The preconception study

The objective is to characterise the health of women from a rural region in Tanzania, before conception. We will examine the prevalence and risk factors of anaemia (including malaria, micronutrient deficiencies and socioeconomic determinants) as well as of hypertension, diabetes and fertility.

The pregnancy study

The objective is to evaluate the effect of anaemia on fetal and placental development. We will evaluate how first-trimester and second-trimester compared with third-trimester anaemia and/or malaria alters fetal and placental development and later neonatal body composition. Placental development will be evaluated as degree of villous branching. Finally, we will characterise how first-trimester, second-trimester and third-trimester anaemia and/or malaria may change the VEGF-A/PlGF balance and the IGF-axis.

The -omics study

The objective of the -omics studies are to explore genetic variation (maternal and fetal), genome-wide DNA methylation and consequent RNA expression in cord blood and placenta from pregnancies affected by anaemia and/or malaria compared with controls. Epigenetic variation established in utero has indeed emerged as a candidate mediator of long-term programming of NCDs.28 29 Characterisation of epigenetic changes after exposure to anaemia may potentially enable us to detect biomarkers to identify groups at risk of developing NCDs.

In this paper we present the overall set-up of the interdisciplinary study, including selection and baseline characterisation of the participating population.

Study setting

The field studies were conducted in Korogwe, Tanga Region, North-Eastern Tanzania (figure 1). The central field site was the Maternity Ward and the Reproductive and Child Health Clinic at Magunga Korogwe District Hospital (KDH). Mobile clinics were set up in 48 villages in the two districts, and 17 dispensaries (including Kwandolwa, Kerenge, Segera, Mgombezi, Makuyuni and Mbaghai) functioned as outreach satellite sites supporting the antenatal care provided by the study team (figure 1). Clinical investigations and sample collection took place at KDH, the dispensaries and mobile clinics in the catchment areas, whereas all follow-up activities during pregnancy were performed at KDH.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Map of Korogwe district and the dispensaries in the study.

Study groups and recruitment

Non-pregnant women were invited to participate in the preconception study. Sensitisation campaigns were held to ensure a sufficient number of participants. A total of 2629 women was screened for inclusion between July 2014 and December 2015. The inclusion criteria were based on a high likelihood of becoming pregnant within the study period and included: age between 18–40 years, a negative urine pregnancy test, not having tried to conceive for more than two consecutive years (regarded as subfertile), no usage of modern contraceptive methods except condoms, not having a child less than 9 months old, living in an accessible area, and on conception willing to attend antenatal care and give birth at KDH. The women were invited to participate in the pregnancy study if conception occurred between July 2014 and March 2016. To ensure early pregnancy inclusion, women were invited to report for pregnancy testing every third month or if they suspected they were pregnant.

The pregnancy study population was supplemented with screening of 445 women in early pregnancy from April 2015 to March 2016. These pregnant women were recruited on a 1:1 basis with a Hb ≤8 g/dL (severe anaemia) or Hb 8.1–10.9 g/dL (mild–moderate anaemia): Hb ≥11 g/dL (non-anaemic), and with a GA of ≤14 weeks based on ultrasound. Follow-up of pregnant women was completed in December 2016 when the last woman in the cohort gave birth.

Patient and public involvement

Patients or the public were not involved in the design of the study concept and protocol, but the research questions and outcome measures were developed considering the women included and disease patterns observed in our previous Strategies TO Prevent Pregnancy-Associated Malaria (STOPPAM) Study conducted in the same region.30 However, the logistic implementation of the present study including recruitment and retaining participants was conducted in close collaboration with the local communities and personnel in the governmental health systems.

Sample size calculations

Sample size calculation was originally based on the observed effect of severe anaemia (Hb ≤8 g/dL) in the second trimester on z-score of birth weight (−0.31) in the STOPPAM Study30 conducted in the same area, with a significance level of 0.05, power of 0.80 and assuming 15% lost to follow-up. With balanced inclusion into the study, we aimed to enrol 480 pregnant women in the first trimester on a 1:2 basis as the severe anaemia group (Hb ≤8 g/dL) versus a group that did not have severe anaemia. The latter group was further divided into two groups (8.1–10.9 g/dL and ≥11 g/dL). However, the percentage of women with severe anaemia was much lower than anticipated. Many of our aims are exploratory, not all depend on birth weight alteration, and the effect on placental development, vascularisation, the IGF-axis and epigenetic alterations might already be observed when moderate–mild anaemia is present, although the exact effect size is not known. It was therefore decided to continue with the study but by including the participants in a 1:1 ratio (Hb <11 g/dL vs Hb ≥11 g/dL). This might have resulted in too little power to detect the actual difference in birth weight, but still helps achieve the other objectives.

In addition, we aimed to include 1500 non-pregnant women irrespective of Hb levels, since it is unknown whether the effect of anaemia on developmental outcomes, including epigenetics, is continuous. Assuming that 30% would become pregnant within the project time line, and that 60% would be identified, we expected to include at least 270 pregnant women.

Socioeconomic status, medical history and clinical examination

The following characteristics were documented at inclusion: ethnicity, socioeconomic status (based on housing, education and occupation), religion, partner characteristics (age, ethnicity, education and occupation), contraceptive methods, gravidity history, medical history and obstetric history, substance abuse, anthropometry, blood pressure (BP), Hb, diabetes, malaria and HIV. Intestinal helminths were tested in a subgroup of 486 women. Women with an average systolic BP ≥140 mm Hg and/or diastolic BP ≥90 mm Hg were scheduled for a follow-up visit to confirm BP status. Diabetes was screened for by measurement of blood glucose and HbA1c. If glucose levels were ≥11.1 mmol/L the woman was scheduled for a fasting blood glucose test.

During pregnancy, at all antenatal visits, the women were monitored for anthropometry, anaemia, infections and hypertensive disorders. Sickle cell trait, thalassaemia and glucose-6-phosphate dehydrogenase deficiency were examined, and an oral glucose tolerance test was done during pregnancy at 32–34 weeks.

All women were offered daily iron and folic acid supplementation from the time pregnancy was recognised and until delivery (combination tablet of 200 mg ferrous sulfate [~43 mg elemental iron] and 400 µg folate per day [Ferrolic–LF, Laboratory and Allied, Mombasa, Kenya]). If anaemia was diagnosed, it was treated with increased supplementation with either two to three combination tablets of iron-folic acid per day or with Hemovit multivitamin syrup (200 mg ferrous sulfate, 0.5 mg B₆, 50 µg B₁₂, 1.5 mg folic acid and 2.33 mg zinc sulfate per 5 mL [Shelys Pharmaceuticals, Dar es Salaam, Tanzania]) in a dose of 10 mL two to three times daily depending on severity.

Blood sample collection

Venous blood was collected in EDTA and serum tubes at the time of inclusion, all antenatal visits, delivery, neonatal visits if cord blood was missed and paternal visits (EDTA tube only). Serum tubes were kept light-protected and all blood samples were kept at 2°C–8°C until processed at KDH within 2 hours of collection.

Hb was measured at the time of inclusion, all antenatal visits and delivery. Bilirubin, ferritin, folic acid, vitamin B₁₂, alanine aminotransferase, C reactive protein and albumin were measured at the time of inclusion, 32–34 weeks of gestation and delivery. Sampling of placental development markers (VEGF-A, Soluble fms-like tyrosine kinase-1 [sFlt-1], PlGF, Pregnancy‐associated plasma protein A [PAPP-A]) was performed at inclusion into the pregnancy study if before GA 11 weeks, at 11–14 weeks, 20–22 weeks and 26–28 weeks of gestation and delivery. Blood sampling for fetal growth markers (IGF-1, IGF-2, Insulin Like Growth Factor Binding Protein 1 [IGFBP1]) was performed at inclusion into the pregnancy study, 11–14 weeks, 26–28 weeks and 32–34 weeks of gestation and delivery in both venous and cord blood.

Ultrasound

Serial, transabdominal ultrasound was used to estimate GA using crown-rump length in the first trimester and head circumference in the second trimester.31 32 If included before week 11 of pregnancy, the GA was estimated again between weeks 11–14. Ultrasound for fetal weight was performed at GA of 20–22 weeks, 26–28 weeks, 32–34 weeks and 37–39 weeks of pregnancy. Additional ultrasound was done if intrauterine growth retardation was suspected and biweekly until delivery for all women diagnosed with gestational diabetes. The Hadlock formula on head circumference, abdominal circumference and femur length was used for weight estimation.33 Placental blood flow was estimated using both uterine and umbilical artery blood flow with pulsatile and resistance index as well as the systolic-diastolic ratio each time fetal weight was estimated. Uterine blood flow was also evaluated at weeks 11–14 and diastolic notching noted at each investigation.34 35

Newborn examination

At the time of delivery and at 4–6 weeks postnatally, weight, length, head circumference, abdominal circumference, mid upper arm circumference, skinfold thickness of biceps, triceps, subscapular and thigh of the newborn were measured. Newborn assessment was performed within 24 hours, and, if delayed due to home delivery, as soon as possible. Z-scores were estimated using a reference chart produced for Tanzania in the STOPPAM Study and the Intergrowth-21st reference charts.30 36–38

Cord blood and placenta samples for genetic analyses

Cord blood samples were collected in EDTA, serum and PAXgene tubes, primarily before the placenta was born. The placenta was processed within 15–20 min after delivery at the hospital, and for home deliveries within 2 hours, when possible. EDTA tubes were placed cold directly after collection and processed within 2 hours, including collection of buffy coat. PAXgene tubes were stored at room temperature for 2 hours, and subsequently at −20°C for 24–72 hours, before storage at −80°C.

The cord and amnion were removed, and the placenta was weighed. Using a biopsy punch, biopsies were collected from each quadrant, from the cord insertion and at the edge of the placenta at the area furthest from cord insertion. Each biopsy was washed with saline solution and divided into the maternal and fetal side. A ~1 cm³ cord tissue sample was collected from the cord end closest to insertion. All biopsies were snap frozen in liquid nitrogen and stored at −80°C. The placentas were macroscopically examined for signs of disrupted development. Membranes, cord, placental size, thickness, shape and parenchyma infarcts were described. Pictures of each placenta were taken to document morphology.

Genome-wide DNA methylation and RNA expression will be performed in cord blood and placenta biopsies using Illumina Infinium MethylationEPIC arrays and RNA sequencing on Illumina Nextseq. Genome-wide association studies will also be performed in both cord blood and maternal blood using CoreExome/OmniExpressExome Illumina arrays.

Stereological placenta samples

Ten full-depth placental tissue blocks were collected using systematic uniform random sampling and vertical sectioning, and conserved in 10% formalin. Isotropic uniform random sampling and Mattfeldt orientator was used to collect slides for paraffin embedding.39 Information of the sample position was collected regarding whether the sample block was from the placental periphery or centre. The slides were orientated so the vertical section contained both the maternal and fetal sides. The tissue was immunohistochemically stained with CD34 to visualise the vessels. Stereological analysis will be applied to assess the placental vascular morphology in placentae from women with or without anaemia and/or malaria during pregnancy. The total villi volume and vessel length, surface and diameter of transport and diffusion villi, will be examined.

Legal and ethical aspects

Explanation about the study was given to village leaders and community members before women were invited. After giving oral information, written informed consent (or thumbprints of illiterate women) was obtained prior to enrolment. All procedures were conducted in accordance with the Declaration of Helsinki. All participants were treated according to Tanzanian national guidelines. Antenatal and perinatal care followed guidelines from The National Road Map Strategic Plan.40 The project assisted all participants in obtaining the best local medical care available if disease was diagnosed.

Patients and public society were not involved in the design of this study.

Statistics

Assumptions of equal variance and normally distributed residuals will be visualised in Q-Q plots and histograms. For univariate analyses, normally distributed data will be presented as mean±SD and compared by Student’s t-test, whereas non-normally distributed data will be presented as median and IQR and compared by the Mann-Whitney U test. Proportions of categorical data will be compared using χ² test. For multivariate analyses on risk factors for poor preconceptional health, and analyses on the effect of anaemia and malaria on fetal and stereological outcomes, multivariate regression analysis will be used. If appropriate, mixed regression modelling will be used on repeated outcome measures including fetal growth trajectories and biomarkers of placental and fetal development. For epigenetic analyses, appropriate bioinformatic analyses will be used taking into consideration the false discovery rates.

Preconception study characteristics

Of the 2629 women screened, 1415 were included in the preconception study (figure 3). Main causes for exclusion were use of modern contraception methods (27%) and age eligibility (26%). Among the 1415 women, 72 did not have venous blood collected at preconception, 34 had an ultrasound dated GA that showed they were already pregnant at time of inclusion (although very early in pregnancy) and 50 never had GA estimated with ultrasound. Additionally, 11 women were later found not to meet all inclusion criteria, leaving 1248 truly preconception women to be included in the final analyses (figure 3 and table 2). Average age, body mass index and BP of the 1248 preconception women were 28 years, 23 kg/m² and 117/75 mm Hg, respectively (table 2). In total, 458 (36.7%) women had anaemia (Hb <12 g/dL) and 34 (3.6%) women were diagnosed as HIV-positive (table 2). Soil transmitted helminths were detected in 13 (2.7%) women.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Overview of study cohort. Number of women screened, included and excluded in the preconception and pregnancy study. ¤: venous blood samples were not collected for 72 women at time of inclusion and were therefore excluded in the final analyses of the preconception study. #: 50 women were excluded because they never had GA estimation with US performed (this included 4 of the 6 women who refused inclusion into the pregnancy study, the 16 women who had an ongoing miscarriage when found pregnant, 18 women who had a miscarriage after inclusion into the pregnancy study and the 12 women who had a miscarriage in between preconception follow-up visits). *34 women were excluded because the US dated GA showed that they were already pregnant at the time of inclusion. GA, gestational age; Hb, haemoglobin; US, ultrasound.

Pregnancy study characteristics

A total of 2995 third monthly visits were conducted for the 1415 women. During these follow-up visits, additional 134 women were excluded for various reasons (figure 3). In total 417 women were later found pregnant and 383 women were followed during pregnancy (figure 3). Further, 155 women were included in the pregnancy study within the first 14 weeks of pregnancy, in a ratio of 1:1 (mild, moderate or severe anaemia versus non-anaemic). Of the 155 women, 7 had severe anaemia (Hb ≤8 g/dL), 71 had mild–moderate anaemia (Hb 8.1–10.9 g/dL) and 77 were non-anaemic (Hb ≥11 g/dL). In total, 538 women were included in the pregnancy study (figure 3).

During pregnancy, 359 (66.7%) women had anaemia (Hb <11 g/dL) of which 85 (15.8%) women had moderate-to-severe anaemia (Hb ≤9 g/dL) and 33 (6.3%) had severe anaemia (Hb ≤8 g/dL). In total, 185 (34.4%) women were diagnosed with malaria and 17 (3.4%) women were diagnosed with HIV during pregnancy. In total, 147 (27.3%) women were non-anaemic and never had malaria throughout pregnancy (table 3).

Birth study characteristics

Of the 538 pregnancies (figure 3), 80 women had a miscarriage (16%), of which 72 were early (≤14 weeks) and 8 were late (>14 weeks, ≤22 weeks)—5 without and 3 with pregnancy outcome data collected at the time of miscarriage (figure 3 and table 3). The birth cohort consisted of 427 women (79%) having given birth to 438 neonates including the 3 women who had a late miscarriage at the hospital and complete outcome data collected.

For the 424 women who had given birth after week 22 of gestation, the average GA was 280 days, and 27/424 (6.4%) women delivered preterm (>22 weeks, <37 weeks). Average birth weight was 3016 g and 10.3% were born with LBW (<2500 g) among singleton live births without congenital anomalies. Twelve deliveries were stillbirths. Neonatal data were collected on 421 of the 426 live births (99%) (table 3 and figure 3). Among the 421 neonates with follow-up data, 13 neonates did not survive until day 28. For the remaining 408, neonatal anthropometric measurements were documented for 400 (98%) (table 3 and figure 3).

Only women with data on time of delivery/late miscarriage were considered potential for paternal follow-up, which was performed for 113 fathers (table 3 and figure 3).