Introduction
Antibodies are critical components of naturally acquired immunity to malaria and are particularly important during the blood stage of the infection where targets include antigens expressed on the surface of the merozoite and the infected red blood cell [
1‐
3]. An understanding of the acquisition and maintenance of the antimalarial antibody response is crucial for improving prospects for successful vaccine development [
2,
4,
5] as well as to guide further design of reliable serological tools for transmission surveillance [
6,
7]. This requires a detailed characterisation of the kinetics of the antibody response following infection, including estimates of the longevity in individuals of different ages and with different levels of prior malaria exposure [
8‐
10].
The longevity of an antibody response is primarily determined by the generation and survival of long-lived antibody-secreting cells (ASCs) [
11‐
13]. For many viral and bacterial infections, a protective and long-lived response is acquired after a single exposure. Furthermore, the half-life of the antibody response to several common vaccine antigens has been estimated to range from approximately a decade, in the case of tetanus toxoid, to life-long and without decay, in the case of the measles vaccine [
14]. However, acquisition of immunity to malaria requires repeated infections [
15,
16] and, although protective antibodies are acquired with time [
1,
17,
18], the antibody response appears to be short-lived particularly in children [
5,
19,
20].
The longevity of long-lived malaria-specific ASCs is difficult to study as they reside in the bone marrow and the secondary lymphoid tissues and are only transiently detectable in peripheral blood following acute infection [
21]. However, the number of bone marrow ASCs correlates with circulating antibody levels in both mice experimentally infected with
Plasmodium chaubaudi [
22] and HIV-infected humans [
23]. In addition, modelling of longitudinal antibody data from highly malaria-exposed children using a bi-exponential decay model has been shown to allow for estimation of the half-life of malaria-specific IgG antibodies and both short- and long-lived ASCs, as well as their proportional contribution to the response [
5]. IgG consists of four different subclasses (IgG
1–4), each with different sets of effector functions and rates of turnover due to the inherent differences in their biochemical properties. The underlying IgG subclass profile may therefore influence the half-life of the antigen-specific total IgG response [
5,
24].
Reliable estimates of decay rates of antimalarial antibody responses require detailed study of the kinetics of the response after infection. However, studies of antibody kinetics in malaria-endemic areas are hampered by difficulties in determining the timing of the latest exposure due to frequent asymptomatic carriage of low-density infections [
25] and the continuous risk of reinfection during follow-up, causing boosting of immune responses [
5]. This can partially be addressed through studies of controlled human malaria infection (CHMI), in which all of the above are carefully monitored and controlled [
26]. However, participants in CHMI trials are typically treated at microscopic patency of blood-stage infection, often before symptoms have occurred [
27,
28]. The immune response observed in a CHMI may thus not fully mirror the response following a symptomatic natural blood-stage infection in which parasitaemia is higher and the inflammatory response more pronounced [
26].
Studying returning travellers, diagnosed with malaria in malaria-free countries, provides a unique opportunity to investigate the kinetics of the antimalarial immune response after a naturally acquired
Plasmodium falciparum infection in absence of re-exposure [
29,
30]. Furthermore, studying travellers with different levels of prior exposure enables investigation of how exposure affects the acquisition of long-lived ASCs and the overall longevity of the response. Previous studies in travellers have indicated short-lived antibody responses with a higher overall antibody reactivity in semi-immune compared to malaria-naïve individuals [
29‐
31]. However, due to infrequent sampling or lack of longitudinal follow-up, it has not been possible to provide quantitative estimates of antibody decay rates. Moreover, there has been little effort to properly quantify the antigen-specific half-life of the different IgG subclass responses or assess how the level of prior exposure affects the ratio of short- to long-lived ASCs.
Here, we investigate the kinetics of the total IgG, and IgG subclass-specific, antibody response to P. falciparum schizont extract and eight recombinant vaccine candidate antigens after a naturally acquired infection in travellers followed prospectively after treatment in Sweden. We use mathematical models to compare how the kinetics and the longevity differ with regard to prior malaria exposure and provide quantitative estimates of the decay of the antibody response as well as the relative contribution from short- and long-lived ASCs.
Discussion
We investigated the kinetics of the antibody response after a naturally acquired P. falciparum infection by studying a cohort of differentially exposed travellers followed for 1 year after treatment in complete absence of re-exposure. We observed greater magnitude, breadth, and longevity of the response to all tested P. falciparum antigens (AMA-1, MSP-119, MSP-2, MSP-3, and RH5) in individuals with previous repeated malaria exposure. Through mathematical modelling of longitudinal data on antigen-specific total IgG and IgG subclass responses, we provided quantitative estimates of the kinetics of antibodies and short- and long-lived ASCs. The half-lives of the antimalarial long-lived ASCs were notably short in comparison with tetanus-specific long-lived ASCs, and although long-lived ASCs were acquired following primary malaria infection, they represented only a small fraction of the total ASCs generated. We observed a greater longevity of the antibody response in previously exposed individuals and show that this can be explained by the maintenance of a greater number of long-lived ASCs.
During the year following infection, geometric mean antimalarial antibody levels of all antigen specificities (total IgG and all IgG subclasses) were consistently higher in individuals with prior malaria exposure. The kinetics of the antigen-specific total IgG response, which is reported to consist predominantly of the IgG
1 and IgG
3 subclasses [
47‐
50], was also best reflected in the subclass kinetics of either IgG
1 or IgG
3. High-level IgG
1 and IgG
3 responses to several merozoite surface antigens have been associated with clinical protection [
47‐
50], and here, previously exposed, but not previously naïve, individuals mounted high IgG
1 and IgG
3 responses to AMA-1 and MSP-2. Furthermore, a large proportion of individuals with previous exposure, but almost none of the previously naïve, mounted detectable IgG
4 responses irrespective of antigen specificity. The role of IgG
4 in malaria is to date incompletely resolved but increased levels have been associated with prolonged or chronic antigen exposure [
24,
48,
51].
Although the half-lives of antibodies and ASCs have previously been estimated from longitudinal antibody data in children living in endemic areas [
5], no quantitative estimates are previously available for primary infections or from cohorts followed in complete absence of re-exposure. Here, the estimated half-lives of
P.
falciparum-specific short-lived ASCs ranged from a few days to a few weeks whereas the half-life of long-lived ASCs range from approximately 2 to 6 years depending on IgG subclass and antigen specificity (2 to 4 years for total IgG, IgG
1, and IgG
3 versus 2 to 6 years for IgG
2 and IgG
4). There was a tendency for the estimated half-life of long-lived IgG
2 and IgG
4 secreting ASCs to be longer than for IgG
1, IgG
3, and total IgG. However, this could be due to biases related to the small number of samples with seroreactivity exceeding the lower limit of detection of the assay, particularly among the previously naïve. The estimated decay rates of antibody molecules as well as both short- and long-lived ASCs are highly consistent with previous estimates for antibodies and ASCs specific for MSP-1, MSP-2, and AMA-1 in African children [
5,
19,
20].
The greater boosting of antibody levels in previously exposed individuals is indicative of a greater number of short-lived ASCs, likely due to the presence of a memory B cell response that upon reinfection rapidly proliferates and differentiates into short-lived ASCs [
52‐
54]. It has previously been demonstrated that malaria-specific memory B cells can be acquired already after a few infections and that they may be maintained, and detected in the peripheral blood, for several years in absence of re-exposure [
55‐
57]. Furthermore, it has been demonstrated that memory B cells can be maintained independently of circulating antibodies and vice versa [
13,
56,
58,
59].
Interestingly, the previously naïve individuals were found to acquire a long-lived component of the antibody response during this primary
P.
falciparum infection. However, this component was small, and in previously naïve individuals, only minor proportions of the ASCs generated in response to the infection were estimated to be long-lived (e.g. 4–10% for total IgG of all antigen specificities). Individuals with prior exposure were estimated to maintain a greater proportion of long-lived ASCs (e.g. 10–56% for total IgG), supported also by the maintenance of two- to ninefold higher antibody levels at the end of follow-up approximately a year after the acute infection. Because of the rapid decay in short-lived ASCs, the antibodies they produce will disappear completely from the circulation within a few months after the infection has been cleared. Any circulating antibodies remaining at the end of follow-up will be maintained entirely by long-lived ASCs. The magnitude of the antibody response at the end of the follow-up will therefore be proportional to the number of long-lived ASC [
22,
23]. The higher levels of antibodies observed in previously exposed individuals compared to previously naïve individuals thus reflect the maintenance of a greater absolute number of long-lived ASCs. The greater longevity of the response in previously exposed individuals is further reflected in the greater antibody breadth (i.e. the number of antigens to which an individual responds), which has been previously associated both with protection from disease and with time since the last infection [
36,
60,
61].
The mathematical model used represents necessary simplifications of the underlying processes that regulate the generation and survival of ASCs [
5]. Differences between previously naïve and exposed individuals are assumed to be due to the differences in pre-infection antibody levels, differences in antibody boosting, and differences in the proportions of short- versus long-lived ASCs [
5,
52‐
54]. In our framework, differences in the longevity of the response could be modelled by varying either the proportion of ASCs that are long-lived or the half-life of the long-lived ASCs or by varying both. However, due to the relatively small sample size, it is not possible to significantly test for differences between the two groups while allowing both of the above to vary. We therefore make the assumption that the half-lives of ASCs are the same in both groups but that they differ in terms of the proportion of long-lived ASCs. Despite this simplification, the current model provides a good fit to the data on antimalarial antibody responses. Furthermore, when fitted to data on IgG antibody responses to TTd, in the same individuals, it reproduces previously published estimates of the half-life of the antibody response to tetanus toxoid of approximately 7–14 years [
14,
46,
62]. This provides a validation of the current model structure and the reliability of the model estimates for the kinetics of the antibody response to malaria antigens.
Because the exposure-dependent differences in antibody kinetics can be described by the differences in the size of the antibody boost and the acquired proportion of long-lived ASCs, we suggest that a more long-lived antibody response to malaria is acquired slowly over time by small consecutive additions to the pool of long-lived ASCs with each repeated infection. The relatively small number of long-lived ASCs generated by each new infection, as indicated by the low proportion of long-lived ASCs in the previously naïve group, contributes to the slow acquisition of a long-lived high titre antibody responses. This is supported by the data from Ghana and the Gambia, where the proportion of long-lived ASCs in children was estimated to increase with age [
5,
19], and could also partly explain the observation of a more rapid seroreversion in children, compared to adults, in endemic areas following interruption of transmission [
63].
To date, most successful vaccines induce protection mediated through antibodies that are maintained following just a few immunisations [
4]. The most advanced malaria vaccine, RTS,S, provides only partial protection, and vaccine efficacy appears to wane quickly as antibody responses decay [
64]. All of the antigens tested in the present study are either current or previous vaccine candidates with clinical trials ongoing for AMA-1 and RH5 [
45,
65]. Most of them, except RH5, are highly polymorphic and have been selected as candidate vaccines partly because of their immunogenicity in the context of natural malaria infection [
4,
66,
67]. However, here we find that despite the high immunogenicity, antibody responses decay rapidly in the absence of re-exposure. It is possible that the high rate of decay together with the highly polymorphic nature of many malaria antigens contributes significantly to the slow acquisition of clinical immunity [
16,
68,
69]. Furthermore, despite an enhanced expansion of parasite-specific antibodies in previously exposed individuals, they did not exhibit clinical protection against malaria disease, suggesting that B cells archived within the memory compartment either have non-protective specificities or that their reactivation is too late to control the acute infection [
16,
68]. RH5, in contrast, is a relatively conserved protein. It is not one of the major targets of naturally acquired antibodies [
70] but a leading blood-stage vaccine candidate that has shown promising results in vitro as well as in non-human primate models [
71,
72]. Nonetheless, the estimated longevity of the response to RH5 was similar to that of the highly immunogenic and much more polymorphic antigens tested, all of which were short in comparison with the response to the well-studied tetanus toxoid vaccine antigen. Accumulating evidence suggests that the immune environment induced by the malaria parasite inhibits the development of a long-lived antibody response. Development of both long-lived ASCs and immune memory are hampered by a dysregulation of the B cell response, where impaired T cell help and germinal centre formation [
73,
74] lead to preferential induction of short-lived ASCs and the generation of the so-called atypical memory B cells [
5,
55,
60,
75,
76]. The decay patterns of malaria-specific antibody responses observed here following natural infection are similar to those observed in African children following vaccination with the RTS,S vaccine [
64]. These findings emphasise the importance of tailoring a malaria vaccine to skew the humoral immune response towards the generation of long-lived ASCs by improving delivery platforms and adjuvants and by optimising dosage and vaccine regimens [
67].
Serology has proven to be a promising tool to monitor medium- and long-term trends in malaria transmission intensity, particularly in low transmission settings [
6]. Available tools have largely been based on the data from cross-sectional epidemiological surveys [
6,
8], but a challenge when modelling cross-sectional serological data is the issue of parameter identifiability [
10]. For example, when data are limited, it may be impossible to distinguish between the scenario in which individuals acquire and lose antibodies rapidly or the opposite in which antibodies are both acquired and lost slowly leading to spurious estimates of transmission intensity [
10]. Improvement of current serological tools for transmission monitoring requires reliable quantitative estimates of the underlying parameters, e.g. the antigen-specific antibody decay rate, as well as the identification of novel serological markers, or combinations of markers, that discriminate between recent or more distant exposure [
7,
61]. In the present context, the difference in antibody kinetics between the exposure groups was greatest for AMA-1 and MSP-2 antigens and these antigens appeared most informative to differentiate individuals based on the prior exposure. However, additional serological markers with shorter half-lives would be needed to accurately predict the timing of the last infection. We suggest that the longitudinal study of differentially exposed travellers may serve as a useful model system to identify such candidate serological markers that can subsequently be validated under field conditions.