P. falciparum infection and maternofetal antibody transfer in malaria-endemic settings of varying transmission

Introduction During pregnancy, immunoglobulin G (IgG) is transferred from the mother to the fetus, providing protection from disease in early infancy. Plasmodium falciparum infections may reduce maternofetal antibody transfer efficiency, but mechanisms remain unclear. Methods Mother-cord paired serum samples collected at delivery from Papua New Guinea (PNG) and the Thailand-Myanmar Border Area (TMBA) were tested for IgG1 and IgG3 to four P. falciparum antigens and measles antigen, as well as total serum IgG. Multivariable linear regression was conducted to assess the association of peripheral P. falciparum infection during pregnancy or placental P. falciparum infection assessed at delivery with maternofetal antibody transfer efficiency. Path analysis assessed the extent to which associations between P. falciparum infection and antibody transfer were mediated by gestational age at delivery or levels of maternal total serum IgG. Results Maternofetal antibody transfer efficiency of IgG1 and IgG3 was lower in PNG compared to TMBA (mean difference in cord antibody levels (controlling for maternal antibody levels) ranged from -0.88 to 0.09, median of -0.20 log2 units). Placental P. falciparum infections were associated with substantially lower maternofetal antibody transfer efficiency in PNG primigravid women (mean difference in cord antibody levels (controlling for maternal antibody levels) ranged from -0.62 to -0.10, median of -0.36 log2 units), but not multigravid women. The lower antibody transfer efficiency amongst primigravid women with placental infection was only partially mediated by gestational age at delivery (proportion indirect effect ranged from 0% to 18%), whereas no mediation effects of maternal total serum IgG were observed. Discussion Primigravid women may be at risk of impaired maternofetal antibody transport with placental P. falciparum infection. Direct effects of P. falciparum on the placenta, rather than earlier gestational age and elevated serum IgG, are likely responsible for the majority of the reduction in maternofetal antibody transfer efficiency with placental infection.


Introduction
During pregnancy, immunoglobulin G (IgG) is transferred from the mother to the fetus, providing protection from disease in early infancy. Plasmodium falciparum infections may reduce maternofetal antibody transfer efficiency, but mechanisms remain unclear.

Methods
Mother-cord paired serum samples collected at delivery from Papua New Guinea (PNG) and the Thailand-Myanmar Border Area (TMBA) were tested for IgG1 and IgG3 to four P. falciparum antigens and measles antigen, as well as total serum IgG. Multivariable linear regression was conducted to assess the association of peripheral P. falciparum infection during pregnancy or placental P. falciparum infection assessed at delivery with maternofetal antibody transfer efficiency. Path analysis assessed the extent to which associations between P. falciparum infection and antibody transfer were mediated by gestational age at delivery or levels of maternal total serum IgG.

Results
Maternofetal antibody transfer efficiency of IgG1 and IgG3 was lower in PNG compared to TMBA (mean difference in cord antibody levels (controlling for maternal antibody levels) PLOS

Introduction
During pregnancy, immunoglobulin G (IgG) antibodies are transferred from mother to fetus. Maternally-derived antibodies provide the neonate with a pre-existing level of passive immunity to help protect from infection and disease. Vaccination of pregnant women can help protect infants against infectious diseases including influenza, tetanus and whooping cough [1]. These strategies rely on adequate maternofetal antibody transfer, but the factors that influence maternofetal antibody transfer are not fully understood. Clinical malaria caused by Plasmodium falciparum infection is uncommon during the first six months of life and, when P. falciparum infections are detected, they tend to be asymptomatic and are of lower parasite density than infections in older infants [2,3,4,5]. Protection against malaria during early infancy has been attributed to many factors [6] including the presence of maternal antibodies in the infant [7], though the relative contribution of maternal antibodies in mediating protection is unclear [8,9]. The humoral immune response is an important component of naturally acquired immunity to malaria and antibodies targeting blood-stage antigens (expressed by merozoites and infected erythrocytes) suppress high parasite densities and progression to symptomatic disease [10]. In malaria endemic areas, individuals develop naturally acquired immunity to P. falciparum infections with age, following repeated infections [10]. Breadth, magnitude and quality of antibody responses are critical, with antibody responses to a broad repertoire of antigenic targets associated with protection against symptomatic malaria in children [11,12,13,14]. Women who would otherwise be relatively immune to clinical P. falciparum malaria are again susceptible to malaria during pregnancy due in part to changes in immune function as well as the appearance of the placenta, a site of sequestration for P. falciparum variants that express PfVAR2CSA on the erythrocyte surface [15]. The predominant antibody isotypes produced in response to malarial antigens are IgG1 and IgG3 [12,16,17,18], which are known to mediate complement deposition and opsonic phagocytosis, mechanisms that have been linked with protective immunity [19,20,21]. Several [32,33,34], four have noted reductions in transfer efficiency for some antibodies but not others [22,23,25,31], while two have reported no association [24,29]. Only one study has investigated the effect of P. falciparum infection on malaria-specific IgG, IgG1 and IgG3, reporting significantly lower maternofetal antibody transfer efficiency in women with any detectable P. falciparum infection at delivery of antibodies specific for some antigens, but not others [31]. Discrepancies in the above findings may be due to differences in antibodies studied, definitions of P. falciparum infection, transmission intensity, antimalarial treatment, location and statistical approaches.
To date, no study has directly reported associations between maternal P. falciparum infections detected in the periphery and antibody transfer efficiency. The relative contribution of gestational age at delivery and maternal sera IgG to associations between P. falciparum infection and reduced maternofetal transfer efficiency are yet to be elucidated. To address these gaps in the literature we sought to assess the effect of maternal P. falciparum infection during pregnancy and placental P. falciparum infection on maternofetal transfer efficiency of IgG1 and IgG3 against several malaria antigens and a non-malaria antigen (measles) in an area of high P. falciparum endemicity in Papua New Guinea (PNG), and an area of low P. falciparum endemicity at the Thailand-Myanmar Border Area (TMBA). We also investigated the extent to which any associations between P. falciparum infection and maternofetal transfer efficiency are explained through associations between P. falciparum infection and early gestational age at delivery or increased maternal sera IgG.

Study population-Alexishafen, Papua New Guinea
Women attending prenatal care at Alexishafen Health Centre, Madang Province, Papua New Guinea (PNG) between August 2005 and September 2007 were recruited into a longitudinal study of malaria and pregnancy following voluntary informed consent [41]. Peripheral blood (5ml) was obtained from women at enrolment, delivery and any additional antenatal clinic visits. At delivery a placental biopsy and 10ml of cord blood were collected. Clinical and demographic data were obtained at enrolment and delivery visits. Presence of microscopic parasitaemia was determined from blood smears by two independent microscopists. A subset of 204 participants with available delivery and cord samples were included in the present study.

Study population-Shoklo Malaria Research Unit, Thailand-Myanmar Border Area
Women attending antenatal clinics of the Shoklo Malaria Research Unit, Thailand-Myanmar Border Area (TMBA), between November 1998 and January 2000 were invited to participate in a chloroquine prophylaxis randomized controlled trial for prevention of P. vivax [42]. P. falciparum episodes were similar in the chloroquine prophylaxis and placebo groups [42]. A total of 118 peripheral maternal blood and paired cord blood samples were utilised from a subset of pregnant women at delivery included in a previously reported study of malaria immunity [43]. At the Shoklo Malaria Research Unit, women are invited to attend an ANC as soon as they become aware of their pregnancy and were then encouraged to attend weekly thereafter.

Method of gestational age assessment
Gestational age was estimated from Ballard scores [44] in PNG, and in TMBA either the Dubowitz method [45] or a calculation based on fundal height [46], which performed well compared to ultrasound in term newborns [47]. Studies conducted in resource-limited settings suggest that Ballard and Dubowitz methods give comparable estimates of gestational age [48,49], however there was no attempt to quality control gestational age assessment between sites. Ballard scores and the Dubowitz method both fall short of estimation of gestational age by early ultrasound assessment [50].

P. falciparum exposures
In PNG, maternal P. falciparum peripheral infection during pregnancy was defined as any light microscopy detected infection in peripheral blood at any point during pregnancy. Women with P. falciparum infections detected were treated with curative doses of chloroquine and sulphadoxine pyrimethamine and all women were given unsupervised prophylaxis [41].
In TMBA, maternal P. falciparum peripheral infection during pregnancy was defined as any light microscopy detected infection in peripheral blood at any point during pregnancy. All women who attend TMBA ANCs were screened weekly for the presence of P. falciparum by light microscopy. When infections were detected they were treated immediately. For P. falciparum or mixed infections, a first infection was treated with quinine or artesunate and all women were given weekly chloroquine prophylaxis [42].

Placental histology
In PNG only, placental histology was performed as described by Rogerson et al [37]. We compared the presence of any parasitaemia in the placenta (pathology class 1-3) to no parasites (pathology class 4-5). As a secondary analysis we compared the presence of parasites and monocytes with malaria pigment in the placenta (pathology class 2) to parasites without malaria pigment-containing monocytes (pathology class 1 and 3) and to no parasites (pathology class [4][5].

Antibody determination at delivery
We measured total serum IgG; IgG1 and IgG3 antibodies to measles antigen (MEV-007-PROSPECbio, Rehovot, Israel) and four P. falciparum antigens (PfEBA175 RII (3D7, amino acid position 146-713), PfMSP2 (3D7, amino acid position 19-249), PfAMA-1 (3D7, amino acid position 25-545) and PfDBL5 (7G8, amino acid position 2003-2270)). We did not assess IgG2 or IgG4 levels, as these isotypes have previously been found to be present at very low levels against the P. falciparum antigens assessed [12,17,18], precluding a reliable assessment of maternofetal antibody transfer. Antibodies to these P. falciparum merozoite antigens are thought to be protective [51,52]. One of these P. falciparum antigens is known to induce IgG1 dominant responses (PfAMA-1) and another is known to induce IgG3 dominant responses (PfMSP2) [18,53]. Antibodies were measured using an enzyme-linked immunosorbent assay (ELISA) as described previously [43]. P. falciparum antigens were coated at 0.5 μg/ml, measles antigen at 1 μg/ml and goat anti-human-kappa at 1/1000. Horeseradish peroxidase (HRP)conjugated goat anti-human IgG (Sigma), mouse anti-human IgG1 (Invitrogen) and mouse anti-human IgG3 (Invitrogen) were used as the secondary antibody (at 1/1000) in the total serum IgG, specific IgG1 and specific IgG3 assays respectively. The tertiary antibody (at 1/ 1000) for specific IgG1 and IgG3 assays was HRP-conjugated goat anti-mouse IgG (Merck Millipore). Assay output was related to a standard curve of pooled sera using a four-parameter logistic nonlinear regression model. TMBA samples had substantially lower levels of malariaspecific antibodies than PNG samples, so immunoassays against P. falciparum antigens were run at multiple concentrations of sera (sera dilutions are provided in S1 Table, range 1/100-1/ 20,000 for P. falciparum antigens). When assay reactivity from a sample was too high to reliably interpolate from the standard curve, then it was rerun at a lower sera concentration. The highest reactivity in an assay was set to 100 arbitrary units (AU). All wash steps were performed using PBS with 0.05% Tween and completed using an automated plate washer within a robotic platform (Perkin Elmer, Waltham, USA). Serum addition was performed using a robotic platform (Perkin Elmer, Waltham, USA). A sample was considered to be seropositive (or in the case of total IgG: elevated) if the AU for that sample was above the mean + 3 standard deviations of those for 15 non-exposed Melbourne blood donors.

Statistical analysis
All statistical analyses were performed using Stata Version 13.1 (StataCorp, College Station, TX, USA). Correlations between cord and maternal antibodies were assessed using Spearman's rank correlation coefficients. Wilcoxon rank-sum tests were used to assess differences in antibody levels between populations. Women who were seronegative for an antibody were not assessed for their ability to transfer that antibody. To assess the impact of study site on antibody transfer, multivariable linear regression of log 2 cord antibody levels was performed with log 2 maternal levels at delivery and study site as covariates, and further adjustment for the potential confounders, gravidity (primigravid/multigravid) and estimated gestational age at delivery (weeks). P. vivax infection was not included in the model.
To assess the impact of P. falciparum infection during pregnancy on cord antibody levels, all analyses were performed separately for each study site, as inherent differences in study design meant that the screening, treatment and recording of infections could not be considered equivalent for the two study sites. Multivariable linear regression was performed assessing the association of infection with log 2 cord antibody levels after adjustment for log 2 maternal antibody levels and gravidity (primigravid/multigravid). Effect modification by gravidity of the association between infection and cord antibody levels was assessed though the fitting of interaction terms between gravidity (primigravid/multigravid) and infection. Three different infection variables were used in the models: peripheral P. falciparum infection at any time during pregnancy (yes/no); placental histology (parasites/no parasites); placental histology (parasites and monocytes containing malaria pigment/parasites without monocytes containing malaria pigment/no parasites). Placental histology data were only recorded in the PNG study.
Where associations were observed between infection and maternofetal antibody transfer efficiency, path analysis was then conducted to investigate potential mediation of the relationship between infection and cord antibody levels through gestational age at delivery and maternal serum IgG. The proportion of the total effect due to an indirect association via gestation age at delivery or maternal serum IgG was then estimated.

Study populations
IgG1 and IgG3 levels were determined for 204 maternal/cord pairs from Alexishafen, Papua New Guinea (PNG) and 118 pairs from the Thailand-Myanmar Border Area (TMBA). Gravidity and gestational age at delivery were higher in TMBA women relative to PNG women (Table 1). P. falciparum antibody seroprevalence, the proportion with detectable P. falciparum infections and the proportion with elevated serum IgG were lower in TMBA women relative to PNG women (Table 1).
Higher levels of malaria-specific IgG1 and IgG3 in maternal and cord sera from PNG than from TMBA Maternal IgG1 and IgG3 levels were strongly correlated with cord IgG1 and IgG3 levels (Spearman's ρ ranges from 0.78-0.96, S1 Fig). Maternal and cord IgG1 and IgG3 levels against P. falciparum antigens were higher in PNG samples, where malaria transmission is higher, than in TMBA samples (Figs 1 and 2, p<0.0001 for all comparisons except PfDBL5 IgG3 (p = 0.16 and p = 0.58 for maternal and cord levels respectively)). IgG1 and IgG3 levels were lower in the cord blood than in the maternal blood for the majority of antigens (Figs 1C and 2C). The cord:maternal ratio of antibodies against all antigens (P. falciparum and measles) was lower in PNG samples than TMBA samples (Figs 1C and 2C). The cord:maternal IgG1 ratio was higher than the IgG3 ratio for each antigen investigated within each study site (Figs 1C and 2C).
Maternofetal transfer efficiency of IgG1 and IgG3 antibodies specific for malaria and measles was lower in PNG women relative to TMBA women To investigate maternofetal antibody transfer efficiency, linear regression was performed, modelling cord antibody levels (log 2 (AU)) with maternal levels (log 2 (AU)) included as a covariate. Maternofetal transfer efficiency of IgG1 and IgG3 antibodies specific for P. falciparum antigens was reduced in PNG compared to TMBA ( age at delivery and gravidity did not substantially alter the magnitude of associations observed ( Table 2).
Maternal P. falciparum infection, detected peripherally, was not associated with antibody transfer efficiency in either study site At PNG and TMBA there was no consistent association between peripheral P. falciparum infection and maternofetal IgG1 or IgG3 transfer efficiency (Fig 3, estimate of adjusted mean difference of antibody levels in cord blood (log 2 AU) for mothers with a peripheral P. falciparum infection versus those without a P. falciparum infection ranged from -0.15 to 0.21 with a median value of -0.05).
Placental P. falciparum infection was associated with lower antibody transfer efficiency in primigravid women, but not multigravid women No data were available on the placental histology of TMBA women, but a study conducted in the Shoklo and Maela camps during the time of this study indicated a low estimated prevalence of P. falciparum placental histopathological changes when early detection and treatment of infection, routine in this population, was performed [40]. Placental histology data were available from the PNG study allowing us to investigate the impact of placental infection on maternofetal antibody transfer efficiency. The presence of placental P. falciparum was associated with reduced maternofetal IgG1 and IgG3 transfer efficiency in primigravid women (Fig 4A, S3 Table) (estimate of mean difference of antibody levels in cord blood (log 2 units) for mother-cord pairs with P. falciparum present in the placenta versus those without placental P. falciparum parasites present ranged from -0.62 to -0.10 with a median value of -0.36). There were no consistent associations between placental infection and maternofetal transfer efficiency in multigravid women (Fig 4B, S3 Table).
In primigravid PNG women associations of placental infections with monocyte infiltrate were greater in magnitude than placental infections without monocyte infiltrate (S2 Fig). Placental P. falciparum infection was associated with lower antibody transfer efficiency in primigravid mothers primarily through mechanisms other than earlier gestational age at delivery or increased maternal IgG Path analysis was performed to assess the proportion of the total effect of placental infection on maternofetal IgG1 and IgG3 transfer efficiency in PNG primigravid women mediated by gestational age and changes in total levels of maternal IgG. The negative effect of placental Table 2

Measles-IgG1
Measles-IgG3 infection in primigravid women was not substantially mediated through induction of shorter gestational age at delivery (proportion of total effect attributable to gestational age mediation ranged from 0% to 18% with a median value of 5%, p ranged from 0.21 to 0.93 for indirect effects) (Fig 5, S4 Table). None of the effect appeared to be mediated through changes in total maternal serum IgG (p ranged from 0.63 to 0.87 for indirect effects).

Discussion
To date, there has been considerable uncertainty regarding the effects of P. falciparum infection during pregnancy on maternofetal antibody transfer efficiency, as well as the mechanisms by which infection may mediate impaired maternofetal antibody transfer. A great strength of this study was the inclusion of women from two separate study sites with differing P. falciparum endemicity and access to antenatal clinics. Importantly, antibody data were standardised to a reference sera pool curve, which enabled direct comparisons of antibody levels across the study populations. The assessment of measles antibodies, which were present at similar levels in both study sites and should not be boosted during a placental P. falciparum infection, strengthened the generalisibility of the findings. Although some studies have included the potential mediating effects of elevated serum IgG and shorter gestation as confounders [22,23,31,33], this is the first study to investigate whether placental infection is mediated via these effects with path analysis. Malaria exposure during pregnancy was captured differently at each study site and therefore separate analyses had to be performed. The prevalence of HIV is very low (<1.5%) in these populations [54,55]) so HIV is unlikely to have a substantial effect on maternofetal transfer at the population level. The lack of placental histology data from the TMBA prevented an assessment of placental infection in that study site. Maternofetal transfer efficiency of some antibodies was reduced in PNG compared to TMBA. Besides the direct effects of malaria, there are several factors that could account for this difference, including maternal genetic differences, gestational age at delivery and altered levels of total maternal IgG [22,23,32]. The median gestational age at delivery of PNG women was 2 weeks shorter than that observed at TMBA. However, differences in gestation cannot fully explain the difference we observed, given the persistence of an association between study site and maternofetal transfer efficiency of some antibodies even after controlling for gestational age at delivery. Very high levels of total maternal IgG are thought to saturate available receptors at the placenta, thereby reducing efficiency of transfer [56], however we did not observe an association between increased maternal IgG and decreased maternofetal transfer efficiency in either population in this study.
Greater maternofetal antibody transfer efficiency at TMBA relative to PNG may be due in part, to the prompt detection and treatment of P. falciparum infections at the TMBA site, preventing the persistence of long-term infections and thereby reducing the risk and consequences of placental infection. Women at TMBA were encouraged to attend antenatal clinics weekly, and were rapidly treated upon diagnosis, so the duration of any Plasmodium spp. infections during gestation is minimal and placental infections are rare [40]. In PNG, policy dictated weekly chloroquine prophylaxis, this was not monitored and so infections with P.  Table. https://doi.org/10.1371/journal.pone.0186577.g005 falciparum may have persisted undetected. Malaria transmission was also substantially higher at the PNG site. At delivery, the majority of women presented with P. falciparum infected placentae in PNG, whereas at TMBA, placental infections are a rare outcome [40].
In primigravid PNG women we observed an association between the presence of P. falciparum parasites in the placenta and reduced maternofetal IgG1 and IgG3 transfer efficiency relative to uninfected women; no association was observed in multigravid PNG women. Primigravid women tend to have placental infections of greater parasite density and more commonly have associated placental inflammation than do multigravid women [57]. Multigravid women are not only to be less likely to present with placental infection, but also less likely to have impaired maternofetal antibody transfer when infected.
There are numerous mechanisms by which placental P. falciparum infection may reduce maternofetal antibody transfer efficiency. Placental P. falciparum infection is associated with a shorter gestational age [35] and shorter gestational age is associated with reduced maternofetal antibody transfer [26,27,28]. In our model, we found that gestational age explained a small proportion of the reduced maternofetal antibody transfer efficiency associated with placental infection. Placental infection has been associated with increased total maternal IgG in some other populations [23,34]; transfer efficiency can be reduced when maternal IgG levels are elevated [56]. However, amongst PNG women we did not observe a reduction in antibody transfer efficiency among those women with increased total serum IgG, as defined by levels greater than the mean plus three standard deviations of Melbourne control levels. The majority of the association between placental infection and reduced maternal transfer efficiency in primigravid PNG women was explained through mechanisms other than gestational age and maternal IgG in the present study. Our results indicate that a large proportion of the observed effect may be due to direct effects of placental infection. Given that placental infections with malariapigment containing monocytes tended to have even lower transfer efficiency than placental infections without monocytes, inflammation at the placenta is likely to be involved. Disruptions to the placental architecture that occur during placental infection [36,58,59,60] may have a negative impact on maternofetal transfer and receptor expression. Placental infection with inflammation seems to impair transplacental transport of glucose and amino acids via a reduction in expression of transport receptors [61,62]; if expression of placental immunoglobulin transport receptors were also reduced then this would explain some of the reduction in antibody transfer efficiency. As IgG1 outcompetes IgG3 for placental receptor binding [63], a reduction in transport receptors would likely have a greater impact on IgG1 transfer relative to IgG3 transfer. Further research is needed to elucidate the precise mechanisms by which placental infection with P. falciparum mediates reduced efficiency of maternofetal transfer, but our findings indicate direct effects of P. falciparum placental infection play a substantial role, emphasising the need for highly efficacious prevention in pregnancy.
Our study was subject to limitations. Our knowledge of infection status over the entire course of pregnancy was incomplete as the infection status of a woman was necessarily restricted to instances when they presented to antenatal care. Infection was detected via microscopy, so sub-microscopic infections were not captured. The infection status of cord blood was not determined in both study settings. Notably in the TMBA study we lacked placental histology data, limiting our analysis of placental histology exposures to pregnant women from PNG. We did not have data available on the outcomes of the children in their first year of life, precluding an assessment of the relationship between maternofetal antibody transfer and infant outcomes.
We have observed that placental P. falciparum infection was associated with reduced maternofetal antibody transfer efficiency in primigravid women and that only a small proportion of this association is mediated by gestational age. Adequate maternofetal transfer of antibodies is essential for maternal vaccination strategies to effectively protect infants; efforts to prevent and treat malaria in pregnancy should continue to be encouraged.