Introduction

Malaria remains a major cause of childhood morbidity and mortality in sub-Saharan Africa. In 2023, Africa accounted for 95% of global malaria deaths, with over three-quarters occurring in children under five years old¹. Alongside malaria, children in these settings face high exposure to a wide array of pathogens - including respiratory, enteric, and helminth infections ², making immune competence critical for survival and long-term health. Despite widespread vaccination efforts, several studies report reduced vaccine efficacy in malaria-endemic regions compared to malaria-naïve populations ^3–5. Although multiple factors may contribute⁶, growing evidence suggests that malaria itself can impair host immunity^5,7,8. Repeated P. falciparum infection has been associated with immunomodulatory changes, including expansion of regulatory T cells ^9,10, regulatory B cells¹¹, and atypical memory B cells with limited effector function^12,13. These alterations promote host tolerance for persistent parasitaemia, resulting from recurrent exposure in endemic settings, but they also suppress effector immune responses¹⁴, potentially to unrelated antigens, including vaccines. While malaria-induced immunosuppression has been described in both experimental and observational studies^7,8,15,16, its duration and broader impact on the developing immune system remain poorly understood. Most prior work has focused on short-term or antigen-specific outcomes, leaving open the question of whether early-life malaria exposure durably attenuates antibody responses over the long term. Conflicting findings across settings highlight the need for context-specific, and longitudinal data studies^16,17.

Here, we address this gap using two longitudinal cohorts from coastal Kenya with contrasting malaria transmission histories. One region (Junju) has experienced a sustained malaria burden, while the other (Ngerenya) underwent a rapid decline in transmission after 2004¹⁸. Intensive weekly malaria surveillance was conducted over more than a decade, enabling precise classification of individual exposure histories. We combined these data with serial serological measurements using a high-throughput in-house protein microarray. This design enabled us to investigate whether clinical malaria in early life leaves a lasting immunological imprint that compromises antibody responses to common childhood pathogens and vaccines. Our findings reveal a durable suppression of humoral immunity linked to malaria exposure in early childhood, with implications for vaccine effectiveness, serosurveillance interpretation, and immune recovery in endemic regions.

Materials and methods

Study setting and cohort design

This study was conducted in Kilifi County, a rural region on the northern coast of Kenya, within the catchment area of the Kilifi Health and Demographic Surveillance System (KHDSS), a long-term population-based platform maintained by the KEMRI-Wellcome Trust Research Programme¹⁹. Serum samples were obtained retrospectively from two well-characterised paediatric cohorts - Ngerenya and Junju - enrolled between 1998 and 2017 as part of annual malaria cross-sectional surveys²⁰. Ngerenya and Junju were selected for their contrasting malaria transmission intensities; Ngerenya experienced a sharp decline in malaria transmission beginning in the early 2000s¹⁸, whereas Junju maintained moderate malaria endemicity throughout the study period, with P.falciparum prevalence approximating 30% during the rainy seasons²¹. All children were visited weekly at home for the detection of febrile episodes, and any child with an axillary temperature ≥37.5°C was tested for P. falciparum parasitaemia using a rapid diagnostic test, with confirmation by microscopic examination of Giemsa-stained thick and thin blood smears. A clinical malaria episode was defined as fever in the presence of ≥2,500 parasites/μL. In addition to active malaria surveillance, serum samples were collected annually from each child and for future serological analysis. The vaccination status of each child for routine childhood vaccines was assessed using digitised immunisation records stored at the KEMRI-Wellcome Trust Research Programme.

Protein microarray antibody profiling

Antibody responses were measured using an in-house protein microarray platform. Recombinant and whole-virus lysate antigens (Supplementary table 1) were reconstituted in a glycerol-based buffer containing 1% Triton X-100 and printed in quadruplicate onto epoxy-coated glass slides using a non-contact microarrayer (Marathon Argus, Arrayjet, Scotland). Slides were fitted into hybridisation cassettes, washed with PBST (0.05% Tween-20 in PBS), and blocked for 1 hour at 37°C using PBST containing 5% BSA. Serum samples were diluted 1:30 in PBST with 5% BSA and incubated on the slides for 3 hours at room temperature. Slides were washed and probed with secondary antibodies: goat anti-human IgG conjugated to Alexa Fluor 647 (SouthernBiotech, cat. no. 2040-03) and goat anti-human IgA conjugated to Alexa Fluor 555 (SouthernBiotech, cat. no. 2050-32), each incubated for 1.5 hours at 25°C. Slides were rinsed, disassembled, and scanned using a GenePix 4300A scanner (635 nm at PMT 600, 532 nm at PMT 700). Mean fluorescence intensities (MFIs) were extracted from duplicate mini-arrays per slide and averaged across four replicate spots per antigen. Data processing and quality control of microarray data were performed in R (version 4.4.2).

Enzyme-linked immunosorbent assay (ELISA)

Measles-specific IgG was quantified using a conventional direct ELISA. Plates were coated overnight at 4°C with 2.30 µg/mL measles antigen diluted in PBS. After blocking with 5% skimmed milk for 1 hour at 37°C, serum samples (1:100 dilution) were added and incubated for 1.5 hours at 37°C. Plates were washed with PBST and incubated with HRP-conjugated secondary antibody (1:100 dilution) for 1 hour. Following additional washes, 100 µL of OPD substrate solution (30 mg OPD in 30 mL PBS with 30 µL H₂O₂) was added and incubated in the dark for 10 minutes. Reactions were stopped with 50 µL of 2.5 M sulfuric acid, and absorbance was measured at 490 nm.

Statistical analysis

All statistical analyses were performed using R (version 4.4.2). Antibody responses between cohorts were compared using the Wilcoxon rank-sum test. P-values <0.05 were considered statistically significant. Data were anonymised and delinked from all personally identifiable information.

Ethics statement

The study was approved by the Scientific and Ethics Review Unit (SERU) of the Kenya Medical Research Institute. All procedures were conducted in accordance with the principles of Good Clinical Laboratory Practice (GCLP).

Results

A natural experiment in coastal Kenya reveals sharply divergent malaria exposure trajectories in early childhood

Between 1998 and 2017, two rural communities in coastal Kenya - Junju and Ngerenya - underwent markedly different transitions in malaria transmission. Both regions experienced a high malaria burden in the 1990s and early 2000s. However, beginning in 2004, transmission in Ngerenya declined sharply and remained near zero for more than a decade, while Junju continued to experience sustained endemicity well into the mid-2010s. To quantify the scale and timing of this transition, we analysed surveillance data collected from August 1998 to April 2017. A total of 1,243 children were followed in Ngerenya and 659 in Junju. In Ngerenya, the proportion of febrile surveillance visits with confirmed P. falciparum parasitaemia declined from 22.4% before 2004 (1,378 of 6,148 visits) to just 1.1% after 2004 (48 of 4,389 visits). In contrast, Junju children continued to experience high rates of febrile malaria, with 17% of visits testing positive between 2007 and 2017 (869 of 5,130 visits) - Fig. 1.

Figure 1

To assess whether the protein microarray platform accurately captured longitudinal P. falciparum-specific immune responses, we examined IgG levels against the plasmodium apical membrane antigen 1 (AMA1) in individual children with known malaria exposure histories. Among children with multiple confirmed episodes of febrile malaria detected via weekly active surveillance, AMA1-specific IgG levels increased sharply over time, tracking closely with the timing of clinical infections (example shown in Fig. 2a), while children from Ngerenya who remained malaria-free throughout follow-up exhibited consistently low IgG levels against AMA1 over more than a decade of surveillance (example shown in Fig. 2b). These patterns mirrored expected immunological trajectories of repeated exposure versus non-exposure, confirming that the microarray platform is capable of detecting meaningful variation in P. falciparum-specific humoral responses. We then extended this analysis to assess longitudinal IgG responses to AMA1 in a subset of 123 children drawn from the two surveillance cohorts. These children contributed a total of 1,194 serum samples, with a mean of 10 samples per child, collected between 2002 and 2017 (Supplementary Fig. 1). This subset was selected on the basis of the relative completeness of their longitudinal follow-up serum sampling and included 58 children from Junju (505 samples) and 65 from Ngerenya (689 samples). AMA1-specific IgG levels diverged sharply between cohorts early in life and remained distinct throughout follow-up. In Ngerenya, levels declined rapidly after 2003, stabilising at lower levels by mid-childhood. In contrast, Junju children maintained elevated levels across all time points (Fig. 2c).

Figure 2

Malaria-exposed children exhibit lower antibody levels to non-malarial antigens

To examine the impact of differential malaria endemicity on the antibody response to non-malarial antigens, we first compared IgG responses to the measles virus among Junju and Ngerenya children. We started by validating the ability of the in-house protein microarray platform to detect biologically meaningful measles-specific IgG responses by examining longitudinal measles levels in individual children with known vaccination histories. Among children with a documented history of receiving all three recommended doses of measles vaccine, we observed a sharp rise in IgG levels following immunisation, followed by sustained levels into later childhood (example shown in Fig. 3a). In contrast, unvaccinated children exhibited consistently low IgG trajectories across all time points (example shown in Fig. 3b). These patterns were reproducible across the cohort and recapitulated expected vaccine-induced versus naïve antibody dynamics, supporting the use of this platform for population-level serological inference. Using this platform, we then compared IgG responses to measles virus among children from Junju and Ngerenya. This analysis was restricted only to children with a documented record of measles vaccination. Despite matched vaccination histories, children from Junju - where malaria transmission remained high - consistently exhibited lower measles-specific IgG titres than children from Ngerenya, where malaria transmission had declined. This difference was evident from the earliest timepoints and persisted throughout childhood. Annual antibody measurements showed that mean measles-specific IgG levels were higher in Ngerenya than in Junju in every sampling year (Fig. 3c). To validate these findings, we measured measles-specific IgG levels by ELISA in a subset of 3-year-old children who had completed measles vaccination at least one year prior. In this assay, Junju children again displayed significantly lower IgG levels than their Ngerenya counterparts (Fig. 3d).

Figure 3

Early-life malaria exposure is associated with long-term suppression of antibody responses

To assess whether the attenuation of antibody responses extended beyond measles, we compared IgG levels to a broader panel of antigens at 10 years of age. This analysis included vaccine-preventable pathogens (Bordetella pertussis, H1N1 influenza virus, rubella virus, and measles virus) and common childhood infections (cytomegalovirus (CMV), Epstein-Barr virus (EBV), herpes simplex virus 1 (HSV-1), coxsackievirus B1). Children from Junju exhibited significantly lower IgG levels for most antigens compared to their Ngerenya counterparts (Fig. 4). This pattern was observed for both viral and bacterial pathogens and was particularly marked for coxsackievirus, EBV, HSV-1, and measles. Antibody levels against P. falciparum AMA1 remained elevated in Junju. Antibody levels to the 2009 pandemic strain of H1N1 were the least differentiated between cohorts, although Ngerenya children still exhibited significantly higher levels than Junju participants (Fig. 4).

Figure 4

To determine whether the attenuation of antibody responses could still be attributed to early-life malaria exposure independent of geographic or environmental differences, we conducted a stratified analysis within the Ngerenya cohort, which experienced a sharp decline in malaria transmission beginning in 2004. Due to stochastic differences in malaria infection around the time of this inflection, children in Ngerenya had different malaria exposure histories despite living in the same area. At the 10-year-of-age time point, sera were available for 62 out of the 65 children that were originally selected in the Ngerenya cohort subset for serological analysis. Of these, 20 experienced one or more episodes of febrile malaria during early childhood prior to the decline in malaria transmission, while 42 remained entirely malaria-free throughout follow-up (Fig. 5a). At 10 years of age, Ngerenya children who had experienced early-life malaria exhibited significantly lower IgG levels to a wide range of antigens compared to their malaria-naïve peers (Fig. 5b). These included responses to Coxsackievirus B1, CMV, H1N1 influenza, HSV-1 and rubella. Differences in antibody level to B. pertussis and EBV did not reach statistical significance.

Figure 5

Discussion

This study demonstrates that early-life exposure to malaria is associated with broad and durable impairments in antibody-mediated immunity to unrelated pathogens and vaccines. By leveraging a natural experiment in coastal Kenya - where malaria transmission diverged sharply between adjacent communities during the early 2000s - we were able to disentangle the immunological effects of malaria exposure from confounding geographic and temporal factors. Our findings reveal that children exposed to malaria in early childhood not only generate lower antibody titres to non-malarial antigens but maintain these attenuated responses well into adolescence, long after malaria transmission has ceased. The attenuating effect of malaria on antibody responses has long been suspected ¹⁵ but difficult to quantify. Prior studies have shown reduced vaccine efficacy in malaria-endemic settings ^3,16, and experimental models have implicated regulatory T cells, atypical memory B cells, and B cell exhaustion as potential mediators of malaria-induced immune suppression^9,12,13,22. However, most human studies to date have focused on short-term immune responses or outcomes in the setting of concurrent parasitaemia. Our data extend this work by demonstrating that transient exposure to malaria in early childhood is sufficient to imprint long-lasting changes on the humoral immune repertoire.

The strength of this study lies in its integration of long-term active malaria surveillance with serial antibody profiling. Children were visited weekly for febrile illness surveillance, allowing for precise documentation of clinical malaria episodes. In parallel, our use of a validated protein microarray platform enabled us to track IgG responses to a wide panel of pathogen and vaccine antigens at multiple timepoints. We first confirmed that the microarray platform captured biologically relevant responses by comparing antibody trajectories in individual children with known measles vaccination and malaria exposure histories. The platform reliably detected both vaccine-induced and infection-associated rises in IgG, providing confidence in the longitudinal patterns observed. We found that children from Junju, a region of persistent transmission, had significantly lower antibody levels to multiple pathogens compared to children from Ngerenya, where malaria transmission declined in the mid-2000s. These differences were most pronounced for antigens such as measles virus, Bordetella pertussis, CMV, rubella virus, and HSV-1. Importantly, all children had comparable vaccination histories and were followed through the same longitudinal infrastructure, minimising the likelihood of differential healthcare access or vaccine uptake.

To test whether early-life malaria exposure specifically contributed to long-term immune suppression, we examined antibody responses within the Ngerenya cohort. Because the transmission decline occurred rapidly, children born just before and after the inflection point experienced markedly different levels of malaria exposure while living in the same geographic area. Among children followed longitudinally, those with even limited early-life exposure to malaria had significantly lower antibody titres at 10 years of age than their malaria-naïve peers. This within-cohort contrast strongly implicates early-life infection as the critical window for immune programming. Taken together, these findings support a model in which malaria exposure during critical developmental windows modulates long-term maintenance of immune memory. This may occur through direct effects on B cell maturation, altered antigen presentation²³, or long-lived changes in lymphoid microenvironments^24–26. While the mechanistic basis of this phenomenon remains to be fully explained, the consistency of the attenuation on multiple antigens and timepoints suggests a generalizable effect rather than antigen-specific phenomenon.

This work has important implications for vaccine policy and infection risk in malaria-endemic regions. Reduced antibody titres to vaccine-preventable diseases may translate into diminished long-term protection, even when vaccine coverage is high. Our findings raise the possibility that children in high-transmission settings may require different immunisation strategies, such as delayed dosing and boosting, to improve vaccine-induced immunity. Moreover, as malaria control efforts continue to shift disease epidemiology, these data highlight the value of longitudinal serological surveillance for understanding the broader immunological legacy of malaria exposure. Our study has several limitations. Although the microarray platform was validated internally and against ELISA, quantitative comparisons across platforms are inherently constrained. Additionally, while the sample size for antibody profiling was modest, the inclusion of dense longitudinal sampling and detailed clinical surveillance adds depth to the dataset. Finally, we cannot fully exclude the influence of other infections or environmental exposures that may have differed subtly between subgroups, although the within-Ngerenya analysis provides strong evidence for malaria as a primary driver of long-term immune attenuation.

In summary, our findings reveal that early-life malaria exposure is associated with long-term suppression of antibody responses to unrelated pathogens and vaccines. This effect is detectable many years after infection, and appears to persist even in the absence of ongoing transmission. As global malaria control efforts continue, understanding the immunological legacy of childhood malaria may be critical for improving vaccination strategies and mitigating susceptibility to other infections.

Data and Code Availability

The datasets and R scripts used for the analysis in this study are available from the corresponding author upon reasonable request. Due to local laws and regulations, individual-level data from the Kilifi Health and Demographic Surveillance System (KHDSS), cannot be made publicly available. All data processing and visualisation were performed using open-source R packages; package versions and analysis pipelines are available upon request.

Supplementary Materials

Supplementary Figure 1

Supplementary Table 1

Acknowledgements

This study was supported by fellowship funding to C.J.S. from the Wellcome Trust (WT105882MA). The funder played no role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript. M.S.S was funded in whole by Science for Africa Foundation to the Developing Excellence in Leadership, Training, and Science in Africa (DELTAS Africa) program [DEL-22-012] with support from Wellcome Trust and the UK Foreign, Commonwealth & Development Office and is part of the EDCPT2 programme supported by the European Union. For purposes of open access, the author has applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.