The focus of enhanced research and malaria control has, understandably, been primarily on P. falciparum malaria, given the mortality and severity of disease associated with this species. However, there is increasing recognition that the risk of infection and the burden of disease due to P. vivax malaria is substantial and, although of limited importance in sub-Saharan Africa, this parasite is often the dominant one in the other major endemic regions of the world. Frequently, P. vivax and P. falciparum occur sympatrically and elimination programmes, in such cases, must take account of the different challenges presented by the two species, especially in terms of persistence of infection, and the complex interactions between the species. This complex balance may be disturbed by vaccination against either P. falciparum or P. vivax in areas where both parasites are prevalent. Eliminating P. falciparum but not P. vivax would be a step forward, but if this was all that was achieved could damage the reputation of a malaria elimination programme.

Malaria elimination means stopping infection. This may be achieved if the measures directed against asexual parasites are fully effective, or by targeting sexual stages directly with drugs or vaccines.

In areas of high transmission, asexual parasite densities are highest in young children, and it is in this age group that microscopic detection of gametocytes is most common. Both asexual blood stage and gametocyte densities then decline with age, though the patterns of decline are somewhat different 12. Epidemiological studies show, however, that transmission of P. falciparum is as dependent on the parasites not detected by routine blood screening as on those that are readily seen. The cumulative evidence for the importance of low-grade asymptomatic infections as a reservoir for infection of mosquitoes is strong. In areas of highly seasonal malaria, where there is often a long dry season in which little or no transmission occurs, persisting very low gametocytaemias are the source from which transmission occurs at the onset of the subsequent rainy season 13. In areas where the endemicity of malaria allows acquired immunity to develop, there are many asymptomatic individuals, particularly adults, who are an important source of infection for mosquitoes 1415.

Detection of gametocytes using molecular techniques, such as reverse transcriptase polymerase chain reaction (RT-PCR) 13 and quantitative nucleic acid sequence-based amplification (QT-NASBA) has shown that gametocytaemias can persist at sub-microscopic levels for months and that the prevalence of gametocytaemic individuals is much higher than was assumed from blood film examinations 16. Of particular relevance to consideration of malaria elimination, Shekalaghe et al 17, in a study of parasite prevalence in an area of low, seasonal transmission in Tanzania, showed that, while microscopically the parasite rates were low (1.9% asexual parasites, 0.4% gametocytes), QT-NASBA revealed much higher prevalence rates (32.5% asexual parasites and 15.0% gametocytes).

These observations indicate the need to adopt a community-based approach to elimination; any selective interventions used should be based more on the focality of malaria 18, rather than on particular groups especially at risk from the clinical consequences of malaria infection, such as young children or pregnant women. Those at low risk clinically may still be important transmitters of infection.

A key factor in the persistence of P. vivax is the fundamental difference in the life cycle shown by P. vivax and Plasmodium ovale, compared with P-. falciparum and Plasmodium. malariae, namely the occurrence of dormant hypnozoites in the liver, from which relapse infections can emerge, weeks, months or years after infection.

There is some evidence from use of PCR techniques, that, as for P. falciparum, the number of P. vivax infections in an affected community is generally significantly underestimated 19. Though the pattern of gametocyte production in P. vivax is different from that of P. falciparum, low grade asymptomatic infections are equally likely to give rise to infectious gametocytes.

The challenges set to vaccine developers 20 by those who drew up the malaria vaccine road map are first, by 2015, to produce a licensed vaccine that has a protective efficacy of more than 50% against severe disease and death from malaria which lasts longer than one year. Secondly, by 2025, to develop a vaccine with a protective efficacy greater than 80% against clinical disease and death that lasts longer than four years.

These targets focus on the prevention of clinical disease, especially its severe and life-threatening form, valid objectives for vaccines that are to be introduced into areas of medium or high transmission. A vaccine that conformed to the 2015 objective, providing protection to half of those vaccinated, would be valuable as part of an integrated control programme alongside vector control and chemo-preventative measures. That level of efficacy would not justify its use alone as an alternative to other means of malaria control.

The expanding programme of experimental vaccine-related research has two broad but overlapping approaches. One is to achieve a much needed better understanding of the nature of protective immune mechanisms against malaria, thus providing a basis for rational vaccine design 21. The other approach is more empirical and involves issues of design and presentation as vaccines of antigens that have been recognised for a long time and which are known to have an important role in the parasite's life cycle.

The primary objective of pre-erythrocytic stage vaccines is provision of a level of protection that prevents any invasion of the blood and hence any clinical malaria. This has been achieved readily in experimental malarias by vaccination with radiation-attenuated sporozoites; this formed the starting point for the extensive investigations into pre-erythrocytic vaccines. Complete protection was also achieved in humans against P. falciparum and P. vivax by exposing them to the bites of mosquitoes that had been irradiated to attenuate their sporozoites. However, this delivery procedure was totally impractical and the research emphasis shifted to the synthetic design and genetic-engineering of sub-unit vaccines 22.

There has, however, been a renewal of interest in the development of attenuated-sporozoite vaccines. Given the high level of sterile protection these whole organism vaccines can induce, this is a very welcome development. Stephen Hoffman established and directs Sanaria Inc. specifically to produce radiation-attenuated sporozoites of P. falciparum from infected mosquitoes in sufficient quantity and in a way that meets the regulatory standards required for their use as a vaccine 23. This has been achieved and phase I/2a clinical trials will begin shortly using irradiation-attenuated sporozoites delivered by intradermal or subcutaneous injection.

Genetic attenuation of sporozoites is also being investigated. Mueller et al 24 produced sporozoites from Plasmodium berghei deficient in the vis3 gene. These gave complete protection when used for vaccination. Sporozoites lacking 6-cysteine secretory proteins required for parasitophorous vacuole formation are equally effective vaccines 25. The immune responses of importance following vaccination with attenuated sporozoites are CD8+T cell-mediated with production of interferon gamma 262728.

Synthetic and genetically-engineered sub-unit vaccines have generally been based on the two surface proteins, circumsporozoite protein (CSP) and thrombospondin-related anonymous protein (TRAP) involved in sporozoite motility and invasion of liver cells. Most of the phase 2 trials of vaccines based on these antigens, using a variety of vaccine constructs, have given poor results and it is perhaps surprising that the RTS, S vaccine candidate has proved to be much more promising. This vaccine is a hybrid molecule expressed in yeast, that consists of the tandem repeat tetra-peptide (R) and C-terminal T-cell epitope containing (T) regions of CSP fused to the hepatitis B surface antigen (S), plus unfused S antigen. The adjuvant ASO2, which consists of an oil in water emulsion containing immunostimulants monophosphoryl lipid A and QS-21, a fraction of Quillaia saponaria, is an essential component of the vaccine. Variants of ASO2 have been tested successfully 29, and an alternative adjuvant, ASO1, where a liposomal formulation replaces the oil in water emulsion, has given very encouraging results 3031.

The first challenge trials with RTS, S/ASO2 gave impressive, but short-lived, protection in naïve adults 32. Similarly, in a trial in Gambian adults, there was greater than 70% protection against parasitaemia in the nine weeks post-vaccination, but the immunity waned rapidly 33. The most comprehensive study of RTS, S/AS02 has been in children in Mozambique, who have been followed up for two years. Over this period, there was 30% protection against clinical malaria and close to 50% protection against severe malaria 34. Most recently, in a small-scale trial in infants designed primarily to test safety and immunogenicity, RTS, S/AS02D had a vaccine efficacy of 65.9% against new infections in the six months of follow-up 35. A series of phase 2 studies preparatory to a large phase 3 trial and potential licensure of this vaccine are in progress.

RTS, S is several years ahead of any other vaccine in terms of assessment of its efficacy in clinical trials. Trials of other pre-erythrocytic stage vaccines, based on CSP, TRAP and other liver stage antigens, several of which have used viral vector delivery systems, have shown some initial promise, but are not sufficiently advanced or effective to be considered yet for evaluation in phase 3 trials 3637.

A range of blood stage antigens have progressed to phase 1 and phase 2 trials. Most are molecules that have been identified as being involved in the process of invasion of erythrocytes by merozoites 38. Promising results from studies with rodent malarias are proving difficult to transfer to human infections 3940414243, though some early encouraging results with MSP-3 antigen have been reported 44. The expectation is that such candidate vaccines might give protection against disease, but not against infection. This was the case with one phase 2 clinical trial of a vaccine containing MSP-1, MSP-2 and RESA antigens which reduced parasite density, but not prevalence of infection 45. It was also strain-specific in its effect and the polymorphism of these antigens, coupled with the variability in invasion pathways P. falciparum can adopt 4647 are a severe challenge to the design of this kind of vaccine. Blocking invasion of reticulocytes by P. vivax merozoites might be a more hopeful strategy, given that the Duffy antigen is thought to be the obligatory receptor on the erythrocyte surface 48 yet, even here an alternative invasion strategy has been proposed 49.

Plasmodium falciparum-infected erythrocytes express highly variable parasite molecules on the red blood cell surface. Naturally-acquired immunity involves variant-specific responses to these antigens and the very complexity of these responses may limit the potential of these antigens as vaccine candidates. However, this variability might be exploitable for specifically targeted vaccines. The P. falciparum-infected erythrocytes that sequester in the placenta of pregnant women have a very selected sub-set of variant surface antigens, notably one coded VAR2CSA, through which they bind to chondroitin sulphate A (CSA) in the placenta. This opens the possibility of designing a vaccine that would be beneficial to pregnant women 50, but which would probably not affect the other variants that circulate and sequester elsewhere using different receptor-ligand interactions. Another postulated approach to vaccination, much less studied, is to block parasite molecules that mediate disease by inducing pro-inflammatory responses. Glycosylphosphotidyl-inositol (GPI) is strongly implicated 51, but any vaccine effect would alleviate symptoms of disease without affecting infection. The immunity induced by such a vaccine might prevent some of the severe complications of malaria mediated by cytokine-induced response to infection. Use of vaccines of this kind, however, has the potential disadvantage of damping down the early clinical features of malaria, such as fever, which could delay the time before a patient sought treatment whilst still allowing parasite multiplication to occur.

The concept of a malaria vaccine that could provide an effective immune response when the antibodies induced had been ingested by blood-feeding mosquitoes, was developed thirty years ago 52. The target antigens of the passively transferred antibodies that blocked transmission were shown to be sexual-stage specific surface molecules (Pfs 48/45 and Pfs 230 in P. falciparum), that are involved in the process of fertilization of macrogametocytes by microgametes. Subsequently, other antigens (P25 and P28 proteins), that are uniquely expressed in the mosquito by zygotes and ookinetes (i.e. after fertilization), were shown to be equally good for induction of transmission-blocking immunity (TBI). The end result in each case is to prevent sporogonic development in the vector.

Experimental studies with animal models have shown that it is possible to induce a highly effective TBI 2253. The most extensive studies have focused on Pfs 48/45, Pfs 230, Pfs 25, Pfs 28 of P. falciparum and orthologues in other Plasmodium species, but more potential vaccine candidates have been identified 54.

The gamete surface molecules (48/45 and 230) are also expressed in gametocytes circulating in the blood. This has made it possible to study the nature and duration of naturally-acquired sexual-stage specific immunity, and has contributed importantly to an understanding of the epidemiology of gametocytes in comparison with that of the much more comprehensively studied asexual blood stages 12. The P25 and P28 proteins are not expressed in gametocytes and hence there is no natural infection-related immunity to them. However, the mRNA that encodes these proteins is measurable in gametocytes and is used as the basis of highly sensitive means of detecting gametocytes 16.

The transmission-blocking activity of sera from vaccinated animals or humans, or from individuals naturally infected, has mostly been assessed by an ex vivo assay, during which mosquitoes feed through a membrane on blood containing gametocytes and a serum under test. Comparison with controls of the numbers of mosquitoes that become infected, and the number of oocysts they carry, gives a measure of the potency of the serum under test.

Experimentally, sera from rabbits, monkeys and mice vaccinated with vaccine candidates Pfs25 from P. falciparum and Pvs25 from P. vivax contained antibodies with transmission-blocking activity. The antibody levels measured by ELISA correlated with both oocyst reduction and the number of mosquitoes that failed to become infected 55; antibody levels persisted at a high level for months after a second or third injection in mice 56. Phase 1 human vaccine trials were also effective 57, though not yet at the level that will be required and can be achieved experimentally.

Alternatives to the membrane-feeding assay are also being assessed for evaluation of transmission-blocking activity. Transmission of the transgenic P. berghei expressing the P25 antigens of either P. falciparum 58 or P. vivax 59 was blocked by antibodies obtained from animal vaccination and phase 1 clinical trials.

Though Pfs48/45 has been clearly shown to induce antibodies that prevent fertilization and correlate with transmission-blocking activity, the conformational nature of the epitopes, and the cysteine-rich nature of the protein has made production of a correctly folded recombinant molecule difficult. Encouragingly, production of a stable, properly folded C terminal portion of the molecule that induces transmission-blocking antibodies has recently been described 6061.

Since elimination of malaria requires complete removal of all parasites, the focus of measures used to accomplish this goal is quite different from that for disease control. Whether the measures used for parasite elimination involve drugs 9 or vaccines, or a combination of the two, what is required are tools that prevent production of gametocytes or that render them non-infective to mosquitoes.

The ideal vaccine would be one that induces complete sterilizing immunity or which is fully effective at blocking transmission. Nothing approaching induction of a sterile immunity has been shown so far and it seems unlikely that this will be achieved with sub-unit vaccines. This does not rule out the use of such vaccines as part of an integrated approach to malaria-elimination, but they are unlikely to induce an anti-parasite immunity of sufficient efficacy to eliminate all parasites. It remains to be seen whether attenuated sporozoites, which, in small-scale studies with a demanding and unusable vaccination regime, did give full protection, will be as effective in the trials now beginning.

The transmission-blocking vaccines currently being tested induce good, but certainly not complete, transmission-blocking immunity. Since the membrane-feeding and the transgenic rodent malaria assays allow sera from clinical trials to be tested for efficacy, it is possible to set up a series of small-scale phase 1/2a trials with different vaccine constructs and vaccination regimes. It should then be possible to make speedy progress towards improving vaccine efficacy and selecting the best vaccine for development. There is an urgency to do this.

There are biological and clinical features of infection that need to be taken into account in designing elimination measures that include vaccines. These are listed in Table 1 and, in particular, involve the interaction between the acquired immune response and the gametocyte infectious reservoir. As discussed earlier in this review, the gametocyte reservoir is grossly underestimated when assessed by conventional means, even in areas of low transmission. The same is true of asexual parasitaemias. In areas of stable transmission, numbers of gametocytes decrease with age more rapidly than do numbers of asexual parasites, but the proportion of individuals who are gametocytaemic is actually higher at lower transmission intensities 12. An increase in gametocyte prevalence has also been seen following control studies 62. For this and other reasons adult carriage becomes as important as that of children in terms of the source of infection for mosquitoes.

Naturally acquired immune responses to the sexual stages of the malaria parasite develop rapidly (in individuals with little or no previous exposure to malaria) and have transmission-blocking activity 63. However, in endemic areas, the ability of sera to reduce transmission decreases in older age groups, corresponding to a reduction in gametocyte numbers 64. The biological significance of this loss of immunity has been reviewed 12 and further supports evidence that adults with few parasites are nevertheless important as a reservoir of infection. There is evidence too, particularly for P. vivax, that a low level of immunity can enhance transmission to mosquitoes 65.

It is frequently stated that, as malaria control and elimination programmes progress, the population no longer has any immunity and becomes highly susceptible to epidemic infections. While this is true to some extent, especially for younger age groups, there is another aspect of immunity to consider. It is well known that those who had immunity to malaria quite rapidly lose their clinical immunity if no longer exposed and may develop a symptomatic infection at low parasite densities on subsequent exposure. However, the anti-parasite immunity they had acquired can persist for many years 666768. In other words, there is immunological memory directed against the infection. The relevance of this to malaria elimination is that, for many years, there may be a proportion of the population capable of supporting low grade infections from which gametocytes can be derived.