Resisting and tolerating P. falciparum in pregnancy under different malaria transmission intensities

This study was conducted between 2010 and 2012 in four sub-Saharan countries (Additional file 1: Figure S1), namely Benin (Allada, Sékou and Attogon), Gabon (Lambaréné and Fougamou), Kenya (Siaya), and Mozambique (Manhiça and Maragra). Pregnant women were enrolled in the context of the Malaria in Pregnancy Preventive Alternative Drugs clinical trial (ClinicalTrials.gov NCT0081121; Table 1) [19, 20]. At enrolment, pregnant women received long-lasting insecticide-treated bed-nets. Following national guidelines in place, HIV status was assessed after voluntary HIV counselling and testing with an HIV rapid test and positive results were confirmed with a second rapid test [19, 20]. Haemoglobin and the syphilis rapid plasma reagin (RPR) test were assessed as part of routine antenatal care on finger-prick-collected capillary blood. Among HIV-infected women, 5 mL of venous blood were taken for CD4 + T cell count by flow cytometry after staining of whole blood with CD3, CD8 and CD4 fluorochrome-labelled antibodies. Flow cytometry acquisition was performed using FACSCalibur (BD Biosciences) and TruCOUNT tubes (Becton Dickinson). Additionally, viral load determination was performed using the COBAS AMPLICOR or AmpliPrep (Roche Diagnostics) devices. Haemoglobin was determined in capillary blood samples using mobile devices (HemoCue and Hemocontrol) [19, 20]. Gestational age at enrolment was determined from symphysis fundal height measurement using a standard tape measure (centimetres) and McDonald’s rule to transform the symphysis fundal height in centimetres into gestational weeks [19, 20]. Following physical examination, recruited women with gestational age ≥ 13 weeks received their first dose of IPTp (either sulfadoxine-pyrimethamine (SP) or mefloquine (MQ) for HIV-uninfected women and either placebo or MQ for HIV-infected women) under supervision. Women allocated to the SP group received standard IPTp (three tablets of the fixed combination therapy containing 500 mg of sulfadoxine and 25 mg of pyrimethamine), whereas participants allocated to the MQ groups received 15 mg/kg of the drug. The second IPTp-SP/MQ administration for HIV-uninfected women, and the second and third administrations of IPTp-MQ/placebo for HIV-infected women, were given at least 1 month apart. All HIV-infected women also received study co-trimoxazole tablets on a monthly basis for daily prophylaxis.

At delivery, maternal haemoglobin was determined and newborns were weighed using weekly calibrated scales (either digital or three beam balances) and their gestational age at birth (except for Kenyan newborns) was evaluated using the Ballard’s score [21]. Newborn weights not captured at birth but within the first week of life were estimated using a linear regression model [22]. Peripheral and placental blood smears, as well as 50 μL of maternal peripheral and placental blood spotted onto filter paper, were collected for parasitological assessments. Tissue samples were also collected from the maternal side of the placenta and placed into 10% neutral buffered formalin. Biopsies were processed, stained and examined following standard procedures [23]. From the first antenatal visit, all study women were indicated to receive ferrous sulphate-folic acid supplements for prevention of anaemia in pregnancy. If a woman was diagnosed with anaemia, she was treated with oral ferrous sulphate 200 mg/8 hours for 3 months or blood transfusion for severe cases. Clinical malaria episodes were treated with oral quinine (first trimester) or artemether-lumefantrine (subsequent trimesters) for uncomplicated malaria; parenteral quinine was used for treatment of severe malaria [19, 20].

Thick and thin blood films, as well as placental biopsies, were read for Plasmodium species detection according to standard, quality-controlled procedures [19, 20, 24]. The quality of the reading across sites was controlled through a quality control program during the study. Past placental infection was defined by the presence of P. falciparum pigment (i.e. hemozoin) without parasite detection on placental histologic examination, and chronic placental infection was defined by the presence of P. falciparum pigment in combination with the detection of parasites [1]. A 30% random selection of paired peripheral and placental blood onto filter papers from HIV-negative women [20] and all the paired filter papers from HIV-infected women [19] were tested for the presence and density of P. falciparum in duplicate by means of a real-time quantitative polymerase chain-reaction (qPCR) targeting 18S rDNA [1, 25]. Parasitemia was quantified by extrapolation of cycle thresholds from a standard curve of P. falciparum ring-infected erythrocytes. Samples without amplification (no cycle thresholds detected) were considered negative, and a density of 2 parasites/μL was assigned if amplification was observed out of the lower range of the standard curve (5 parasites/μL). A negative control with no template DNA was run in all reactions. P. falciparum infections were considered submicroscopic if parasites were detected by qPCR but not by microscopy [1]. A quality check program was established to ensure comparable performance of qPCR techniques in different laboratories [25].

A random selection of 50% of peripheral plasma samples collected at delivery from HIV-uninfected women in the extremes of the malaria transmission spectrum (n = 170 from Benin and n = 170 from Mozambique) was tested for IgG levels against P. falciparum recombinant VAR2CSA domain (DBL3X from 3D7 strain), and the merozoite surface protein-1 (MSP-1₁₉; 19-kD fragment, 3D7 strain) using a multiplex suspension array panel (xMAP^TM technology) and the Luminex® 100/200™ System (Luminex Corp., Austin, TX, USA) [1]. Briefly, magnetic carboxylated microspheres (MagPlex^TM-C, Luminex) were coupled with recombinant protein. After blocking with bovine serum albumin in phosphate-buffered saline, microspheres were sequentially incubated with 100 μL of plasma (dilutions 1:500, 1:20,000 and 1:800,000 in duplicate for each sample), 100 μL of biotinylated anti-human IgG (diluted 1:2500) and 100 μL of streptavidin-conjugated R-phycoerythrin (diluted 1:1000). The plate was immediately read using Bio-Plex Manager version 4.0, and at least 50 microspheres per analyte were acquired per sample. Crude mean fluorescent intensity was exported with background fluorescence from blank wells already subtracted. PfHRP2 [26] was quantified in plasmas available from women with peripheral qPCR-detected P. falciparum infection at delivery (n = 42 in Benin and n = 13 in Mozambique) using a commercial HRP2 antigen-capture ELISA (Malaria Ag CELISA kit; Cellabs) [27].

Pregnant women were included in the analysis if they had all information on IPTp type of treatment, HIV infection, age, parity, newborn weight, maternal haemoglobin levels, as well as qPCR results in peripheral and placental blood at time of delivery. Women were classified as primigravid (first pregnancy) and multigravid (at least one previous pregnancy). Age was categorised as < 25 and ≥ 25 years on the basis of median maternal age in the study population. Preterm birth was defined as a gestational age < 37 weeks. Participant’s baseline characteristics, parasitological outcomes, antimalarial IgG levels and PfHRP2 levels were compared between study areas by univariate analysis and logistic or lineal regression models. Changes in maternal haemoglobin levels and birthweight with increases in qPCR-peripheral parasite density were assessed by linear regression. The impact of P. falciparum infections on haemoglobin levels, birthweight and preterm birth was assessed in logistic regression models. Continuous outcome variables (qPCR-parasite densities, antimalarial IgGs and HRP2 levels) exhibiting a skewed distribution were log transformed. Regression models were estimated through a backward stepwise approach with a significance level for removal from the model of 0.1 and a significance level for addition of 0.05, adjusted by type of IPTp drug and baseline covariates at recruitment (season, age, gravidity, gestational age, anaemia, literacy, RPR result and mid-upper arm circumference (MUAC)), as well as CD4 + T cell count in the case of HIV-infected women [19, 20]. The modification of the associations by study area was assessed by including interaction terms into the regression models and combining the coefficients plus the interaction and the standard error by the delta method. P values of less than 0.05 were considered to indicate statistical significance. Stata version 13 was used for data analysis and GraphPad version 5.01 for graphical depiction of data.

A total of 1714 pregnant women, 946 (55%) HIV-uninfected and 768 (45%) HIV-infected, were included in this study (Table 1 and Additional file 1: Figure S1). The 3428 peripheral and placental filter papers from these 946 HIV-uninfected and 768 HIV-infected women were tested for the presence and density of P. falciparum by qPCR [25, 28]. Covariates at recruitment were similar in the study subgroup and the rest of the women participating in the randomised trial [19, 20] (Additional file 1: Table S1). Prevalence of qPCR-detected P. falciparum at delivery, either in peripheral blood at delivery or placental blood (maternal qPCR infection in Table 2), was 41% (143/349) in Benin, 11% (28/257) in Gabon and 6% (21/340) in Mozambique (P < 0.001) among HIV-uninfected women, and 9% (32/351) in Kenya and 4% (15/417) in Mozambique (P < 0.001) among HIV-infected women (Fig. 1a and Table 2). Similar trends were observed for maternal microscopic infections (Fig. 1b and Table 2). The relationship between microscopy positivity and qPCR parasitaemia in the different study sites is presented in Additional file 1 (Table S2). The proportion of samples that were negative by microscopy but for which qPCR indicated a parasitaemia above 200 parasites/μL was not different between sites (Additional file 1: Table S3).

Microscopic P. falciparum infection detected at delivery in peripheral blood and/or the placenta of Mozambican women was associated with a reduction of 1.17 g/dL (95% CI –2.09 to –0.24, P = 0.014) in their haemoglobin levels, but this reduction was not observed among Beninese nor Gabonese women (P = 0.581 and P = 0.154, respectively; Fig. 2a). Microscopic infections were also associated with larger drops in haemoglobin levels from recruitment to delivery in Mozambican (–1.66 g/dL, 95% CI –2.68 to –0.64, P = 0.001) and Gabonese (–0.91 g/dL, 95% CI –1.79 to –0.02, P = 0.045) women, but not among Beninese women (P = 0.213; Fig. 2c). No impact of microscopic P. falciparum infection in the mother was observed on the birthweight, irrespective of study area (Fig. 2e). However, past or chronic placental P. falciparum infection, defined by the presence of P. falciparum pigment (i.e. hemozoin) on placental histologic examination, was associated with a reduction of 247.2 g (95% CI –479.1 to –15.2, P = 0.037) in the birthweight of Mozambican babies, which was not observed in Benin (P = 0.569) and Gabon (P = 0.486). Given the strong association between birthweight and gestational age at delivery (Additional file 1: Table S4), the regression model assessing the relationship between birthweight and infection was adjusted by preterm birth, showing a loss of statistical significance (P = 0.192). However, past or chronic placental P. falciparum infection were associated with an increased risk of preterm birth in Mozambican newborns (OR = 7.05, 95% CI 1.79 to 27.82, P = 0.005), but not among Beninese (P = 0.785) and Gabonese newborns (P = 0.639). Microscopic P. falciparum infection was not associated with reductions in haemoglobin levels of HIV-infected women from Kenya and Mozambique nor with reductions of birthweight (Fig. 2b, d, f). Finally, no adverse clinical impact was observed for P. falciparum submicroscopic infections, defined by the detection of P. falciparum by qPCR but not by microscopy, irrespective of the HIV-infectious status of the pregnant women (Additional file 1: Figure S2). Adjusting variables that remained in the final regression models and interactions assessed with parity, age and IPTp treatment are detailed in Table S5 and S6 of Additional file 1.

The proportion of peripheral submicroscopic infections among HIV-uninfected women was the highest in Benin (83%; 82/106), followed by Mozambique (60%, 9/15) and Gabon (55%, 12/22, P _adjusted = 0.033; Fig. 3a). Similar trends, although not statistically significant, were observed for placental submicroscopic infections (82% (78/95), 58% (10/17) and 71% (15/21), respectively, P _adjusted = 0.674; Fig. 3b). Among pregnant women with a qPCR-detected infection either in peripheral or placental blood, the proportion of submicroscopic infections in any of the two compartments was the highest in Benin (94%, 124/132), followed by Gabon (82%, 18/22) and Mozambique (74%, 14/19, P _adjusted = 0.028). PfHRP2 concentrations, indicative of overall parasite biomass [26], were higher in pregnant women with a peripheral qPCR-detected infection from Mozambique (n = 13) than from Benin (n = 42, P = 0.048; Fig. 3c). HRP2 levels decreased with parity of pregnant women in Benin (0.13, 95% CI 0.02 to 0.79, P = 0.028) but not in Mozambique (1.09, 95% CI 0.09 to 12.21, P = 0.945). In contrast, age was not associated with differences in HRP2 levels among Beninese (P = 0.052) and Mozambican women (P = 0.123). Levels of IgGs against DBL3X from VAR2CSA and MSP1 were higher in women from Benin than in women from Mozambique (P < 0.007; Fig. 3d). IgG responses against MSP1 increased with parity of Beninese pregnant women (2.39, 95% CI 1.49 to 3.83, P < 0.001) but not among Mozambican pregnant women (P = 0.401). IgG responses against DBL3x increased with parity of Beninese and Mozambican pregnant women although the magnitude of the increase was higher among Beninese (4.77, 95% CI 3.08 to 7.38, P < 0.001) than Mozambican women (1.86, 95% CI 1.23 to 2.79, P = 0.003; Fig. 3e). Age was not associated with differences in IG responses irrespectively of antigen and study area (P = 0.223 for MSP1 and P = 0.171 for DBL3X in Benin and P = 0.959 for MSP1 and P = 0.380 for DBL3X in Mozambique). No differences in covariates at recruitment were observed between the women with plasma available for serological tests and the rest of the women. The proportion of submicroscopic infections did not differ among HIV-infected women from Kenya and Mozambique (Fig. 3a, b).

Doubling qPCR parasite densities in peripheral blood of HIV-uninfected Mozambican women was associated with a reduction of 0.16 g/dL (95% CI –0.29 to –0.02, P = 0.023) in their haemoglobin levels. In contrast, haemoglobin levels remained unaffected by increasing parasite densities in Beninese (P = 0.287) and Gabonese women (P = 0.381; Fig. 4a and Additional file 1: Figure S3). Similarly, doubling qPCR parasite densities in peripheral blood of HIV-uninfected Mozambican women was associated with a larger drop in the haemoglobin levels from recruitment to delivery (–0.29 g/dL, 95% CI –0.44 to –0.14, P < 0.001), while no relationship was found in Beninese (P = 0.076) and Gabonese women (P = 0.176; Fig. 4c and Additional file 1: Figure S3). No association was found between qPCR parasite densities and birthweight irrespectively of the country of women, nor with pregnancy outcomes in HIV-infected women (Fig. 4).