According to the World Health Organization, an estimated 228 million cases of malaria occurred worldwide, with 405,000 deaths, in 2018 (1
). Plasmodium falciparum
accounted for the majority of cases. In areas with high transmission of malaria, children under 5 years and pregnant women are particularly susceptible to infection, illness, and death (1
). In sub-Saharan Africa, 11 million pregnant women exposed to malaria infections in 2018 would have delivered about 872,000 children with low birth weight (1
). The increased susceptibility of pregnant women to malaria compared to nonpregnant adults is due, in part, to the physiological adaptation of the maternal immune system to prevent rejection of the fetus as a semiallogeneic transplant (2
). In this context, an important role is played by cytokines, the levels of which change during the different trimesters of pregnancy.
During the first trimester, proinflammatory cytokines act as mediators of the embryo-maternal paracrine dialogue associated with apposition, attachment, and invasion (3
). In this process, members of the interleukin-6 (IL-6) cytokine superfamily are required at the early implantation stage. While leukemia inhibitory factor (LIF) is expressed at the highest concentrations in the endometrial glands and is strongly associated with normal implantation, IL-11 acts on endometrium luminal epithelium to facilitate blastocyst attachment and implantation (5
). The IL-12/IL-15/IL-18 system cooperates with the IL-1 system and other cytokines, such as vascular endothelial growth factor (VEGF) and epidermal growth factor (EGF), via interactions with endometrial leukocytes and natural killer cells in local angiogenesis and tissue remodeling (7
). Cellular immune responses also play an important role with human decidual natural killer cells that promote trophoblast invasion through the production of IL-8, interferon (IFN)-inducible protein 10 (IP-10), VEGF, and placental growth factor (8
Once implantation has taken place, successful pregnancy is characterized by a cytokine balance that is tipped toward a predominantly T helper 2 (Th2)-type response. Progesterone contributes to support IL-3, IL-4, IL-5, and IL-10 production, which inhibits Th1 responses and favors allograft tolerance in women (9
). In addition, in the systemic circulation of healthy pregnant women, IL-4, IL-6, IL-10, and IL-13 production progressively increases, while the serum levels of most Th1-type cytokines decrease significantly after the second trimester compared with the levels observed during the first trimester (10
Given the importance of the timely regulation of cytokine networks during pregnancy, the dysregulation of this system at both local and systemic levels is invariably characteristic of adverse pregnancy outcomes, including spontaneous abortion, preterm labor, pre-eclampsia, and intrauterine growth restriction (IUGR) (11
). Thus, Th1-type response alterations at implantation sites have been associated with placental tissue damage in some inflammatory diseases. Such is the case when P. falciparum
infection occurs during pregnancy, leading to a pronounced proinflammatory response in the placenta with deleterious effects on both maternal and fetal health (13
So-called placental malaria (PM) is characterized by the sequestration of parasite-infected erythrocytes in the maternal intervillous blood spaces of the placenta. The parasite-derived protein VAR2CSA, which is transported to and expressed on the surface of infected erythrocytes, mediates this sequestration (15
). VAR2CSA binds to chondroitin sulfate A (CSA) on the proteoglycan syndecan-1, which is expressed by syncytiotrophoblasts (16
). The infected erythrocytes that cause PM elicit Th1-type cytokine production in the placenta, where elevated levels of proinflammatory cytokines, such as tumor necrosis factor alpha (TNF-α), IL-2, IFN-γ, and IL-1β, are observed, especially in primigravidae (17
). Increased placental levels of chemokines have also been described, leading to monocyte infiltration along with the recruitment of macrophages, cytotoxic T cells, B cells, and granulocytes into the placenta (18
). T cell proliferation and enhanced macrophage phagocytic activities, aimed at controlling parasite proliferation, are a feature of the inflammatory response during PM (17
The persistence of infected erythrocytes, accompanied by the excessive accumulation of leukocytes in the intervillous spaces of the placenta, are ultimately responsible for the pathological outcomes of PM. In this context, numerous studies have investigated the relationships between cytokine production during PM and pregnancy outcomes (21–25
). High levels of placental TNF-α were associated with several pregnancy outcomes, including severe anemia, low birth weight (LBW), and IUGR (21
). Likewise, high expression of IL-8 associated with IUGR was found in placentas of multigravidae (21
). Furthermore, in a previous study, our findings revealed an association between babies with low birth weight for their gestational age (GA) and low placental plasma levels of IFN-γ and IL-5, independent of gravidity (25
However, in peripheral blood, parasite-specific immune responses may be decreased due to the trafficking of cells out of the circulation and the reduction of peripheral parasitemia as the development of the placenta is completed (26–28
). Low levels of cytokines were detected in systemic plasma of pregnant women at delivery, and these were not significantly related to pregnancy outcomes (23
). Immune responses in the context of malaria during pregnancy are generally underexplored, and longitudinal studies could lead to a better understanding of the relationship between peripheral immune markers and poor pregnancy outcomes.
Here, the relationships between P. falciparum infections and peripheral plasma cytokine activity during pregnancy, on one hand, and the consequences of different immune equilibria on pregnancy outcomes, on the other hand, were investigated. Therefore, circulating plasma levels of IL-4, IL-5, IL-10, IL-12p70, and IFN-γ were measured at different gestational ages in a cohort of Beninese women to identify biomarkers potentially of prognostic and/or diagnostic value for pregnancy-associated malaria (PAM) and/or for poor pregnancy outcomes.
The primary aim of the study described here was to determine whether the levels of circulating cytokines, measured at different moments during a longitudinal study of pregnancy-associated malaria in a cohort of Beninese, reliably reflected the different pregnancy outcomes recorded. Our basis for the study was an observation made in our own earlier study of cytokines in placental plasma at delivery. In that study, we found that low levels of both IFN-γ and IL-5 were associated with the delivery of SGA babies and, thus, could represent markers of such a poor pregnancy outcome (25
). Other studies, including our own, have also consistently shown that high levels of IL-10 present in placental plasma are closely associated with the occurrence of infections with P. falciparum
Here, multivariate analyses were performed to determine (i) whether circulating cytokine concentrations at different time points of pregnancy were associated with either P. falciparum infections or poor pregnancy outcomes; (ii) the predictive value of circulating cytokine levels at inclusion with respect to P. falciparum infection and poor pregnancy outcomes; and (iii) the associations between episodes of P. falciparum infection and of anemia during pregnancy and the levels of cytokines at delivery.
Considering IL-10, multivariate analyses showed that high concentrations are strongly associated at each time point during pregnancy with P. falciparum
infections. In addition, increased IL-10 levels observed at delivery were associated with P. falciparum
infections that had occurred during pregnancy. Thus, these observations are consistent with the results of our own previous studies as well as those of others (25
). IL-10 is a key immunoregulatory molecule during infection. That IL-10 exerts direct effects on CD4+
T cells, inhibiting proliferation and the production of IL-2, IFN-γ, IL-4, IL-5, and TNF-α, is well documented (34–36
). Thus, IL-10 can suppress both Th1- and Th2-type responses. That capacity is consistent with our observations here of declining concentrations of both Th1- and Th2-type cytokines, significant in the case of both IL-5 and IL-12p70, in the face of sustained infection-induced IL-10 levels throughout pregnancy. Placental infection with P. falciparum
is characterized by parasite-infected erythrocyte sequestration associated with often pronounced inflammation leading to detectable proinflammatory markers in peripheral blood (37
). Thus, as a direct reflection of infection in the placenta, the inflammatory response and ensuing pathology can lead to poor pregnancy outcomes. We speculate that the strong IL-10 response observed in association with infection is a driving force in suppressing both potentially protective as well as pathological immune responses to maintain a healthy pregnancy. Thus, parasite persistence in the placenta is an unintended consequence of maternal efforts to control inflammatory responses. Separately, we did notice lower levels of IL-10 at ANV4 compared with the levels found either at inclusion or at delivery, an observation that was consistent across all groups, regardless of their infection or other status (Tables 2
). We speculate that this particular finding is related to women’s changing overall immunological status during pregnancy that is thought to progress from an initial predominantly proinflammatory state to a more Th2-dominated noninflammatory environment before reverting to a proinflammatory state late in pregnancy prior to parturition (4
). Thus, the comparatively lower levels of IL-10 seen at ANV4 would correspond to the noninflammatory period during which, logically, the necessity for IL-10 is reduced. Perhaps somewhat paradoxically, the circulating concentration of IL-10 at delivery also was found to be significantly lower in women who experienced multiple infections during pregnancy (Table 5
, lower rows). This finding could reflect the fact that, upon detection, infections were always immediately treated, as stipulated in the study protocol. Therefore, in the case of those with multiple infections, it is plausible that proinflammatory responses, and the associated anti-inflammatory IL-10-led responses, will have had insufficient time to be well-established and, hence, to persist in the case of repeated treatments. This would be entirely consistent with our published findings concerning the association between submicroscopic infections and circulating cytokine levels (33
We also observed increased IL-10 levels associated with P. falciparum
infection in primigravid compared with multigravid women. In high-transmission areas, primigravidae are at greater risk of placental infection, whereas such a gravidity effect is less marked in low-transmission areas and absent from areas with epidemic malaria (38
). Gravidity-specific immunity, naturally acquired through consecutive pregnancies, contributes to resistance to placental infection with P. falciparum
, reduces parasite densities, and prevents disease in areas of high and stable transmission. Thus, among primigravidae who lack specific acquired immunity, severe anemia and the delivery of underweight babies are more frequent. In addition, although the risk of maternal infection is reported to be highest during the second trimester (40
), data from our own earlier studies suggest that P. falciparum
infections can be particularly harmful at the beginning and at the end of pregnancy (30
). This is consistent with our current observation concerning the association of P. falciparum
infection with higher circulating levels of IL-10 in primigravid compared with multigravid women. Multivariate analysis of the predictive value of the cytokines assessed here did not reveal elevated IL-10 at inclusion to be associated with the risk of P. falciparum
infection during pregnancy. In the STOPPAM study conducted in Tanzania, the combination of increased IL-10 and IP10 with decreased RANTES was predictive of infection with P. falciparum
regardless of gestational age (31
). Overall, the data suggest that IL-10 can be useful as a diagnostic rather than a prognostic marker of P. falciparum
infection during pregnancy.
In the context of IL-12p70, increased levels of this proinflammatory cytokine at inclusion were associated with a significantly increased prospective risk both of P. falciparum
infection during pregnancy as well as of preterm birth. On the other hand, increased levels at the inclusion of IFN-γ, a Th1-type proinflammatory cytokine, were associated with a significantly decreased prospective risk of P. falciparum
infection during pregnancy, whereas when infections were present (at inclusion and at delivery), the levels of this cytokine were slightly but significantly elevated. These seemingly contrasting associations for two different proinflammatory cytokines are not easily resolvable. IL-12 is required for optimal IFN-γ production in vivo
during immune responses, particularly during bacterial or parasitic infections (41
). Circulating monocytes are the primary source of bioactive IL-12p70, in response to which NK cells readily produce IFN-γ, activating macrophages and enhancing their bactericidal activity and providing a mechanism of T-cell-independent macrophage activation during the early phases of innate resistance to infections (42
). Separately, the ability of IL-12 to induce and maintain antigen-specific Th1-type responses is essential to control infections with many microbial pathogens (41
). We can only speculate that pregnancy-related and P. falciparum
infection-related modulation of the production of different cytokines, including but not exclusive to Th1- and Th2-type responses, differentially affect their systemic levels depending on gestational age, for example, but also on the degree of chronicity of the infection. Successful pregnancy is characterized by a shift of the Th1/Th2 balance with a decrease of Th1-type cytokines and an increase in Th2-type cytokines. High circulating levels of IL-10 would be expected to inhibit monocyte activation and Th1-type cytokine synthesis, while low systemic levels of IL-12 in the face of elevated IL-10 levels are a feature of severe malaria in children (44
). Furthermore, downregulated IL-12p70, resulting from reduced transcription of its component IL-12p40, has been shown to be correlated with the uptake of P. falciparum
hemozoin pigment and elevated levels of IL-10 in children with severe malarial anemia (46
). Whether such mechanisms influence responses during the predominantly uncomplicated infections associated with pregnancy remains unknown.
We were particularly interested in the circulating cytokine profiles associated with the risk of preterm birth or small-for-gestational-age babies independently of P. falciparum
infection and anemia. As already mentioned above, multivariate analysis revealed a significantly increased risk of preterm birth associated with increased levels of IL-12p70 at inclusion into the study, while the same association was also found for increased IL-5 and IL-10 levels at ANV4. Conversely, increased levels of IL-12p70 and IL-4 at ANV4 were found to be associated with a significantly reduced risk of preterm birth, as were increased levels of IL-5 at delivery. Many studies have reported that proinflammatory cytokines, such as IL-12, are implicated in the pathogenesis of poor pregnancy outcomes (29
). However, most of those reports focused on peripheral or placental levels of cytokines at delivery. In the longitudinal study reported here, we know that between inclusion and ANV4, the women received 2 curative doses, spaced at least 1 month apart, of the antimalarial drug combination sulfadoxine-pyrimethamine. Thus, at least in those harboring P. falciparum
infections when taking the treatment, it is plausible that the clearance of the parasites during that interval will have led to a change in the circulating cytokine profile, possibly explaining the contrasting associations observed at the 2 different time points during pregnancy.
We found that an increased risk of delivering an SGA baby was associated with increased levels of IL-5 at inclusion but that increased levels of the same cytokine at delivery were, conversely, associated with a significantly reduced risk of having a baby who was SGA, with the obverse clearly being that low levels of IL-5 were associated with an increase in such a risk. Thus, the latter finding is entirely consistent with the observation in our earlier study, namely, that low levels of IL-5 in placental plasma equated to an increased risk of having an SGA baby. In that study, we also observed the same association with increased risk of SGA babies for low levels of placental IFN-γ, but this was not the case for the levels of circulating IFN-γ assessed here. Nonetheless, it is striking that the circulating levels of the same cytokine, IL-5, measured at delivery appear here to distinguish two groups with distinctly different poor pregnancy outcomes, i.e., those with low circulating levels of IL-5 at delivery were at greater risk of either PTB or SGA. Whether the immunosuppressive activity of infection-induced IL-10 plays a determining role in this scenario and precisely how the suppression of a Th2-type cytokine might be mechanistically linked to either outcome of pregnancy are questions that remain open to speculation.
There are some limitations to our study that deserve mention. Although data from women found to be seropositive for HIV were excluded, the scope of the STOPPAM study did not allow for the detection of other coinfections, i.e., those of bacterial or helminth origin, for example, which could have influenced the cytokine measurements. In addition, we were unable to include an assessment of the possible influence of submicroscopic infections with P. falciparum, since we did not have relevant data for all women at all time points. It is also plausible that intermittent preventive treatment during pregnancy (IPTp) with sulfadoxine-pyrimethamine could, through the clearance of parasites, have modified cytokine concentrations. In the latter context, it is salient to note that, on average, the STOPPAM participants received their first and second doses of IPTp at 21 and 26 weeks of gestational age, respectively, placing the second dose approximately 8 weeks prior to the ANV4 sampling for the study here. We speculate that such an interval is sufficiently long to preclude any direct influence of the treatment on cytokine concentrations.
MATERIALS AND METHODS
The participants for this study were drawn from the EU-funded STOPPAM (Strategies to Prevent Pregnancy-Associated Malaria) study cohort. The STOPPAM study was conducted in Benin from 2008 to 2012 and involved the recruitment of 1,037 pregnant women who were monitored until delivery. One of the primary stated goals of STOPPAM was to quantify the pathophysiological and immunological effects of PAM. The study was approved by the ethics committees of the Faculty of Health Sciences of the University of Abomey-Calavi in Benin and of the Institute of Research for Development (IRD) in France.
Participants were enrolled in three health centers with maternity clinics (Comé, Akodeha, and Ouedeme Pedah), located in the district of Comé in Mono Province, southwestern Benin. Comé is predominantly urban, while the two other centers are in rural settings and are located 10 km apart. The area is considered moderate for malaria transmission, with two peaks during the rainy seasons (April to July and October to November) (49
Study design and population.
Women with a gestational age of fewer than 24 weeks who were resident for more than 6 months within 15 km of the health center and who were planning to deliver in the hospital were included after having given written informed consent (50
). They were monitored at each scheduled monthly antenatal visit (ANV) for clinical and biological information. From the second trimester onwards, and according to the national guidelines at the time, they received two doses of sulfadoxine-pyrimethamine (SP) as part of intermittent preventive treatment during pregnancy (IPTp) spaced at least 1 month apart.
The current study involved a subgroup of 400 women selected according to the pathological outcomes of pregnancy in order to quantify cytokine levels in peripheral plasma samples. In this context, 4 groups were defined: women who experienced P. falciparum infection during pregnancy, mothers who had anemia during pregnancy, women with preterm births (PTB), and women who delivered babies who were small for their gestational age (SGA).
P. falciparum infection during pregnancy was defined as a positive thick blood smear after microscopic examination of peripheral blood. Examination for P. falciparum infection was routinely conducted at each ANV and during emergency visits. All infected women were treated with quinine or SP per national guidelines.
Maternal anemia was assessed at each ANV and was defined as a hemoglobin level of <10 g/dl. PTB babies were those born at a gestational age of <37 weeks, as defined by ultrasound examination conducted before 24 weeks of GA. SGA babies were those with fetal growth restriction classified according to sex-specific charts for birth weights in less than the 10th percentile of fetal weight for gestational age (51
). Samples from confirmed HIV-positive patients were not used in this study.
Venous blood samples were collected from the women at each ANV in vacutainers with citrate phosphate dextrose adenine anticoagulant. After centrifugation, plasma samples were separated, aliquoted, and stored at −80°C until further use.
Plasma cytokine quantification.
Plasma samples collected at inclusion (INC), ANV4 (4 months after inclusion), and at delivery (DEL) were thawed and were analyzed for cytokine detection. Levels of IL-4, IL-5, IL-10, IL-12p70, and IFN-γ were assessed using Cytometric Bead Array (CBA) enhanced sensitivity technology (BD Biosciences, San Diego, CA, USA) according to the manufacturer’s instructions. Data were acquired by flow cytometry (BD FACSCalibur) and analyzed using FCAP Array software (BD, Soft Flow Hungary). The lower limits of detection were 0.072 pg/ml, 0.034 pg/ml, 0.007 pg/ml, 0.006 pg/ml, and 0.007 pg/ml for IL-4, IL-5, IL-10, IL-12p70, and IFN-γ, respectively.
We considered three periods during the pregnancy follow-up: INC (enrolment in the study), ANV4, and DEL. We chose inclusion for the first period of pregnancy instead of trimester categorization, because few women were included in the study during the first trimester of their pregnancy.
The study population was stratified into subgroups according to clinical and other characteristics that comprised (i) infection with P. falciparum, (ii) maternal anemia, (iii) intrauterine growth restriction (assessed by SGA), and (iv) PTB.
Cytokine concentrations of the different groups were compared with a control group comprised of mothers who had no plasmodial infection, no anemia, and babies who (i) were delivered at term and (ii) had adequate birth weight for gestational age. In addition, cytokine concentrations were compared according to the number of P. falciparum infections that occurred during pregnancy. The nonparametric Kruskal-Wallis test was used for these comparisons. Bonferroni’s correction was applied to correct for the high number of statistical tests. To identify independent associations between cytokine levels and either PAM or pregnancy outcomes, we performed multivariate analysis. Unsegregated data were used instead of segregated subgroups to determine the independent effect of PAM or each pregnancy outcome, adjusted on the basis of each of the others, on the cytokine concentrations. We then investigated the influence of PAM or anemia at a given time point on cytokine levels measured at the same time, and the association between the occurrence of PAM or anemia during pregnancy and the cytokine concentrations observed at delivery, using linear regression models. For the latter, cytokine concentrations at delivery were adjusted for the presence of P. falciparum infection at delivery. Subsequently, we investigated the influence of cytokine concentrations at inclusion on the occurrence of PAM or anemia during pregnancy using logistic and Poisson regression models. Here, cytokine levels were also adjusted for the presence of P. falciparum infection at inclusion. Finally, the relationships between PTB and SGA with cytokine levels measured during pregnancy and at delivery were determined using logistic regression models. All the tests were performed including at the first step with all the factors described. Covariates with P values of <0.2 were included in multivariate models as well as other factors of interest (P. falciparum infection, anemia, PTB, SGA, and cytokine levels at INC, ANV4, and DEL) for a stepwise analysis to find the best model. In the final multivariate analysis, the association between cytokine concentration at a given time point and P. falciparum infection or pregnancy outcome was validated if a P value of ≤0.05 was obtained. The occurrence of PAM or anemia is defined by the presence of PAM or anemia (at least one episode of PAM or anemia) and the number of episodes of P. falciparum infection or anemia during pregnancy. Other factors included in the models as covariates were gravidity, maternal age, gestational age, education level, IPT, number of ANV, birth weight, and use of bed nets. Some factors were used as categorical variables (number of anemia events, number of infections, maternal age, gestational age at inclusion, gravidity, number of IPTp, number of ANV, and birth weight). In all cases, we performed separate analyses for each time point. All statistical analysis was performed using Stata/IC version 13 (StataCorp LP, College Station, TX).