Background
The density of malaria parasites in an individual has implications for the ability to detect the infection, since standard diagnostic test such as microscopy and rapid diagnostic test (RDT) can miss low density infections [
1]. Parasite density is considered to be mainly shaped by host factors, in particular immunity [
2]. Naturally acquired immunity against
P. falciparum develops after repeated exposure, with a faster acquisition in high transmission areas [
2]. However, most infections in low transmission settings, where immunity is expected to be minimal, are characterized by their low parasite densities [
1]. This observation challenges the idea that host immunity is the only determinant of epidemiological malaria features.
Parasite factors such as the number of co-infecting parasite variants that compete within the human host, intrinsic growth rate, virulence, sexual commitment, and fitness may also play a role in shaping parasite densities under different intensities of transmission [
3‐
5].
Plasmodium parasites have developed sophisticated mechanisms to adjust to host conditions [
6] and seasonal fluctuations [
5]. However, most of the evidence supporting this adaptability has been obtained from in vitro and animal models, leaving the role of parasite factors in shaping human infections and their potential to undermine the efficacy of malaria control programs to be determined [
4]. In addition, most of longitudinal data on malaria in Africa is based on symptomatic cases [
7] and lacks precise estimations of parasite densities, which makes it difficult to study the epidemiological signatures of these parasite adaptations.
Pregnant women attending antenatal care (ANC) clinics constitute an attractive group for monitoring changes in malaria prevalence in the population [
8]. A few studies have explored parasite densities in this group, in particular in low transmission settings or during marked declines in transmission, but none have included direct measures of immunity. Infections acquired before pregnancy are maintained at low density by antimalarial immunity developed after life-long encounters with
P. falciparum [
9,
10]. These infections, together with those acquired in pregnancy, multiply to high densities in primigravidae from the end of the 12th week of gestation [
9,
11] when the placenta selects
P. falciparum parasites expressing VAR2CSA that binds to chondroitin sulfate A [
12]. Antibodies against VAR2CSA are acquired in a gravidity-dependent manner, and appear to protect multigravid women against severe malaria [
9‐
11]. This pregnancy specific immune response reflects exposure in very specific time windows (i.e., the pregnancy period) and might be utilized to gain a better understanding of the role of immunity in controlling parasite densities.
There is little knowledge on the relationship between intensity of malaria transmission, parasite densities, and immunity, especially among pregnant women and in contexts of declining transmission. Here, we aimed to describe parasite densities in pregnancy, and how gravidity and humoral immunity affect these, during a period of declining malaria in southern Mozambique. To address this, we undertook a 3-year prospective observational study among pregnant Mozambican women at first ANC visits in southern Mozambique. We applied molecular methods to quantify
P. falciparum infections and antibody levels in women from a historically moderate transmission setting (Ilha Josina) [
13], a setting that rapidly transitioned from moderate to low malaria transmission after an elimination initiative (Magude) [
14], and a historically low malaria transmission setting (Manhiça). Using an interaction analysis, we tested the hypothesis that pregnancy-specific immunity is the major factor controlling parasite densities in malaria-exposed multigravidae, whereas immune-independent parasitological factors dominate parasite densities in anti-VAR2CSA-naïve primigravidae.
Methods
Study area and population
This 3-year prospective observational study was conducted between November 2016 and November 2019 at ANC clinics in Manhiça Sede District Hospital, Ilha Josina Health Center (Manhiça District), and Magude Sede Health Center (Magude District), in Maputo Province, southern Mozambique (Additional File
1: Fig. S1). The region is characterized by its subtropical climate with a hot and rainy season from November to April and a cool and dry season during the rest of the year. Overall, malaria transmission is low in Manhiça District, with rapid diagnostic test (RDT)-based parasite rates (PR
RDT) in the community of 6.1% and 1.4% in 2017 [
15] and 2018 (B. Galatas, personal communication), respectively. PR
RDT in Ilha Josina, a river island in the confluences of the Incomati River with historically high-to-moderate transmission levels [
13], was 18.1% in 2017 and 3.7% in 2018 (B. Galatas, personal communication). Magude District, located 102 km from Manhiça, is a low transmission area (PR
RDT: 2.6% in 2017 [
15] and 1.4% in 2018 [
14]) resulted from the deployment of an intensive package of malaria elimination interventions since 2015 [
14]. Indoor residual spraying was deployed annually in the study areas.
Recruitment and data collection
Pregnant women at their first routine ANC visit were invited to participate in the study if resident in the study area. A fingerprick blood drop was collected onto filter papers (dried blood spots [DBS]) with the participant’s consent. A brief form, including visit date, age, gravidity, gestational age based on the fundal height measurement, area of residence, and recent movements was completed. HIV status was recorded from the maternal health card. In case of an unavailable record, an HIV serological rapid test was done according to standard procedures for voluntary counseling and testing. Maternal hemoglobin levels were determined by HemoCue® Hb 201 + (HemoCue AB, Angelholm, Sweden).
Parasitological determinations
DNA extracted from DBS was used for detection and quantification of
P. falciparum in duplicate using real-time quantitative PCR (qPCR) assay targeting the 18S ribosomal RNA gene [
16]. Experiments were repeated if the qPCR efficiency or the positive control were out of the range obtained by the average ± 3 standard deviations (SD; Additional File
1: Fig. S2). Positive DBSs were used for Illumina next generation sequencing of PCR amplicons targeting 101 single nucleotide polymorphisms (SNPs) in the parasite genome [
17]. The 101-SNP barcode was used to estimate the number of distinct genotypes within the infection (complexity of infection [COI]) using the program COIL [
18].
Immunoglobulins G determinations
Immunoglobulin G (IgG) eluted from 5760 available DBS were quantified using the xMAP technology and the Luminex 100/200 System (
https://www.luminexcorp.com) [
19]. The multiplex suspension array panel included VAR2CSA antigens (Duffy binding-like recombinant domains DBL3-4 [
20] as well as peptides targeting the NTS region [P1] and ID1 [P8 and PD] [
19]), non-pregnancy specific antigens (19-kDa fragment of the merozoite surface protein-1 [MSP1
19] [
21], region II/F2 of erythrocyte-binding antigen-175 [EBA175] [
22], full-length
P. falciparum reticulocyte-binding homologue protein 2 and 5 (PfRH2 [
23] and PfRH5 [
24])), and biomarkers of recent
P. falciparum exposure [
25] (gametocyte exported protein 18 [GEXP18], acyl CoA synthetase 5 [ACS5] ag3, early transcribed membrane protein 5 [ETRAMP5] ag1 and heat shock protein 40 [HSP40] ag1). Information about antigens, procedures for reconstitution of DBS and quality control, bead-based immunoassay, and data normalization are described in Additional File
1: Fig. S2 and Additional File
2: Supplementary Methods [
26,
27].
Sample size calculation and data analysis
We estimated that the molecular detection of
P. falciparum in all DBS collected from Ilha Josina (
n = 250/year) and a random selection of approximately 2700 samples per year in both Magude and Manhiça would allow for the estimation of the 95% confidence interval (95%CI) of annual
P. falciparum positivity rates between 20 and 5% in each of the 3 sites, with a margin of error lower than or equal to the expected positivity rate. Age was categorized as either younger than 18 years or 18 years and older, season as rainy (November to April) or dry (May to October) and year of recruitment from November to October in 2016–2017 as year 1, 2017–2018 as year 2,or 2018–2019 as year 3. Women reporting to have changed their residence area during their pregnancy, as well as women with missing information on gravidity, residence area or HIV, were dropped from the analysis. Detectable infections (Di) were defined as those with parasite densities equal or above 100 parasites/μL, which is the detection limit commonly used for standard RDTs [
28], and expressed relative to the total of infections detected by qPCR (proportion of Di [pDi]).
Categorical and continuous data were compared between sites, gravidity groups, and time periods using Fisher’s exact test and Student t-test, respectively. Multilevel mixed-effects regression models with random intercept at site level were estimated to assess associations with qPCR positivity and pDi (logistic) and log-transformed parasite densities and mean fluorescence intensities (MFIs) as well as hemoglobin levels (linear). All models were estimated using transmission level (site) and gravidity (first pregnancy and two or more previous pregnancies) as independent variables and were adjusted for season, HIV status, residence in village or rural area, and place where molecular analysis was conducted (Mozambique or Spain). The hypothesis that the control mechanisms of P. falciparum infection in pregnancy are gravidity-dependent was assessed including an interaction term between gravidity and transmission levels (by site and period) in the models. Secondary analyses, which included genetic complexity of infections and secundigravidae, were assessed using linear and mixed-effects regression models, respectively. The significance level was set at 0.05. All analyses were performed using the Stata statistical program version 16 (Stata Corporation, College Station, TX, USA).
Discussion
This is the first study to provide epidemiological and immunological evidence of non-immune factors that influence parasite densities when malaria burden falls from high-to-moderate to low levels. Between November 2016 and November 2019, P. falciparum qPCR-positivity rates at first ANC visit in southern Mozambique declined from 28% in Ilha Josina, 7% in Manhiça, and 5% in Magude to 13%, 2%, and 2%, respectively. Decreasing transmission in these three different epidemiological settings were followed by changes in parasite densities which are gravidity-dependent. Among secundigravidae in Ilha Josina, parasite densities and the relative abundance of detectable infections increased with declining malaria, as expected from les malaria exposure in the previous pregnancies. However, falling parasite rates were associated with declining parasite densities among anti-VAR2CSA-naïve primigravidae, without major changes in pregnancy-specific and general immunity. Therefore, factors other than acquired immunity produce less detectable, lower-density infections among pregnant women over short but marked declines in malaria burden.
Trends of parasite densities among multigravid women are consistent with the role of pregnancy-specific immunity in controlling parasite densities [
16,
29]. The proportion of detectable infections was lower among multigravid women from areas with recent high-to-moderate transmission (Ilha Josina and Magude), compared to primigravidae in respective regions. As
P. falciparum rates fall in Ilha Josina, parasite densities and detectability in secundigravidae increased to levels observed in primigravidae. These changes in a short period of 3 years are accompanied by reductions in antimalarial antibodies, both pregnancy-specific and generally, in secundigravidae [
16], presumably reflecting a reduction in parasite exposure in the previous first pregnancy and a waning of acquired immunity [
30]. In contrast, no gravidity effect or change during the study period was observed in parasite densities and pregnancy-specific immunity among women from the historically low transmission setting in Manhiça.
Parasitemia in first pregnancy follows the opposite patterns of multigravid women, pointing to differences in the factors that determine parasite densities among women without pregnancy-specific immunity. When
P. falciparum rates were at their highest levels (first year of the study), parasite densities and detectability, as well as the infection-associated reduction in hemoglobin, were higher among primigravidae from Ilha Josina, than those from Magude and Manhiça. Thereafter, parasite densities among primigravidae from Ilha Josina decreased as parasite rates dropped. This shift in parasite densities was not accompanied by changes in VAR2CSA-specific antibodies, which were similarly low among primigravid women in the three settings [
31], as expected from the limited VAR2CSA exposure at the pregnancy onset (i.e., before the placenta develops [
9,
11]). It can also not be explained by general anti-malarial immunity acquired before of pregnancy, as levels of antibodies against non-pregnancy-specific antigens were correlated with parasite rates, which was expected to affect densities reversely. More recent infections in Ilha Josina [
32] are also unlikely to explain the higher observed parasite densities, as this effect is expected to manifest independently of gravidity and in the rainy season, when the force of infection is larger, compared to the dry season. Instead, the data suggests the role of factors other than acquired immunity in regulating parasite densities in primigravid women from Ilha Josina during marked epidemiological transitions. The higher number of genetically different parasites among infected primigravidae from Ilha Josina compared to those from Manhiça is in line with the increase in parasite clones and multiplication rates as transmission increases suggested from previous studies [
3,
33]. The immunity developed during previous pregnancies may mask these non-immune factors in multigravid women. Further studies are needed to identify the molecular mechanisms of the immune-independent control of parasite densities and their potential impact on balancing parasite replication and gametocyte production, as suggested by evolutionary and experimental models [
3,
4].
The results of this study have several public health implications. First, the proportion of
P. falciparum infections that can be detected with currently available diagnostic tools varies substantially depending on the intensity of transmission and gravidity. The proportion of detectable infections in primigravidae increased from 50 to 83% in areas with a PfPRq
PCR of 4% and 28%, respectively. In contrast, detectable infections in multigravidae decreased from 44 to 21% with increasing parasite rates. Therefore, levels and trends of malaria burden, as well as fertility rates, are expected to affect the sensitivity to detect
P. falciparum infections and as a consequence the efficiency of screen-based strategies [
34], as well as the contribution of pregnant women in sustaining transmission [
35]. Second, the adverse effects of
P. falciparum infections over periods in which malaria prevalence declines may increase among multigravid women as immunity wanes and parasite densities increase, but decrease among primigravidae as non-immune factors reduce parasite densities. Third, the results of this study highlight the value of monitoring malaria in pregnancy as an independent measure of malaria burden [
8] that can capture geographical and temporal malaria trends. Fourth, the identification of factors other than acquired immunity which determine densities as transmission changes [
5] may allow the improvement of surveillance approaches and the development of new antimalarial interventions. Finally, incorporating non-immune factors in mathematical models of malaria in pregnancy [
9,
34,
35] may improve their efficacy in describing the host-parasite biology and malaria transmission dynamics.
This analysis has several limitations. First, inaccessibility of the placenta at booking visits prevents the direct study of parasite dynamics in that organ; however, parasite densities in peripheral and placental blood have been shown to follow similar trends [
16,
29]. Second, the low number of
P. falciparum infections in the low transmission settings of Magude and Manhiça limited the ability to stratify women with two or more pregnancies in the analysis. Third, genotyping of parasite infections was not successful for low-density infections, which represents a large proportion of available samples. Fourth, age-dependent effects might have been attributed to pregnancy-specific immunity, given the correlation between age and gravidity. However, previous modeling data [
34] suggests that pregnancy, rather than age, is the major determinant of the observed patterns. Fifth, it is not possible to generalize the observations of this study to non-pregnant individuals, although similar reductions in parasite densities with declining transmission have been observed in the general population [
36,
37]. Finally, the study did not assess genetic, behavioral, socio-economic, and environmental factors which may contribute to the different malaria burden [
32]. However, these are not expected to vary substantially in the study populations, since they were relatively closely located in the south of Mozambique.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.