Cord blood hepcidin: Cross-sectional correlates and associations with anemia, malaria, and mortality in a tanzanian birth cohort study

Hepcidin, the master regulator of bioavailable iron, is a key mediator of anemia and also plays a central role in host defense against infection. We hypothesized that measuring hepcidin levels in cord blood could provide an early indication of interindividual differences in iron regulation with quantifiable implications for anemia, malaria, and mortality-related risk. Hepcidin concentrations were measured in cord plasma from a birth cohort (N = 710), which was followed for up to 4 years in a region of perennial malaria transmission in Muheza, Tanzania (2002-2006). At the time of delivery, cord hepcidin levels were correlated with inflammatory mediators, iron markers, and maternal health conditions. Hepcidin levels were 30% (95% confidence interval [CI]: 12%, 44%) lower in children born to anemic mothers and 48% (95% CI: 11%, 97%) higher in placental malaria-exposed children. Relative to children in the lowest third, children in the highest third of cord hepcidin had on average 2.5 g/L (95% CI: 0.1, 4.8) lower hemoglobin levels over the duration of follow-up, increased risk of anemia and severe anemia (adjusted hazard ratio [HR] [95% CI]: 1.18 [1.03, 1.36] and 1.34 [1.08, 1.66], respectively), and decreased risk of malaria and all-cause mortality (adjusted HR [95% CI]: 0.78 [0.67, 0.91] and 0.34 [0.14, 0.84], respectively). Although longitudinal measurements of hepcidin and iron stores are required to strengthen causal inference, these results suggest that hepcidin may have utility as a biomarker indicating children’s susceptibility to anemia and infection in early life. The U.S. National Institutes of Health International Clinical Studies Review Committee of the Division of Microbiology and Infectious Diseases approved the study procedures, and the institutional review boards of the Seattle Biomedical Research Institute and the National Institute for Medical Research in Tanzania provided ethical clearance. Participating mothers provided written informed consent for themselves and their newborn child. Mothers enrolled during labor were reconsented at the first follow-up visit. Prompt care was provided to sick children in accordance with Tanzanian Ministry of Health protocols. Blood smear results and hemoglobin concentrations were evaluated during the study visits, and health workers had full access to the results for clinical decision-making. All subsequent laboratory measurements were performed on deidentified samples. The Mother-Offspring Malaria Study, initiated in 2002 as a prospective birth cohort study of malaria, has been described in detail previously.22 A total of 1,045 pregnant women (1,075 offspring) between the ages of 18 and 45 years who presented for delivery at the Muheza Designated District Hospital in the Tanga region of Tanzania between September 9, 2002, and October 13, 2005, were invited to participate in the investigation; children were followed up until May 18, 2006. To be eligible for this study, children had to be 1) born to human immunodeficiency virus (HIV)–negative mothers, 2) sickle cell disease free, 3) a singleton birth, and 4) followed for a minimum of 28 days (Figure 1 ). At the time of hepcidin measurement, plasma samples were no longer available for 20% of the originally recruited sample. However, no substantial differences in baseline variables were found between children with and without measured hepcidin; this suggests that, although the missing hepcidin measurements decreased statistical power, their absence was unlikely to bias results (Supplemental Table 1). Selection of the study sample. Trained project nurses and assistant medical officers administered questionnaires to mothers and collected clinical information using standardized forms. Maternal peripheral blood samples and placentas were collected at delivery, and placental blood was extracted from placental tissue by mechanically pressing full-thickness placental tissue. Cord blood samples were collected immediately after parturition from venous umbilical blood vessels using routine procedures for cord clamping and vessel cannulation. Blood samples were collected in ethylenediaminetetraacetic acid for anticoagulation and fractionated by centrifugation at 3,000 g for 3 minutes. Plasma samples were frozen at −70°C until the immunoassays were performed. Clinicians monitored children’s health statuses during sick visits and at routine visits occurring on a biweekly basis during the first 12 months of life and a monthly basis for any follow-up beyond the first year. Children’s hemoglobin was measured at sick visits and during routine visits at approximately 3, 6, 12, 18, 24, 30, 36, 42, and 48 months of age. Parasitemia by P. falciparum was determined after counting 200 leukocytes on Giemsa-stained thick blood smear of a sample collected by heel or finger prick during child visits. In both the children under 4 years of age and pregnant women, all-cause anemia and severe anemia were defined as hemoglobin concentrations < 110 and < 70 g/L, respectively.23 Malaria was defined as the detection of P. falciparum parasitemia in peripheral blood. Severe malaria was identified clinically as a P. falciparum-positive blood smear with one or more of the following symptoms: severe anemia (i.e., hemoglobin concentration < 50 g/L), respiratory distress (i.e., respiratory rate > 50 breaths/minute in neonates and > 40 breaths/minute in older children with two of the following: nasal flaring, intercostal indrawing, subcostal recession, and grunting), > 1 convulsion episode in 24 hours, coma (i.e., Blantyre score < 3), prostration (i.e., inability to sit upright in a child normally able to do so or drink in a child too young to sit), hypoglycemia (i.e., blood glucose < 40 mg/dL), renal failure (i.e., urine output < 12 mL/kg/day), hemoglobinuria, jaundice, and shock (i.e., cold extremities, rapid heart rate, and/or systolic blood pressure < 50 mmHg). Placental malaria was defined as the detection of P. falciparum parasitemia in placental blood. The 25-amino acid bioactive isoform of hepcidin was detected in cord blood plasma using the hepcidin-25 (human) EIA kit (Bachem, Bubendorf, Switzerland) as previously described.24 A dilution series of standard hepcidin-25 was run on each plate with values ranging from 25 to 0.05 ng/mL. Sample values were interpolated using logistic 4-parameter nonlinear curve fitting in GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA). If hepcidin readings initially fell outside the linear portion of the standard curve, samples were rerun at appropriate dilution. If samples exhibited a high coefficient of variation (> 15%) between duplicates, they were rerun, and the new value was substituted. Soluble inflammatory and iron markers in cord, maternal peripheral, and placental blood plasma were measured using commercially available multiplex, bead-based platforms (BioPlex®; BioRad, Irvine, CA) and custom-made assay kits as previously described.25 Samples that did not produce detectable concentrations of a given marker were assigned a value of half the limit of detection of that marker. Sickle cell trait was determined by cellulose acetate paper electrophoresis (Helena Laboratories, Beaumont, TX), and α-thalassemia was determined by polymerase chain reaction.26 The statistical approach used to describe the cross-sectional correlates of hepcidin was adapted from the methodology used by the Fibrinogen Studies Collaboration.27 Positively skewed continuous variables were loge-transformed before analysis. Cross-sectional correlations between hepcidin and continuous traits were first quantified by Pearson’s r. Mean levels of loge-transformed cord hepcidin were plotted against the mean for each eighth (i.e., corresponding to approximately one half of a standard deviation) of continuous traits. This approach allowed assessment of the shape of any association with cord hepcidin without assuming linearity a priori. Univariate linear regressions were then used to evaluate mean percent differences in hepcidin per level of categorical variables and by standard deviation of continuous variables; the mean percent differences were estimated with the formula (eβ − 1) × 100%, where the β coefficient represents the mean difference in loge-transformed hepcidin level. The coefficient of determination (r2) from a linear regression was used to quantify the proportion of variance in cord hepcidin explained by the measured correlates. For the prospective analyses, the primary exposure was hepcidin measured from cord plasma, which was evaluated both continuously and in tertiles (histogram provided in Supplemental Figure 1; cutoffs provided in Supplemental Table 2). Associations of cord hepcidin with repeated measurements of hemoglobin (N = 6,121 visits) and parasitemia (N = 5,493 visits with ≥ 1 parasite per 200 white blood cells) were evaluated using linear mixed-effects models with village- and child-specific random effects, quadratic time trends, and adjustment for placental malaria status, cord blood levels of loge ferritin and loge C-reactive protein, and concurrent malaria infection status. Fractional polynomial models of best fit were used to visualize the nonlinear associations between hemoglobin concentrations and age for children with above and below median cord hepcidin levels. For survival analyses, Cox proportional hazard models were used to calculate the hazard ratios (HRs) for the times to the first episode of anemia, severe anemia, malaria, severe malaria, and child death. Models were adjusted for potential confounders, including village of residence, placental malaria status, and cord blood levels of loge ferritin and loge C-reactive protein. To allow HRs to be compared informatively across any pair of hepcidin tertiles and without depending on the precision within any arbitrarily selected baseline group, 95% confidence intervals (CIs) were estimated from floated variances.28 All statistical analyses were performed using Stata, version 12 (StataCorp LP, College Station, TX). P values were from two-sided tests.