A seven-year study on the effect of the pre-erythrocytic malaria vaccine candidate RTS,S/AS01 E on blood stage immunity in young Kenyan children

Background: RTS,S/AS01 E, the most advanced malaria vaccine confers partial immunity. The vaccine-induced pre-erythrocytic immunity reduces exposure to blood-stage parasites, delaying acquisition of antibodies to blood-stage antigens. However, the duration of this effect is unknown. Methods: We measured, by enzyme-linked immunosorbent assay, IgG-antibodies to 4 Plasmodium falciparum blood-stage antigens (AMA1, MSP1 42, EBA175, and MSP3) on 314 children randomized to receive RTS,S/AS01 E or Rabies vaccine at 5 – 17 months of age in a phase 2b trial in Kenya, and thereafter participated in a 7-year study of the duration of vaccine immunity. Results: Antibody levels to MSP1 42, AMA1 and EBA175 were slightly lower among the RTS,S/AS01 E recipients, relative to the Rabies-control vaccinees, during the first 48 months of surveillance. Irrespective of vaccine arm, antibody levels to merozoite antigens were positively associated with the risk for malaria. However, this was only apparent at high levels for EBA175 and AMA1 and was not evident after adjusting for heterogeneity in malaria-exposure. Among children with asymptomatic parasitaemia, antibody levels were associated with reduced clinical malaria. Conclusions: The reduction in levels of antibodies to blood-stage antigens induced by vaccination with RTS,S/AS01 E can last for several years. In absence of asymptomatic infection, anti-merozoite antibody levels were unreliable correlates of clinical immunity.


Introduction
Despite the recent gains in malaria control, the disease remains a major public health risk, with 216 million cases and 445,000 deaths associated with malaria in 2016 1 . Progress in malaria control has stalled and may have reversed in some areas 2 .
RTS,S/AS01 E is the most advanced candidate malaria vaccine and is based on the circumsporozoite protein (CSP) that targets the pre-erythrocytic cycle of Plasmodium falciparum in humans. Vaccination with RTS,S/AS01 E has been partially efficacious against malaria in phases II and III trials in Africa 3,4 . RTS,S/ AS01 E induces pre-erythrocytic immunity. In contrast, naturally acquired immunity to malaria is largely dependent on antibodies to blood-stage parasites including the merozoite stage. Although there are no unambiguous correlates of natural immunity 5 , antibodies to merozoite antigens have been associated with protection through multiple mechanisms including the inhibition of erythrocyte invasion and replication 6 , complement-dependent mechanisms 7 , and enhancement of uptake and clearance by circulating phagocytes 8,9 . Antibodies to antigens expressed on the surface of infected red blood cells (iRBCs) have also been associated with immunity, which could inhibit or reverse sequestration of iRBCs, inhibit formation of rosettes, and promote opsonization of iRBCs for uptake by phagocytes 10-12 .
Antibodies to malaria parasites are acquired as a result of exposure. As such, interventions like insecticide impregnated bed nets and RTS,S/AS01 E -vaccination that reduce exposure to blood-stage antigen will affect the rate at which antibodies to merozoite and other blood-stage antigens are acquired. Previously, we and others demonstrated that RTS,S/AS01 E and RTS,S/ AS02 vaccinations reduced blood stage antibody levels, likely as a result of reducing the exposure to blood stage parasites due to induction of partial pre-erythrocytic immunity 13,14 . However, the duration of this effect remains unknown. It is important to determine the duration of this effect as RTS,S/AS01 E vaccination could delay the development of naturally acquired immunity, increasing the possibility of continued susceptibility in older children after the waning of the vaccine induced immunity 15 .
In this study, we aimed to determine the durability of the previously reported reduction in antibody levels to merozoite antigens in children receiving RTS,S/AS01 E vaccination, relative to Rabies control vaccines 13 . We analysed plasma samples collected from children during a seven-year extended follow up of a phase IIb randomized, controlled trial of RTS,S/AS01 E among young children in Kilifi, Kenya, examining antibodies to 4 different merozoite antigens by enzyme-linked immunosorbent assay (ELISA). We then analysed the effect of RTS,S/AS01 E vaccination on the acquisition of these antibodies and tested for potential correlations between antibody levels and protection from clinical malaria episodes.

Methods
Study design 447 healthy Kenyan children aged 5 -17 months were randomized in a 1:1 ratio to receive 3 doses at monthly intervals of either RTS,S/AS01 E or Rabies vaccine in a phase 2b trial, to evaluate the efficacy and safety of RTS,S/AS01 E against clinical malaria episodes by P. falciparum infection. Details have been published elsewhere 3 .

Monitoring for episodes of clinical malaria
The primary end point was a clinical episode of malaria, defined as an axillary temperature of >37.5°C, with a P. falciparum parasite density of 2500 parasites/microlitre of blood. Active surveillance was implemented with weekly home visits, where children were screened for fevers associated with P. falciparum parasites, both during the trial and the extended follow up period. A parallel passive surveillance was implemented by field workers residing in the study villages and health care staff in local health facilities.
Asymptomatic infections were detected by both microscopy and blood-smears during the cross-sectional data and sample collecton surveys described below.

Blood samples
Vaccines doses were given at month 1, 2 and 3. Blood samples were initially taken (1) 20 . In brief, each antigen was coated onto high absorbance plates (Immulon4 HBX) at a con-centration of 0.5 micrograms/mL and stored at 4°C overnight. The plates were washed 3 times in phosphate-buffered saline (PBS) with 0.05% Tween 20 (PBS-T) and blocked for 3 h with blocking buffer (1% w/v dried skimmed milk powder in PBS-T). After 3 additional washes, 100 microlitre of each plasma sample were added to duplicate wells at a final dilution of 1/1000 in PBS-T. The next day, after 5 washes, 100 microlitre of horse radish peroxidaseconjugated antihuman IgG (DAKO) at a dilution of 1:5000 in blocking buffer was added to each well, and plates were incubated for 3 h. The plates were then developed using H 2 0 2 as substrate and OPD (Sigma) as the colorimetric indicator for 20 min in the dark. Plates were read at 492 nm on a Molecular Devices Versa Max ELISA reader. Tests were repeated if duplicate optical density (OD) values for an individual plasma sample varied by more than a factor of 1.5. A pool of serum samples from an area in Africa where malaria is highly endemic was titrated on each plate and acted both as a positive control and provided values for a standard curve for converting optical density (OD) readings into arbitrary units, minimizing inter-plate and inter-day variations. A 3-parameter sigmoid ligand binding model was used to least-squares fit a curve to the values of the hyperendemic serum sample pool, and this was used to calculate sample antibody levelss on each plate.

Statistical analysis
Antibody scores from ELISAs were expressed relative to the OD readings obtained from the hyperimmune standard, with a score of 1000 scaled to be the maximum reactivity seen at the lowest dilution used in the hyperimmune standard curve, and then log-transformed before analysis. Student's T-test with comparison of means and non-parametric analyses with comparisons of medians and rank sum tests were used to compare groups. For the prospective association with malaria risk, the antibody levels were split into deciles, and a Poisson regression analysis was conducted with the unit of analysis being the period of time after each antibody level was estimated, hence including up to 10 observations per child, using the clustered sandwich estimate in Stata 15 (StataCorp LLC). We used the exposure index, as previously described 21 , to estimate exposure to malaria based on geographical location. n=278) months of the third dose of vaccination were tested for antibody levels. The antibody levels varied widely, with the majority of the children being unresponsive (i.e. lower than the lowest value on the straight part of the sigmoid curve based on the dilution of the hyperimmune standard serum), while the rest had values lying within the straight part of the hyperimmune standard curve ( Figure 1).

Results
Anti-merozoite antigen antibody levels split by RTS,S/ AS01 E vaccination Geometric mean antibody levels for all the 4 merozoite antigens increased with age, irrespective of vaccination group, but this was more apparent for 3 of the 4 antigens, and less apparent for AMA1 ( Figure 2). There were indications of seasonal variation during the first year of sampling when 4 samples were collected per child, as previously described 13 , but it was not possible to assess seasonality once sampling was scaled back to 1 sample per child per year, timed to occur in the dry period just before the main transmission season. Antibody levels for AMA1, EBA175 and MSP1 42 diverged after vaccination, with levels being higher among the Rabies control vaccinees than the RTS,S/AS01 E vaccinees at months 12, 24 and 36, 24 and 36, and 6, 12, 24, 36 and 84 of the third dose of vaccination for AMA1, EBA175 and MSP1 42 , respectively ( Figure 2). Thus, the divergence was temporal for AMA1 and EBA175, as the differences in the median antibody levels reduced with time and were similar by 48 months of the third dose of vaccination. In contrast, anti-MSP1 42 antibody levels were still higher among the Rabies-control than the RTS,S/AS01 E vaccinees at 84 months of the third dose (the last time point of sampling) with statistical significance (Table 1). Similar patterns were seen on non-parametric analyses with medians.   unknown. Here, we investigated the longevity of the reduction of antibody levels to four blood stage antigens after an extended follow up of the vaccines and controls for up to 7 years post-vaccination. We found that immunization with RTS,S/AS01 E and the associated clinical protection resulted in the reduction of antibody response to MSP1 42 , AMA1 and EBA175 antibody levels but not for MSP3. While the antibody levels for AMA1 and EBA175 among RTS,S/AS01 E vaccinees were below those measured in the Rabies control vaccinees during the first 48 months of monitoring, antibodies to MSP1 24 remained lower in the RTS,S/AS01 E vaccinees than in the controls throughout the study period. This latter, persistent difference was statistically significant except at the very last timepoint, when statistical significance was only marginal, considering that there are multiple comparisons by timepoint and by adjuvant (p=0.019).
In this study, antibody levels to four specific merozoite antigens were not associated with clinical protection. Rather anti-merozoite antibody levels were positively associated with the risk of clinical malaria for the group as a whole. The most likely explanation for this is that antibody responses are markers of exposure, and therefore represent ongoing risk of future exposure to malaria, and this interpretation is supported by the fact that the positive association was reduced after controlling for the exposure index. It is possible that higher antibody titers might have been protective (i.e. those above a protective threshold 22 ).
In the presence of asymptomatic infection, anti-EBA175 antibodies at all the deciles were higher than the lowest (i.e. nonreactive) decile, and some of the higher deciles for AMA1 and

Antibody levels and subsequent risk of clinical malaria
Antibody levels were split into deciles, which were then tested for prospective associations with protection from malaria in the transmission period after each, but before, the next sampling time-point. Pre-existing antibody levels for the 4 different merozoite proteins were not associated with clinical immunity (Figure 3). Instead, the incident rate ratio for clinical malaria increased with rising antibody levels. This relationship was most apparent at higher antibody levels (>5 th decile) for AMA1 and EBA175 (irrespective of vaccine arm). However, the incident rate ratios for the effect of antibodies on clinical malaria reduced after controlling for heterogeneity in malaria exposure using an exposure index, suggesting that these anti-merozoite antibodies are markers of exposure, rather than immunity.
Furthermore, when all the data were stratified by asymptomatic-parasite positivity at sampling by microscopy, the highest levels for AMA1 and MSP3, and all the of levels for EBA175 above the non-reactive group, were associated with reduced rate ratios for clinical malaria, among the children with asymptomatic parasitaemia at the time of sampling. Associations between higher antibody levels and increased incident rate ratios were maintained among the children without asymptomatic parasitaemia (Figure 3).

Discussion
We and others reported previously that RTS,S/AS01 E vaccination resulted in a reduction in antibody levels to blood stage malaria antigens 13,14 . However, the duration of this effect is MSP3 antibodies were associated with protection from clinical malaria (irrespective of the vaccine arm). This finding is consistent with several previous studies where analyses of single antigen-specific antibody responses within whole populations demonstrated no protective effect of antimalarial antibodies, but the same antibodies were associated with clinical immunity when parasite-positive individuals were analysed separately 22,23 . Our analysis involves only four antigens and there is evidence that the breadth of antibody positivity is also important for protection 24,25 .
An RTS,S/AS01 E induced reduction in blood stage immunity will have implications for the outcomes of vaccination if the vaccine is deployed for routine use among African children. If vaccination resulted in delayed development of natural immunity, then some of the gains of the vaccination may be offset by delayed susceptibility as the vaccine induced immunity wears off. Studies done to date on Phase II trials have suggested this possibility with a three-dose vaccine regimen 15 , although the effect may be countered by a fourth dose 26 . We show here that antibodies to blood stage immunity are reduced after vaccination with RTS,S/ AS01 E , which is consistent with induction of pre-erythrocytic immunity leading to a reduced incidence of blood-stage parasitaemia. However, antibodies induced by natural exposure to the four blood stage antigens tested were not consistently associated with immunity to malaria and there is no widely accepted or consistent immunological marker for immunity to malaria. It will be important to combine RTS,S/AS01 E with other malaria control measures like insecticide treated nets for protecting individuals from malaria, and further clinical evaluations of the four-dose vaccine regimen should include long-term follow up in the implementation trials.

Ethical considerations
The study protocol and its subsequent amendments received ethical and scientific approval from the Kenyan Medical Research Institute National Ethics Committee. The study was overseen by an independent data-monitoring committee and local safety monitors and was conducted in accordance with the Helsinki Declaration of 1964 (revised 1996) and Good Clinical Practice guidelines. Written informed consent in the local languages (Swahili or Giriama) was required from parents/guardians for participation.

Author information
FMN and PB conceptualized the study, supervised and managed the collection of immunology data, analysed and interpreted the data, and wrote the paper. JM performed the antibody measurements, JW conducted and supervised surveillance for malaria, and sample collection, PN conducted and supervised the the vaccine trial, supervised and sample collection. KM and PB obtained the funding, and supervised the overall conduct of research. CD was involved in antibody measurements and interpretation of the data. PB conceptualized and provided supervision for the study, supervised and managed the clininical trial, analysed and interpreted the data, prepared the metadata and wrote the paper. All the authors reviewed the manuscript.  In this study, the authors take advantage of a rich collection of longitudinally sampled plasma from an RTS,S malaria vaccine trial to address questions of blood-stage antibody responses after vaccination. In their prior work, they showed that RTS,S vaccinees have reduced antibodies against four merozoite antigens during the 12 months after vaccination (Bejon JID 2013). Here, as a follow-up, they compare antibody levels in RTS,S and control vaccinees at multiple intervals up to 84 months and assess the relationship between Pf-specific antibodies and prospective malaria risk during the interval prior to the following antibody time point.

Grant information
Overall, the unadjusted data supports (albeit with modest significance at most time points) the main conclusion that humoral immunity to tested blood-stage antigens is hampered or delayed by RTS,S vaccination, and this reduction in immunity is extended for years after vaccination. As the authors mention in the Discussion, they only test four merozoite antigens that may not elicit naturally protective antibody responses, and so the connection to RTS,S-associated long-term reductions in protective blood-stage immunity still remains a question.
As shown in their previous work, the differences in antibody responses after vaccination was affected by RTS,S vaccine-induced differences in prior malaria episodes. Thus, it would be important to show if the significant difference between comparisons in Fig 2 and Table 1 still hold after adjusting for prior episodes. Related to this, and to provide more conservative interpretations of their results, if one were to account for multiple comparisons (three comparisons per time point), only EBA175 and MSP1 have significant differences at 24, 36, and 60; and at 8.5, 24, 36, respectively.
In the Discussion, the statement "In the presence of asymptomatic infection, anti-EBA175 antibodies at all the deciles were higher than the lowest (i.e. nonreactive) decile" is redundant and unclear as written. Was the intention to state "…[malaria protection] for anti-EBA175 antibodies at all the deciles [was] higher…," given the lower IRR in the higher deciles for EBA175? The subsequent clause, "and some of the higher deciles for AMA1 andMSP3 antibodies were associated with protection from clinical malaria (irrespective deciles for AMA1 andMSP3 antibodies were associated with protection from clinical malaria (irrespective of the vaccine arm)" is not supported by the plots in Fig. 3 given most of the 95%CI bars cross unity (the lone exception being the highest decile for AMA1).
It is notable that the lone antigen that clearly did not show a significant group difference in Fig. 2 was MSP3, which, interestingly, was also the only antigen expressed as FVO, with the others being 3D7. As the authors are aware, the RTS,S vaccine was most effective against the vaccine strain (3D7) (Neafsey et al NEJM 2015). Thus, a possible explanation for the lack of difference for MSP3 is similarity in incidence of malaria episodes caused by non-vaccine strains. This possibility should be added to the discussion. It would be interesting to compare antibody data for 3D7 antigens vs. antigens from heterologous strains, especially a genetically divergent one such as FVO.
Minor Comments: To harmonize with the main text, it would be more useful for the reader if Fig. 1 showed the best fit curve (fitted to standards) for each antigen with sample values superimposed to provide a sense of how many samples were on the linear versus lower and upper plateaus. 4 or 5 parameter logistic models are more frequently used for ELISA dilutional standard curves as they most often give the best fit. The authors should briefly mention why they opted for a less conventional model.
Also, can the authors explain the rationale for using Poisson regression to estimate malaria risk at each interval in this study when Cox regression was used in their previous study? Did some of the underlying assumptions change when the evaluation was extended from 12 m to 84 m?
Although the timing of the sampling can be somewhat inferred from Fig. 2, Table 1, and the text, it would be helpful for readers if there was a Fig. showing the plasma sampling intervals.
In the Results section, the authors use "age" in the first reference to Fig. 2. However, Fig. 2 plots use months from the 3rd vaccine dose on the X axes. "Age" should be changed to months after vaccine dose as the latter is more accurate.
In the statistical analysis section, please indicate that deciles were determined on a per time point basis if this is the case.
For the AMA1 in Fig. 2, the y axis could be re-scaled to better visualize the significant difference between groups at the circled time points. Currently, lines for both groups appear superimposed.
Note that that there is a discrepancy in Fig. 2: Text legend states shaded regions are 95% CI whereas For Fig. 3, the main text and legend mention deciles were used but only 5 categories are shown for AMA1 and 8 categories for the other 3 antigens. If the lowest deciles (5 for AMA1 and 2 for the others) were combined into a single reference group, please indicate in the legend. Also, it might be clearer if X axis was labeled as categorical deciles (0-10%, 11-20%, etc) as the current X axis can be assumed to be continuous.

Typographical notes:
For the last sentence of ELISA methods, extra "s" in "levelss" In order to test the hypothesis that immunity to clinical malaria is delayed by RTS,S, this manuscript details the findings from a study of serum ELISA antibody levels of blood stage antigens found in blood from RTS,S trial participants or Rabies vaccine controls. As this cohort was studied over 7 years, and active surveillance data is available, it is an invaluable indication of the effects of this vaccine on infection, clinical episodes and immune parameters affected. IgG to MSP1-42, (3D7), MSP3 (FVO), PfEBA175RII (3D7), and AMA1 (3D7) were used as an indication of exposure and/or immune status and found to represent (recent) exposure best, though they do accumulate with time since vaccination/age. Please describe the extent of exposure of participants using averages or distribution of EIR or episodes. Perhaps the result would depend on this range. All I see is "Exposure Index, as previously described...based on geographical location". I suggest you break the y-axis to show the normal (are they all normal?) part of the curve more I would say values below the "linear part of the sigmoid curve" of the hyperimmune standard Fig 1 result: curve, were considered unresponsive. Are these the data points marked as a concentration of 2 in Figure  1?
The increase with age is apparent for AMA1 for participants with highest levels, the difference The legend could be more descriptive of the data and less of the method. Define IQR, CTl vs Vac in Fig 2: legend. Also, more could be done to make the figure print well in black and white-change shape of symbols and lines in one group. Could significance be indicated on the graph itself?
Text -the sentence describing the data is hard to understand. "Antibody levels for A, E, M Table 1: diverged after vaccination. Levels were higher for Rabies vaccinees than RTS,S vaccinees for AMA1 at 24 months, for EBA at 24-72 months, and for MSP1 8.5, 24, 36 and 84 months. (Unless p<.05 is not the cutoff?) But as is, only two sets of dates are listed with three antigens listed "respectively". The table itself could be improved by adding stars for significance, and by separating antigens with a thicker line for easier readability.
Great analysis, please state the meaning of IRR-Incident Rate Ratio in the legend, or on the Can you tell from your data, if the larger error bars for MSP1 slide positive group, are due to the faster decline of these antibodies compared to other specificities? It will be great when robust neutralizing assays have been developed that could distinguish effective blood stage antibody from ELISA positive/exposure induced.
In the last sentence, maybe you mean among the children with asymptomatic parasitemia? or without symptomatic parasitemia?
Please check the run-on sentence "In the presence of asymptomatic infection, anti-EBA175 Discussion: antibodies at all the deciles were higher than the lowest (i.e. non-reactive) decile (I'm already lost), and some of the higher deciles for AMA1...". I would add a sentence about the positive side-reduced exposure leading to reduced clinical episodes. How does that balance out with the potential for increased episodes from delayed immunity, are they predicted to be worse episodes, as in later severe disease?

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? No source data required

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed. Competing Interests: 1.

3.
Reviewer Expertise: Immunology of malaria I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard. do not. They can then proceed by stating if the antibody titers detected in this study are either above or below that threshold level and then proceed to speculate if there is merit on the statement that the administration of the RTS,S vaccine directly (or indirectly) impairs the development of the P.f. specific antibodies. It's worth noting that it is the same team who reported a 53% efficacy for the RTS,S vaccine with 38 episodes of clinical malaria in the RTS,S arm compared to 89 episodes observed in the control group vaccinated with a Rabies vaccine that was associated with higher anticircumsporozoite antibody titers (Bejon 2008 ). This detail needs to come in their et al., discussion.
Minor comments: The authors need to justify the choice of Rabies vaccine in the control group either by referring to their previous works or other related works.
There is need for a paragraph on the limitations of the study in the Discussion section. The authors raise the point about malaria transmission control measures and vaccine approaches being effective in controlling the pre-erythrocytic stages of the malaria infection but, not only failing to confer robust antibody mediated immunity, but might even be compromising/impairing the development of the P.f. antibody-mediated immunity against the blood stage in the affected children. Maybe they need to point out how a combination of these measures might have a compounded effect on the development of merozoite-stage immunity.
In the discussion section the authors propose that the administration of the RTS,S vaccine should be combined with the provision of other malaria control measures like ITNs. They also suggest that the four-dose vaccine regimen should include long-term follow up in the implementation trials. Adding a paragraph on informed speculations on what could be achieved by combinations of vaccine candidates (especially one specific for the sporozoite stage and one specific for the merozoite stage) might add weight to their argument. This is in light of their observation that RTS,S might actually be predisposing the recipient children to acquiring blood stage malaria infection presumably due to the impaired development of this stage-specific immunity.
They only introduce about the four-dose in passing in the discussion section. Could they please expand on that and say more on what is being proposed to be implemented?
The fact that seasonality was not entirely eliminated as a potential confounded in this second report needs to be expanded in the discussion section with appropriate reference to how seasonality was observed to affect the outcome of the vaccine in the previous study/report.
In the introduction please consider change the sentence to "...based on the circumsporozoite protein (CSP) that targets the pre-erythrocytic STAGE (not cycle) of the P. falciparum life cycle in humans." 10.

13.
14. 15. humans." The main objective was to determine the duration of the effect of the RTS,S vaccine. Has this been achieved and reported appropriately? Could they do this better by referring to what they had reported in their JID paper (Bejon , 2011 ) and build on that in this report? et al.
Since they are using pool serum samples from healthy controls, could the authors clearly state how blood samples were collected from the study participants, what volume, in what tubes, and what anticoagulant was used?
The plasma samples were collected at different stages of the seven year duration implying some were stored for longer than others. What measures were put in place to account for a possible deterioration of the antibodies which could account for the reported decline of antibody titers with time?
The legend for Figure 1: "Distribution of anti-merozoite antibody levels. Antibody levels were measured by ELISA" requires more details.