Periconceptional multiple-micronutrient supplementation and placental function in rural Gambian women: A double-blind, randomized, placebo-controlled trial

Background: Maternal micronutrient deficiencies are commonly associated with clinical indicators of placental dysfunction. Objective: We tested the hypothesis that periconceptional multiplemicronutrient supplementation (MMS) affects placental function. Design: We conducted a double-blind, randomized, placebo-controlled trial of MMS in 17- to 45-y-old Gambian women who were menstruating regularly and within the previous 3 mo. Eligible subjects were pre- randomly assigned to supplementation with the UNICEF/WHO/United Nations University multiple micronutrient preparation (UNIMMAP) or placebo on recruitment and until they reached their first antenatal checkup or for 1 y if they failed to conceive. Primary outcome measures were midgestational indexes of utero-placental vascular-endothelial function [ratio of plasminogen-activator inhibitor (PAI) 1 to PAI-2 and mean uterine-artery resistance index (UtARI)] and placental active transport capacity at delivery [fetal to maternal measles antibody (MMA) ratio]. Results: We recruited 1156 women who yielded 415 pregnancies, of which 376 met all of the inclusion criteria. With adjustment for gestational age at sampling, there were no differences in PAI-1 to PAI-2 or MMA ratios between trial arms, but there was a 0.02-unit reduction in UtARI between 18 and 32 wk of gestation (95% CI: 20.03, 20.00; P = 0.040) in women taking UNIMMAP. Conclusions: Placental vascular function was modifiable by periconceptional micronutrient supplementation. However, the effect was small and supplementation did not further affect other variables of placental function. This trial was registered at www.controlled-trials. com as ISRCTN 13687662. The study was a double-blind, randomized, placebo-controlled trial of periconceptional MMS assessed on 3 primary outcomes. The first was the ratio of maternal plasminogen activation inhibitor (PAI) 1 (a marker of endothelial activation) to PAI-2 (a marker of placental function) at 18–22 wk of gestation (13). In a healthy pregnancy, the concentrations of both biomarkers should increase progressively, but their ratio should decline as placental mass, and PAI-2 production, increases (14). The second primary outcome was uterine-artery Doppler waveform at 18–22 wk of gestation [a surrogate marker of placental perfusion that correlates with trophoblast invasion (15)]. The uterine-artery pulsatility index (UtAPI) and uterine-artery resistance index (UtARI) were also measured at 28–32 wk of gestation, and diastolic notching was noted. These indexes quantify systolic and diastolic components of the flow velocity waveform in a specific blood vessel over a single cardiac cycle. The higher the values, the greater the downstream vascular resistance. The third primary outcome was the ratio of the delivery concentrations of fetal to maternal measles antibody (MMA), which was used as a proxy marker of placental transport capacity. The micronutrient supplement (Lomapharm) was a coated tablet containing 15 vitamins and trace elements, formulated to a composition specified by the UNICEF/WHO/United Nations University (UNICEF/WHO/UNU) for use by pregnant women in the developing world and known as the UNICEF/WHO/UNU international multiple micronutrient preparation (UNIMMAP) (16). UNIMMAP contains the following: vitamin A (800 retinol equivalents), vitamin D (200 IU), vitamin E (10 mg), vitamin C (70 mg), thiamin (1.4 mg), riboflavin (1.4 mg), niacin (18 mg), pyridoxine (1.9 mg), cobalamin (2.6 mg), folic acid (400 μg), iron (30 mg), zinc (15 mg), copper (2 mg), selenium (65 μg), and iodine (150 μg). The placebo was manufactured to be indistinguishable from the supplement (Lomapharm). The study took place between March 2006 and June 2008 in the Kiang West region of The Gambia among 33 villages under demographic surveillance by the United Kingdom Medical Research Council (MRC) field station staff at Keneba. Kiang West comprises 750 km2 of savannah scrub and is bordered on 3 sides by the River Gambia and its tributaries and on the fourth by a partially surfaced road. The population consists of ∼14,000 individuals, predominantly ethnic Mandinka, who live by subsistence farming. Nutritional status and morbidity patterns in this community have been well described and are largely defined by a distinct tropical seasonality, with a long dry season from November to June followed by a period of intense and daily rainfall between July and October (17–19). Local HIV seroprevalence at the time of the study was ∼1% (20). In this area, the incidence of low birth weight was 13%, PTB was 12%, and 25% of newborns were small for gestational age (<10th centile weight for gestational age) with FGR (18). Local maternal and child primary health care is provided by government nurse trekking teams, supported by the clinical staff of MRC Keneba. Most women deliver at home under the supervision of traditional birth attendants (TBAs) who undergo basic training in clean birth practices and recognition of common pregnancy complications (21). Secondary and tertiary health care services are provided at the Royal Victoria Teaching Hospital (RVTH) in Banjul, which is 4 h by road from Keneba. Women aged 17–45 y and residing in Kiang West during the recruitment phase of the study and who were registered in the MRC Keneba demographic database and not concurrently enrolled in other intervention studies were eligible to take part in the trial. Eligible women (and their guardians for those <18 y old) were invited to attend a recruitment clinic in their home villages. They were provided with oral and printed study information by a nurse-midwife (NM) and asked to give written consent if they wished to take part. Those who used no contraception, were not known to be pregnant, and had experienced menses in the past 3 mo and were not breastfeeding were included. Those who were severely anemic (hemoglobin concentration <7 g/dL) were treated before being considered for re-recruitment at a later date. Women with a subsequent multiple pregnancy or obvious fetal abnormality were excluded, although they were clinically followed because nonnutritional factors determine placental function in such pregnancies (22). Eligible subjects identified in the demographic survey were randomly assigned before recruitment to 1 of 4 color-coded groups. An independent researcher allocated 2 color codes to intervention and 2 to placebo groups and held the allocation key. Randomization was completed by an MRC Keneba statistician (AJF) by using computer-generated permuted blocks of 8, stratified by maternal age. Researchers involved in recruitment, data collection, or analysis were blind to subjects’ color codes, and all staff and participants were blind to the allocation key. The NM recorded a brief obstetric, medical, and educational history and collected a 10-mL venous blood sample from each participant (Sarstedt Monovette). Whole blood was analyzed for hematology and the presence of malaria parasites. Plasma was separated within 4 h of venipuncture, frozen at −80°C, and later analyzed for vitamin A and ferritin concentrations. Weight, height, left midupper arm circumference (MUAC), and systolic/diastolic blood pressure were measured by a trained fieldworker. Women were issued tablets corresponding to their randomized color-coded intervention group and were asked to take 1 tablet daily until told to stop by a member of the trial team. Participants were seen at least fortnightly by a fieldworker in their village and were resupplied with tablets according to their color groups. The number of tablets apparently consumed between visits was used as a measure of compliance (compliance = number of tablets apparently consumed/number of days enrolled in study). Information on morbidity and menstrual patterns was recorded. Women were asked to notify fieldworkers of missed menstrual periods and to provide urine samples for pregnancy testing by an NM when pregnancy was suspected (QuickVue One-Step hCG Urine Test; bioMérieux). With a positive pregnancy test result, the woman’s study intervention was withdrawn and replaced with 60 mg elemental iron and 250 μg folic acid daily, in accordance with Gambian national antenatal policy. These women were referred to the study antenatal clinic (ANC) at MRC Keneba for pregnancy confirmation and follow-up. Those without a positive pregnancy test result continued with supplementation until July 2007 (a period of 12–14 mo), at which time the supplementation was discontinued. Hematologic outcomes for nonpregnant participants are reported elsewhere (23). Women who migrated from the study area for 6 consecutive weeks were defined as lost to follow-up. Fetal deaths were classified as miscarriages (spontaneous loss of pregnancy before 24 completed weeks of gestation) or stillbirths (no signs of life at birth after 24 completed weeks of gestation). These were reported by TBAs and cross-checked by the fieldworker and/or the NM. In cases of uncertainty, confirmatory sonography and urinary pregnancy testing were completed at the ANC in MRC Keneba. Viable pregnancy was confirmed and dated by using transabdominal ultrasound at the ANC booking clinic. Women without a viable pregnancy at the ANC were assumed to have miscarried and were censored from the study. Women with multiple pregnancies or fetal anomalies were also censored. All women were offered continuing clinical care, and those who required specific obstetric management were referred to RVTH. Women with sonographically dated singleton pregnancies were invited and offered transport to attend the ANC for complete clinical review at 18–22 wk and 28–32 wk, including anthropometric measurements, urinalysis, blood pressure assessment, external obstetric examination, detailed fetal sonography, and venipuncture. These tasks were completed by the NM, and sonography was completed by one of the authors (SO). Pregnancies of >24 wk of gestation at booking could not be accurately dated by ultrasound and were dated by maternally recalled last menstrual period (LMP). Because maternally recalled LMP is imprecise in this population (24), a restricted analysis of delivery outcomes alone was used for this subgroup. At ANC booking, a 5-mL blood sample was drawn and analyzed for hemoglobin concentration, presence of malaria parasites, and syphilis serology. As required by the Gambian Government Ethics Committee, each subject was offered serologic testing for HIV, with pre- and posttest counseling. HIV-positive participants were referred to the MRC HIV clinical service in Fajara. Women who delivered before attendance at the HIV clinic were offered maternal and infant nevirapine prophylaxis, in keeping with the national guideline at the time. An uncuffed 10-mL venous blood sample was drawn at the 18–22-wk and 28–32-wk review. Aliquots were prepared, stored, and analyzed within 1 h of venipuncture, as described for recruitment samples, except for ferritin concentration, which was not measured at 18–22 wk. Additional aliquots of chilled citrate-plasma aliquot and serum were analyzed for PAI-1 and PAI-2 concentrations and placental hormones [human prolactin (hPL) and human chorionic gonadotropin (hCG)], respectively. The subject was given directly observed intermittent preventative treatment with sulfadoxine-pyrimethamine at 18–22 wk and 28–32 wk. TBAs reported impending deliveries to their local resident fieldworker. A 10-mL cord blood sample taken from a large vein on the fetal side of the placenta of live-born infants. The placenta was transported as soon as it was collected to MRC Keneba, cleaned, the membranes trimmed, and the cord cut close to its insertion. Placental weight was recorded to the nearest 10 g by using a digital balance. Cord serum and whole-blood aliquots were prepared. The concentration of cord serum MMA was measured. Whole blood was analyzed for hematologic variables and the presence of malaria parasites. Within 72 h of delivery, an NM examined the mother and infant. Maternal weight, MUAC, urinalysis, and blood pressure were recorded. An uncuffed 10-mL venous blood sample was drawn from the mother and aliquots prepared and analyzed as for the cord samples. Neonatal anthropometric measurements were recorded. Dubowitz scoring was conducted to improve the detection of preterm infants, as part of standard clinical risk assessment. Maternal height, weight, MUAC, and blood pressure were measured by pairs of fieldworkers at recruitment and by 1 of 2 NMs at scheduled antenatal visits and at delivery. Standard techniques were used for each measurement. Weight was measured (to the nearest 0.1 kg) by using daily calibrated, digital scales (Tanita Corporation). Height (to the nearest 0.1 cm) was measured by using a daily calibrated stadiometer (Leicester height measure; Seca) and BMI was estimated [weight (kg)/height (m)2]. Underweight was defined as a BMI (in kg/m2) <18.5. MUAC (to the nearest mm; left arm) was measured with graduated tapes (Henley Medical Supplies). Triplicate blood pressure was measured (automated Omron 705IT device; Omron) and measurements replicated to within 5 mm Hg. Hypertension was defined as systolic blood pressure >139 mm Hg or diastolic blood pressure >89 mm Hg. Neonatal anthropometric measurements were performed by 1 of 2 NMs. Birth weights were measured (to the nearest 20 g) with sling and portable spring balances (CMS Weighing Equipment), and these were regularly checked with standard weights. Birth length (to the nearest 5 mm) was measured by using neonatal length mats (TALC Teaching Aids). Head circumference (to the nearest mm) was measured with graduated tapes (Henley Medical Supplies). Fieldworkers and NMs were trained and cross-compared in anthropometric techniques as part of their employment induction with the MRC, which was repeated at the start of this study. Refresher training was undertaken regularly. Hemoglobin measurements at pregnancy booking used a hemoglobinometer (HemoCue B). Analyses on venipuncture samples included hemoglobin concentration, white blood cell count, mean cell volume, mean cell hemoglobin, and reticulocyte count (Cel-Dyn 3700 Analyzer; Abbott Diagnostics). Anemia was defined as hemoglobin <12 g/dL in nonpregnant women and <11 g/dL in pregnant women. Malaria blood films were prepared and stained with Giemsa stain, and 100 high-power microscopic fields were examined to determine parasite count against 200 white blood cells. PAI-1 and PAI-2 antigen concentrations were measured with mouse-monoclonal antibody–based ELISAs (Elitest; Hyphen-BioMed) as was MMA in maternal and fetal blood (IBL International), hPL, and hCG (Addenbrookes Hospital). Vitamins A and E were measured by HPLC (25). Vitamin A deficiency was defined as a serum retinol concentration <0.7 μmol/L and vitamin E deficiency as a serum α-tocopherol concentration <12 μmol/L. Ferritin was measured by instrumental immunoassay (Dimension Xp; Siemens). Iron deficiency was defined as a plasma ferritin concentration <15 μg/L. Obstetric sonography was conducted by one of the authors (SO) who had certified training in standard methods (25). The ultrasound machine was an ACUSON Antares with a CH6-2 (5.71-MHz) transducer (Siemens). Gestational age was derived from sonographic measurement of the fetal crown-rump length or biparietal diameter. At the 18–22-wk and 28–32-wk review, fetal biometry and utero-placental blood flow were assessed by sonography and Doppler velocimetry of the right and left uterine arteries (UtAs), umbilical artery, and fetal middle cerebral artery (MCA) waveforms. The resistance and pulsatility indexes were calculated for each vessel and for the peak systolic velocity (PSV) of the MCA. The presence of diastolic “notching” in either of the UtAs was recorded. A mean UtARI >0.55 and bilateral notching, or a mean UtARI >0.65 and unilateral notching, were defined as “high resistance” waveforms (26). Women with high-resistance waveforms were monitored for pre-eclampsia and/or FGR and, if necessary, referred to RVTH. Intraobserver variation in sonographic variables was not estimated. We estimated that a sample of 200 women in each arm of the trial would generate 90% power to detect a 20% difference in the mean PAI-1 to PAI-2 ratio at 18–22 wk of gestation between supplementation groups at a 5% level of significance. This was based on variance parameters (population mean ratio: 0.40; SD: 0.23) drawn from a trial of antioxidant vitamin supplementation in British women at risk of pre-eclampsia (27) and allowed for 10% loss to follow-up. This sample size had 90% power to detect a change in mean UtARI equivalent to 0.35 SDs at the 5% significance level [assuming a mean UtARI of 0.30; SD: 0.01 (26)]. To detect a difference in mean MMA transplacental transfer ratio of 0.1, with 90% power at the 5% level of significance, ∼80 mother-infant pairs were required in each arm of the trial, assuming a mean transfer ratio of 1.01 and an SD of 0.18 (28). It was considered that this sample size would be readily available within the main pregnancy cohort. On the basis of MRC Keneba demographic survey data, we assumed that 16% of women aged between 17 and 45 y in this community would conceive in any given year. This suggested that at least 2400 women would be required over the course of the study to satisfy the largest estimate of required sample size, 400 pregnancies. We identified 3206 eligible subjects in the survey. Data were double-entered into a computerized database (MS-Access; Microsoft) and verified within 48 h of collection. Scheduled data validation checks were made. A copy of the raw data set was maintained with the MRC data management team. All available data on singleton pregnancies without fetal anomalies were included and analyzed according to the original randomization. Every effort was made to complete the data set for each subject, but specific outcomes on individual subjects were occasionally unavailable (e.g., on women who had withdrawn their consent, migrated out of the study area, or otherwise failed to attend within the predefined timeframe of a particular endpoint). In these cases, the remaining data were included in all other analyses. Continuous data were analyzed by using linear regression, assuming equal variance. Outcomes were regressed on a single predictor, trial arm. To improve the precision of primary analyses, multiple regression was used to adjust for the effects of gestational age at point of sampling, because gestational age is an a priori predictor of all pregnancy outcomes. Evidence for interaction between treatment effect and gestational age at point of sampling was also sought for primary outcome measures. Continuous data that were not normally distributed were log-transformed before analyses of their geometric means. Those that were not rendered normally distributed by this transformation were analyzed by using nonparametric tests. Binary responses were analyzed by logistic regression. Data collected on the same individuals at ≥2 time points were analyzed by using random-effects models fitted by generalized least squares. Although the study was not powered for subgroup analyses, we went on to assess treatment effects on the primary outcomes by season of conception (wet-hungry compared with dry-harvest) by fitting a season × treatment interaction term into the models. This stratification was justified as an exploratory analysis on the basis of previous data showing that PTB and small-for-gestational-age birth show strongly divergent patterns of seasonality in Kiang West (18). Significance was set at P < 0.05. Statistical analyses were carried out by using Stata 9.1 (StataCorp). All women and their infants were offered free routine clinical services, including infant vaccinations. Transport was provided by MRC ambulance. Malaria parasitemia was treated with chloroquine or quinine and sulfadoxine-pyrimethamine in accordance with national policy. Pre-eclampsia was defined by a gestational systolic blood pressure >140 mm Hg or a diastolic blood pressure >90 mm Hg in a previously normotensive woman, in conjunction with dipstick proteinuria. Women with pre-eclampsia were referred to RVTH for further management. This trial was approved by the Scientific Coordinating Committee of MRC Laboratories, The Gambia, and by the MRC/Gambian Government Ethics Committee (L2005.111v2 SCC 1000). An independent trial monitor and data safety and monitoring board assessed trial activity at regular intervals, under the auspices of Good Clinical Practice guidelines (29). MRC Keneba offers free primary health care in collaboration with the Gambian Government Lower River Divisional Health Team. Apart from the structured clinical contacts outlined (during which subjects were given transport and a meal), no additional benefits were provided to trial participants. The trial was registered as ISRCTN 13687662 (www.controlled-trials.com/isrctn/pf/13687662).