Background
A vaccine against
Plasmodium falciparum has long been sought and is a badly needed tool in the fight to eliminate and eradicate malaria. Experiments indicate that parasites and clinical disease can be controlled through passive transfer of antibody [
1,
2], although specific clinical markers of protection have not yet been elucidated. A partially protective pre-erythrocytic vaccine is currently in Phase 3 trials [
3]. However, a "second generation" more highly protective vaccine is needed. Inclusion of blood stage antigens in a multi-stage vaccine would likely lead to higher levels of protection against clinical disease, and would also provide protection against epidemic malaria if pre-erythrocytic immunity is incomplete [
4].
AMA1 is one of the leading blood stage vaccine candidates, and was evaluated in a Phase 1-2b trial in malaria-exposed children in Mali [
5,
6]. This trial evaluated a vaccine combining two allelic forms of AMA1 (AMA1-C1), adjuvanted with Alhydrogel. The vaccine was moderately immunogenic and no overall impact of vaccination on malaria parasitaemia or disease was seen [
6]. Allele-specific effects were not demonstrated [
7]. As per the analytic plan for the Phase 2 (biologic impact) part of the study, 16 secondary outcomes were examined for data up to day 154. Four of the 16 secondary outcomes examined were related to haemoglobin (Hb); all showed trends towards a negative impact of vaccination with AMA1 relative to the comparator group, with two of these statistically significant at the 0.05 level (mean Hb during clinical malaria, unadjusted p = 0.004, and number of episodes with Hb < 8.5 g/dL, unadjusted p = 0.029), although differences were no longer significant after correction for multiple tests. Additional analyses using all subjects with available data from day 1 until day 364 showed significance of Hb effects: both minimum Hb [p = 0.0121] and mean Hb during clinical malaria events [p = 0.0044] were significantly lower in the AMA1 group, and the incidence of Hb < 8.5 g/dL [p = 0.0096] was more frequent in the AMA1 group. An extended follow-up period from November/December 2007 until January 2008 was added due to concern about the possible impact on anaemia events, and an imbalance in serious adverse events related to malaria. However, no significant differences in Hb endpoints were found using only the extended follow-up data [
6].
This paper describes further analyses of the differences between the AMA1-C1 vaccine and the control (Hiberix) groups in the study described in [
6]. Additional analyses focused on the repeated incidence of anaemia and explored possible confounders or mechanisms of action of the effect of vaccination on Hb. Variables included were: baseline Hb, as lower Hb at the start of the malaria transmission season has been shown to be associated with anaemia during clinical malaria [
8]; Hb S, C, and alpha-thalassaemia, as variant Hb is known to be protective against severe malaria and Hb S has been shown to delay time to first clinical malaria event [
9‐
12], and these variables could also affect the risk of anaemia. Similarly, G6PD status is a potential confounder for anaemia [
13] and was included in the model. Age is a risk factor for development of anaemia, with younger children more vulnerable [
14], therefore age was added. Data on 24 subjects from the Phase 1 part of the study was also included (see [
5]).
Discussion
This intensive exploration of the relationship of AMA1 vaccine with anaemia was motivated by secondary results uncorrected for multiple comparisons, so results should be interpreted not as a confirmatory study but as the continuation of an exploratory study. Nevertheless, despite these cautions on over-interpretation, effects were still seen when more subjects were included, more potentially confounding variables were included, and more statistically sophisticated analyses were performed. The strengthening of the effect after inclusion of baseline haemoglobin may be because of the uneven randomization at baseline, i.e. children enrolled in the comparator group had lower baseline haemoglobin and more anaemia, thus partially masking the strength of the association of vaccination of AMA1 and anaemia in the initial analysis. The imbalance in anaemia was seen in higher grade as well as lower grade events, although numbers were small and this difference was not statistically significant. Clearly the possible impact on anaemia should be closely evaluated in future clinical trials of AMA1 in malaria-exposed populations.
In theorizing about the possible causes of this apparent anaemia effect, the first possibility that should be considered is that the comparator Hemophilus influenza B vaccine (Hiberix) was somehow protective against anaemia, perhaps through a general improvement in health status. However, this vaccine primarily protects against invasive Hib disease in the first year of life, and also would be expected to exert a level of herd immunity to non-vaccinated children in the village [
20]. No previous studies of this class of vaccines which showed protection from anaemia were found in a literature search, and a protective effect of Hiberix is unlikely.
If the increase in anaemia in the group receiving the AMA1 vaccine is a true effect of vaccination, what possible mechanisms could explain this? Experimental
falciparum challenge studies in
Aotus monkeys have shown profound anaemia in some cases, and partial control of infection resulting in persistent low grade infection has been postulated to be a possible mechanism for this effect [
21]. In the Phase 2 trial parasite densities were similar between the vaccinated and control groups, and time to first event (malaria infection, and fever at varying levels of parasitaemia) was similar between the groups. Thus partial control of parasitaemia appears to be an unlikely explanation in this case.
The aetiology of malaria anaemia is not well understood [
22,
23]. Three mechanisms are thought to interact to cause anaemia: lysis of infected and uninfected red blood cells primarily during acute infection, sequestration (also occurring during acute infection), and bone marrow suppression. The nadir in haemoglobin typically occurs 4-5 days after treatment, and recovery from anaemia can be prolonged. No difference in clinical evidence of haemolysis (such as jaundice or splenomegaly) was seen between the groups in this study, although these were not study endpoints and it is possible that a difference occurred but was not detected. Coombs tests, haptoglobin, and reticulocyte counts were not done. Sequestration is mediated by adherence factors, possibly related to antibody. No association was seen here between peak (day 42) or induced (day 42-day 0) anti-AMA1 antibody levels and minimum hemoglobin. Although there is no evidence showing a qualitative difference in vaccine-induced anti-AMA1 antibody compared to the infection-induced antibody judged by surface plasmon resonance technique or in vitro growth inhibition assay (unpublished observation), there may be some qualitative difference which these assays are not able to detect, possibly contributing to increased lysis or adherence. Pro-inflammatory cytokines are also thought to play a role in malaria anaemia, with excessive production of Th1 cytokines thought to be beneficial during the early stages of a plasmodial infection and Th2 cytokines likely to be protective in chronic infection or in the recovery period [
24,
25]. The alum adjuvant used in this vaccine induces a Th2 biased response, and it is theoretically possible that when given with a blood stage antigen the immune response is adversely shifted, although how this would act remotely from the time of vaccination is not clear.
Other efficacy field trials have been conducted with blood stage vaccines. A Phase 2b trial in Kenyan children of FMP1 (MSP1) adjuvanted with AS02 showed no overall protection despite high levels of induced antibody [
26]. An imbalance was seen in the number of cases of transient low haemoglobin (9 in 200 vaccinees vs. 2 in 200 controls) but no significant difference in the risk of Hb < 8.0 g/dL was seen and time to first episode of Hb < 8.0 did not differ between the groups, although there was a trend towards an increase in the vaccine group (Hazard ratio 1.53 (95% CI 0.90, 2.59, p = 0.11). Results of a Phase2b trial in Malian children of FMP2 (AMA1) adjuvanted with AS02 also showed no overall protection despite high levels of antibody, although strain specific effects were seen in the first year (5
th Multilateral Initiative on Malaria, November, 2009; and 2009 annual meeting of the American Society of Tropical Medicine and Hygiene). No imbalance in anaemia was seen, although these results are not yet published and additional analysis is pending. A Phase 2b trial in Malian children of MSP3 adjuvanted with alum is ongoing (ClinicalTrials.gov Identifier: NCT00652275), and a Phase 2b trial of a GLURP/MSP3 fusion protein also adjuvanted with alum is planned (5
th MIM Pan-African Malaria Conference, 2009, Symposium 14).
While it is possible that the association of anaemia with vaccination with AMA1-C1/Alhydrogel in this study is an artifact, and no mechanism is apparent, the robustness of these findings and the imbalance in anaemia seen in one other field trial of a blood stage vaccine [
26] are cause for concern. The impact of vaccination on anaemia outcomes should be closely examined, particularly for field trials of blood stage vaccines in malaria-exposed children.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
RDE designed the study, coordinated study execution and analysis, and wrote the paper. MPF designed the study, analysed the data, and co-wrote the paper. IS and AD conducted the field study. KM performed the immunologic analysis. MAG and AG performed clinical laboratory analyses. MSS conducted the field study and was responsible for the study database. OKD and DD designed the study and supervised conduct in the field. All authors reviewed the manuscript and approved the final draft.