Research Recommendations from the National Institutes of Health Workshop on Predicting, Preventing and Treating Preeclampsia

Introduction

Preeclampsia (PE) is a pregnancy-specific, multisystemic disorder that occurs in roughly 5% of pregnancies in the United States.^1,2 The rates of PE in the United States have been steadily rising over the past 30 years with PE being a leading cause of maternal and fetal/neonatal morbidity and mortality.³ Women with PE, as well as their offspring, are at greater risk for chronic diseases, including cardiovascular disease (CVD) later in life.^4,5 Despite being the leading cause of maternal death and a major contributor to maternal and perinatal morbidity, drug treatment to prevent PE is at best minimally effective, and current management therapies have significant limitations. In fact, at present, delivery is considered the only effective intervention for treating PE⁶ and even then PE may continue in the postpartum period or present de novo.⁷ The lack of progress in identifying new therapeutic targets for the treatment of PE is due in part to a paucity of animal models that address the heterogeneity of human PE as well as lack of understanding of basic disease mechanisms. There are no models of spontaneously developing PE in animals. Developing novel animal models to mimic human PE, identifying biomarkers to predict PE, conducting basic studies to better understand disease mechanisms, and ultimately, developing novel therapeutic strategies are clearly needed.

The National Heart, Lung, and Blood Institute (NHLBI), in collaboration with the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), held a workshop to examine the topic of Predicting, Preventing and Treating Preeclampsia on May 21–22, 2018, in Bethesda, MD. Speakers were chosen for expertise in general areas identified as important and cutting edge by the organizers. Specific topics discussed during the workshop were: i) pathophysiology of PE; ii) immunological origins of PE; iii) how to define subsets of PE; iv) biomarkers; v) nutritional factors; vi) genetics and epigenetics; vii) animal models; viii) collaboration, harmonization, sharing large data and samples sets in the study of PE; ix) periconceptional contributors to pathophysiology; x) clinical risk factors for PE; xi) novel therapies for the prevention and treatment of PE; xii) long-term cardiovascular consequences of PE and xiii) patient perspectives. Based on these discussions, recommendations for future research were developed in the following general areas: 1) Pathophysiology of PE; 2) Impact of PE on long-term health outcomes; 3) Biomarkers and diagnostic tools; 4) Translational research and novel treatment strategies. The purpose of this report is to summarize the scientific research recommendations arising from deliberations of the invited speakers and meeting attendees. These research recommendations are summarized in Table 1 (which is keyed to the manuscript text) and are presented in no particular order of priority. The goal of the recommendations is to provide guidance regarding future research on PE and its short- and long-term consequences.

Table1:

Summary of recommendations from the NIH workshop on “Predicting, Preventing and Treating Preeclampsia”

1. Pathophysiology	2. Long-term health outcomes	3. Biomarkers/Diagnostics	4. Translational research
a. Elucidate heterogeneity of mechanisms and pathobiology.	a. Characterize short-term trajectory with surrogate and intermediate endpoints through observational study.	a. Acquire adequate data to capture the spectrum of PE heterogeneity.	a. Develop new animal models of PE.
b. Understand normal pregnancy physiology and pathobiology leading to PE.	b. Leverage existing cohorts and incorporate PE in calculating CVD risk for mother and offspring.	b. Balance discovery science, clinical trials and biospecimen collection.	b. Support development of novel therapeutics and extend “Accelerating Medicines Partnership” to include PE.
c. Dyad/follow-up research to understand future risk.	c. Investigate mechanisms that underlie the progression to CVD following PE.	c. Standardize data and harmonized datasets for phenotype/genotype correlation.
d. Identify mechanisms by which risk exposures lead to PE.	d. Test the impact of patient and provider awareness and education on improved outcomes.	d. Combine and use existing data sets through cloud computing, machine learning, innovative bioinformatic tools and statistical methods.
e. Determine biological and environmental mechanisms of resilience.	e. Examine how management of CVD risk factors in post-partum impacts future CVD outcomes.	e. Acquire data from preconception, pregnancy, postpartum and inter-pregnancy.
f. Understand pathophysiology of term PE.	f. Investigate the bidirectional relationships between PE-PTSD and PE-cognitive function.	f. Apply new technologies to understand pregnancy physiology.
		g. Establish a “Pregnancy Genome Atlas”.
		h. Establish a hierarchical “omics” discovery approach.

RESEARCH RECOMMENDATIONS

1. Pathophysiology of Preeclampsia

Our understanding of the pathophysiology of PE has increased dramatically in the last 30 years. Preeclampsia is no longer considered merely a hypertensive disorder with renal dysfunction. Rather, abundant evidence now supports PE as a pregnancy-specific multisystemic syndrome.⁸ Placental syncytiotrophoblast stress results in increased inflammatory activation with resulting endothelial and metabolic dysfunction.^9,10 However, other than aspirin prophylaxis to reduce preeclampsia in the 20% of women with early onset preeclampsia who are risk stratified by a combination of biophysical and biochemical screening,¹¹ and calcium in low calcium intake populations, this knowledge has not translated into clinically relevant prevention or prediction strategies.¹²

The workshop considered possible explanations for this disparity as potential targets for future research. The initial gap to be addressed is the possibility, supported by clinical, biochemical and epidemiological findings, that PE is not a single disorder and that different pathways may converge on a common syndromic endpoint.¹² Identifying these possible disease subtypes may allow for development of novel biomarkers and therapies. Other potential targets only recently receiving attention are pre-pregnancy or early first-trimester pregnancy changes.¹³ These may be particularly beneficial, since based on experience with other disorders, this is a time in pregnancy when therapy might be most effective.

It was also recommended that different strategies to study PE be considered. The changes in mother and baby that persist after pregnancy (or perhaps were present before pregnancy) could inform relevant PE pathophysiology. In addition, studying disorders known to increase PE risk could elucidate pathways and mechanisms responsible for PE. For example, why do sleep disorders (e.g sleep disordered breathing, SDB) or obesity increase PE risk? And understanding why some women with pre-existing risks for PE, or with placental changes associated with PE are resilient, and do not develop the syndrome or its components (e.g. fetal growth restriction or FGR), could provide useful insights. Finally, we felt it important to address the most common form of PE, PE at term, responsible for much of maternal and infant mortality. The pathophysiology of term PE is not explained by abnormal placentation and has largely been ignored. To address the gaps related to better understanding the pathophysiology of PE, the following recommendations were proposed:

1a. Elucidate heterogeneity of mechanisms/pathobiology resulting in PE and related maternal and neonatal outcomes.

Based on the difficulty in predicting and preventing PE and the heterogeneity of clinical presentation, a reasonable assumption is that PE is not a single disorder and that there may be different pathways converging on a common syndromic endpoint. Identifying such disease “subtypes” would allow directed studies of prediction and prevention. Several principles should guide the search for such subtypes, including careful phenotyping guided by standardized data collection to facilitate sharing. Data sharing should be the rule for all studies and strongly encouraged. Studies should be designed, and data presented in a manner that emphasizes rather than obscures different subsets.¹⁴

One potential approach to discern different subtypes of PE is through mechanistic phenotyping.¹² This could be assisted by comparing clinically accepted subtypes (e.g. early vs. late onset PE or PE with vs. without fetal growth restriction). Longitudinal metabolic, inflammatory and angiogenic assessments of both the maternal and fetal phenotypes are required. Innovative approaches to longitudinal assessments (exosomes, microvesicles, cells within the cervix) should also be explored to complement studies of the placenta at delivery. Discovery approaches should be applied to maternal and fetal biomaterials. These should include samples from other placental syndromes (preterm birth, FGR, stillbirth) to generate new hypotheses including the possibility that there may be subtypes of these syndromes that share pathophysiological pathways with each other and PE.

1b. Understand normal pregnancy physiology and early pathobiology predisposing to PE.

Accumulating evidence suggests that direct causal or predisposing factors for PE may reside before, or in very early pregnancy.¹⁵ In this context, promising new targets for investigation are the endometrium and corpus luteum. Impairment of endometrial maturation in the secretory phase and during early pregnancy has been observed in women who develop PE.^15,16 Thus, obtaining and studying the relevant tissues from these two critical time periods of pregnancy is paramount. However, obtaining these invaluable tissues is a formidable challenge and registries or other strategies to identify the availability of these tissues for sharing are critically needed. In addition to the endometrium, the corpus luteum may also play an important role in the physiological changes of the maternal circulation during the luteal phase and early pregnancy.¹⁷ Future research should validate the physiological role of the corpus luteum and consider its deficiency as a causal or predisposing factor for PE. The molecular pathologies identified in these studies should be used to inform circulating biomarkers that identify these changes noninvasively to facilitate early identification of women at risk for developing disease. Animal studies may also provide valuable insights.¹⁸

Nutritional factors have the ability to affect almost every aspect of pregnancy.¹⁹ There are several events that occur in very early pregnancy, crucial to the pathophysiology of PE, that could be affected by nutrition. Evidence indicates that in the first eight weeks after conception, during histiotrophic nutrition, the conceptus develops in a low oxygen environment and receives all nutrients from those available in the decidua.²⁰ During the transition to nutrient delivery through blood flow from spiral arteries, the oxygenated arterial blood increases oxygen tension and antioxidants could be vital in mitigating potential damage.²¹ However, except for folic acid, our understanding of the potential beneficial impact of nutrients in early pregnancy is limited. Most studies examining the impact of nutrition in PE begin after the critical events of early placentation and embryogenesis (>12 weeks of gestation), are observational, and/or have PE as a secondary outcome. Randomized controlled trials of the influence of periconceptional nutrition (specific nutrients or dietary patterns) on subsequent PE are needed.

1c. Dyad/follow-up research to elucidate PE pathophysiology and future risk.

Women who develop PE as well as their offspring are at increased risk for developing chronic CVD and hypertension later in life.^22,23 What is not known is which women and infants will have these later outcomes. Dyad studies, that include clinical and biological data, will be important to understand and identify those individuals at greatest risk for later health disorders, thereby enabling targeted prevention, surveillance and allocation of clinical care resources.

Not only will dyad studies identify individuals at greatest risk for later health disorders, but dyad studies may also inform the pathogenesis and subtypes of PE. For example, studying women who take longer to resolve elevated blood pressure or proteinuria, or the infants that display more elevated blood pressures might focus attention on key pathways that were involved in the pathogenesis and potential targets for prevention. They could also clarify different disease subtypes related to maternal, placental and/or fetal factors. Preeclampsia that “looks” the same in pregnancy may show distinctions in long-term health implications as well as risk for subsequent pregnancies. Dyad studies can also provide insight into the resilience question as to why some women do not develop later CVD.

1d. Identify pathophysiological mechanisms by which risk exposures lead to PE.

Emerging literature links both extrinsic and intrinsic exposures to the development of PE. Specifically, extrinsic, environmental factors such as exposure to air and noise pollution, are associated with PE.^24,25 Infections have also been associated with PE and may increase a woman’s risk.^26,27 Sleep disordered breathing (SDB), characterized by recurrent episodes of airflow limitation and intermittent hypoxia during sleep, has recently emerged as an independent, intrinsic risk factor for PE.²⁸ Research is needed to determine whether SDB-triggered oxidative stress, elevated sympathetic tone, inflammation, or other mechanisms contribute to maternal, placental, and or/fetal pathobiology in PE. Exploring the mechanisms by which extrinsic and intrinsic factors increase the risk for PE may provide insights into modifiable targets for the prevention and management of PE.

1e. Determine biological and environmental mechanisms of resilience.

Based on the interaction of the maternal and fetal genomes, a modification of research strategies to understand diseases of pregnancy should be considered. In this context, one may explain placental defects that lead to fetal growth restriction without the maternal systemic disease of PE as a maternal genetic or environmental resistance to the disease.^12,29 Comparison of mothers resistant and susceptible to PE-induced hypertension may reveal pathways that receive a “toxic” signal from the placenta and initiate the local and systemic immune and inflammatory responses in the mother. Mechanisms that lead to resistance to maternal hypertension or fetal resistance to growth restriction may elucidate novel treatment strategies for PE.

1f. Understand pathophysiology of term PE.

Most PE (>80%) occurs at term (>37 weeks) and the number of women and infants dying from PE during that period, especially in low-resource settings is substantial. The pathophysiology of term PE is not well defined, but it is known that abnormal spiral remodeling with subsequent placental malperfusion, such as occurs in PE in early gestation is rarely present in term PE. However, it has been recently hypothesized that the final pathophysiology, syncytiotrophoblast stress, is the same in early and term PE. In the latter, stress comes, not from abnormal vasculature, but from a placenta that has reached the limits of its reserves and no longer functions to sustain fetal and placental demands.^9,10 Understanding syncytiotrophoblast stress and the phenomenon of placental senescence could inform longitudinal measurements of syncytiotrophoblast stress to predict PE and other late pregnancy abnormalities and improve their prevention (by better timing of delivery). Additional hypotheses to explain term PE should be encouraged and tested.

2. Impact of Preeclampsia on long-term health outcomes

Epidemiologic evidence linking data for individual women across decades has firmly established a link between the development of hypertension during pregnancy and an elevated risk of hypertension, CVD and renal disease later in life.^30,31 Risk ratios for these outcomes are about 2-fold higher in women with PE, and as high as 8-fold for early onset PE requiring delivery prior to 34 weeks.³² Also, rates of chronic hypertension two to five years after affected pregnancies are 50% following early-onset PE, 39% following gestational hypertension, and 25% following late onset PE.³³ By comparison, hypertension rates in women with normotensive, term births are very low (1%) 2 to 5 years after delivery.³⁴

There are gaps in the evidence that have hampered the ability to leverage the diagnosis of PE to improve the cardiovascular health of young adult and older women. For example, little is known about brain health, cognitive function, depression and other mental health issues following PE, or how the trauma of severe pregnancy complications may further impair cardiovascular health in affected women. Furthermore, despite recognized increased frequency,³⁵ recurrence,³⁶ and severity of PE in African American women, little is known about the long-term patterns of CVD risk for this group of women following PE. To address the knowledge gaps related to the impact of PE on long-term health outcomes, the following recommendations were proposed:

2a. Characterize the short-term trajectory with surrogate/intermediate endpoints through observational study.

Large, retrospective epidemiologic studies suggest that PE may be one of the earliest clinically-identifiable markers of a woman’s heightened risk of future CVD. Low birth weight, preterm birth, delivery of a small for gestational age infant, recurrent miscarriage, and PE are all associated with future CVD.³⁷ The development of a risk score derived from pregnancy history, added to currently available CVD risk scoring systems, could identify women at high-risk for CVD who need to be followed more aggressively. Including a brief pregnancy history “clinometric” and obtaining birth weight in existing cohort studies, as applicable, would help determine whether the “clinometric” does or does not make a difference in identifying individuals at higher risk of CVD.

2b. Leverage existing cohorts to calculate a risk score for younger and older women that incorporates PE and other pregnancy data as CVD risk factors. In addition, utilize these cohorts to assess offspring consequences and risk for CVD.

Identification and access to datasets that have sufficient clinical data to facilitate longitudinal follow-up of affected women are needed. This approach will allow for the confirmation of both the exposure (PE) and outcome (CVD) based upon accepted clinical criteria and will also provide longitudinal data with respect to cardiovascular risk factors and their treatments before, during, and after affected pregnancies. The results of these studies will allow for the assessment of whether a positive association between a history of PE and future CVD is independent of traditional risk factors. In addition, they may provide information regarding the effects of the treatment of traditional risk factors on future CVD in women with PE.

Another recommendation is to utilize studies with CVD endpoints, and retrospectively seek a diagnosis of PE. Studies/trials with stored biological samples should be utilized to determine the “natural history” of CVD risk following complicated pregnancies. In addition, cohorts that also follow offspring health after PE are needed. There are intergenerational risks for PE, suggesting that in utero exposure to PE may contribute to risk. Studies that investigate the mother-child dyad, PE and CVD risk are needed to identify optimal periods for interventions.

2c. Investigate mechanisms that underlie the progression to CVD following PE.

Obesity, hypertension, and dyslipidemia are risk factors for PE, as well as CVD.³⁸ Lifestyle interventions have the potential to reduce these traditional CVD risk factors, although more work is needed to demonstrate the efficacy of interventions in women of reproductive age. A recent meta-analysis assessed the effect of diet and lifestyle interventions on metabolic risk factors and the risk of PE. A total of 18 randomized clinical trials were reviewed and studies that focused on diet were found to have been effective in reducing the risk of PE.³⁹ What is not known are the optimal timing of the intervention, dose, delivery method, and context (e.g., clinic based, online, home-based).

Excessive gestational weight gain is associated with PE risk.⁴⁰ A question remains if weight gain represents fat accrual or fluid retention. Future research should examine whether reducing excessive gestational weight gain in accordance with the 2009 Institute of Medicine Guidelines (IOM) could prevent PE. Previous lifestyle interventions have had a challenging time reducing excessive gestational weight gain to the 2009 IOM guidelines, so future research should also address whether prevention of obesity before pregnancy may help prevent PE. Weight gain is also a leading risk factor for the development of SDB, which is itself a risk factor for cardiometabolic disease.⁴¹ The contribution of SDB to PE and subsequent CVD risk in women with excessive weight gain/retention needs to be delineated, as well as the potential to mitigate PE and related CVD risk by treating SDB.

In addition, investigating how risk factor management after PE may improve subsequent pregnancy health, short term cardiometabolic health, and long-term CVD risk is needed. These approaches, testing known and suspected predisposing factors, should be complemented by studies, including discovery-based studies, searching for unrecognized biological propensities of women with prior PE for later in life CVD.

2d. Test whether new models of care that incorporate patient/provider awareness, education, and coordination lead to better short, intermediate and long-term outcomes.

Intermediate outcomes of CVD are needed to study the role of postpartum management of CVD risk. As an example, excessive gestational weight gain has been linked to both PE and an adverse cardiometabolic profile as early as 12 months postpartum.⁴² Therefore, interventions in the postpartum period targeting prevention of gestational weight retention, such as physical activity, can be implemented in women after preeclamptic pregnancies, with the goal of reducing their immediate cardiometabolic and future CVD risks. This is especially relevant to women in whom the association between metabolic syndrome and CVD is stronger than in men.⁴³ Other intermediate outcomes may include subclinical findings of increased cardiovascular risk and atherosclerotic burden, namely coronary artery calcifications⁴⁴ and carotid artery intima-media thickness,⁴⁵ which have been associated with a previous history of PE as well as long term CVD events.

2e. Examine whether aggressive management of CVD risk factors in post-partum women leads to better CVD outcomes.

Studies are needed that would inform optimal methods and approaches for the timely diagnosis and treatment of hypertension after preeclamptic pregnancies. A recent study of women with severe PE one-year after delivery indicated that office blood pressure measurements diagnosed 24% of women to be hypertensive versus 41.5% when using ambulatory blood pressure monitoring.⁴⁶ Better recognition and treatment of hypertension after preeclamptic pregnancies may lead to improved cardiovascular outcomes and early treatment may have special value. This is particularly important given the evidence that hypertension contributes to more CVD events in women relative to men.⁴⁷

2f. Investigate the bidirectional relationship between PE-Post-Traumatic Stress Disorder (PTSD) and PE-Cognitive function.

A previous study using VA-covered deliveries demonstrated that military women with a history of PTSD are at increased risk for PE.⁴⁸ In turn, women with PE may be at risk of experiencing postpartum PTSD.⁴⁹ PTSD is a major psychological problem and future research should be directed at recognition, prevention and treatment of this disorder in pregnant women and women who have had PE. PTSD, a known disruptor of neuroendocrine health, is also associated with increased risks for CVD and cardiovascular mortality.⁵⁰ Future studies should address the bidirectional relationship between PTSD and PE, which may exhibit a cumulative effect on the future CV health of affected women. Subject to increased screening and clinical intervention, PTSD may be a modifiable risk factor for CVD in women after preeclamptic pregnancies.

There is a growing body of evidence suggesting that women with PE may experience cognitive decline later in life.^51,52 It is possible that these women are at an increased risk of dementia through increased risks for CVD later in life. This association may be further mediated by the presence of white matter lesions which have been demonstrated on neuroimaging studies in women with severe forms of PE.⁵³ Longitudinal studies of neurocognitive health after preeclamptic pregnancies are critically needed, aiming to identify women at risk for whom early screening and intervention strategies would be advisable.

3. Biomarkers and diagnostic tools

Currently, identification of biomarkers or development of diagnostic tools to predict PE has proven of limited value, as there is often substantial overlap between normal pregnancy and PE.^54–58 This is especially true when assessing low-risk primiparous woman for whom factors to stratify risk are limited to clinical features.⁵⁹ Moreover, PE is a heterogeneous condition with likely multiple pathways converging on a common series of clinical findings making it unlikely that any one biomarker will predict all cases of PE.¹ The definition of PE itself is problematic as it is descriptive of the clinical syndrome and therefore not particularly useful in understanding pathophysiology and identifying subtypes. Hence to better capture and describe the spectrum of heterogeneity, the importance of ongoing acquisition of clinical data and biological samples is evident. Biobanking of pregnancy samples faces several unique challenges due to the rarity of different conditions, the gestational timing of sample availability and the need to sample fetal, maternal, and even paternal tissues.

Beyond utility in early gestation, biomarkers such as angiogenic factors, soluble fms-like tyrosine kinase-1 (sFlt-1) and placental growth factor (PIGF) and their ratios have potential value in the prediction and risk stratification for PE and other adverse outcomes at other time-points during pregnancy.^60–62 They have the potential to alter clinical care similar to the use of troponin for chest pain and myocardial infarction. Despite introduction of late pregnancy measurements of these biomarkers into clinical practice in Europe, concerns related to the precise clinical utility remain to be addressed. There is also a dearth of longitudinal studies of biomarkers that span preconception, pregnancy, postpartum/lactation and a subsequent pregnancy. Despite an increasing knowledge of the pathophysiology of PE we still lack fundamental knowledge of the physiologic changes of pregnancy, e.g. plasma volume changes⁶³ as well as contemporary simple methods to measure these changes as aids to diagnosis and clinical management. Similarly, our knowledge of placental physiology, which clearly plays an early and important role in many pregnancies impacted by PE, lags behind that of other organs. The placenta is second only to the brain in terms of functional complexity as demonstrated by the number of expressed genes. And placental pathologies probably cause more human morbidity and mortality than any other organ.⁶⁴

PE is heterogeneous and a highly heritable syndrome, with both maternal and fetal genes being implicated.^65,66 In no other field do we need to study one patient with a unique genome living inside another with a related genome. Additionally, the paternal genome and associated epigenetic modifiers, such as sperm microRNAs, may influence pregnancy outcomes.^67,68 There is an urgent need for better conceptual and bio-informatics tools to measure interactions between the maternal and fetal genomes, the role of the paternal genome, critical trans-generational effects, and placental somatic variation (mosaicism). The following recommendations were proposed to address these gaps and challenges:

3a. Acquire adequate data and biological samples to capture the spectrum of PE heterogeneity.

One of the major challenges in studying PE heterogeneity is the lack of clinical phenotyping, linked to genomic/molecular/biochemical phenotyping, especially in studies with adequate sample sizes to yield statistical power. Consortia of investigators which optimally would include high-, low- and middle-income countries are needed to address the question of whether PE shares common phenotypes in these regions. Universal biobanking and clinical data warehousing protocols should be developed, spanning pregnancy and peri-pregnancy periods. Steps to achieve this goal would include 1) a national inventory of large scale biobanking efforts, 2) supporting efforts of groups already undertaking this, particularly those representing diverse and rural areas of the country and 3) supporting harmonization and providing the architecture to allow for clinical data and biosample sharing.

3b. Balance discovery science, clinical trial and biospecimen collection.

Beyond utility in early gestation, biomarkers such as sFlt-1 and PlGF have potential value in the prediction and risk stratification for PE and other adverse outcomes at other time-points during pregnancy.^61,62 Well-designed, multi-center trials are needed to test the clinical utility of angiogenic and other biomarkers in improving pregnancy outcomes balanced with the cost, the need for additional surveillance and unnecessary intervention resulting from false positive screening. Importantly, a pragmatic trial incorporating discovery science and biospecimen collection that could be applied across the United States with attention to low-resource settings would be valuable.

3c. Facilitate data combination by standardization of data collected into harmonized databases with defined minimal/optimal data sets for phenotype/genotype correlation.

Existing data sets with diverse patient numbers, variables and formats and collected in incompatible databases are extremely difficult, if not impossible, to combine. Clinical trials as well as basic science studies should incorporate standardized/harmonized databases that allow easy combination of data for meta-analyses with pre-defined minimal and optimal datasets to study phenotype/genotype correlation.¹⁰ Agreement on a standard data dictionary would further ease compatibility and combination.

3d. Leverage, combine and utilize existing and new data/biospecimen sets by use of cloud computing, machine learning and innovative bioinformatic tools/statistical methods.

Large repositories of data and samples from previous perinatal research studies funded by NIH and others are often available but overlooked/inaccessible to investigators. Efforts should be made to open these repositories. Existing medical record data (e.g. hospital and Ob/Gyn clinic records) should be examined for the reliability of diagnoses and used if possible and innovative methods for data collection and analysis, e.g., crowdsourcing, could be utilized. Existing databases are often difficult to combine hence efforts should be made to design and use programs that allow their combination together with innovative statistical methods to analyze them.

3e. Acquire data from the preconception period, through pregnancy, postpartum and inter-pregnancy.

The majority of work in preventing and treating PE is limited by having data only obtained during or immediately after pregnancy. However, the most critical deficiency of data about PE in pregnancy outcomes research, are data informing maternal condition and environment prior to pregnancy. There is a wide range of inter-individual normal values for biomarkers, coupled with changes across the menstrual cycle, gestation, lactation, and mammary involution. The data available across these unique physiological states is most often cross-sectional, which does not allow tracking biomarker changes over time, within an individual woman. However, to our knowledge, important biomarkers related to PE, such as PlGF and basic health and nutrition biomarkers (e.g., blood pressure, hemoglobin, insulin) have not been measured longitudinally in the same women across the reproductive spectrum. Longitudinal studies that span preconception, pregnancy, postpartum/lactation, and a subsequent pregnancy would be crucial for understanding the pathophysiology of PE, before and after its manifestation, and could greatly aid our understanding of healthy pregnancy in general. Given the difficulty and expense of studying a disorder before pregnancy with a prevalence of less than 10%, innovative approaches to using existing data from other studies that have “inadvertently” collected such data should be encouraged.

3f. Apply new technologies and appropriate analytical strategies to address basic questions in pregnancy physiology.

There are fundamental aspects of pregnancy physiology, let alone PE, which are still unknown. This is, in part, due to the challenges of studying early pregnancy and lack of current or new technologies. Many observations forming the basis of our current concepts of PE (e.g. failed spiral artery remodeling and its consequences and frequency) have only recently^9,21 or have never been reassessed with current technology. For example, we know that plasma volume expands across gestation and women who develop PE appear to have lower expansion.⁶³ However, older studies used poor definitions, primitive technologies, and inadequate analytical strategies to identify subgroups. Thus, it is not surprising that a simplistic approach of expanding plasma volume was not successful.⁶⁹ New methods for measurement of plasma volume and other such physiological parameters and appropriate analytical strategies allowing recognition of subtypes will not only improve our understanding of the relationship with hypertensive disorders of pregnancy, but may also allow their measurement as part of individualization of care.

3g. Establish a “Pregnancy Genome Atlas”.

Genetic variations that are associated with high risk pregnancies are discrete markers, present at all stages of the disease progression and are inexpensive to measure.⁷⁰ Genotyping can be used to properly distinguish subgroups with similar underlying disorders, but more importantly, genotypes can be used to integrate longitudinal observations. Millions of humans have now undergone exome or even genome sequencing in various studies or for clinical indications. There is an urgent need to collect pregnancy history from persons being sequenced for other reasons, so that adverse pregnancy outcomes can be correlated with sequence data. In parallel, it will be important to apply exome and genome sequencing to well-characterized PE cohorts. Usually the genes implicated in human pregnancy disorders have unknown and evolutionary divergent functions. A “Pregnancy Genome Atlas” (analogous to the recently completed Cancer Genome Atlas funded by the NIH⁷¹) would advance understanding of the third of our genome normally active only during pregnancy.

3h. Establish a hierarchical “omics” discovery approach.

Next-generation sequencing technologies and newer genomic, proteomic, transcriptomic, metabolomic, and other “omic” approaches are yielding valuable insights, but a comprehensive understanding remains elusive. In achieving a better understanding of PE, there is a logical hierarchy of the “omic” approaches from genome to transcriptome, proteome and metabolome, that can be addressed longitudinally in a hierarchical discovery manner in combination with clinical data. “Human knockouts” with gene deletions or truncating mutations and patients carrying major pathogenic mutations can be studied longitudinally, using both hypothesis-based and “omic” discovery approaches.

4. Translational research and novel treatment strategies

Translational research has led to novel approaches in early diagnosis and prognostication of PE,⁷² prevention,⁷³ and novel therapies.⁷⁴ Description of the molecular physiology and etiology of PE lags far behind that of other syndromes due to challenges in investigating developmental pathology and high diversity of mammalian reproductive physiology. A range of different animal models (mouse, rat, ovine and primate) exist for studying the pathophysiology of PE. These models do not describe the role of the placenta in the origins of PE, the pre-existing susceptibility, and the post-pregnancy consequences nor have they led to the development of new therapeutic approaches. Integration of data from multiple models is required to develop a global picture of the disease and may be required for each molecular subtype of PE. Animal models nonetheless provide a useful tool for hypothesis testing and examining potential pathophysiological mechanisms.

A wide variety of preventative strategies e.g. calcium supplementation,⁷⁵ vitamins C and E⁷⁶, fish oil, antihypertensives,⁷⁷ salt restriction, low dose aspirin,⁷⁸ low molecular weight heparin⁷⁹ have either failed or had minimal effect to prevent PE. This can be attributed to either the wrong timing, choice of therapy or the heterogeneous nature of PE masking any effect. More recent trials, in which metformin was used, had conflicting results regarding PE prevention.^80,81 Further, a recent meta-analysis of metformin use which suggested a high probability for reduced incidence of hypertensive disorders of pregnancy was marred by the small number of studies, low quality of evidence and clinical heterogeneity that precluded generalizability of results.⁸² Trials of statins^64,74 or other therapies, that were chosen based on biological plausibility of the therapeutic response and known safety data in pregnancy, are still preliminary or are yet to be reported. Transitioning a novel therapeutic from basic science to clinical trials faces significant regulatory challenges. The importance of novel therapies is amplified by the knowledge that PE has consequences for both the later cardiovascular health of the mother and the offspring. This further highlights the urgent need for overcoming the hurdles and identifying new therapeutics for the different phenotypes of PE. The following recommendations were proposed to address these gaps and challenges:

4a. Develop animal models for PE.

Different animal models exist for studying the pathophysiology of PE including, but not limited to the RUPP, infusion of sFlt-1 or vasopressin and several genetic models.⁸³ However, each of these models recapitulates only certain aspects of the pathophysiology of PE and not the full spectrum of pathophysiological features of the disease. Development and characterization of new spontaneous or genetic models for different phenotypes of PE e.g. that are mediated by placental dysfunction are needed. In addition, academic/industry partnerships should be encouraged to utilize existing and novel animal models for preclinical efficacy and toxicology studies of new therapeutics.

4b. Support infrastructure for development of novel therapeutics for PE.

While translational research has led to novel approaches in early diagnosis and prognostication of PE,⁷² prevention,⁷³ and therapies,⁷⁴ the challenge of developing novel therapeutics for the treatment of pregnancy-related disorders, including PE remains. Another challenge is access to a pipeline of compounds or therapeutics (especially those neglected by companies). Extension of the Accelerating Medicines Partnership (a public-private partnership with the National Institutes of Health (NIH), the U.S. Food and Drug Administration (FDA), and the private sector to identify new approaches to drug development) to include pregnancy disorders would facilitate development of novel therapeutics for PE. Development and access to novel animal models of different PE phenotypes would facilitate safety and efficacy studies for development of phenotype specific therapeutics. Pragmatic strategies such as the employment of pharmacometric tools (i.e. population pharmacokinetics or physiologically based pharmacokinetics and dose finding studies) and the application of innovative clinical trial designs in safe PE research should be undertaken.

Conclusions

The workshop convened by the NHLBI and NICHD identified several scientific challenges and questions related to research on predicting, preventing and treating PE, is summarized in Table 1. Concerted efforts were made to provide balance among the scientific areas explored at the meeting, yet it is acknowledged that the expertise of the invited expert panel and meeting attendees may have unintentionally steered the recommendations and that there are many other areas in which research is warranted. Nonetheless, the recommendations provided can help guide future research on better understanding the pathophysiology and long-term consequences of PE as well as development of novel biomarkers and therapies for PE.