Placental tumor necrosis factor alpha but not gamma interferon is associated with placental malaria and low birth weight in Malawian women

Malaria in pregnancy predisposes to maternal anemia and low birth weight (LBW). We examined the possible roles of the cytokines tumor necrosis factor alpha (TNF-α) and gamma interferon (IFN-γ) in these adverse outcomes. We measured cytokine concentrations in placental, peripheral, and cord blood plasma in relation to malaria parasitemia and placental monocyte accumulation in 276 Malawian women. Maternal hemoglobin concentration, human immunodeficiency virus status, and infant birth weight were determined. Concentrations of TNF-α in placental blood were correlated with densities of Plasmodium falciparum-infected erythrocytes (P < 0.0001) and of intervillous monocyte infiltrates (P < 0.0001) on placental histology. Peripheral blood TNF-α concentrations were relatively low and were weakly associated with malaria. TNF-α concentrations were higher in placental blood, where they were strongly associated with malaria. Placental plasma TNF-α levels were higher in women who had LBW babies (P = 0.0027), women with febrile symptoms (P < 0.0001), and teenage mothers (P = 0.04) than in other women. The presence of TNF-α in cord blood was not associated with malaria infection. IFN-γ levels were infrequently elevated, and elevated IFN-γ levels were not associated with poor pregnancy outcomes. Placental production of TNF-α, but not of IFN-γ, may be implicated in impaired fetal growth in Malawian women. Women attending Queen Elizabeth Central Hospital, Blantyre, Malawi, for delivery were screened for malaria infection by examination of thick films from placental and peripheral blood. Giemsa-stained films were examined by counting parasites against 200 leukocytes, and parasitemia was determined by using an assumed leukocyte count of 6,000/μl. Women with malaria infection (peripheral and/or placental parasitemia of >1,000/μl on thick-film microscopy) and uninfected controls (matched for gravidity and for age within ±2 years, but with peripheral and placental thick blood films negative for malaria) were recruited into the study following witnessed informed consent. A placental parasite density of ≥1,000/μl was seen in 77.5% of infected women in this study, but in only 48.9% of infected women delivering in the hospital (S. J. Rogerson, E. Chaluluka, L. Njiragoma, M. Kanjala, P. Mkundika, and M. E. Molyneux, unpublished data). Blood films were examined by a second microscopist, with a third read to resolve conflicting results, as described elsewhere (23). Final parasitemia was the average of two agreeing counts. Babies were weighed at delivery; LBW was defined as <2,500 g. Venous blood was collected from a peripheral vein and cord blood was collected from umbilical veins by venipuncture. Placental blood was collected by incising the maternal surface of the placenta (which was cleaned with sterile gauze and normal saline) and aspirating blood welling into the incision with a sterile transfer pipette. Heparinized (first 40 cases) or EDTA plasma was separated within an hour of delivery and frozen at −70°C until it was aliquoted for assay. Placental tissue biopsy specimens (2 by 2 by 1 cm) were collected into neutral buffered formalin. After fixation, blocks were wax embedded, and sections stained with Giemsa stain or hematoxylin and eosin were prepared by standard procedures. Maternal hemoglobin concentrations were measured with a Hemocue (Angelholm, Sweden) hemoglobinometer. Placental sections were examined by a standardized approach, and parasite and monocyte densities were determined by counting 500 intervillous-space cells, including infected and uninfected erythrocytes and leukocytes (23a). Parasitemia was expressed as the number of infected erythrocytes per total number of erythrocytes. Monocyte density was calculated as the number of monocytes detected per total erythrocytes and leukocytes counted. Each was expressed as a percentage. Malaria pigment (hemozoin) in monocytes and in fibrin was assessed by using a semiquantitative scale (score, 0 to 4) and recorded as present or absent. For the first 120 samples, HIV testing was performed on coded samples, without patient identifiers. For subsequent samples, HIV testing followed voluntary counselling and testing, performed the day after delivery. Plasma or serum was tested by a Serocard rapid test for HIV types 1 and 2 (HIV-1 and -2) (Trinity Biotech, Dublin, Ireland), and results were confirmed by either HIV-1 and -2 ELISA (Ortho-Clinical Diagnostics, Neckargemund, Germany), Vironostika HIV Uni-Form II (Organon Teknika, Boxtel, The Netherlands), or Determine (Abbott Laboratories, Amadora, Portugal). Cytokine assays were performed on consecutive placental plasma samples available from women enrolled in the study. Placental and maternal or cord blood cytokine concentrations were measured for TNF-α and IFN-γ by using enzyme-linked immunosorbent assay kits from R&D Systems, Abingdon, United Kingdom, according to the manufacturer's instructions. Additional cord blood samples were assayed for TNF-α by using the human TNF-α Duoset (R&D Systems, Minneapolis, Minn.). Only samples collected into EDTA were analyzed for IFN-γ. Limited peripheral blood samples were assayed owing to lack of reagents. For TNF-α, assays were performed on peripheral, placental, and cord blood samples from 20, 241, and 141 individuals, respectively. For IFN-γ the numbers were 32, 187, and 53 samples. From information supplied by the manufacturer, the upper limits of normal were defined as 15.6 pg/ml for both IFN-γ and TNF-α. Data were entered into Microsoft Access and transferred to Stata 6.0 (Stata Corp., College Station, Tex.) for analysis. Normally distributed data were compared by Student's t test, and nonnormally distributed data were compared by the Wilcoxon rank sum test. The studies were approved by the College of Medicine Research Committee, University of Malawi.