsVolunteers were recruited from the periurban community of Bakau, which is on the coast of The Gambia. Approval was obtained from the Joint Gambian Government/Medical Research Council Ethics Committee and the Central Oxford Research Ethics Committee. Written, informed consent was obtained from all volunteers, after obtaining initial community consent, having discussion with volunteers in the local languages, and disseminating information sheets and consent forms translated into local languages in Arabic script. The trial was conducted in accordance with Medical Research Council, UK, guidelines for the conduct of clinical trials

Potential volunteers underwent thorough clinical evaluation and were screened for hematological (full blood count), renal (plasma creatinine and urinalysis), and hepatic (plasma alanine aminotransferase [ALT]) dysfunction. A total of 20 semi-immune, healthy adults, 18–45 years old, were enrolled. An independent safety monitor based in The Gambia monitored the study

For comparison between malaria-exposed and -naive individuals, ELISPOT data from assays performed on volunteers from the United Kingdom are included in the analysis. The UK study is reported in full elsewhere [11]. The UK DNA/MVA group consisted of 9 volunteers (3 volunteers in each group), who received either 1, 2, or 3 1-mg immunizations of DNA ME-TRAP followed by 2 immunizations of 3×10⁷ pfu of MVA ME-TRAP [12]. In these 3 groups, effector T cell frequencies were equivalent, and, therefore, the vaccination regimen received by UK DNA/MVA vaccinees were comparable to those received by Gambian DNA/MVA vaccinees. The UK MVA-alone group consisted of 5 volunteers who received 3 immunizations 3×10⁷ pfu of MVA ME-TRAP (identical to the immunizations given to the Gambian MVA-alone group)

The 2 study vaccines were DNA ME-TRAP and MVA ME-TRAP. The individual epitopes that constitute the ME string are described in detail elsewhere [9]. The clinical vaccines were manufactured by contract manufacturers (DNA ME-TRAP by Qiagen and MVA ME-TRAP by IDT). Regulatory approval for prior UK phase 1 studies of these vaccines was obtained from the UK Medicines Control Agency (MCA). The Gambian government accepts MCA approval in the absence of a Gambian regulatory authority

The present study was a phase 1 open-label trial. All vaccinations in both the UK group and the Gambian group were administered at 3-week intervals. A total of 12 volunteers received 2 1-mg immunizations of DNA ME-TRAP intramuscularly, followed by 2 immunizations of 3×10⁷ pfu of MVA ME-TRAP intradermally; 8 volunteers received 3 immunizations of 3×10⁷ pfu of MVA ME-TRAP intradermally. Eligible volunteers were allocated to each of these 2 groups alternately, in order of enrollment. Each volunteer was observed for at least 1 h after vaccination. Study visits were scheduled for 1, 2, 3, 7, and 28 days after each vaccination and for 8–10 weeks after the final vaccination. A total of 6 screened, healthy, unvaccinated volunteers were bled 3 times on days 0, 56–84, and 140–156 to assess, in the Gambian population, background variation in effector T cell responses over the study period

Each Gambian volunteer had 30 mL of venous blood drawn from an antecubital vein, on the following 5 occasions: screening (day −28 to day −7); 1 week after the first vaccination (in the MVA group only; day 7); 1 week after the second vaccination (in both groups; day 28); 1 week after the third vaccination (in both groups; day 49); 1 week after the fourth vaccination (in the DNA/MVA group only; day 70); and 8–10 weeks after the final vaccination (day 142–156). Complete blood counts and ALT and creatinine assays were performed according to the standard operating procedures of Medical Research Council Laboratories, The Gambia

ELISPOTs were performed on Millipore MAIP S45 plates with MabTech antibodies, in accordance with the manufacturer’s instructions. Quantities of 4×10⁵ freshly isolated peripheral blood mononuclear cells (PBMCs) were incubated for 18–20 h on the ELISPOT plates, in the presence of 25 μg/mL of peptides, before being developed. The number of spot-forming cells (SFCs) were counted by the AutoImmun Diagnostika system. Individual 8–17-mer epitopes were used for epitopes from the ME string, whereas 20-mer epitopes overlapping by 10 were used to span TRAP, with both T9/96 and 3D7 strains of TRAP spanned in their entirety. Because of cell-number limitations, peptides were assayed in pools, and cells were assayed in duplicate for each pool. Cell separations were performed on frozen cells by use of Miltenyi Biotech magnetic cell sorting beads and were checked by costaining and fluorescence activated–cell sorter analysis

Recombinant T9/96 TRAP and 3D7 TRAP were coated overnight at 4°C on Nunc immunoplates, at a concentration of 4 μg/mL. Standard ELISAs were performed, with incubations for 2 h at 37°C

Peptide pools 5–9 spanned 3D7 TRAP, and 10–13 spanned T9/96 TRAP (figure 1). Spots were summed, across relevant pools, and the “no peptide” negative-control spot counts were subtracted from the requisite number of times (figure 2A–2D). Arithmetic means, geometric means, medians, and interquartile ranges (IQRs) were derived; all 4 are provided for initial immunogenicity values presented in the text, to allow for evaluation of the distribution of these summed-response data. Groups of individual summed responses were compared by the Mann-Whitney test. This test may not have power to detect some true differences but was used because variance between groups was often markedly dissimilar. It was not always possible to obtain normality via conventional transformations

ResultsA total of 18 of 20 vaccinees completed the study protocol: 2 withdrew their consent after the second DNA immunization; neither had experienced any adverse events. After a total of 24 DNA ME-TRAP immunizations, there were no local or systemic adverse events. The MVA vaccine was well tolerated (table 1). All the tabulated adverse events were mild (no interference with activities of daily living). The 1 moderate adverse event, transient limitation of arm abduction, occurred after MVA vaccination. MVA causes a characteristic local reaction after intradermal administration, with redness and induration peaking at 48–72 h. In 5 of 18 subjects who received first immunizations of MVA ME-TRAP, there was a blister of <2 mm in diameter at the center of the indurated lesion, which, in all cases, healed without complications over 1–3 weeks. Analysis of the hematology and biochemistry safety assays revealed no adverse events

The 6 unvaccinated volunteers had geometric mean effector T cell frequencies of 5.3, 7.2, and 6.6 SFCs/10⁶ PBMCs, at the beginning, middle, and end of the study, respectively. Effector T cells were induced in vaccinees with specificity for all peptide pools; TRAP-specific frequencies were higher than ME-specific frequencies (figure 1). After DNA ME-TRAP immunization, effector cell frequency showed a small but non–statistically significant increase. MVA ME-TRAP immunization induced much greater effector T cell frequencies. Immunogenicity of MVA ME-TRAP after prior DNA ME-TRAP (in terms of vaccine-induced T9/96 TRAP–specific effectors) was >3-fold higher in Gambian volunteers than in UK volunteers (arithmetic mean, 175.4 vs. 51.4; geometric mean, 69.8 vs. 19.8; median, 172.0 vs. 31.2; IQR, 23.4–239.4 vs. 15.6–131.0 SFCs/10⁶ PBMCs; P=.03) (figure 2A and 2C). Immunogenicity was also higher in Gambians who received MVA ME-TRAP immunization without prior DNA ME-TRAP than in UK adults who received the same regimen (arithmetic mean, 55.4 vs. 17.2; geometric mean, 23.0 vs. 11.0; median, 28.8 vs. 15.2; IQR, 10.6–105.6 vs. 3.4–40.7 SFCs/10⁶ PBMCs; P=.10) (figure 2B and 2D). T9/96 TRAP effector T cell frequency after MVA boosting was higher in Gambians who received prior DNA ME-TRAP than in Gambians who received MVA without prior DNA (geometric mean, 69.8 vs 23.0 SFCs/10⁶ PBMCs; P=.07) (figure 2A and 2B). After the second MVA immunization in the Gambian DNA/MVA group, the geometric mean effector T cell frequency was 63.3 SFCs/10⁶ PBMCs, compared with 69.8 SFCs/10⁶ PBMCs after the first MVA immunization. In both the UK DNA/MVA group (data not shown) and the Gambian MVA-alone group (figure 2B), effector T cell frequency kinetics showed a decay from 7 days after the first MVA immunization to 7 days after the second MVA immunization. In the group with the strongest responses (DNA/MVA-vaccinated Gambian adults), geometric mean effector frequencies at the 8–10 week follow-up were 70.6% of the peak frequencies

To assess the ability of a malaria vaccine to induce a cross-reactive T cell response, we evaluated the effector response to the 3D7 strain of TRAP, a heterologous strain with 6% sequence variance at the amino-acid level, compared with T9/96. Cross-recognition of responses induced in Gambians was far in excess of that seen in UK volunteers (vaccine-induced response to 3D7 after DNA-then-MVA immunization in Gambian adults vs. that in UK adults; geometric mean, 65.1 vs 5.5 SFCs/ 10⁶ PBMCs; P=.01) (figure 2A and 2C). Before vaccination, T cell responses in the group of Gambians who received MVA without prior DNA were higher to 3D7 than to T9/96. In this group, immunogenicity was correspondingly greater for 3D7 than for T9/96 (vaccine-induced responses; geometric mean, 25.9 vs 15.0 SFCs/10⁶ PBMCs; P=.16) (figure 2B)

The induced effectors are of both CD4⁺ and CD8⁺ T cell subsets (figure 3). Most responses were either CD4⁺ or mixed CD4⁺ and CD8⁺ responses, but pure CD8⁺ responses were also seen

A total of 12 of 20 Gambian volunteers had titers of anti-TRAP antibodies (to both or either of T9/96 and/or 3D7 strains) that were statistically significantly above titers in malaria-naive individuals. There was no statistically significant increase or decrease of antibody titers after vaccination (data not shown)

DiscussionMost vaccines in widespread use have been developed and formulated for optimal antibody induction. The prime-boost approach outlined in the present report is an example of a new approach that targets the maximization of T cell immunogenicity. Potent T cell induction is likely to be necessary to vaccinate effectively against intracellular organisms such as HIV, Mycobacterium tuberculosis and liver-stage P. falciparum and for cancer immunotherapy

The ability of MVA vaccines to amplify preexisting T cell responses induced by priming with DNA vaccines in animal models suggests that the vaccines may be more immunogenic in African individuals who have been primed by natural exposure to malaria. Elsewhere, in a mouse model, it has been shown that immunogenicity of a single dose of a recombinant vaccinia vaccine is not protective but that prior exposure to malaria sporozoites boosts this immunogenicity to protective levels [13], resulting in protection that is T cell mediated. Our findings confirm this prediction. The immunogenicity of an MVA malaria vaccine (with or without DNA priming) was of greater magnitude in previously exposed Gambian individuals than in malaria-naive individuals from the United Kingdom. There are genetic and environmental differences between malaria-naive and -exposed individuals in this study, other than their status with regard to previous exposure to malaria, but we feel that these differences are unlikely to account for the reported altered immunogenicity. Genetic and, in particular, HLA differences between the populations might have been most evident in responses to the HLA class I restricted peptides of the epitope string, but these contributed very little to the overall response level (figure 1). Interestingly, even with sporozoite priming, further priming by DNA immunization is still necessary for optimal T cell induction by MVA immunization

The greatly increased cross-recognition demonstrated in Africans in this study provides encouragement for further work with T cell–inducing DNA and recombinant viral vaccines. Lack of strain transcendence has long been viewed as a potential obstacle to malaria vaccination. The Gambian volunteers who received MVA without prior DNA vaccination (figure 2B) had higher prevaccination effector T cell frequencies to the 3D7 strain of TRAP than to the T9/96 strain. After immunization, frequencies of T cells specific for 3D7 TRAP increased more than did those specific for T9/96, which is consistent with the well-recognized immunological phenomenon termed “original antigenic sin” [14]

Not only CD8⁺ T cells, but also high frequencies of CD4⁺ T cells, were induced by DNA/MVA and MVA immunizations. Although CD4⁺ T cells have traditionally been considered to be helper T cells for either antibody production or CD8⁺ T cell cytotoxicity, it is now clear that many CD4⁺ T cells have effector activity. Directly cytotoxic CD4⁺ T cell clones confer protection in murine adoptive-transfer experiments [15], and such cytotoxic CD4⁺ T cell clones are present in attenuated sporozoite immunization–protected humans [16]

In mouse models, contraction of ∼90%–95% of the effector T cell pool by apoptosis occurs after infectious challenge over the course of 2 weeks (for CD8⁺ T cells) or 7 weeks (for CD4⁺ T cells), in 1 model [17]. We examined the kinetics of such contraction in humans. We demonstrated the persistence of a residual memory pool with rapid effector function 8–10 weeks after final vaccination, with frequencies at this time >50% of the peak frequencies

In all groups studied, the second MVA immunizations, which were performed 3 weeks after the first MVA immunizations, show no increase in immunogenicity. In fact, effector T cell frequencies show a reduction that is consistent with natural decay kinetics, from the peak frequency 7 days after the first MVA immunization. This may be because, at this interval, a host immune response to the highly immunogenic MVA vector prevents infection of host cells and recombinant protein expression after the second MVA immunization. A study is underway to evaluate immunogenicity of the second MVA immunization after a 12-month interval

It has been difficult to induce strong effector T cell immune responses through vaccination. We present a safe strategy that is immunogenic for effector T cell induction in malaria and is likely to be applicable to other fields. Prime-boost approaches with viral-vector boosts have been seen as an excellent option for improving the efficacy of prophylactic DNA vaccines in diseases in which T cell responses are protective. The confirmation of the ability of MVA to boost preexisting T cell responses in humans implies that MVA and the DNA/MVA combination may also be effective for immunotherapy of chronic infections and tumors. Examples that may merit evaluation are TB infection before onset of disease, hepatitis B virus–infected individuals at risk of disease progression, and immunotherapy of HIV-positive and melanoma patients. The first field-efficacy study of prime-boost immunization against malaria in Africans is underway

• Received February 10, 2003.
• Accepted May 7, 2003.

• © 2003 by the Infectious Diseases Society of America