Prof. KH Nicolaides, Harris Birthright Research Centre for Fetal Medicine, King’s College Hospital, Denmark Hill, Golden Jubilee Wing, London SE5 9RS, UK. Email Kypros@fetalmedicine.com
Objective To assess the maternal central haemodynamics in normotensive women with pregnancies complicated by severe fetal growth restriction (FGR).
Design Cross-sectional study.
Setting A tertiary referral fetal medicine unit.
Population The study groups comprised 107 women with normal singleton pregnancies and 20 with singleton pregnancies complicated by FGR at 25–37 weeks. In the latter group, assessment was carried out within 10 days prior to their delivery. All the women were normotensive, without any medical problems.
Methods Two-dimensional and M-mode echocardiography of the left ventricle.
Main outcome measures Maternal left ventricular systolic and diastolic function.
Results In the FGR group, compared with the normal group, there was increased total vascular resistance (TVR), reduced systolic function characterised by lower cardiac output, stroke volume, heart rate, ejection time and septal and lateral long-axis shortening. Mean arterial pressure (MAP) was not significantly different between the groups.
Conclusions Severe FGR is associated with reduced maternal systolic function and increased TVR but no change in MAP. TVR may be a useful tool in the classification and management of FGR. The findings suggest that in FGR, there is increased blood viscosity due to lack of intravascular space expansion.
Fetal growth restriction (FGR) is a common cause of neonatal morbidity and mortality and is now increasingly recognised as a risk factor for cardiovascular and metabolic disease in later life.1–3 Although multiple aetiologies exist, defective trophoblastic invasion of the maternal spiral arteries is a universally accepted pathophysiological finding.4
Normal placentation is thought to trigger a fall in systemic vascular tone and to create a state of intravascular volume depletion. This provokes an increase in plasma volume, heart rate (HR) and hence cardiac output (CO). To cope with this physiological stress, the maternal heart increases its compliance, its contractility and its size, thus modifying the function of the left ventricle. This reorganisation results in a 50% increase in CO and uteroplacental perfusion.5–7
There is some evidence that impaired placentation is associated with altered maternal cardiovascular function. Bosio et al.8 demonstrated that pregnancies complicated by pre-eclampsia are characterised by a hyperdynamic circulation, which crosses over to a low CO high peripheral resistant state at the onset of pre-eclampsia. In normotensive pregnancies with FGR, there is a contradictory evidence about maternal haemodynamics, with some studies reporting reduced CO and left ventricular (LV) diastolic function in the first and third trimesters9–11 and others reporting no difference in LV function between normal pregnancies and those affected by FGR.12 Accurate interpretations cannot be made from the published literature due to the small number of subjects studied. Disparity may also be explained by differences in the employed echocardiographic methodologies and inconsistencies in the definition of FGR. Classification of FGR based on birthweight less than 10th or 5th percentile is likely to include a proportion of small-for-gestational-age (SGA) fetuses.13
Doppler velocimetry of the fetal circulation is a useful tool in the assessment of fetal compromise and prediction of hypoxemia and acidemia. Classification based on the temporal changes in the fetal circulation will select truly compromised fetuses.14–16 Assessment of maternal haemodynamics in such a clearly defined homogenous group will provide useful information, which may correlate with the fetal outcome. The aim of this study was to assess the maternal central haemodynamics in true FGR pregnancies.
This was a cross-sectional cohort study of maternal cardiac function in 20 women with singleton pregnancies complicated by FGR at 25–37 weeks of gestation. All women were assessed within 10 days prior to their delivery. We also examined 206 pregnant women with normal singleton pregnancies attending routine antenatal care at 10–40 weeks of gestation. All women were healthy, with no adverse medical history and not on any medication. Gestational age (GA) was calculated from the last menstrual period and confirmed by ultrasound biometry in the first trimester. Normal fetal anatomy was demonstrated by serial ultrasound scans in a fetal medicine unit; fetuses with chromosomal abnormalities, genetic syndromes and infections were excluded. All women gave a written informed consent to participate in the study, which was approved by the Research Ethics Committee of King’s College Hospital, London.
The selection criteria for FGR were the following: normotensive pregnancies, estimated fetal weight less than the 3rd centile for GA,17 fetal asymmetry (defined as the ratio of head circumference to abdominal circumference above the 95th centile), cerebral redistribution (defined as the ratio of umbilical artery to middle cerebral artery pulsitility index greater than the 95th centile) and oligohydramnios (defined as amniotic fluid index below the 5th centile).
The women were studied after a rest period of 15 minutes in the left lateral decubitus position. Measurements were obtained when three consecutive electrocardiogram measurements of the HR demonstrated a variation below 10%. Two examiners (J.E.A.K.B. and N.A.K.) carried out all measurements, and in all parameters, three cardiac cycles were averaged. Two-dimensional and M-mode echocardiography was performed using a 3.5-MHz transducer (Toshiba Aplio CV and Powervision 7000—SSA 380; Toshiba Corporation, Tokyo, Japan) according to the guidelines of the American Society of Echocardiography.18 Stroke volume (SV) was computed as the product of the cross-sectional area of the LV outflow tract and the velocity time integral of the pulsed Doppler subaortic waveform measured in the five-chamber view. CO was calculated as the product of HR and SV.19 Total vascular resistance (TVR) was calculated as TVR = MAP × 80/CO.
Long-axis function of the left ventricle was evaluated by two-dimensionally guided M-mode recordings through the mitral valve annulus, using the apical four-chamber view for the lateral and septal side.20–22 Diastolic function was assessed by measuring the LV filling dynamics in the apical four-chamber view. Transmitral flow velocities were recorded, with the sample volume positioned adjacent to the tips of the mitral leaflets in diastole. Peak velocity of early atrial filling (E), peak velocity of late atrial filling (A) and ratio of peak E to peak A were measured.23,24
Blood pressure measurements were performed using a mercury sphygmomanometer (Accoson Dekamet, AC Cossor & Son (Surgical) Ltd, London, UK) according to the recommendations of the British Hypertension Society.25 Mean arterial pressure was calculated from the equation: MAP = (BP systolic + (2 × BP diastolic))/3.
The Kolmogoroff–Smirnoff test was used to assess normality of the distribution of the data. Unpaired t test or chi-square test, where appropriate, was performed in order to examine the differences in the demographic characteristics between the examined populations.
Multiple regression was used to assess the significance of differences between the FGR and the normal pregnancies at 25–37 weeks. For these comparisons, we used the data of only 107 normal pregnant women examined at the same gestational range as those in the FGR group (25–37 weeks). The independent variables were GA, GA2, group (normal = 1 and FGR = 2), interaction (interaction = GA × group). When the term ‘interaction’ was statistically significant, it indicated not only that the difference between measurements of pregnant women with FGR and normal pregnancies was statistically significant but also that this difference changed with gestation. In our multiple regression models, we investigated the independent contribution of maternal race, age and height.
For each of the variables examined in the pregnancies complicated by FGR, the observed value was subtracted from the normal mean for gestation, based on the regression equation of the normal singleton pregnancies. This difference was then expressed as a percent difference between the two groups and its 95th CI was calculated.
Reproducibility between the two examiners was analysed in six nonpregnant women of childbearing age by Bland and Altman’s26 95% limits of agreement method, with differences expressed as mean ± 2 SD.
For power calculation, CO was considered the main outcome measure. Using the SD of the normal pregnancies, recruitment of 19 women with FGR and 95 women with normal pregnancies was calculated to allow detection of a difference of 1 l in CO between the two populations, with 90% power at the 5% level.
The statistical package SPSS 8.0 (SPSS for Windows, Rel. 8.0.0. 1997; SPSS Inc., Chicago, IL, USA) was used.
The demographic characteristics of the normal and FGR groups are compared in Table 1. In the FGR group, there were two intrauterine deaths at 26 and 29 weeks and 18 live births at a mean gestation of 32 weeks (range 27–38 weeks). The mean interval before maternal cardiac assessment and delivery was 4 days (range 0–10 days).
Table 1. Demographic characteristics of the study populations (mean and 1 SD)
Normal pregnancies, n= 107
FGR pregnancies, n= 20
Maternal age (years)
Maternal height (m)
Maternal weight (Kg)
Caucasian, % (n)
Afro-Caribbean, % (n)
Nulliparity, % (n)
Gestation at delivery (weeks)
In the normal group, CO increased with gestation (CO = 3.8 + 0.2 × GA − 0.004 GA2, P < 0.0001, R2= 0.10) to a maximum of 30 weeks and thereafter declined towards term. Similarly, SV increased with gestation (SV = 62.8 + 1.8 × GA − 0.03 GA2, P= 0.02, R2= 0.04) to a maximum of 30 weeks and thereafter declined towards term. HR increased linearly with gestation (HR = 75 + 0.26 × GA, P= 0.007, R2= 0.04), while ejection time (ET) decreased linearly with gestation (ET = 330.0 − 1.1 × GA, P < 0.0001, R2= 0.08). Long-axis shortening at the septal (Sept = 13.6 + 0.2 × GA − 0.005 × GA2, P < 0.0001, R2= 0.10) and lateral annulus (Lat = 15.4 + 0.2 × GA − 0.006 × GA2, P < 0.0001, R2= 0.08) both increased with gestation until 20 weeks and subsequently decreased slowly to term. Transmitral E/A ratio decreased linearly with gestation (E/A = 1.8 − 0.01 × GA, P < 0.0001, R2= 0.10). MAP and TVR decreased in parallel to a minimum at 20 weeks, thereafter slowly increasing to term (MAP = 100.5 − 1.8 × GA + 0.04 × GA2, P < 0.0001, R2= 0.06) (TPR = 1595.3 − 50.3 × GA + 0.9 × GA2, P < 0.0001, R2= 0.13).
Multiple regression analysis examining the differences between the FGR and the normal groups showed a significant contribution for GA but not for maternal race, age or height (Table 2). In both groups, CO, SV, septal and lateral long-axis shortening increased with gestation to a maximum at mid-pregnancy and thereafter declined to term. For all four variables, the measurements were lower in the FGR group than in the normal group (Table 2; Figures 1 and 2). HR increased (Figure 1), while ET and transmitral E/A ratio decreased (Figure 3) linearly with gestation. HR and ET were lower in the FGR group than in the normal pregnancies, while E/A ratio was similar between the two groups. MAP and TVR decreased with gestation to a nadir at about mid-pregnancy and thereafter increased towards term (Figure 3). MAP was similar between the two groups, while total peripheral vascular resistance was higher in the FGR group than in the normal group.
Table 2. Results of the multiple regression analysis between the FGR and the normal groups
Coefficient P values
Mitral valve E/A
Table 3 gives the difference between the FGR and the normal groups as a mean (95% CI) percent difference of the latter.
Table 3. Percent difference between the FGR and the normal pregnancy groups (mean and 95% CI)
Inter-observer variability between the two examiners for CO, SV, HR, septal and lateral long-axis shortening, ET and E/A were: 0.02 (−0.37 to 0.42), 0.86 (−3.455.1 to 7.0), 0.17 (−2.77 to 2.438), −0.27 (−2.4 to 1.86), −0.27 (−1.68 to 1.14), 2.33 (−17.56 to 22.23) and 0.03 (−0.58 to 0.64), respectively.
The findings of this study demonstrate that in pregnancies complicated by severe FGR, compared with normal pregnancies, maternal systolic cardiac function is reduced, TVR is increased and diastolic function is not altered.
In our FGR group, compared with the normal pregnancies, there was a higher prevalence of women of black race and of younger age. Studies comparing healthy, young black women with white adult women have shown differences in cardiac structure and function between the two racial groups. Hinderliter et al.27 compared LV function and structure between 62 black and 71 white healthy young adults. Average daytime blood pressure (determined by ambulatory monitoring during a typical work or school day) and resting blood pressure were similar between the two groups. However, LV wall thickness was significantly greater in black than in white subjects, both in men and women. Black subjects also had a higher resting systemic vascular resistance and lower resting CO, even after controlling for subject body surface area. There were no significant differences between black and white subjects in LV mass or fractional shortening, although some studies do report higher values for LV mass in black subjects than for white subjects.28 Nevertheless, in our study, the differences between the two groups are present even after controlling for racial differences. On the contrary, in a study comparing younger with older women (range 16–88 years), CO indexed to patient’s body surface area (cardiac index) has been shown to decrease slightly with advancing age, while total peripheral resistance increased with age.29 Thus, LV performance at rest measured by cardiac index is slightly lower in older than in younger clinically normal adults. However, these data involve women at the higher age limits and it is doubtful whether they would be applicable in a small age difference in a young, healthy population. Nevertheless, even if these differences were relevant to our population, the impact of the lower maternal age in the FGR group would lead to an underestimation rather than an overestimation of the differences.
Our findings are in agreement with those of Vasapollo et al.,10 who compared maternal haemodynamics between 21 women with normal pregnancies and 21 women with FGR at 25–36 weeks. However, our finding of no difference in diastolic function between the FGR and the normal groups contradicts their result of diastolic dysfunction in the FGR group compared with the normal group. Diastolic function describes the ability of the left ventricle to maintain adequate filling volumes and maintain a good CO without substantial increase in the intracavity pressure.30 This was examined by transmitral flow velocity profile, where E/A ratio normally decreases in pregnancy as a result of increased LV mass and compliance. Vasapollo et al.10 reported a significant reduction in E/A ratio, as evidence of diastolic dysfunction in FGR compared with normal pregnancies.10 Yet, in the prementioned study, the systolic and diastolic blood pressures were higher than in our patients, reaching ranges approximating pregnancy-induced hypertension. This on its own could be the cause of the reported LV diastolic dysfunction. Furthermore, if preload is reduced in this population and MAP is not different from the normal pregnancies, then there would be no obvious pathophysiological mechanism to explain the reported diastolic dysfunction.
The pattern of changes in the maternal cardiac function in the normal pregnancies shows, as previously published,21 an increase in CO (as a result of an increase in HR and SV), reaching a peak early in the third trimester to levels of about 50% above than those of the nonpregnant controls and a small decline towards term. Similar changes occur in SV, with a 30% increase by mid-pregnancy, while HR shows a linear rise with advancing gestation to about 15% above than those of nonpregnant controls. We assessed only longitudinal and not transverse LV systolic function as, in our previous study,21 the former seemed to reflect the impact of increasing preload and afterload on the LV during pregnancy sooner than the latter. Long-axis shortening, as with CO and SV, increased with gestation until mid-pregnancy and thereafter declined towards term, mirroring the changes in MAP. MAP and TVR initially decreased with gestation, reaching a nadir at about mid-pregnancy, with a subsequent increase until term to levels higher than that of nonpregnant controls. Contrary to systolic function, diastolic LV function, as assessed by the E/A ratio, progressively declined with gestation, reflecting the negative impact on the LV on the increase in preload and afterload with the concomitant LV hypertrophy during pregnancy.
In FGR, LV systolic function is reduced. This is shown by the lower CO, which in turn is explained by a lower SV and HR. In addition to this, we have found that long-axis shortening is reduced. Anatomically, the left ventricle is organised into longitudinal fibres that are located in the subendocardial and subepicardial layers of the myocardium, extending from the apex to the base of the heart, and circumferential fibres, which are found in the mid wall and base of the ventricle, being responsible for longitudinal shortening and transverse function, respectively.31 Long-axis shortening is a more sensitive measure of LV systolic function than transverse axis function (i.e. ejection fraction and fractional shortening),31 as the subendocardial layer is more susceptible to ischaemia and blood pressure changes than the circumferential layer, hence reflecting changes in preload and afterload in advance of circumferential fibres. That explains why long-axis dysfunction becomes evident before transverse axis dysfunction.31–33
LV systolic function is dependent on preload, myocardial contractility and afterload. Preload during pregnancy is mainly dependent on plasma volume expansion that starts as early as in 5 weeks of gestation and peaks at about 30 weeks to levels of about 40% above those of nonpregnant women.34,35 There are reports of an insufficient plasma volume expansion in FGR.36 Salas et al. measured plasma volume by the modified Evans blue dye method in 30 women with FGR and compared the findings with those of 26 normal pregnancies. They found that not only the plasma volume in FGR pregnancies was significantly lower compared with that of normal pregnancies but also expansion of plasma volume, comparing values during pregnancy and 3 months postpartum, was less than 50% of that of the normal pregnancies.36 The reduced plasma volume expansion in FGR pregnancies has been advocated to be evident from as early as 4–5 weeks of gestation. Duvekot et al.9 examined early in the first trimester ten women who subsequently had normal pregnancies and six women who subsequently developed FGR. They found a smaller left atrial diameter, an index of preload and thus plasma volume expansion in pregnancies that went on to develop FGR. Similarly, Vasapollo et al.10 have also shown reduced left atrial and LV end-diastolic diameters in FGR compared with normal pregnancies, indicating reduced preload. Therefore, reduced preload is a significant determinant of reduced systolic function and CO in FGR pregnancies. Myocardial contractility is difficult to assess since all ultrasonographic parameters used to assess it (such as ejection fraction, fractional shortening and long-axis shortening) are also preload and afterload dependent and therefore do not independently reflect myocardial contractility. Although our data show impaired long-axis shortening and Vasapollo et al.’s10 data show impaired ejection fraction and fractional shortening in the FGR group, these observations are not adequate to support a hypothesis of impaired contractility of the LV in the women with FGR pregnancies.
Afterload, as depicted by TVR, is dependent on vascular tone and blood viscosity. In normal pregnancy, an early fall in systemic and renal vascular tone creates an intravascular volume depleted state. This vasodilation activates the renin–angiotensin–aldosterone system and results in 40% increase in plasma volume.34,37 The increase of the red cell mass lags behind the increase in plasma volume, resulting in the physiological anaemia of pregnancy, with a concomitant reduction in blood viscosity and improved blood flow in the sluggish placental circulation. Therefore, normal pregnancy is characterised by a drop in peripheral resistance and blood viscosity, both of which contribute to reducing the burden of the left ventricle and improve oxygen supply to the peripheral maternal tissues and the placenta. On the contrary, hypertension,38 high haematocrit and blood viscosity have been associated with FGR.39–41 However, at constant blood pressure, blood flow is inversely related to the viscosity of blood due to the stasis created by the increased cellular constituent of blood; therefore, high viscosity reduces flow and increases peripheral resistance. In our data, we only included normotensive women with FGR and there was no difference in MAP between the normal and the FGR pregnancies because the purpose of our study was to examine a group of pregnancies with FGR without superimposed pre-eclampsia. Therefore, the elevated peripheral resistance observed in the FGR group in our data must be due to the reduced haemodilution and increased blood viscosity. The importance of the TVR measurement as a tool for screening and guiding antenatal management in SGA pregnancies needs further evaluation.
Our study has shown that maternal cardiac function in FGR, compared with normal pregnancies, is characterised by reduced CO and long-axis shortening and increased TVR. The extent of lack of intravascular volume expansion in these women becomes more evident if one notes that their CO is identical to that of nonpregnant controls.21 To date, there has been only one study that has sound methodology in selecting women and assessing maternal cardiac function.10 Previous studies have given conflicting results due to either suboptimal categorisation of women with FGR9,11 or the use of techniques nonvalidated in pregnancy.13 In this study, according to stringent criteria, we have investigated pregnancies complicated by true FGR, reporting maternal haemodynamics within the 10 days prior to their delivery in order to correlate these events with the maximum degree of fetal disease. Further research is needed to establish differences in maternal cardiac function between true FGR, SGA fetuses and pre-eclampsia complicated by FGR and to assess the value of these findings in the management of high-risk pregnancies.
This study was funded by the Fetal Medicine Foundation (registered charity 1037116).