Enhancing Prediction of Adverse Pregnancy Outcomes in Africa Through Partnerships and Innovation

Background Despite a significant global reduction in maternal and child deaths over the last decade, several key areas are still woefully neglected. The burden of maternal mortality remains unacceptably high in Sub Saharan Africa (SSA) where 66% of all global maternal deaths occur and the maternal mortality ratio is \~500/100,000 live births. Over 80% of preterm births (\~12 million) occur in SSA and Asia and nearly 3 million neonatal deaths occur globally each year, accounting for more than half all under-five child mortality . There is also a silent global epidemic of stillbirths with 2.6 million occurring every year. The risk of perinatal death is approximately 8 times higher in low-and-middle-income countries (LMIC), particularly those of SSA, than in high-income-countries (HIC). The Great Obstetrical Syndromes (GOS) collectively are a major contributor to maternal and perinatal morbidity and mortality. GOS are pregnancy disorders caused by defective placental development and include pre-eclampsia, fetal growth restriction (FGR), unexplained stillbirth and spontaneous preterm birth. Due to this common mechanistic pathway, these disorders frequently co-exist: FGR is present in \~50% of stillbirths and commonly accompanies pre-eclampsia. Placental maldevelopment in GOS leads to stress later in gestation resulting in a systemic imbalance of soluble fms-like tyrosine kinase 1 (sFLT-1) and Placental growth factor (PlGF). These are potential markers to predict, diagnose and follow disease in GOS.

Attempts to identify pregnant women at risk of these adverse outcomes have had little success. Those described have mainly been identified by opportunistic secondary analyses of data collected for other reasons. To address this challenge, prospective studies have been conducted mainly in HIC to develop better screening tests. However, pregnancy cohort studies in HIC do not adequately represent the very different biological and socio-economic conditions found in SSA. There are some prospective cohorts in SSA to identify screening tests/predictors for adverse pregnancy outcomes but majority of these are ongoing in and others are inconclusive. The WHO Alliance for Maternal and Newborn Health Improvement (AMANHI) Study is a large cohort of over 10,000 pregnant women and their babies in SSA and South Asia aimed at studying the interactions between genes and a wide range of varying environmental exposures on key pregnancy and birth outcomes. Prospective pregnancy cohorts in SSA will shed light on determinants, mechanisms and solutions for the excess burden of maternal and perinatal morbidity and mortality. Therefore, the investigators aim is to investigate the clinical, biological and social antecedents of adverse pregnancy outcomes in prospective cohorts of Ugandan women, with a focus on the GOS.

Work which has led up to the project As an academic obstetrician working at Makerere University, the PI's main research interest has been the enigmatic disorder of pregnancy, pre-eclampsia, which is particularly common in African women. For the PI's Wellcome-funded PhD project, the PI undertook a genetic study of immune system genes: "Killer Immunoglobulin-like receptor (KIR) and Human Leucocyte Antigen-C (HLA-C): a case-control study of pre-eclampsia". Participants were recruited in Mulago Hospital, the teaching hospital for Makerere University. This was the first genetic case-control study on pre-eclampsia among indigenous Africans despite African ancestry being a predisposing factor to pre-eclampsia. This study revealed interesting genetic similarities but also differences from comparable studies performed in Europeans. A genetic region marked by the KIR2DS5*006 gene was associated with protection against pre-eclampsia among Ugandan women; this region is unique to populations in SSA. Follow-up work has revealed that unique African KIR2DS5 allotypes act as activating receptors for the C2 HLA-C ligand that is expressed by the fetal placenta. For the PI's post-doctoral work, the PI recruited \~2000 more pre-eclampsia cases and similar control numbers to continue this genetic study in collaboration with scientists in Cambridge, Denver and Stanford. The analysis is currently ongoing and the genetic and clinical results will contribute to understanding the fundamental mechanisms that underpin pre-eclampsia. Importantly, this work has generated a valuable resource of DNA to study other genetic variants associated with pre-eclampsia in African women in the future.

Despite the insights provided by these studies, their retrospective nature means that there is limited potential to translate knowledge gained into clinically useful predictive tests. Through the PI's ongoing Future Leaders-African Independent Research (FLAIR) fellowship, awarded by the Royal Society and African Academy of Sciences, the PI is beginning to address this challenge by prospectively collecting data and biological samples from a cohort of 1,500 pregnant women from early pregnancy at Kawempe Hospital. The main aim is to determine whether women who go on to experience pre-eclampsia can be identified early in pregnancy. The PI's mentor on this project, Professor Gordon Smith, has successfully performed a similar study in Cambridge and the PI is aligning the PI's protocols with his Pregnancy Outcome Prediction Study (POPS). This approach has the advantage of allowing direct comparisons between the two very different clinical settings and study populations.

Although the ongoing pregnancy cohort study has brought a new dimension to the PI's research programme, some challenges remain. Firstly, due to limited resources the study only focuses on pre-eclampsia and not the other GOS with which pre-eclampsia frequently coexists due to the common underlying aetiology of defective placentation. Studying all these conditions together provides leverage in identifying the determinants, mechanisms and solutions for the excessive maternal and perinatal morbidity and mortality and stillbirths in SSA. Secondly, similar prospective pregnancy cohorts recruited in early pregnancy are rare in SSA despite the excessive maternal and neonatal morbidity and mortality here. Thirdly, this is a personal fellowship so opportunities for training and mentoring younger scientists are limited.

The proposed project will address these challenges by prospectively collecting data and biological samples from a larger cohort of pregnant women from early pregnancy at Kawempe and Mulago Hospitals. The aim is to determine whether women who go on to experience these complications (pre-eclampsia, stillbirth, FGR, preterm birth) can be identified early in pregnancy. The advantage of performing this study in these two hospitals is the large numbers of patients delivering there with severe complications (death, HELLP syndrome, eclampsia, peripartum cardiomyopathy, renal failure, intracranial haemorrhage, pulmonary oedema). Investigators will also incorporate multidisciplinary training and mentoring opportunities as well as focussing on the regional and global partnerships already established in maternal and newborn health. An innovative aspect will be the use of artificial intelligence in data analysis. While genetic and obstetric/clinical factors can be bioinformatically and experimentally determined, socio-demographic factors are less tractable and may require elaborate data models to quantify. AI and machine learning techniques present a great opportunity to both identify and calibrate socio-demographic parameters important in the aetiology of GOS. Using AI techniques, investigators will seek to calibrate the intersection between socio-epidemiological and clinical/biological predictors of GOS.

The investigators intermediate goal is to develop better methods of screening in SSA with the aim of targeting interventions for women at high risk of pre-eclampsia and other GOS. Characterising the diagnostic effectiveness of candidate screening tests is an essential prerequisite for a randomised clinical trial (RCT) of screening and intervention. Hence, the proposed study as is seen an essential first step for the ultimate goal: to determine targets for therapeutic interventions (biological or sociodemographic) which will be tested with an RCT.

Statement of the problem Pre-eclampsia, as well as other GOS, is a major cause of to maternal and perinatal morbidity and mortality, especially in SSA. The cause of the GOS is still unclear which makes prevention difficult and so mainstay of treatment is early diagnosis and management in order to minimize complications to the mother and her newborn. However, early diagnosis and treatment is difficult in SSA partly because of late presentation and health system inadequacies which cause delay in provision of treatment. There is therefore need to identify women at high risk of developing the GOS so that these are prioritized for close follow up in the antenatal clinic and in labour in order to improve outcomes for both mother and baby.

Most studies conducted in sub-Saharan Africa (SSA) to investigate the risk of adverse pregnancy outcomes have been case-control in nature. There are very few cohorts recruited from early in pregnancy in a SSA population, who are most at risk of maternal and newborn mortality and severe morbidity. Despite the insights provided by case-control studies, these fail to provide insights into events and biomarkers present in early pregnancy that determine and predict GOS- when preventive interventions could best be implemented. To address this challenge, prospective studies have been conducted in HIC to develop screening tests that perform better. However, pregnancy cohort studies in HIC do not adequately represent biological and socio-economic conditions found in SSA.

Therefore, in this project the PI aims to investigate the clinical, biological and social antecedents of pre-eclampsia and other GOS in a prospective cohort of Ugandan women. Data generated from this prospective pregnancy cohort in SSA would have the potential to shed light on determinants, mechanisms and solutions for the excess burden of maternal and perinatal morbidity and mortality in the region.

Research Questions

1. What exposures (epidemiological, social, demographic, clinical, laboratory) linked to the different adverse pregnancy outcomes - pre-eclampsia, stillbirth, preterm birth and fetal growth restriction?
2. What is the positive predictive value of screening using biomarkers (soluble Flt-1 and placental growth factor) and serial ultrasound for the different adverse pregnancy outcomes in (1)?
3. What is the intersection between socio-epidemiological, clinical and laboratory parameters of the GOS using artificial intelligence?

General aim of the research To enhance prediction of adverse pregnancy outcomes in Africa through partnerships and innovation

Specific objectives

1. To determine the exposures (epidemiological, social, demographic, clinical, laboratory) linked to the different outcomes - pre-eclampsia, stillbirth, preterm birth and fetal growth restriction.
2. To determine the positive predictive value of screening using biomarkers (soluble Flt-1 and placental growth factor) and serial ultrasound for the different adverse pregnancy outcomes in (1) above.
3. To generate a resource of data and biological samples to facilitate hypothesis driven and discovery based approaches to identifying novel predictors and/or mechanisms of pre-eclampsia and the other GOS.
4. Calibrate the intersection between socio-epidemiological, clinical and laboratory parameters of the GOS using artificial intelligence.

Significance of the Study and impact The study will update clinical-epidemiological risk factors for pre-eclampsia and other GOS in SSA, allowing women at risk to be selected for close pre-natal follow up. The biomarkers will be translated into point of care diagnostics. The high numbers of pathological pregnancies will facilitate rapid development of high-performance algorithms (based on clinical-epidemiological risk factors) and tests (based on biomarker identification and validation) through RCTs of screening/interventions. Importantly, although the performance of tests will be context-specific to SSA, the results will inform approaches for HIC. Indeed, the investigators clinical setting allows study of adverse pregnancy outcomes that are rare in HIC and yet of global health importance, such as stillbirth and complications of severe pre-eclampsia. Indeed, these occur more commonly in women with recent African ancestry in USA and elsewhere. This study will establish a much-needed resource describing in detail, from early pregnancy, the course and outcomes of pregnancy in women of SSA. The investigators are aware of very few studies in SSA that provide meticulous phenotyping from early pregnancy to term. To achieve this, the investigators study will use serial ultrasonography to allow accurate dating and assessment of fetal growth, complemented by rigorous collection of biochemical and clinical data. Biological samples will provide a resource to identify African-specific mechanisms of disease and may identify useful novel biomarkers for adverse pregnancy outcomes caused by underlying placental dysfunction. Longer-term studies from the established cohort will be planned for the future such as those for predictors of future health of the mothers and their offspring following adverse pregnancy outcomes.

Study setting The study will be conducted at Kawempe National Referral Hospital (KNRH) and Mulago Specialised Women and Neonatal Hospital (MSWNH), some of the main teaching hospitals for Makerere University. The two hospitals are located in Kawempe Division of Kampala district and together have over 28,000 deliveries a year (60-100 deliveries per day). KNRH offers outpatient antenatal care services from Tuesday to Thursday to women with both high risk and low risk pregnancies. On average 200 women are seen per antenatal clinic day and of these 100 women are first-time attendees and \~50% of women present at a gestational age of less than 18 weeks. Approximately 25 nulliparous women attend the clinic in the first trimester on an antenatal clinic day. MSWNH offers outpatient antenatal care services from Monday to Friday to women with both high risk and low risk pregnancies. On average 50 women are seen per antenatal clinic day and of these 30 women are first-time attendees and \~50% of women present at a gestational age of less than 18 weeks. Approximately 10 nulliparous women attend the clinic in the first trimester on an antenatal clinic day. This setting allows getting sufficient numbers of pre-eclampsia adverse outcomes that are rare in HIC, including eclampsia and stillbirth. At both hospitals, all women found to have hypertension are sent to the labour ward for further care which may include delivery. After delivery all women, including normotensives, are admitted to the postnatal ward from where they are subsequently discharged. Upon discharge women are advised to return to the postnatal clinic at 6 weeks or earlier in case of any concerns or complications. Women who have had complications such as pre-eclampsia, postpartum haemorrhage and obstructed labour are usually advised to return to the postnatal clinic after a week. Most of the services at the KNRH are free but some of the paid for services are ultrasound scan, CT scan and X-ray. Most of the services at the MSWNH are paid including ultrasound scan, but for the study related procedures these costs will be incurred by the project. The research tests (biomarkers) will be done at the Translational Lab at Mulago (which is a collaboration between the Department of Obstetrics and Gynaecology and Infectious Diseases Institute). The baseline (routine) lab tests will be carried out by a commercial lab that will be selected through the Makerere University procurement process.

Study design: This will be a prospective cohort study.

Study endpoints

The study outcomes are:

Primary: pre-eclampsia, stillbirth, FGR, spontaneous preterm labour Secondary: Conditions associated with severe pre-eclampsia: eclampsia, HELLP syndrome, renal failure, pulmonary edema, intracranial haemorrhage, peripartum cardiomyopathy, maternal death, neonatal death prior to discharge, postpartum haemorrhage.

Case definition for pre-eclampsia This will be according to the international definition.

1. New onset hypertension of 140/90 mmHg or more, on more than one occasion at least 4 hours apart
2. At 20 weeks of gestation or more assessed by ultrasound scan
3. Proteinuria of +1 or more by dipstick on at least one occasion

Case definition for spontaneous preterm labour This will be according to the international definition. Preterm labour will be defined as labor or delivery that occurs before 37 completed weeks of gestation but after viability (28 weeks of gestation for the investigators study). Spontaneous preterm labour will refer to preterm labour which is not induced due to medical indications or otherwise.

Case definition for fetal growth restriction (FGR) This will be according to the international definition. FGR will be defined as babies with birth weights less than the 10th percentile for gestational age.

Case definition for stillbirth A stillbirth is the death or loss of a baby before or during delivery. Sample size determination This will be a prospective cohort study conducted at Mulago and Kawempe Hospitals, with >25,000 deliveries/year. Daily antenatal clinics see \~150 new patients/day allowing recruitment of significant numbers of severe GOS outcomes rare in HIC, including eclampsia, cardiomyopathy and stillbirth. The study is strategically designed as a prospective cohort providing opportunities to investigate multiple potential exposures and outcomes. A sample size of 4,000 gives sufficiently precise confidence intervals for all adverse events up to an incidence of 0.1%.

Based on analysis of electronic medical records (EMR) data over a 12 month period, the numbers of subjects with the outcomes of interest (based on estimates from PROGRESS study in Table 1 and Table 3; 21,000 deliveries/year at Kawempe Hospital (unpublished denominator data from PREPARE study).

Study participants Study population Nulliparous pregnant women attending the antenatal clinics at Kawempe and Mulago Hospitals.

Sampling procedure and data collection:

All women identified as nulliparous from the antenatal register or antenatal charts will be approached and then screened for eligibility using a checklist. Recruitment into the study will take place 5 days a week and study participants will be consecutively recruited. Specifically, participation in the study will involve the following:

1. The screening for potential study participants will be done by the PI's research assistants and for those willing to participate in the study an obstetric ultrasound scan will be performed to determine the gestational age of the pregnancy and the viability of the fetus. To be eligible the gestational age should be below18 weeks and the fetus has to be alive. It is feasible to recruit 30 participants per week.
2. Information about the study will be given to all eligible participants and those who agree to participate will sign a consent form. After obtaining written informed consent a study number will be allocated to the enrolled participant and then demographic data (age, marital status, religion, ethnicity and that of parents, the ethnicity of spouse and his parents, address, and educational level) and clinical history will be collected using an interviewer-administered questionnaire. In addition, a physical examination will be performed on the participants and also recorded in the questionnaire. Data will be entered electronically.
3. At enrolment 15ml of blood will be taken from the participant. Seven (7) mls will be used to carry out tests that should be routinely done for all pregnant women and these include; complete blood count, renal function tests, liver function tests, malaria, HIV test and syphilis test. Eight (8) mls will be used to test for biomarkers (soluble Flt-1 and PlGF). In addition, 10 mls of urine will be taken for proteins, cells, casts and glucose.
4. Participants will be followed up at three time points in pregnancy (at 20, 28 and 36 weeks of gestation) and at the time of delivery. The visit windows are as follows (in weeks); 20 (19-22), 28 (26-30) and 36 (34-38). Investigators shall try to make these visits to coincide with regular antenatal care visits for participant's convenience. Participants will still be expected to attend any other visits scheduled by the hospital team that cares for them, outside the research related visits.
5. At each antenatal follow up visit, more clinical information and a physical examination will be performed (in particular blood pressure and urine protein measurement). Participants will also have further scans and blood sample collection of 10 mls which will be used to get serum and plasma to be used for analysis of biomarkers (soluble Flt-1 and PlGF) that predict pre-eclampsia. At the 20 weeks visit an additional 3mls of blood will be taken off and stored for DNA extraction in future. Separate consent will be sought for storage of this whole blood and the remaining serum and plasma (after the initial analysis), such that it can be used for future studies. For example, plasma can be subsequently processed for the analysis of micro particles, cell free DNA and cell free RNA.
6. The scan at 20 weeks will take place as normally and is routinely done to assess the fetal anatomy. Any findings from that scan will result in the same care as for non-study participants. Investigators will obtain some further measurements at the time of the scan (fetal biometry, Doppler flow velocimetry and amniotic fluid volume) but this will not affect the participants' care.
7. Participants will have research scans at 28 and 36 weeks. These scans are not normally part of routine care. Investigators will mask the results of scan measurements (fetal biometry, Doppler flow velocimetry and amniotic fluid volume) to clinicians and participants. Masking is justified as there is no evidence, even in high income settings, that revealing these results from non-clinically indicated scans is effective in reducing the risk of perinatal complications. However, investigators will reveal five scan diagnoses which will lead directly to changes in clinical management: (1) placenta praevia, (2) fetal anomaly, (3) severe oligohydramnios and (4) breech presentation at 36 weeks and (5) intrauterine fetal death (IUFD). In case of fetal anomaly or IUFD, the standard of care will be followed in delivering the news and managing the patient thereafter. This requires that the team of doctors and midwives managing the patient delivers the news and then offer counseling thereafter. In case of congenital anomalies that are not compatible with extrauterine life (eg anencephaly), the mother will be advised about termination of the pregnancy. Otherwise, she is counselled to carry the pregnancy till delivery. In situations of these adverse outcomes the investigators research team will work with the hospital team to beef up the ongoing counseling to the patient.
8. At delivery, research midwives will record details of labour, delivery and pregnancy outcome including data on duration of labour, mode of delivery and intrapartum complications. Routine data from the baby (weight, sex, Apgar score) will also be collected. They will take samples of the umbilical cord for DNA (a 2cm piece), 10mls of maternal blood for serum and plasma samples for a viral screen and samples of the placenta to make formalin-fixed blocks for histology of placentas and membranes. 10 mls of cord blood will be taken for isolation of serum. All samples taken off at the time of delivery will be used for future studies and additional consent will be sought from participants.
9. Study participants and their babies will be reviewed 6 weeks after delivery as a routine and not for research purposes.

There will be training on the study processes and measurements before the study begins and then monthly for the first 3 months and on quarterly basis thereafter. All devices will be calibrated and maintained as per local/manufacturers' requirements before the study begins and during the course of the study. To ensure reliability of measurements, clearly written down standard operating procedures (SOPs) will be developed. The SOPs will be signed off before the first participant is recruited. In case any abnormality is detected in the blood and urine tests during any of the study visits, participants will be referred for appropriate care.

To minimise loss to follow-up for study participants, investigators shall mainly rely on regular phone calls to the participants.

Ethical considerations:

Before the start of the research project, approval to conduct the study will be sought from the School of Medicine Research and Ethics Committee at Makerere University and finally the Uganda National Council for Science and Technology (UNCST). Permission to conduct the study will be sought from management of Kawempe National Referral Hospital and Mulago Specialized Women and Neonatal Hospital. The participants will give written informed consent to participate in the study. Participants will give additional written informed consent to sample storage for future studies and for re-contact during future studies. Investigators shall ask for participants' mobile phone numbers and those of their next of kin or friend and this is for purposes of easily contacting them in case there is need. These phone contacts will be only known to the researchers closely working on this study. Women diagnosed with pre-eclampsia or any other pregnancy complication will be managed according to the Kawempe and Mulago Hospital protocols.

Consent forms will be securely kept by the PI and separate from the data collection tools. Withdrawal from the study will not jeopardize health care and this will be provided free to all women. Participants will receive 8.33 USD (30,000 Ug Shs) as a modest transport refund at each follow up visit and 1.375 USD (5,000 Ug Shs) as compensation for their time at each follow-up visit. The compensation will be in kind as refreshments for participants.

Reporting of adverse events and disease related events When a maternal death occurs in the study it will be considered as a serious adverse event (SAE) and will be reported to the School of Medicine IRB within 7 days as is stipulated in the guidelines. The IRB form for recording, evaluating, and assessing causality of SAEs is provided in the appendices.

Data management Data will be entered electronically by the research assistants using tablets. Data cleaning will be carried out by the PI assisted by the study coordinator. The cleaned soft data will then be duplicated and stored in safe custody while the duplicate copy will be used for further analysis using STATA software and R. A confidential register of the details for each participant will be made, only accessible to the research assistants and the PI. It is only the study numbers that will appear on the participants' study documents except the consent forms. Consent forms will be securely kept by the PI and separate from the other data collection tools.

Data analysis:

The use of a prospective cohort study design allows multiple subsequent analytic approaches. The three main methods of analysis will be (1) analysis of the entire cohort, (2) analysis as case cohort studies, (3) analysis as nested case control studies. As examples, for exposures (objective 1) or biomarkers/ultrasound findings (objective 2) available for all participants, analysis of the entire cohort will be used and logistic regression analyses will identify epidemiological, clinical, and laboratory parameters predictive of pre-eclampsia or other GOS. Associations will be estimated using risk ratios and 95% confidence intervals.

In a case cohort design, all women experiencing the outcome of interest will be identified and a random sample of the cohort also obtained. All women who develop pre-eclampsia or any GOS will be compared to a random sample of the cohort who will have not developed that outcome of interest by the time of delivery. Investigators shall assess the relationship between the sFLT1:PlGF ratio with and without serial ultrasound data, and the risk of GOS. Analysis will be performed using logistic regression as for objective 1, but the data from the comparison group will be statistically weighted by the inverse of the sampling fraction to allow calculation of screening statistics, such as positive predictive value, sensitivity and specificity.

The nested case control approach will be used where assays are expensive or laborious, such as in studies comparing the placenta in cases of stillbirth, and when using DNA sequencing-based approaches to detect pathogens (viral, bacterial or eukaryotic).

Investigators will focus on AI techniques to extract and calibrate features, variables and risk factors from the socio-epidemiological data to be collected. The data will then be used in training AI-based model(s) that integrate socio-demographic and clinical/pregnancy outcomes on the one hand, with GOS outcomes on the other to simultaneously ascribe their dynamic effect sizes in varying circumstances. That calibration could inform early interventions through helping optimize service delivery models to improve the prevention, tracking, management and treatment of women with GOS. This would not only help quantify the relative contributions of biological versus socio-demographic factors to GOS but also bolster the implementation of the problematic last mile in the mitigation of the GOS.

Quality control Given the extensive experience of the PI in such studies, she will train and supervise the research assistants (midwives and lab technicians) to make sure that the study samples are collected within the stipulated time as indicated in the proposal. Standard operating procedures will be developed (SOPs) for the processes involved in the study such as blood pressure measurement, drawing of blood, placental sampling. Only accredited laboratories will be used to perform the desired tests.