Fetal Responses during Placental Malaria Modify the Risk of Low Birth Weight

Study subjects and clinical procedures.The subjects included in the present study were individuals participating in the Mother-Offspring Malaria Studies (MOMS) Project in Muheza district, northeastern Tanzania, between September 2002 and October 2005. Mother-infant pairs were recruited during delivery hospitalization at the Muheza Designated District Hospital. Clinical information, including antimalarial agent use during pregnancy to prevent malaria, was obtained by history and physical examination, as well as review of medical records. Pregnant women were eligible for the study if they were between 18 and 45 years old and had no clinical evidence of chronic or debilitating illness, such as recent significant weight loss or chronic diarrhea. Cases of twin or triplet gestation, human immunodeficiency virus infection, or early neonatal death were not included in the analyses in this study. LBW was defined as weight at birth of <2,500 g.

Protocols for procedures used in this study were approved by the International Clinical Studies Review Committee of the Division of Microbiology and Infectious Diseases at the U.S. National Institutes of Health. Ethical clearance was obtained from the institutional review boards of Seattle Biomedical Research Institute and the National Institute for Medical Research in Tanzania. Written informed consent was obtained from each mother before she entered the study.

Peripheral blood was obtained by venipuncture from women at delivery and anticoagulated with citrate phosphate dextrose. Cord blood samples were obtained by clamping the cord and cannulating umbilical vessels immediately after delivery. After removal of the umbilical cord and fetal membranes, placental blood samples were obtained by manual compression of the placental tissue in a grinder. Placental and cord blood samples were anticoagulated with EDTA. All samples were collected by trained nurses. Plasma was obtained by centrifugation at 3,000 × g for 3 min and was stored frozen at −70°C until it was thawed on the day that cytokine assays were performed.

Parasitemia was defined as identification of any parasites in a placental blood smear or a cord blood smear by microscopy. Thick and thin smears were prepared for all samples; thin smears were fixed with methanol. Blood slides were stained for 10 min in 10% Giemsa stain, washed in tap water, air dried, and then examined using light microscopy at a magnification of ×100. Ten thousand red blood cells were examined in the thin smear before it was concluded that a placental blood slide was negative.

Plasma analyte assays.Each plasma sample was analyzed using a multiplex, bead-based platform (BioPlex; Bio-Rad, Irvine, CA) and custom-made assay kits as previously described (8). For each serum sample, all analytes were assayed in a single day, thus eliminating freeze-thaw cycles. All pipetting and sample identification were performed with a bar code-enabled, high-speed pipetting robot (Megaflex; Tecan, Research Triangle Park, NC). The detection limits for the different soluble factors were as follows: IFN-γ, 0.04 pg/ml; TNF-α, 0.10 pg/ml; IL-10, 0.02 pg/ml; ferritin, 0.07 ng/ml; leptin, 1.28 pg/ml; and CRP, 0.01 μg/ml. The levels of soluble factors were adjusted to account for dilution in anticoagulant at the time of sample collection.

Statistical analyses.Analyses were performed using Statview 5.0.1 (SAS Institute, Cary, NC). Differences in the proportions of placental blood samples with detectable IFN-γ from PM-positive (PM⁺) and PM-negative (PM⁻) women were analyzed by a chi-square test. Differences between groups in the levels of soluble factors or parasite densities were analyzed by a Mann-Whitney test or Kruskal-Wallis test, according to the number of groups. An unpaired t test was used to test for differences in maternal age and infant birth weight between the PM⁺ and PM⁻ groups. Factorial analysis of variance was used to identify interactions between PM or gravidity on the relationship between soluble factors and birth weight. Univariate linear regression and multivariate linear regression were used to analyze the relationship of cytokines and other soluble factors to birth weight. Levels of IFN-γ, TNF-α, IL-10, leptin, and ferritin, as well as parasite density, were log transformed to obtain normal distributions for regression analyses.

Demographic characteristics of the study population.After twins and triplets, cases of human immunodeficiency virus infection, and early neonatal deaths were excluded, 808 mother-infant pairs remained for analysis. Demographic characteristics of the cohort stratified by PM and gravidity are shown in Table 1. A total of 103 study participants had PM at delivery, while parasites were detected in cord blood smears of five participants. Primigravid and multigravid women with PM were significantly younger than their PM⁻ counterparts. PM was associated with lower birth weight in all gravidity groups (P < 0.05 for all groups), but the smallest newborns were delivered by first-time mothers. Mean birth weight did not vary significantly based on the mother's history of taking antimalarials as preventive treatment during pregnancy (data not shown), possibly due to the known high rates of drug-resistant parasites in the study area. According to the Dubowitz score, the gestational age of LBW newborns was not significantly different from that of other newborns (data not shown), suggesting that intrauterine growth retardation was the primary cause of LBW in this cohort. Infected primigravid women had significantly higher parasitemia levels (median percentage of red blood cells infected, 2.9%; P = 0.0002) than infected secundigravid women (median, 0.6%) or multigravid women (median, 0.9%).

PM alters maternal but not fetal levels of inflammatory markers.PM was associated with significant increases in the levels of inflammatory markers (including IFN-γ and TNF-α) in the placental blood of primigravid and multigravid women but not in the placental blood of secundigravid women (Table 2). Placental IFN-γ was detected more frequently in samples from PM⁺ primigravidae than in samples from PM⁻ primigravidae (68.2% versus 48.4%; P = 0.02) or multigravidae (84.2% versus 48.9%; P = 0.003) but not secundigravidae (56.7% versus 54.8%; P > 0.999). Similarly, the levels of ferritin, a marker of inflammation, as well as iron stores, were significantly increased in placental blood of infected primigravid and multigravid women but not in placental blood of secundigravid women. The level of anti-inflammatory cytokine IL-10 was increased in placentas of infected women in all gravidity categories.

The levels of TNF-α, IL-10, and ferritin were also increased in the maternal peripheral blood of women with PM. Unlike the pattern seen in placental blood, the increases occurred in secundigravid women as well as in primigravid and multigravid women. The peripheral blood IFN-γ levels in PM⁺ women were not significantly different from those in PM⁻ women.

The placental leptin levels in PM⁺ samples were not different from those in PM⁻ samples for any of the gravidity groups. The peripheral leptin levels were significantly lower in PM⁺ primigravid women than in PM⁻ primigravid women, but this relationship was not observed for other gravidity groups.

In general, the fetal levels of cytokines and other soluble factors were not altered during PM. PM was related to a small but significant increase in the level of cord blood IL-10 in secundigravid women. PM did not significantly alter cord blood levels of TNF-α, IFN-γ, leptin, or ferritin in any of the gravidity groups. The levels of the inflammatory marker CRP were elevated (>8.2 μg/ml) in only 17/457 (3.1%) cord blood samples tested, and elevated levels of CRP were seen in similar proportions of PM⁺ cases (4/62, 6.4%) and PM⁻ cases (13/395, 3.3%) (P = 0.22).

Gravidity and PM modify relationships of soluble factors to LBW.We examined the relationship between all measured analytes and birth weight by using univariate linear regression. Placental IL-10 had a significant inverse relationship with birth weight (β = −0.114, P = 0.002), while cord leptin had a significant positive relationship with birth weight (β = 0.277, P < 0.0001). Cord ferritin, placental IFN-γ, and placental TNF-α were not significantly related to birth weight in these aggregate analyses.

PM and gravidity are known to influence relationships between inflammatory cytokines and birth weight in malaria-exposed individuals, and the relationship is limited to first pregnancies (12). We performed factorial analysis of variance to examine whether gravidity and PM modified relationships between soluble factors and birth weight in this cohort. Interaction terms involving gravidity and/or PM significantly or nearly significantly modified relationships between birth weight and several soluble factors, including cord ferritin (significance of term PM * parity, P = 0.04) and placental TNF-α (significance of term placental TNF-α * PM, P = 0.07). We therefore stratified our analyses by gravidity (Fig. 1) and by PM status (Fig. 2).

• Open in new tab
• Download powerpoint

FIG. 1.

Cord blood levels of (A) leptin and (B) ferritin for normal-birth-weight (open boxes) and LBW (shaded boxes) neonates. 1, primigravidae (188 normal-birth-weight babies and 20 LBW babies); 2, secundigravidae (170 normal-birth-weight babies and 5 LBW babies); 3, multigravidae (345 normal-birth-weight babies and 12 LBW babies). Each box plot indicates the median (horizontal line) and interquartile range (box), whereas the whiskers indicate the 10th and 90th percentiles. The differences between two groups were analyzed by the Mann-Whitney U test. Associated P values are shown.

• Open in new tab
• Download powerpoint

FIG. 2.

Placental blood concentrations of (A) TNF-α, (B) IFN-γ, and (C) IL-10 for normal-birth-weight (open boxes) and LBW (shaded boxes) neonates. For the PM⁺ group, there were 78 normal-birth-weight babies and 9 LBW babies; for the PM⁻ group, there were 603 normal-birth-weight babies and 26 LBW babies. Each box plot indicates the median (horizontal line) and interquartile range (box), whereas the whiskers indicate the 10th and 90th percentiles. The differences between two groups were analyzed by the Mann-Whitney U test. Associated P values are shown.

In analyses stratified by gravidity, LBW was related to significant or nearly significant reductions in levels of cord leptin and cord ferritin in all gravidity groups (Fig. 1). Although the placental levels of TNF-α, IFN-γ, and IL-10 were elevated in primigravid and multigravid women delivering LBW babies compared to women delivering normal-birth-weight babies (data not shown), the differences were not significant.

In analyses stratified by PM status, LBW was strongly related to inflammatory cytokines when the mother was PM⁺. The placental levels of TNF-α, IFN-γ, and IL-10 were all significantly and substantially increased in PM⁺ mothers delivering LBW babies compared to women delivering normal-birth-weight babies (Fig. 2). Conversely, the cord levels of both leptin and ferritin were significantly decreased in LBW babies compared with normal-birth-weight babies, and this was observed for both PM⁻ and PM⁺ groups (data not shown).

Placental IFN-γ is specifically related to decreased birth weight.We performed multivariate regression analysis of the factors that were related to birth weight in the univariate analyses (Table 3). For multivariate analysis, we stratified by gravidity and PM status, since these variables interacted with several plasma factors and modified their relationship to birth weight. Because one-half of women had undetectable placental IFN-γ levels (which therefore could not be normalized by transformation), we analyzed the groups with detectable and undetectable IFN-γ levels separately.

In PM⁺ primigravidae with detectable placental IFN-γ, cord blood ferritin was strongly related to increased birth weight (β = 0.80, P < 0.0001), while placental IFN-γ was related to decreased birth weight (β = −0.39, P < 0.01). These relationships were not seen in other PM⁺ women (Table 3). In multivariate analysis that excluded IFN-γ data but included all PM⁺ primigravidae, cord blood ferritin was still related to increased birth weight (β = 0.41, P = 0.01), while placental TNF-α was related to decreased birth weight (β = −0.50, P = 0.03). The effect of cord blood ferritin was different in different groups. When placental inflammatory cytokines were related to decreased birth weight (such as IFN-γ or TNF-α in PM⁺ primigravidae or TNF-α in PM⁻ secundigravidae without detectable IFN-γ), ferritin had a positive effect on birth weight. Conversely, ferritin had an inverse relationship with birth weight in PM⁻ secundigravidae with detectable IFN-γ, in whom none of the cytokines had an effect on the outcome (Table 3).

In PM⁻ women, cord blood leptin levels were significantly associated with increased birth weight in all gravidity groups. Leptin levels were not related to birth weight in PM⁺ women.