Background

In malaria endemic areas, immunity is acquired with age and exposure to Plasmodium parasites, reflected by the progressive decrease in the rates of clinical malaria and eventually also of parasite densities [1]. In such settings, mostly in sub-Saharan Africa, younger children are at higher risk of developing the most severe forms of malaria and of dying as a result of infection. Different levels and/or types of naturally acquired immunity could influence the severity of a malaria episode [2]. Partial or suboptimal immune responses in young children may contribute to the severity of disease presentations [3,4,5,6].

Antibodies to the blood stages of the parasite life cycle, which are responsible for the disease symptoms and pathogenesis, increase with age [7] and are involved in immunity [8], but the precise targets of protection have not been fully elucidated. A high number of parasite antigens are polymorphic and/or variant, giving rise to different “strains” or genotypes of parasites that differ in time and space and may be associated with certain disease outcomes. Antibodies binding to polymorphic epitopes may not broadly recognise the alternative antigenic types, leading to immune evasion. If immunity to Plasmodium falciparum was essentially strain-specific [9], younger people would be less able to control the infection and disease, and older people would become immune after being exposed to a sufficiently large number of strains circulating in the community [10, 11]. Evidence for strain-specific immunity comes from experimental infections in monkeys [12] and humans [13,14,15], including the use of malariatherapy in the treatment of neurosyphilitic patients [16]. A primary infection by one parasite strain elicited an immune response protecting against that strain but not against infection by a different strain [17].

The merozoite surface proteins MSP-1 and MSP-2 of P. falciparum are potential targets of naturally acquired immunity and malaria vaccine candidates, but they also display antigenic diversity [18]. Antibodies recognizing both type-specific and conserved determinants have been detected in immunized animals [19] and humans [20]. In exposed humans, circulating IgGs recognise specifically one or other polymorphic antigenic types such as those present in the block 2 of MSP-1 [21] or in the dimorphic fragments of MSP-2 [22], but responses against polymorphic fragments of MSP-2 have not been investigated in detail during natural infections, particularly in children with severe malaria (SM). Some seroepidemiological studies in endemic areas have found protective associations between IgG to MSP-1 [7, 23,24,25,26,27] or MSP-2 [28, 29] and parasitaemia and/or symptomatic malaria episodes, but others have reported no correlations or even the opposite [30].

In some instances, antibodies against one of the genotypes were associated with protection (e.g. MSP-2 type IC1/3D7) whilst antibodies to other genotypes correlated with risk (e.g. MSP-2 type FC27) [31]. Vaccination with certain MSP-1 or MSP-2 genotypes followed by challenge with heterologous parasites has resulted both in protection [32] or no protection [33]. Vaccines based on MSP-1 or MSP-2 including variable regions have failed in clinical trials performed in malaria endemic regions possibly due to immune escape [34, 35], while others based on more conserved antigens, such as Rh5, show high promise [36].

New disease episodes have been associated with parasite genotypes to which the child has not been previously exposed [37, 38] while asymptomatic infections could be associated with genotypes previously encountered [39] or with presence of antibodies to conserved epitopes. Similarly, it was hypothesized that severe forms of malaria could be found in children with antibodies to polymorphic epitopes that do not match the infecting parasite genotypes, resulting in inefficient control of parasitaemia. Previous studies found either a similar degree of previous exposure to malaria in mild and severe disease [40, 41], or differential antibody recognition of certain antigens by clinical presentation [5, 42]. However, studies have rarely assessed antibodies for specificity against the particular parasite genotype(s) infecting the individual. In Amazonia, parasite density rather than the presence of allele-specific antibodies matching the MSP-2 type in infecting parasites was a major predictor of symptom severity [43]. In Papua New Guinea (PNG), children with malaria illness had higher levels of IgG to MSP-2 IC1/3D7 matching the infecting parasite genotype than to the heterologous FC27 type, and IgG to MSP-2 IC1/3D7 was significantly lower among children with cerebral malaria (CM) who died than among children who survived CM. Antibodies did not differ between CM and uncomplicated malaria (UM) [44]. Those prior studies were conducted outside of sub-Saharan Africa, where the heaviest malaria burden occurs, and did not include patients with severe malarial anaemia (SMA), who may behave differently from CM or UM patients.

Here, allele-specific IgG responses to MSP-1 and MSP-2 on admission were assessed in relation to the fresh parasite isolates collected at the time of admission, to test whether children with UM were more likely to have antibodies specific for the infecting P. falciparum genotype than children with SM (CM or SMA). In addition, antibody responses at convalescence in children who had sequelae and/or P. falciparum recurrent infections after discharge were analysed, to evaluate whether those negative outcomes were more common in the absence of antibodies recognising the infecting genotype.

Methods

Patients and samples

Children under age 12 years with SM were admitted to the Malaria Research Project and Wellcome Trust Centre (MRP), Department of Paediatrics, Queen Elizabeth Central Hospital, Blantyre, Malawi, during two malaria transmission seasons (January-June, 1996 and 1997). Clinical, demographic and parasitological data were recorded for all patients, including state of consciousness, history of seizures, duration of symptoms and treatment prior to admission. Finger prick samples were used to determine parasitaemia from thick and thin blood films and haematocrit levels. The Blantyre Coma Score (BCS) [45], was used to assess the level of consciousness. Among SM patients, 126 were diagnosed as CM (BCS < 3), 59 as SMA (Hb < 5 g/dl or PCV < 16%), and 53 as both CM and SMA. Control patients with UM (n = 81; fever with no recent history of coma or convulsions, and BCS of 5) were recruited from three sources: (i) ambulant children screened for enrolment in studies of novel antimalarial therapy at Ndirande Health Centre, Blantyre (ii) patients admitted to the hospital with a final diagnosis of UM and (iii) children attending the hospital for blood examination for malaria parasites with a clinical diagnosis of UM.

Five millilitres of venous blood in lithium heparin or EDTA were collected at presentation, centrifuged, buffy coat depleted, and plasma was stored at − 70 °C till the laboratory measurements performed in 1998. Plasma samples were obtained from 352 children. Participants were invited to return at 1 month, or earlier if they were unwell, and convalescent blood samples were obtained from 220 children in total from all study groups. Informed consent was obtained from parents or guardians, and the study obtained ethical approval from the Malawi National Health Sciences Research Committee (Study Approval Number P.11/07/593).

Negative control sera were obtained from 50 Scottish adults who had not been exposed to malaria. A pool of sera from African immune adult individuals was used as a positive control in all tests.

Parasite genotyping and production of recombinant proteins

Genotyping of P. falciparum MSP-1 and MSP-2 polymorphic alleles by a nested PCR is detailed in the accompanying manuscript [46] and reported elsewhere [21]. Expression, purification and characterization of recombinant proteins used in this study have been reported elsewhere [21]. Briefly, genomic DNA from cultured parasite isolates was used as template for PCR amplification of specific fragments of the genes. PCR products were cloned and expressed in Escherichia coli as recombinant proteins fused to the C-terminus of glutathione S-transferase (GST) of Schistosoma japonicum using the pGEX-2 T vector. GST fusion proteins were affinity purified by adsorption on glutathione agarose beads (Sigma, UK), and their concentration and purity were estimated by the Bradford protein assay kit (Bio-Rad, UK) and Coomassie blue-stained sodium dodecyl sulphate–polyacrylamide gel electrophoresis. The GST protein alone was purified from cultures transformed with pGEX-2 T vector (without MSP-1 or MSP-2 inserts) and used as a control antigen in ELISAs. All the proteins included here were shown to reflect the antigenic structure of P. falciparum-derived native proteins.

P. falciparum MSP-1

Constructs were derived from polymorphic and conserved regions of MSP-1. Fusion proteins based on N-terminal sequences included two block 1 constructs, from the MAD20 and Palo Alto variants, respectively, and five constructs representing the three main block 2 types: K1 (Palo Alto and 3D7), MAD20 (MAD20 and Well), and RO33 [21] (Fig. S1). Location of primers used for genotyping of P. falciparum MSP-1 are also shown in Fig. S1. In addition, two constructs representing the C-terminal regions were included: a GST fusion protein containing most of block 17 (N1631 to N1726 of the Wellcome isolate) corresponding to the 19 kDa fragment [47], and a baculovirus-expressed recombinant protein corresponding to the 42 kDa fragment of the CAMP isolate, donated by Dr. J. Lyon [5].

P. falciparum MSP-2

Constructs were derived from allelic sequences representing the two major dimorphic types of the protein, IC1/3D7 (13 constructs) and FC27 (9 constructs), designated here as type A and type B, respectively (Fig. S2). Primers used for amplification of MSP-2 DNA and derivation of additional MSP-2 proteins are shown in Table S1. Eight MSP-2 type A constructs were derived from polymorphic R1 repeat regions of the gene (primers 5/7) of isolates T9/96, ThaiTn, 7G8, T9/102, CH12/12, RO33 and T9/94. Four MSP-2 type A constructs were derived from dimorphic regions of the gene (primers 8/6) of isolates RO33, CH150/9 and T9/102, and one protein represented almost the full-length MSP-2 (primers 13/14) of the T9/96 isolate. With regard to MSP-2 type B, two constructs were derived from polymorphic R1 regions consisting of 32-mer amino acid repeats (primers 5/3 or 13/3) of the K1 isolate, 3 constructs were derived from R2 regions consisting of 12-mer amino acid repeats (primers 12/6) of isolates K1 and T9/105. Two constructs represented almost the full-length protein (primers 13/14) of isolates Dd2 and GF88/175, and 2 constructs were derived from conserved regions encoding the N-(primers 13/16) and C-terminus (primers 17/14) of MSP-2 of the K1 isolate.

Enzyme-linked immunosorbent assay (ELISA)

Wells of microtiter plates (Immunolon-4, Dynatech) were coated overnight at 4 °C with 0.05 μg/100 μl of antigen per well in 0.1 M carbonate (Na2CO3-NaHCO3) buffer (pH 9.6). Three sets of plates containing, respectively, panels of recombinant proteins derived from MSP-1 and MSP-2 types A and B were prepared. In parallel, antigen control wells containing GST or buffer alone were set up in each plate. The plates were washed 3 × with PBS-0.05% Tween 20 and blocked for 5 h at room temperature with 200 μl of blocking buffer (1% wt/vol skimmed milk powder in PBS-Tween) per well. At the same time, Malawian test plasma and European control sera were diluted 1/500 in the blocking buffer and incubated at room temperature for 5 h. Plates were washed as above, 100 μl of diluted sample/well was added to duplicate wells, and incubated overnight at 4 °C. After washing, IgG bound to the wells was detected with horseradish peroxidase-conjugated rabbit anti-human IgG (Dako Ltd. High Wycombe, UK) at 1/5000 dilution in PBS-Tween for 3 h at room temperature. After washing, the reactions were developed with 0.012% H2O2 as the substrate and o-phenylendiamine (Sigma) as the chromagen (100 μl per well). The reactions were stopped after 10 min with 20 μl of 2 M H2SO4 per well. Optical density (OD) was measured at a wavelength of 492 nm.

Statistical analysis

Specific serological reactivity was calculated by subtracting OD values for the GST or buffer controls from the value obtained for the recombinant protein to obtain specific OD values. Positive samples were then defined as those giving a specific OD above the cut-off. The cut-off for each antigen was taken as the mean + 2 standard deviations of the European control samples.

Data was introduced into the SAS statistical analysis package (SAS Institute, 1990) to analyse the correspondence between the MSP-1 or MSP-2 types of the infecting P. falciparum isolates and the specificity of the IgG responses mounted by participants. A positive IgG response to MSP-2 A (IC1/3D7 type) was defined as the presence of IgG recognizing one or more than one of 13 fusion proteins based on MSP-2 type A alleles. Likewise, a positive IgG response to MSP-2 B (FC27 type) was defined as the presence of IgG recognizing any of 7 non-conserved constructs representing MSP-2 type B sequences. A trait called “concordance” was defined as a match between the type-specificity of positive IgG responses and any parasites detected in the acute or convalescent stages of the illness. Thus, an IgG response was defined as concordant when the specificity of the IgG matched the type of the infecting parasite found in the blood of the same patient; a response was defined as discordant when the specificity of the IgG detected did not correlate with the type of any parasite present. Children who had no detectable IgG were defined as non-responders and were excluded from the concordance analysis. The relationship between parasitaemia, multiplicity of infection, age and antibody levels was analysed by PROC GLM. The relationship between concordance, disease status and outcome was analysed by PROC GENMOD (categorical linear model), using a binomial distribution for the dependent variable. Significance was defined at the 5% level.

Results

Participant characteristics are summarized in Table 1. Seropositivity (%) of IgG to each MSP-1 and MSP-2 construct tested for the acute and convalescent visits is presented in supplementary material. Antibodies to all the polymorphic, dimorphic and conserved fragments of MSP-1 and MSP-2 were detected, with diverse seroprevalences (Tables S2 and S3). A total of 343 of 352 acute samples (97.4%) and 216 of 220 convalescent samples (98.2%) had detectable antibodies to one or more antigens, with 17 non-responder children (4.8%). Seropositivity consistently declined between acute and convalescent samples except for MSP-1 block 2 RO33 (Tables S2 and S3). A negative association was found between multiplicity of infections and IgG levels to MSP-1 C-terminus (PROC GLM, 19 kDa p = 0.036, 42 kDa p = 0.048). There was a trend (p = 0.14) towards higher breadth of antibody responses in more severe CM compared to less severe UM cases at convalescence (Fig. S3) while no difference was noted on admission. The IgG responses to the polymorphic and dimorphic protein regions were further analysed here in relation to the infecting P. falciparum genotypes defined by PCR in the same patients [46].

Table 1 Clinical and parasitological data of the patients studied
Full size table

Type-specific antibodies to polymorphic block 2 of MSP-1

The higher prevalence of IgG to block 2 K1 type correlated well with the predominance of infecting parasites of the K1-type in the same children [46]. The correspondence between the infecting parasite and the specificity of the IgG response to the block 2 of MSP-1 at the individual level was analysed in those children whose plasma contained detectable antibodies to proteins representing one of the three polymorphic types: K1, MAD20 and RO33.

In acute samples, parasite typing and serological analyses were done in 345 children (Table 2). Overall, antibodies more frequently matched the infecting type (example in Fig. S4) than not (example in Fig. S5). Fifty-three (15.4%) plasma samples had IgGs to the K1 type of MSP-1, i.e. reacted positively with either Palo Alto and/or 3D7 proteins. IgGs specific for the K1 type matched the presence of an infecting parasite of the K1-type in 44 serum samples (83% concordance), whereas a mismatch was found in 9 samples (17%). 25/350 (7.1%) samples had IgGs specific for the MAD20 type, i.e. reacted positively with MAD20 and/or WELL type proteins. Among those, the presence of MAD20-specific antibodies correlated with the presence of a MAD20 type parasite in 17 children (68% concordance), and a mismatch was observed in 8 patients (32%). Finally, 20/348 samples (5.7%) contained IgGs reacting to the RO33 protein; 15 children had RO33-type parasites (75% concordance), and 5 children did not (25%). Despite the low prevalences of IgGs recognizing one type of block 2, specific antibody responses were detected in 77.5% of children who had block 2 antibodies at presentation. Although mixed infections were very common in these patients [46], antibody responses to more than one type of block 2 were very rare (Fig. S4D).

Table 2 Concordance between MSP-1 and MSP-2 genotypes and IgG type-specific responses
Full size table

Among the 307 discharged children (survivors), a total of 190 children came for a follow up visit 1 month later, and additional blood samples were obtained from 31, 11 and 4 children at 2, 3 and 4 months after admission, respectively (218 samples). Among those, 154 (71%) were parasitaemic by microscopy and 64 (29%) were negative. Ninety-seven (63%) of the positive samples were obtained from asymptomatic children, 44 (28.6%) from children with UM, 10 (6.5%) from children with persisting neurological sequelae and 3 (2%) from children readmitted to hospital with SMA. 79/97 (81.4%) asymptomatic infections were undetected by microscopic examination but became apparent by subsequent PCR typing. In these infected cases, anti-MSP-1 block 2 antibodies detected in convalescent plasma were compared to the parasites found during acute infection and also to the parasites found at convalescence. 34/234 convalescent plasmas (14.5%) had K1-specific IgG, and this matched the presence of an infecting parasite of the K1-type in 32/34 cases (94% concordance). A mismatch was found in only 2 samples. 11/235 convalescent samples (4.7%) contained MAD20-specific antibodies. The presence of MAD20-specific IgGs correlated with the presence of an infecting parasite of the MAD20 type in 8/11 of the convalescent children (73% concordance), whereas a mismatch was observed in 3 patients. Finally, 22/233 convalescent plasmas (9.4%) contained IgGs reacting with the RO33 protein. In 21/22 children (95.5% concordance) a parasite of the RO33 type had been detected in acute and/or convalescent phase, whereas in only 1 sample there was a mismatch between parasite type and antibody response.

Overall, IgGs to MSP-1 block 2 genotype(s) present in acute and/or convalescent stages were detected in the majority (91%) of convalescent responders. In the asymptomatic individuals, a number of genotypes appeared to be controlled below the disease threshold level in the presence of MSP-1 type-specific IgGs possibly produced or boosted in response to concurrent infection(s).

Type-specific antibodies to dimorphic regions of MSP-2

The higher seroprevalence of IgG to type A than to type B MSP-2-specific regions was consistent with the higher prevalence of parasites of the MSP-2 A type in the area [46]. In 279 children, the IgG response to the two dimorphic types of MSP-2 on admission was compared to the infecting parasite genotypes (Table 2). Among those, 219 (78.5%) contained IgGs to MSP-2 type A. In 181 of these children (82.6%), anti-MSP-2 type A antibodies matched the presence of a parasite of the MSP-2 type A (Fig. S6), but in 38 children (17.4%) there was a mismatch (Fig. S5B). 208 children (73.5%) had antibodies to MSP-2 type B; antibodies matched the parasite in 108 patients (52%) and did not in 100 patients (48%).

At convalescence (n = 192), 142 children (74%) had IgGs to MSP-2 type A. In 130 (91.5%) there was a match between anti-MSP-2 antibodies and parasites of the MSP-2 type A in acute and/or convalescent phases, whereas in 12 children there was a mismatch (8.5%). 136 children (72%) had IgGs to MSP-2 type B; in 100 patients (73.5%) there was a concordant response, whereas in 36 patients (26.5%) there was a discordant response. Convalescent plasmas were more likely to contain antibodies matching an infecting parasite of the MSP-2 type B than acute samples (p < 0.001).

Antibody responses to both A and B types were common in individual children, as mixed infections containing both genotypes were frequently encountered in these patients [46]. Some of the double positive responses could be explained by cross-reactive antibodies recognizing C-terminal conserved sequences, but in most cases IgGs were directed to group-specific regions that are conserved only within each allelic family (Fig. S6C). Overall, 68% of acute and 83% of convalescent patients made specific antibodies matching the MSP-2 type(s) of the infecting parasite(s).

Relationship between type-specific antibody responses and malarial disease severity

Comparison of antibody levels to conserved antigen epitopes between the different disease manifestations (UM, CM, SMA) has been previously reported [5]. To test whether the presence or absence of IgGs specific for the infecting parasite genotype in acute infections had an influence on the severity of the disease, the correspondence between antibody responses and parasite type(s) was compared among groups of children with the different clinical presentations. In addition, type-specific antibody responses were evaluated at convalescence. The concordances between the presence of a parasite bearing a particular MSP-1 and MSP-2 genotype and the presence of antibodies of the corresponding specificity for each disease group are shown in Table 2.

Figure 1 shows all the data from the children in whom IgG responses to MSP-1 and MSP-2 antigens could be compared to the parasites detected and PCR-typed in their blood at acute and convalescence sampling, stratified by disease diagnosis and outcome.

Fig. 1
figure 1figure 1

Heatmaps of levels IgG antibodies by antigen and diagnosis, indicating MSP-1 block 2 and MSP-2 dimorphic types of infecting parasites and disease outcomes, in acute (A) and convalescent (B) samples, and heatmaps of concordance of IgG antibody response by antigen type and disease groups. Each row corresponds to a patient. Columns correspond to antibody responses to different antigen fragments on the left and categorical variables (diagnosis, outcome, parasite types) on the right. Continuous antibody levels (OD) and categorical concordance variables (matched or unmatched comparing anti-MSP-1 and MSP-2 IgG responses vs. respective parasite types) are scaled in green–red colour intensities. For MSP-1 block 2, K1-type parasites are depicted in green, MAD20-type in blue, and RO33-type in purple. For MSP-2, A-type parasites are depicted in red and B-type in orange (white represents absence of the genotype)

Full size image

Regarding antibodies to the MSP-1 block 2 type, the low numbers of seropositive children in each disease group precluded a meaningful statistical analysis. However, there were no apparent differences in the ability to mount strain-specific immune responses to this polymorphic region of MSP-1 among patient categories. In all groups, most of the few responding patients (between 6 and 15% of total) recognized the infecting parasite and produced antibodies of the appropriate specificity in the acute and convalescent stages.

The analysis of the relationship between MSP-2 antibody specificity and disease status revealed some differences among patient groups. Among children infected with parasites bearing MSP-2 type B, matching anti-MSP-2 type B antibodies were significantly less prevalent during acute disease in children with CM than in those with SMA (χ2 test, p < 0.005) or UM (χ2 test, p < 0.025). CM patients with anaemia followed a similar trend as children with CM alone, i.e. lower percentage of matches for MSP-2 type B than SMA alone (p < 0.05). Among children infected with parasites of the MSP-2 type A on admission, those with SMA had anti-MSP-2 type A antibodies in acute and convalescent stages less frequently than those with UM (p = 0.06) or CM, which might have been driven by the lower ages of children with SMA relative to other groups, but the differences were not statistically significant.

Patients who died did not differ in the specificity of antibody responses compared to those who survived (p > 0.05). The relationship between antibody-parasite genotype concordance for MSP-2 and disease outcome in patients who survived is shown in Table 3. Children who suffered neurological sequelae, as a result of CM, were more likely to have antibody responses of the wrong specificity than children who recovered from any form of disease. However, there were relatively few patients with sequelae and the differences were not statistically significant.

Table 3 Concordance of antibody specificity to MSP-2 and disease outcome in all patients
Full size table

In summary, patients with milder malarial disease appeared to have a higher proportion of matching antibody responses to MSP-2 than those who developed severe symptoms (Fig. 1). Children who were infected with parasites of the MSP-2 type B and developed CM produced antibodies that were the least concordant to MSP-2 type B parasites present at the acute stage. Children infected with parasites of the MSP-2 type A who presented with SMA produced antibodies that were the least concordant to MSP-2 type A parasites in acute and convalescent stages.

Discussion

An analysis of antibody responses to polymorphic regions of P. falciparum MSP-1 and MSP-2 in comparison to the individual’s infecting genotypes in Malawian paediatric patients with distinct malaria clinical presentations is reported here. Previously, it was shown that children with SMA had very low levels of IgG to the conserved C-terminus of MSP-1 and MSP-2, and to full-length AMA-1, while children with CM had significantly higher levels of IgG to those antigens than children with UM [5]. Here, a step further was undertaken to assess whether a lack of IgGs specific for the MSP-1 and MSP-2 alleles present in the acute stage infection could be associated with SM.

In most patients, the specificity of IgG response to polymorphic regions of MSP-1 and MSP-2 was found to correlate with the parasite genotyped by PCR, thus antibodies were detected early in the course of the disease. Overall, concordance for block 2 of MSP-1 was between 70 and 87%, consistent with a longitudinal study in Sudan [48]. Concordance for MSP-2 was lower, 77% for type A and 52% for type B, which may be partially explainable by a lower sensitivity of detection of the PCR that could have underestimated low-density parasite genotypes. As antibodies at the time of patients’ presentation can reflect both past and present infections, discordances found in some acute and/or convalescent samples might be due to the persistence of IgGs produced in past infections with different parasites. Although antibody levels tended to decrease soon after drugs had cleared parasites, patients did not usually serorevert after a month (i.e. they remained seropositive). In addition, as parasite genotypes can appear and disappear from the peripheral blood every 48 h [39], some discordances could be due to sequestration of particular genotypes in mixed infections with distinct synchronous populations.

Infections with MSP-2 type B parasites induced IgG of the matching specificity less frequently than infections with the type A. It is possible that parasites bearing MSP-2 type A or B vary in their ability to induce immune responses. The observation that MSP-2 type B was more common in adults than in children [49], and also more frequent in symptomatic than asymptomatic individuals [50], led to the suggestion that parasites with MSP-2 type B must be more virulent than those with type A. It was argued that MSP-2 type B parasites may evade the immune system more successfully than type A parasites in semi-immune individuals. In addition, antibodies specific for dimorphic regions of MSP-2 type A have been associated with lower malaria morbidity [28]. However, another study found that high levels of IgG to MSP-2 type B were not significantly associated with a lower likelihood of clinical episodes caused by parasites bearing type B compared to type A antigen, and vice versa for IgG to MSP-2 type A [15]. It is possible that parasites with type A MSP-2 do not survive as well as parasites with type B due to a more efficient type-specific acquired immunity. Data presented here indicate that the immune system may respond less efficiently to MSP-2 type B during the acute phase, which might lead to more parasite proliferation and morbidity. Indeed, MSP-2 type B infections were associated with higher parasitaemias in these patients [46]. A putative immunological advantage of parasites with type B MSP-2 might be related to distinct structural features. A less diverse repeat structure in MSP-2 type B compared to type A probably reflects separate evolutionary strategies of the two allelic families [51]. The conservation of a general structural organization in MSP-2 type B alleles could be explained by functional constraints, e.g. involvement in receptor-ligand interactions with the host. Indeed, future analyses should go beyond antibody levels and also measure neutralizing and non-neutralizing functionality [42]. One study reported strain-transcending Fc-dependent killing of P. falciparum by MSP-2 allele-specific human antibodies, with unexpectedly high level of heterologous antibody-dependent parasite inhibition and opsonic phagocytosis [52].

It was anticipated that only antibodies specific for the current infection, and thus concordance, would increase during convalescence. As expected, IgGs at follow up matched the infecting parasite(s) better than those measured in acute infections, particularly for MSP-2 type B infections. The presence of type-specific antibody responses did not prevent many patients (70%) from being infected. However, children with persistent infections at convalescence had very low parasite densities, detectable only by PCR and not by microscopy, and mild (30%) or no symptoms (63%) of malaria. This suggested that type-specific antibodies had some anti-parasite role at controlling parasitaemia below the disease threshold level and that children were in the process of acquiring the status of premunition rather than these IgGs just being an indication of exposure to prior infections. A feature of acquired immunity to malaria is that it is never complete or sterile, and it has been shown that immunity strongly restricts the growth of diverse parasite genotypes [37, 39].

The general reduction in MSP-1 and MSP-2 IgG seropositivity within a month was surprising; it was expected that IgGs would be higher at convalescence than during acute disease since it takes some time for the host to respond to an infection. However, only IgG to MSP-1 RO33-type block 2 followed this prediction. Previous studies have reported that IgGs to MSP-1 are short-lived, especially in children [30, 48, 53]. It could also be that children are brought to hospital days after the infection and symptoms onset, and these two dates are separated by an unknown and variable amount of time. Thus, data suggest that infected children produce IgGs that decay after the parasites have been cleared below disease threshold levels.

No differences between disease groups were detected regarding antibody responses to block 2 of MSP-1, despite being a target of strain-specific immunity [13, 14], largely due to the low prevalence of IgG responses to any given genotype. To complement this work, it would be interesting to also test the specificity of antibodies to the dimorphic regions of MSP-1 (blocks 6-16), using recombinant proteins representing the two major families (MAD20 and Well/K1), in analogy to the study by Früh [53]. Concerning responses to MSP-2, children who developed SM symptoms differed from children with UM in their ability to mount specific antibodies to the infecting parasites in the acute phase of the disease. A decreased ability to respond to parasites of MSP-2 type B may predispose a child to develop CM. As a whole, the CM group had very high prevalence of anti-MSP-2 type B IgGs (80%). Therefore, the low frequency of responders to type B was a characteristic of CM patients infected with parasites of this type. Similarly, a decreased ability to respond to infecting parasites of MSP-2 type A may partially contribute to SMA. The lower frequency of anti-MSP-2 type A positive responses among SMA patients was apparent only in those who were infected with parasites of type A, since 88% of plasmas from the SMA group as a whole had IgG to MSP-2 type A. This observation fits with conclusions from a study in PNG that found that antibodies to MSP-2 type A were associated with less anaemia [28]. The majority of SMA patients had antibodies matching MSP-2 type B infecting parasites. Thus, high levels of IgG against MSP-2 type B did not seem to confer any protection against anaemia [44].

No differences in the specificity of antibody responses in relation to the outcome of the disease were found comparing between those who died and those who survived. This was in contrast to Al-Yaman [44], who reported lower anti-MSP-2 type A antibody levels among fatalities. In this study, patients who survived and suffered from neurological sequelae after CM had less concordant IgG responses to MSP-2 than patients who had a full recovery or who were asymptomatic. Despite this trend, there were too few patients with sequelae to draw a clear association.

Beyond MSP-1 and 2, analyses of naturally acquired allele-specific antibody responses to other merozoite vaccine candidates like AMA-1 have been performed in patients with different clinical presentations. In Kenya, no clustering of allelic haplotypes with disease severity was observed and antibody reactivities to three AMA-1 alleles were highly correlated [54]. Thus, antibodies to both conserved and allele-specific epitopes of blood stage P. falciparum targets showing amino acid sequence diversity may contribute to protection against disease and severity.

The study had some limitations. There were many fewer MSP-1 than MSP-2 antibody data points, and presumably this affected power and thus precision of conclusions about MSP-1 genotypes and gaps in antibody response. In addition, without a pre-infection sample it was not possible to attribute the antibody response to current or prior infection. The fact that antibody prevalence to the genotypes did not change much between presentation and convalescence could be because much of what we detected was pre-existing with possibly infection-induced boosting. The study also has strengths, like the distinct analysis between CM and SMA, who are often wrongly bundled together as SM despite being totally different clinical manifestations of P. falciparum malaria. However, the biological interpretation of the findings in the CM + SMA group remains challenging.

Conclusion

Low antibody response to one or the other allelic family of MSP-2, in particular, might differently predispose a child to develop one or the other severe forms of malaria, but the reasons why this might be the case remain unknown. In addition to immune escape, it is conceivable that different regions and/or allelic forms of the MSPs are processed and presented to the immune system by different pathways and, perhaps, result in divergent immunopathological processes. For instance, it is possible that parasites bearing MSP-2 type A or B vary in their ability to induce inflammatory cytokines (e.g. TNF) implicated in causing severe pathology. The possibility of a differential Th1/Th2 pattern of immune responses induced by each MSP-2 dimorphic type merits investigation as it may have implications for malaria vaccine development. Nevertheless, the study hypothesis could be more directly evaluated measuring pre-existing/acquired IgG responses to parasite proteins more directly linked to SM, such as clonally variant antigens on the infected erythrocyte surface (PfEMP1), which are also highly polymorphic and targets of malaria immunity.