Hemodynamic assessment is assuming an increasingly significant role in the understanding, prediction and management of a growing range of cardiovascular conditions. Evolving techniques that allow assessment of both the heart and vasculature have been pioneered in the non-pregnant population. Obstetrics is fast catching up, and these methods are now finding a wide range of applications in the care of pregnant women. They can provide valuable information that helps us to understand the physiology of normal pregnancy. Perhaps more importantly, they are starting to shed light on the pathophysiology of various pregnancy conditions, in particular pre-eclampsia (PE), fetal growth restriction and gestational diabetes mellitus. Furthermore, understanding the hemodynamic changes during normal and complicated pregnancy may help to improve our understanding of the associated increased risk for cardiovascular disease after complicated pregnancy.
There is increasing evidence that hemodynamic assessment may be used to guide therapeutic management, such as the choice of a particular type of antihypertensive medication in the management of PE or pregnancy-induced hypertension. Moreover, some techniques show promise for the prediction of pathology (again, PE and fetal growth restriction in particular) in the first, second or third trimester. Some studies have even investigated whether it might be possible to predict these conditions even prior to pregnancy, the hypothesis being that at least a proportion of women who develop these pathologies during pregnancy may be predisposed by pre-existing vascular abnormalities.
Levels of angiogenic markers, such as placental growth factor (PlGF) and soluble fms-like tyrosine kinase-1 (sFlt-1), have been shown to be related to the development of PE. The interrelationship between maternal hemodynamics and these angiogenic markers is particularly illuminating when attempting to understand the enigmatic pathophysiology of PE and/or fetal growth restriction. Validation and cross-comparison studies are helping to establish the relative utility and accuracy of the various methodologies.
Why is this important now?
In western countries, more women than men die of cardiovascular disease[1, 2], making this an important public health issue in women. In addition to the gender-independent classical risk factors seen in both men and women, in women the occurrence of PE is associated with a two-to-seven-fold increased risk of cardiovascular disease. Though this has been known for a number of years, new methods of assessing maternal hemodynamics now offer the opportunity to help answer the age-old question as to whether this is because these women already had cardiovascular pathology prior to pregnancy and PE represents an ‘unmasking’ of this pre-existing condition (analogous to gestational diabetes), or whether PE arises de novo in otherwise healthy women and damages the cardiovascular system, leading to the long-term increase in risk.
Cardiovascular function and adverse pregnancy outcome
Classic studies from the early 1990s suggested that cardiac output is higher in women who develop PE compared with those who are normotensive. In contrast, in those pregnancies that are destined to be complicated by poor fetal growth, very early first-trimester cardiac output changes are less marked compared with those in women who go on to have normal-sized babies. Later studies, from the 2000s, have lent further credence to this hypothesis, suggesting that fetal growth restriction is associated with high total peripheral vascular resistance and low cardiac output, while PE is associated with the reverse[6, 7]. From this has developed the concept, not investigated fully as yet, that therapy targeted at vasodilatation and increasing intravascular volume might lead to increased fetal growth and a better outcome in early growth restriction.
The key to these functional changes may be the way in which the heart adapts structurally to the increased demands that the mother's circulation faces in pregnancy. In contrast to the physiological eccentric left ventricular (LV) remodeling seen in healthy pregnancies, pre-eclamptic pregnancies are characterized by the less favorable concentric LV remodeling. The additional increase in left ventricular mass (LVM) in women with PE does not always resolve after delivery[9-11]. In fact, Melchiorre et al. have shown that, up to 1 year postpartum, 40% of former PE patients have structural or functional cardiac abnormalities, consistent with heart failure Stage B (HF-B). As described in this issue of the Journal, Ghossein-Doha et al. recently studied, in a cross-sectional cohort, the degree to which PE and conventional but modifiable cardiovascular risk factors are associated with asymptomatic structural and functional cardiac abnormalities postpartum. They found that, after adjusting for conventional risk factors, PE remained associated independently with a four-fold increased risk of having subclinical HF-B. Moreover, after PE, prehypertension aggravates the risk for HF-B four-fold, while other components of metabolic syndrome do not.
Pregnancy and cardiac remodeling
Normotensive and hypertensive pregnancies provide two very different models to study cardiac remodeling in response to different patterns of change in both preload and afterload. Cardiac remodeling is defined as a change in size, shape and/or structure of the heart and serves as an important compensatory mechanism to maintain the pumping capacity of the heart in response to either volume or pressure overload. It can be divided geometrically into either eccentric or concentric remodeling, based on the ratio of LV wall thickness to end-diastolic volume, the so-called relative wall thickness (RWT). Eccentric remodeling, as seen in healthy athletes and normal pregnancy, and defined as RWT < 0.43, is determined primarily by volume overload and is, in most cases, physiological. Concentric remodeling, as seen in hypertensive heart disease and PE, and defined as RWT ≥ 0.43, is determined mainly by pressure overload and is, in most cases, maladaptive and detrimental. Pressure and volume overload-induced stresses stimulate various signalling pathways essential for the induction of a hypertrophic response of the cardiomyocyte. Cardiac remodeling is determined not only by volume and pressure load, but also by neurohormonal factors.
Normotensive pregnancy is primarily a state of increased volume load driven by the need for the developing fetus to receive adequate blood, oxygen and nutrient supplies. During the first trimester, cardiac output increases by around 40%, in concert with a fall in total peripheral vascular resistance and a rise in cardiac preload. During normal pregnancy, indicators of preload, such as ventricular volumes, left atrial size and plasma volume, increase progressively, while determinants of afterload, such as decreased vascular resistance and heart rate augmentation, can be observed. As a response to the physiological hemodynamic demand that accompanies normal pregnancy, morphological changes leading to eccentric hypertrophy are triggered. These morphological changes in cardiac remodeling are reversible, with no long-term adverse effects on cardiac function[9, 19].
LV concentric remodeling is often paralleled by a subnormal, poorly expanding plasma volume, elevated cardiac afterload and cardiovascular sympathetic overactivity. This aberrant cardiac geometry and function triggered by PE may persist postpartum, especially after preterm PE. Melchiorre et al. showed that, at 1 year postpartum, the proportion of patients with aberrant LV geometric remodeling was significantly higher in the PE group compared with the control group (72% vs 24%). Early-onset PE (developing before 34 weeks of gestation) in particular relates to persistent residual postpartum concentric LV remodeling.
It is likely that the high prevalence of structural and functional cardiac abnormalities postpartum is at least in part a result of unfavorable cardiac adaptation in response to hemodynamic changes during PE, although baseline differences cannot be excluded. Therefore, it is likely that preventing unfavorable cardiac adaptation during pregnancy by modifying aberrant hemodynamics may decrease the prevalence of unfavorable cardiac profiles postpartum. Insight into cardiac adaptation and the related molecular profiles during complicated pregnancy may provide opportunities to improve clinical approaches in order to predict and prevent PE-related LV concentric remodeling.
Identifying the pre-stage of cardiovascular disease
Although using the term HF-B is subject to debate, this definition aids in clustering and identifying asymptomatic structural and cardiac abnormalities in these relatively young women. This is particularly relevant because there is an increasing understanding that cardiovascular diseases are generally progressive, proceeding through asymptomatic to symptomatic stages, and that the progression from the asymptomatic HF-B to the symptomatic HF Stage C increases mortality five-fold. Since therapeutic intervention during the asymptomatic phase of cardiac impairment can improve the long-term prognosis more effectively than does initiating intervention at a symptomatic stage, identifying this stage in relatively young subpopulations may have a major public health benefit[21, 22].
One of the primary determinants of HF is blood pressure. It has been suggested that the cardiac effect of blood pressure follows a J-curve. Thus, mildly increased blood pressure, high enough to lead to diagnosis of prehypertension after PE yet below the threshold for diagnosing hypertension, may play an important role in the increased prevalence of structural abnormalities in this young female population. Prehypertension, defined as systolic blood pressure between 120 and 139 mmHg and/or diastolic blood pressure between 80 and 89 mmHg, has been identified as a potent predictor of chronic hypertension. Data from the Framingham Heart study indicated that, compared with blood pressure < 120/80 mmHg, prehypertension was associated with an increased risk of myocardial infarction (relative risk (RR), 3.5) and coronary artery disease (RR, 1.7). Ghossein-Doha et al. found that two-thirds of women identified initially as normotensive 1 year after PE who went on to develop hypertension fulfilled the criteria for prehypertension. Earlier studies showed that prehypertension accelerates the development of LV hypertrophy and cardiac diastolic dysfunction[27, 28]. Certain antihypertensive medication (mainly angiotensin-converting enzyme inhibitors and angiotensin-receptor blockers) has a protective and inhibitory effect on adverse cardiac remodeling. Whether treating prehypertension with blood pressure-lowering medication intervenes in the progressive increase in LVM index and RWT, resulting in a decreased prevalence of HF, remains to be elucidated.
Though perhaps not studied as comprehensively as the heart, vascular function has been assessed in relation to established disease or its prediction. Various characteristics of vascular function can be assessed, using measures of arterial stiffness, wave reflection, microcirculation and endothelial function (either by flow-mediated dilation (FMD) or carotid intima–media thickness). Using one or more of these methodologies, a range of blood vessels, including the aorta, brachial artery, carotid artery, ophthalmic artery and uterine artery (uteroplacental circulation), has been studied. In addition, the maternal microcirculation can now be assessed reproducibly, as can retinal vascular responses.
Arterial endothelial function is assessed classically by FMD. The technique, though reproducible, takes several minutes and proxy methods include the augmentation index (AIx) and pulse-wave velocity. Arteries found to be relatively stiff in the first trimester are associated with a higher risk of PE and small-for-gestational-age (SGA) babies. Established PE is also associated with stiffer arteries as measured by AIx and uterine artery impedance[32, 33]. This raises the intriguing possibility that uterine artery impedance is related as much to a mother's systemic arterial function as to the ‘downstream’ placental bed.
The maternal venous system must not be forgotten. Though it is more difficult to assess, and venous assessment is less reproducible than is arterial, there are distinct differences between healthy and pre-eclamptic pregnancies[35, 36].
Studies in this issue of the Journal
This special issue of the Journal brings together some of the cutting-edge research in this field, which we hope will excite clinicians and academics about the future potential of these new and evolving techniques. The studies included can be considered broadly under five categories: physiology, therapeutics, prediction, postpartum and validation/cross-comparison.
Guy et al. reported that maternal cardiac function (cardiac output, stroke volume and total peripheral resistance) is influenced by maternal characteristics and proposed the use of multiples of the median to adjust for these correlations. Moreover, Iacobaeus et al. have demonstrated, in a longitudinal study, that, during normal pregnancy, the volume expansion necessary for sufficient fetal growth is accommodated by early and marked changes of the maternal vascular system. These hemodynamic changes seem to be dependent on normal adaptive endothelial and vascular function. In their letter, Staelens et al. describe renal venous Doppler sonography in health and disease: the measurements may be difficult to obtain and influenced by extrinsic factors, but the waveforms may give clues about the vascular abnormalities underlying, for example, HELLP syndrome or systemic autoimmune conditions.
Stott et al. suggest that assessment of individual maternal hemodynamics may be useful when selecting the type of antihypertensive drug to be used in pregnancy at different stages of treatment. In women starting antihypertensive drug therapy, they could predict those women who would respond to labetalol and those who would require additional vasodilator treatment. Similarly, during the acute phase of labetalol therapy, they used maternal hemodynamic measurements to identify women whose hypertension was unlikely to be controlled using labetalol alone and so were likely to require additional vasodilator therapy. The use of serial hemodynamic monitoring to guide treatment in these women significantly reduced the rates of severe hypertension and allowed women to be triaged into those with low vascular resistance, who could be treated successfully with labetalol alone, and those with high vascular resistance, who would need additional vasodilator therapy.
Ambrozic et al. evaluated lung and cardiac ultrasound for assessment of fluid tolerance and fluid responsiveness in severely pre-eclamptic patients before and after delivery. They found that severe PE is associated with an increase in extravascular lung water, which could in part be caused by disturbed diastolic LV function. Excess lung water can be identified in severely pre-eclamptic patients before the appearance of clinical signs by using lung ultrasound. The role of these findings in clinical practice still needs to be determined.
Initial studies focused predominantly on the first- and second-trimester prediction of PE and fetal growth restriction[43, 44]. More recently, however, there have been studies investigating prediction in the third trimester (35–37 weeks' gestation) of later development of PE and even studies attempting to predict these pregnancy pathologies prior to conception.
In a cohort of around 3000 pregnancies, Guy et al. assessed the prediction of PE/pregnancy-induced hypertension using maternal cardiac function at 35–37 weeks' gestation. In women who developed term PE, total peripheral resistance and mean arterial pressure were increased, and cardiac output was decreased, compared with in those who remained normotensive throughout pregnancy. However, the assessment of maternal cardiac function at 35–37 weeks' gestation is unlikely to improve the performance of screening for PE provided by maternal factors and mean arterial pressure alone. This group found similar results when assessing pregnancies with SGA or large-for-gestational-age (LGA) fetuses. In SGA pregnancies, maternal cardiac output and heart rate were reduced, while total peripheral resistance was increased, and in LGA pregnancies, cardiac output and heart rate were increased, while total peripheral resistance was reduced, compared with women with normal-sized babies. However, again, maternal hemodynamic assessment is unlikely to increase the prediction of SGA or LGA over and above a combination of maternal characteristics and fetal ultrasound biometry.
Dragan et al. evaluated whether a cut-off of 38 for the sFlt-1/PlGF ratio is of predictive value for PE in singleton pregnancy at 30–37 weeks' gestation and found that, in routine screening of singleton pregnancy, the performance of sFlt-1/PlGF > 38 is modest for prediction of delivery with PE at < 1 and < 4 weeks and poor for prediction of delivery with PE at ≥ 4 weeks.
In a prospective cohort of women followed from prior to conception, Mahendru et al. measured maternal cardiac output prior to pregnancy and in mid-pregnancy and report an association between the change in this parameter during pregnancy and birth weight. Mid-trimester cardiac output is not related to fetal size in the mid-trimester, hence cardiac output was not being ‘driven’ by a large placenta or fetus. Rather, the cardiac output was hypothesized to be causative in determining the eventual fetal size. This may provide evidence for the first direct link between cardiovascular function and birth weight.
McKelvey et al. performed a prospective cohort study in nulliparous women, measuring total uterine artery blood volume flow rate by transabdominal ultrasound at 12, 20 and 24 weeks of gestation and found a significant and positive correlation between total uterine blood volume flow in the first and second trimesters and birth weight and gestational age at delivery, implying a potential role for this parameter in prediction of SGA and preterm birth.
A systematic review of studies using carotid intima–media thickness reported the presence of atherosclerosis at the time of PE and at least 10 years after pre-eclamptic pregnancy. Furthermore, microcirculatory dysfunction, evidenced by a reduction in reactive hyperemia index value and increased arterial stiffness, persisted in previously pre-eclamptic women assessed between 6 months and 4 years postpartum, particularly in those who had early-onset PE. In the same cohort, persistent subclinical contractile impairment involving the whole heart was also reported. Moreover, pregnancies affected by fetal growth restriction still had demonstrable evidence of subclinical endothelial dysfunction 6 months after childbirth. Interestingly, HF-B occurs in a quarter of formerly pre-eclamptic women and, after adjustment for conventional risk factors, PE remained associated independently with HF-B[12, 54], while prehypertension increased the risk for HF-B a further four-fold. However, over 60% of those who have HF-B 4 years postpartum recover, while around 20% of previously pre-eclamptic women with a normal echocardiogram in the first year postpartum will develop HF-B subsequently.
Validation and cross-comparison
The structure and function of the heart can be assessed by echocardiography (either non-invasively, by the transthoracic route, or invasively, by the transesophageal route), or by more invasive procedures such as a Swan–Ganz catheter. Heart function may be assessed by one of a range of cardiac output monitors, including the Ultrasound Cardiac Output Monitor (USCOM®), Non-invasive Cardiac Output Monitor (NICOM®) and an inert gas rebreathing method (INNOCOR®). The issue with there being so many different methods is validation. Cornette et al. found that cardiac output measured in pregnant women non-invasively using echocardiography and invasively using pulmonary artery catheterization showed excellent correlation. This suggests that, given its non-invasive nature and ready availability, echocardiography could be considered as the reference for validation of other methods of measuring cardiac output in pregnant women. Vinayagam et al. demonstrated good agreement between USCOM and NICOM cardiac output monitors when compared with echocardiography, specifically in the third trimester. Both devices have good repeatability/reproducibility and can be utilized by healthcare professionals with different levels of experience. However, individual values do differ significantly among the various devices used, and it is therefore imperative that device-specific reference ranges in pregnancy and postpartum are developed.
These fascinating papers surely show that this field of clinical research is coming of age. Few areas within women's imaging have quite so many crosscutting links: cardiology, vascular physiology, engineering, pharmacology and bio-informatics are the most obvious but there are doubtless others. We hope that this issue of the Journal will inspire many more studies and encourage multidisciplinary collaborations in the emerging world of maternal hemodynamics.