Antibody responses to Plasmodium falciparum and Plasmodium vivax blood-stage and sporozoite antigens in the postpartum period

During pregnancy a variety of immunological changes occur to accommodate the fetus. It is unknown whether these changes continue to affect humoral immunity postpartum or how quickly they resolve. IgG levels were measured to P. falciparum and P. vivax antigens in 201 postpartum and 201 controls over 12 weeks. Linear mixed-effects models assessed antibody maintenance over time and the effect of microscopically confirmed Plasmodium spp. infection on antibody levels, and whether this was different in postpartum women compared with control women. Postpartum women had reduced Plasmodium spp. antibody levels compared to controls at baseline. Over 12 weeks, mean antibody levels in postpartum women increased to levels observed in control women. Microscopically confirmed P. falciparum and P. vivax infections during follow-up were associated with an increase in species-specific antibodies with similar magnitudes of boosting observed in postpartum and control women. Antibodies specific for pregnancy-associated, VAR2CSA-expressing parasites did not rapidly decline postpartum and did not boost in response to infection in either postpartum or control women. After pregnancy, levels of malaria-specific antibodies were reduced, but recovered to levels seen in control women. There was no evidence of an impaired ability to mount a boosting response in postpartum women.

Pregnant women in malaria endemic settings may possess antibodies to a broad range of antigens but are still susceptible to P. falciparum and P. vivax infection in pregnancy 16 . In the case of P. falciparum, this is largely because they lack antibodies to placental binding P. falciparum isolates that express VAR2CSA, a variant surface antigen expressed on the surface of the infected erythrocyte involved in placental sequestration 17,18 . As VAR2CSA is primarily encountered by the immune system during pregnancy, VAR2CSA antibodies are typically absent or at low levels prior to first pregnancy and increase with exposure through multiple pregnancies 19 . Antibody responses to other blood-stage antigens may also boost upon exposure in pregnancy 20 . There is currently a lack of consensus as to how antibody responses, particularly VAR2CSA antibodies, are maintained or decline postpartum. Previous studies have reported increases 21,22 , no change 23,24 and decreases [22][23][24][25][26] in P. falciparum antibody levels postpartum compared to pregnancy depending on study site and antigen. However, no study included a control group of non-pregnant women to enable comparisons, nor had available data on the presence of Plasmodium infection between serological measurements [21][22][23][24][25][26] . Importantly, most studies assessed antibodies at one time-point postpartum [21][22][23][24]26 , thereby not taking into account potential fluctuations and limiting inferences about postpartum antibody persistence. Additionally, past research has only considered P. falciparum targets in African settings.
To address the paucity of data and investigate whether the postpartum period comprises a period of humoral immunity transitioning to a non-pregnant state, we determined P. falciparum and P. vivax antibody levels at multiple time points in postpartum and control women living at the Thai-Myanmar border. We sought to determine whether there are differences in boosting and maintenance of Plasmodium spp. specific antibodies in postpartum women compared with control women.

Methods
Ethics statement. The  Study design and population. This study investigated 201 postpartum cases and 201 non-postpartum (and non-pregnant) controls over a 12-week period of study. Individuals with at least 3 sera samples available, or who experienced a microscopically confirmed P. falciparum infection, were selected from a larger case-control study of malaria in the postpartum period (described previously 9 ). Briefly, pregnant women attending Shoklo Malaria Research Unit (SMRU) antenatal clinics located in North-Western Thailand from November 2007 to September 2009 were invited to participate and were asked to find a non-pregnant female of similar age and location to act as a control. During the follow-up period all women were tested for the presence of Plasmodium infection via weekly light microscopy of blood smears and completed questionnaires on behavior. Monthly blood samples (~200 μ l) were collected to assess haematocrit and serology. The first measurement (baseline) was obtained at first postpartum visit (median days since delivery: 13, interquartile range: 9-16). Microscopically confirmed P. falciparum infections were treated with mefloquine and artesunate. Microscopically confirmed P. vivax infections were treated with chloroquine. Antibody determination. We measured antibodies to eight P. falciparum antigens (merozoite (PfEBA140 RIII-V , PfEBA175 RII , PfEBA175R III-V , PfAMA-1, PfMSP2, PfRh2); P. falciparum infected erythrocyte: PfDBLα ; sporozoite (PfCSP), all 3D7 alleles) and four P. vivax antigens (merozoite (PvAMA-1 (Palo Alto), PvMSP1 19 (SalI), PvDBP (SalI)); sporozoite (PvCSP (Belem and PNG)) and one VAR2CSA expressing P. falciparum parasite strain (CS2) 27 (Supplementary Table 1). These antibodies represent different lifecycle stages and are biomarkers of exposure and protective immunity 28,29 .
1462 sera samples from 402 women were assayed by high-throughput ELISA and flow cytometry. Nine seroreactive pools from Thailand and Papua New Guinea individuals were included as positive controls for standardisation and 37 non-exposed Melbourne donors acted as negative controls. ELISA was conducted as described previously 20 . The locations of samples on plates were randomised across the entire cohort and for each assay all samples were processed on the same day to minimise batch effects. Reactivity from the nine positive controls (in quadruplicate) was used to adjust for any inter-plate batch variability. Antigen and sera concentration are provided in Supplementary Table 1. Testing for IgG binding to the surfaces of CS2 infected erythrocytes was conducted as described previously 30 . Data were acquired by flow cytometry (FACSVerse, BD Biosciences) and analysed using FlowJo (FlowJo LLC, OR, USA). Assay output was derived by subtracting the mean fluorescence intensity (MFI) of uninfected erythrocytes from the MFI of trophozoite-infected erythrocytes. Seropositivity was defined as an OD or MFI exceeding the mean plus three standard deviations of negative controls.
Statistical analysis. Statistical analyses were performed using Stata Version 13.1 (StataCorp, College Station, TX, USA). Correlations of antibodies at baseline were assessed using Spearman's rank correlation coefficients. Antibody levels were analysed as log transformed continuous variables centred on the median value (log 2 ([OD or MFI] + 0.001) − median(log transformed value)). Overall P. falciparum and P. vivax merozoite immunity scores were derived by calculating the average of the log transformed (centred) values for each antibody targeting species-specific merozoite antigens. To investigate if postpartum status was associated with species-specific antibody responses, linear mixed-effects modeling (with random intercept and slope) of log 2 Scientific RepoRts | 6:32159 | DOI: 10.1038/srep32159 antibody levels versus time was performed with interaction terms between postpartum status and time (weeks). This enabled us to compare the slope of antibodies over time in postpartum women with control women.
Potential confounders, selected a priori using a causal diagram 30 , included clinic attended (Mawker Thai/ Wang Pha/Walley/Mu Ler Chai), age (years) and history of working outdoors. In the case of PfVAR2CSA expressing parasites, a binary variable of gravidity ≥ 3 (note, gravidity of 0, 1 and 2 had similar PfVAR2CSA antibody levels) was included as a confounder instead of age, as gravidity is more predictive than age for placenta-adherent parasite immunity, and age and gravidity were highly collinear. In a subgroup analysis of postpartum women, detectable (by light microscopy) species-specific infection during pregnancy (yes/no) was included as a covariate. To assess the impact of changing haematocrit on antibodies over time, an interaction term between time and haematocrit change (≥ 4%) was assessed. To assess whether antibodies were boosted in response to infection, species-specific infection during follow-up was included as a time-varying variable with presence of infection (yes or no) at any visit prior to the corresponding antibody level measurement. The effect of heterologous species infection was assessed for each antibody response but only incorporated into the final model if p < 0.05. Given that we have reported all statistical comparisons conducted, p values were presented unadjusted for multiple comparisons 31 . Associations were interpreted based on the magnitude and direction of effect in addition to confidence intervals and p values.  (Table 1). Bednet use was high in both postpartum and control women (> 93%); postpartum women were less likely to report a recent history of working outdoors than control women (60% versus 72%). During follow-up, 22 postpartum and 32 control women experienced a microscopically confirmed P. falciparum infection; 67 postpartum and 50 control women experienced a microscopically confirmed P. vivax infection. One postpartum woman and four control women experienced a co-infection.  Antibody dynamics in postpartum and control women. Availability of multiple serological measurements for each woman allowed us to assess postpartum antibody dynamics and relate these findings to control women. Due to the large number of merozoite responses determined and the considerable collinearity between merozoite responses, an average P. falciparum merozoite response and P. vivax merozoite response were generated for ease of interpretation. In the absence of an infection, P. falciparum and P. vivax antibody levels appeared relatively stable in postpartum and control women during the 12-week follow-up period ( Supplementary Figures 1  and 2). Residual variances were significantly lower in those who did not experience species-specific infection than in those who did experience species-specific infection (likelihood-ratio test, all p < 0.001, except PfVAR2CSA p = 0.05). Many postpartum and control women who were PfVAR2CSA seropositive at baseline maintained their responses over the course of the study (Fig. 2). PfVAR2CSA seropositive women with a documented infection during pregnancy had a significant negative slope of antibodies over time (− 0.06 (95% Confidence Interval: − 0.11, − 0.02; p = 0.01) while in the other PfVAR2CSA seropositive women there was no significant decline in antibodies (p = 0.50).

Characteristics
In order to assess differences in antibody responses at baseline and over time in postpartum and control women, multivariable linear mixed-effects models were constructed for each antibody response (Table 2;  Supplementary Tables 2 and 3). Age and a history of working outdoors, both indicators of past exposure, were associated with increased antibody levels to P. falciparum and P. vivax targets. Adjusted mean levels of antibodies to PfVAR2CSA expressing parasites were 0.29 (95% CI 0.15-0.42) higher in women with at least three prior pregnancies compared to those with two or fewer prior pregnancies (Table 2, p < 0.001). At baseline, postpartum women had lower mean levels of P. falciparum and P. vivax antibodies targeting merozoites, sporozoites and infected erythrocytes compared to control women (adjusted mean difference 0.16 to 0.32 (Table 2)).
Mean antibody levels did not change significantly over time in control women ( Table 2, all p > 0.14). In contrast, the mean level of antibodies in postpartum women increased during the follow-up period; with a 0.01-0.03 mean increase per week follow-up ( Table 2). As such, over the course of the 12 weeks of follow-up the initial difference in merozoite, sporozoite and infected erythrocyte antibody levels between postpartum and control women reduced over time as mean antibody levels of postpartum and control women converged (Fig. 3). Adjusting for haematocrit in the models did not alter these observations. Similarly, in a subset analysis of postpartum women accounting for species-specific infections during pregnancy (which were associated with increased baseline antibody levels), mean postpartum antibody levels increased with time (p = 0.10 for Pf merozoite immunity, p < 0.05 for all others) (Supplementary Table 4).

The effect of species-specific infection on P. falciparum and P. vivax antibody levels in postpartum and control women.
To investigate the effect of species-specific infection on antibody levels, and whether postpartum status modified this effect, the previous models were expanded to incorporate species-specific infection, detected by light microscopy, during follow-up as a time-varying variable ( Table 3 and Supplementary  Table 5). There was strong evidence of boosting with species-specific infection in antibody responses to Pf merozoites, Pv merozoites, PfCSP, PvCSP, and PfDBLα (mean increase 0.85 (95% CI: 0.57, 1.  To investigate the impact of heterologous species-infection on antibody levels, models were examined incorporating heterologous-species infection during follow-up as a time-varying variable. Infections with one species did not boost antibody responses against blood-stage antigens (

Discussion
This is the first study comparing humoral immunity to malaria over time between postpartum women and control women. Antibody responses were relatively stable in the absence of infection but more variable in those who experienced microscopically detectable infections during follow-up. Postpartum women had reduced antibody levels to P. falciparum and P. vivax antigens after delivery when compared with control women, however, postpartum levels recovered to control levels after 12 weeks. The majority of antibodies were boosted in response to species-specific infection, but there was no evidence of any boosting of antibodies against VAR2CSA-expressing parasites in the postpartum period. There were no detectable differences in the magnitude of boosting in response to infection in postpartum and control women.
The observation that antibody levels to Plasmodium spp. antigens increased in postpartum women over 12 weeks of follow-up suggests that humoral immunity is transitioning back to normal levels after pregnancy. This concurs with two studies in Africa that observed increased antibody levels at a single time-point (1 month and 6 weeks) postpartum compared to pregnancy 21,32 . Levels of circulating antibodies against merozoites and infected erythrocytes have declined during pregnancy in other studies in Africa and the Pacific 25,33,34 ; the increase observed postpartum is likely reflective of a return to pre-pregnancy levels. Antibodies are preferentially transported to the fetus during pregnancy 35 , so post-pregnancy antibody levels may then increase to pre-pregnancy homeostatic levels. The haemodilution that occurs during pregnancy is followed by a period of haemoconcentration after delivery 36 ; this change would be expected to result in increased antibody concentrations. Haematocrit levels in this cohort of postpartum women increased over 12 weeks postpartum 9 , however postpartum antibody levels increased even when adjusting for increasing haemconcentration suggesting that other factors play a role in changing antibody levels, such as the cessation of maternofetal antibody transfer.
Antibody responses specific to P. falciparum and P. vivax blood-stage antigens increased with homologous species-specific infections in concordance with numerous studies including a study in pregnant women from the same population 20 . We found no strong evidence of species-transcending boosting of immunity against blood-stage antigens. Antibodies against PfCSP and PvCSP appeared to boost in response to both species of infection, suggesting that some of the humoral response directed against CSP demonstrates species-transcending recognition, a finding that concurs with evidence from animal models 37 . We found no evidence of differences in the boosting of blood-stage or sporozoite antibodies in response to infection in postpartum women compared with control women indicating that the postpartum humoral immune system responds to Plasmodium spp. infection similarly to control women.
Less than 10% of women were seropositive at baseline for antibodies against VAR2CSA expressing parasites, reflecting the low endemicity of P. falciparum in this population; and the frequent monitoring and early treatment   19 , antibodies against VAR2CSA were higher in women who had experienced more pregnancies. Whether VAR2CSA-specific antibody responses are maintained at adequate levels outside of pregnancy in the relative absence of VAR2CSA expressing parasites has been the matter of some uncertainty. It has also been proposed that antibodies to VAR2CSA may rapidly decay postpartum 38 . However, we found that antibodies to VAR2CSA only decayed rapidly amongst those recently exposed in pregnancy, while in other individuals VAR2CSA antibodies were relatively stable. This agrees with observed biphasic decay in antibody responses post antigen exposure 39 and recent research investigating VAR2CSA-specific B cell memory which found that memory can be maintained for many years in the absence of antigen exposure 40 . Importantly, our study found no evidence of any boosting of antibodies against VAR2CSA-expressing parasites with infections in postpartum or control women, which would seem to support the paradigm that exposure to VAR2CSA is minimal outside of pregnancy. Our data supports the notion that antibody responses against VAR2CSA acquired in earlier pregnancies can be maintained through to future pregnancies. Strengths of this study included the recruitment of a control group, which enabled direct comparisons between postpartum antibody responses with other women; and the weekly sampling for Plasmodium infection by light microscopy. However, it is probable that some women had submicroscopic infections 41 ; which may have accounted for the few fluctuations observed in uninfected women. Submicroscopic carriage of parasites may also assist in the maintenance of antibody levels and differences in prevalence of submicroscopic infections between postpartum and control women may exist, though this could not be assessed in the present study. The frequent monitoring and prompt treatment of women in this study is a higher standard of treatment than many women would receive elsewhere; it is possible that untreated infections would result in a greater magnitude of antibody boosting. Additionally, this study took place in an area of low P. vivax and P. falciparum transmission, so findings may not be generalisable to areas of higher endemicity. Antibodies in this study were measured via ELISA and flow cytometry, and as yet there are no agreed upon standard or defined threshold for protection, so the clinical relevance of the reduced levels of antibodies postpartum is yet to be determined. A transient reduction in malaria-specific antibody levels would be expected to increase susceptibility to clinical malaria in some women, but as the correlates of antibody-mediated protection remain poorly defined 42 , further research is needed to assess the clinical implications of the reduction observed.
In conclusion, we observed that postpartum women had lower levels of antibodies to a variety of malaria antigens after delivery but were in the process of recovering to control levels. Despite slightly lower antibody levels, postpartum women showed no signs of impaired boosting in response to infection. Evidence from population