The main finding of this study is that AMA1 antibody responses induced by vaccination of malaria-naïve human volunteers with a single AMA1 variant are biased towards the vaccine allele. This is in agreement with previously published results from a similar human Phase I vaccine trial 
and observations in rabbits 
. Approximately half of the IgG induced by vaccination of human volunteers reacts in ELISA with heterologous AMA1 variants, similar to what has been observed in rabbits 
. The functionality of the antibody response (GIA titre) to heterologous AMA1 expressing strains reflects differences in the quantity of IgG; suggesting that, although absolute antibody levels to heterologous AMA1 variants are lower, these antibodies are, on per mass basis, equally capable of inhibiting parasite growth.
The data presented here also demonstrate that both adjuvant and dose can have a profound impact on antibody levels, while other aspects of the antibody response like avidity, domain recognition, breadth and subclass distribution, are much less influenced by adjuvant or dose. This is, with exception of the subclass distribution, in agreement with what has been observed in rabbit studies 
The data presented here show that the widely used adjuvant Alhydrogel yields relatively low antibody levels, and increasing the antigen dose from 10 to 50 µg only marginally improved antibody responses. The data confirm that Alhydrogel is not potent enough to induce high levels of functional antibodies in malaria naïve subjects, as was previously reported 
. The low potency of Alhydrogel at inducing a functional response is even more pronounced when the functionality of the antibody response is evaluated on strains expressing heterologous AMA1 alleles. Therefore adjuvants more potent than Alhydrogel are definitively required for the induction of functional antibody responses to AMA1 variants not included in a vaccine.
Recently, vaccination induced efficacy against the AMA1 vaccine allele formulated with AS02 in a Phase IIb study was estimated at 64%, whereas overall vaccine efficacy was estimated at 17%, indicating the importance of covering AMA1 variants 
. Thus, the challenge for an AMA1-based vaccine appears to be with covering AMA1 polymorphism. Theoretically, two options would be available: i) The induction of a broad antibody response and ii) The maximisation of heterologous antibody responses by using potent adjuvants or prime boost strategies combined with a potent adjuvant, such that the response induced with a single variant would be high enough to also be sufficiently functional against heterologous strains. As the latter may not be possible with adjuvants currently available, a more practical approach would be the combination of the induction of the broadest response possible with the highest response possible.
The breadth of the antibody response can be improved by vaccination with a mixture of several AMA1 variants, either naturally occurring 
, or artificial ones 
. Three artificial diversity covering (DiCo) AMA1 sequences 
have recently been produced under cGMP and will enter clinical testing in the near future.
The magnitude of the antibody response can be improved by the use of potent adjuvants, as shown here. A novel proprietary adjuvant, CoVaccine HT™, has yielded promising results in rhesus macaques and rabbits 
. Moreover, P. knowlesi
AMA1 formulated with CoVaccine HT™ has induced protection against blood-stage challenge with P. knowlesi
in the rhesus macaque model and the degree of protection was correlated with GIA titre 
. It would therefore be interesting to combine DiCo AMA1 with CoVaccine HT™ in a Phase I trial.
Antibody avidity was determined with a sodium thiocyanate elution ELISA and average antibody avidities ranged between 0.6 and 0.9 M for the various antigens. These values are lower than what was previously observed in rabbits (0.9 to 1.4 M) 
. Of note is that the average avidity to the heterologous CAMP AMA1 was higher than the homologous antigen. This could possibly be explained by the fact that half of FVO AMA1-specific antibodies bound to CAMP AMA1 and that this fraction may bind with higher avidity. Conversely, avidities to the heterologous 3D7 AMA1 were lower than the homologous avidities. This may represent antigenic relatedness of the respective AMA1 molecules, with CAMP being more close to FVO AMA1 and 3D7 more distant.
The antibody response to AMA1 appears to be mainly directed against domains I and II, as competition with a construct comprising these domains removes about 80% of antibodies bound. Competition with a domain II-III construct removes a similar amount of antibodies as the I-II construct, suggesting that the majority of the response would be directed against domain II. This is, however, not supported by the depletion obtained by a domain II construct. Of note here is that only constructs including domains I-II induce functionally active antibodies in rabbits 
, suggesting conformational authenticity. The Domain II-III and the domain II constructs both failed to elicit functionally active antibodies in rabbits 
. The results obtained here warrant the statement that the majority of the antibody response to AMA1 is directed against domains I and II and a further subdivision for these two domains is not possible with the data hitherto obtained. The importance of domains I and II in the antibody response is in agreement with what has been found in rabbits 
The subclass distribution found in vaccinated malaria naive volunteers was similar to what was observed in exposed children 
and reflects the expected subclass distribution for a protein antigen 
. Antigen dose or adjuvant only marginally influenced the subclass distribution.
In conclusion, vaccination with a single allele AMA1 vaccine induces a humoral immune response that is biased towards the vaccine allele. The magnitude of the response can be enhanced by a potent adjuvant, in contrast other parameters of the humoral response like breadth, avidity and subclass distribution appear much less influenced by the adjuvant. Future vaccine development should focus on improving both breadth and magnitude of antibody responses to AMA1.