British Journal of Obstetrics and Gynaecology
May 1999, Vol106, pp. 453466
Doppler fetal circulation in pregnancies complicated by
pre-eclampsia or delivery of a small for gestational age
baby: 2. Longitudinal analysis
K.Harrington Consultant and Senior Lecturer, M. 0. Thompson Clinical Research Fellow,
*R. G. Carpenter Statistician, *M. Nguyen Sratistician, TS. Campbell Professor
Academic Department of Obstetrics and Gynaecology,The Homerton Hospital, London; *Departmentof Medical Statistics, London
School of Hygiene and Tropical Medicine; tSt. George’sHospital, London
Objective To observe the longitudinal changes in growth, and associated Doppler measurements, of the
fetal circulation in pregnancies with a normal outcome and those complicated by pre-eclampsia, birth
of a small for gestational age baby, or a combination of these complications.
Methods Two hundred and ninety-two women had serial ultrasound scans performed from the 24th
week of pregnancy. Measurements obtained included: the abdominal circumference, umbilical artery
pulsatility index, the middle cerebral artery pulsatility index and time-averaged velocity, and the thoracic aorta pulsatility index and time-averaged velocity. Outcome measures included the birth of a
small for gestational age infant, pre-eclampsia or a combination of these complications.
Results One hundred and sixty-seven pregnancies ended in the normal birth of an appropriately grown
infant at term, while 105 had a complicated outcome. They were divided into three categories: preeclampsia only (i.e. with the birth of an appropriately grown fetus, n = 13); small for gestational age
only with no evidence of pre-eclampsia, n = 55; and pre-eclampsia complicated by the birth of a small
for gestational age infant, n = 37. The group with pre-eclampsia complicated by small for gestational
age was closest in characteristics to true clinical intrauterine growth restriction. A reduction in the
abdominal circumference mirrored a rise in the umbilical artery pulsatility index and preceded
changes in the middle cerebral artery and thoracic aorta. The greatest degree of change in the fetal circulation was observed during the three weeks preceding delivery. Ratios of the values obtained from
the fetal and umbilical circulation demonstrated the greatest deviation from normal.
Conclusions A reduction in fetal growth velocity preceded changes observed in the fetal circulation.
Ratios of the fetal Doppler parameters provided the clearest evidence of deterioration in the fetal
condition. The information provided may be of use in the diagnosis and management of the
growth-restricted fetus.
INTRODUCTION
Intrauterine growth restriction (IUGR) and preeclampsia, or proteinuric pregnanc y-induced hypertension, are common complications of pregnancy which
are associated with a failure of normal placental invasion and development. The consequences of this placental dysfunction may include an altered resistance to
blood flow in the uterine circulation, as well as adaptive changes in the fetal circulation, and carry a significantly increased risk of morbidity and mortality for
both the mother and the
The introduction of real-time ultrasound has enabled
the observation of changes in fetal growth during
Correspondence: Mr K. F. Hanington, Academic Department of
Obstetrics and Gynaecology, The Homerton Hospital, Homerton
Row,London E9 6SR, UK.
0 RCOG 1999 British Journal of Obstetrics and Gynaecology
pregnancy, thereby making the antenatal detection of
fetuses at increased risk of IUGR possible4. Smallness,
however, is an observation not a diagnosis, and not all
small for gestational age babies suffer from IUGR and
its associated risks of stillbirth, perinatal mortality and
neonatal handicap5. Even though the failure of a fetus to
attain or exceed its expected growth potential may result
from numerous different pregnancy complications6, the
final common pathway most often encountered in practice is via uteroplacental insufficiency.
Pre-eclampsia is unique to humans and has not been
shown to develop spontaneously in other animal
species. Previous animal studies have therefore
depended on the induction of hypertension and uteroplacental insufficiency by experimental means to simulate
this essentially human complication of p r e g n a n ~ y ~ . ~ .
Similarly, embolisation studies on animal placentae
have been used to acquire information about umbilical
453
454
K. H A R R I N G T O N ET A L .
blood flow changes’. We cannot assume that the results
of artificially created deprivation in animal studies accurately reflect the changes that occur in human pregnancy
complicated by true uteroplacental insufficiency.
Doppler ultrasound allows a noninvasive assessment
of fetal haemodynamics’O.”.Doppler investigation of
the umbilical arteries provides information concerning
perfusion of the fetoplacental circulation, while Doppler
studies of selected fetal organs are valuable in detecting
the haemodynamic rearrangements that occur in
response to fetal hypoxia and anaemia.
In order to make the best use of the information
obtained from Doppler studies of the fetal circulation,
clinicians need to be aware of temporal changes in the
fetal circulation, and how these changes correlate with
changes in fetal size and other tests of wellbeing. There
have been reports of longitudinal data from the fetal circulation using Doppler ultrasound, but these studies
were commenced after a diagnosis of fetal growth
restriction or compromise was already established1”14.
The aim of this study was to record longitudinal
changes observed in fetal growth as well as fetal
Doppler waveform indices in pregnancy. We also sought
to compare the changes that occurred in pregnancies
that ultimately developed pre-eclampsia or resulted in a
small for gestational age baby with those pregnancies
that resulted in a vaginal birth at term of an appropriately grown baby.
METHODS
This study was conducted in the fetal assessment unit
of an inner city teaching hospital. Two hundred and
ninety-two pregnant women had serial ultrasound
scans from 24 weeks of gestation. The observations on
167 pregnancies, which ended with a term birth of an
appropriately grown baby, were used to define the normal range. This group has previously been describedls.
The study group contained 125 high risk pregnancies
referred to the fetal assessment unit for investigation
and monitoring that subsequently developed complications. The indications for referral included: suspected
IUGR, oligohydramnios, abnormal uterine artery
Doppler studies (persistent notching), hypertension, or
a poor obstetric history. Multiple pregnancies and
those with a fetal structural or chromosomal anomaly
were excluded from the study. The outcome for each
pregnancy was obtained by examining the labour ward
records, and neonatal intensive care unit records. The
outcome was unknown in two high risk cases, leaving
123 pregnancies available for analysis. The results
were not revealed to clinicians or patients, and no
action was taken on the basis of the fetal Doppler
ultrasound findings. Ethics committee approval was
obtained for the study.
The investigations were performed using an Acuson
128 (Mountain View, California, USA) with 3.5 and
5 MHz linear and curvi-linear transducers. The women
rested for 10-15 minutes in a semi-recumbent position
before the ultrasound investigations were commenced.
Initial measurements included the fetal atdominal circumference and the amniotic fluid volume as previously
describedI6,recorded as < or 2 2 cm in depth. The different methods for the identification of the fetal vessels and
obtaining measurements also have been described previously”. Doppler flow velocity waveforms were
obtained from the middle cerebral artery, thoracic aorta,
and umbilical artery. The values of the pulsatility index
and time-averaged velocity were recorded. The results
of Doppler investigations were not revealed to the clinicians involved in the women’s care.
Outcome definitions
The pregnancies were divided into different categories
depending on the outcome. A normal outcome was
defined as one where an appropriately grown fetus was
born at term in the absence of maternal hypertension or
vaginal bleeding in pregnancy. he-eclampsia was
defined as maternal hypertension (blood pressure
> 140/90 mmHg) on at least two separate occasions in
the presence of proteinuria (> 300 mg in 24 h)17. Infants
defined as small for gestational age (SGA) were those
born with a birthweight below the 10th centile based on
the charts currently in use in the hospitalIR.
Pregnancies complicated by pre-eclampsia, which
resulted in the birth of an appropriately grown fetus,
were assigned to the proteinuric pregnancy-induced
hypertension (PP1H)-only group. Pregnancies complicated by the birth of an SGA baby, but which had no
evidence of pre-eclampsia were placed in the SGA only
group. Pregnancies complicated by both pre-eclampsia
and an SGA baby were allotted to the PPIH + SGA
group. The division of pregnancies with a complicated
outcome into these three categories allowed us to
examine differences in the relation between fetal
Doppler in pre-eclampsia, with or without the added
complication of an SGA fetus.
Table 1. The intra-operator variability for the Doppler indices.
UA = umbilical artery; PI = pulsatility index; MCA = middle
cerebral artery; TAV = time averaged mean velocity.
Test
UA PI
MCAPI
MCATAV
Aorta PI
Aorta TAV
Coefficient of variation (%)
5.6
6.1
1.4
4.9
7.2
0 RCOG 1999 Br J Obstet Gynaecol 106,453-466
LONGITUDINAL A N A L Y S I S OF DOPPLER FETAL CIRCULATION
In view of our objectives, 18 pregnancies which had
complications that did not fall into any of the categories
under study (i.e. not complicated by pre-eclampsia and
with an appropriately grown baby) were excluded from
the control group. They were composed of women who
were delivered spontaneouslybefore 37 weeks of a baby
with a birthweight that was appropriate for gestational
age, or required delivery by caesarean section for reasons such as two previous caesarean deliveries, breech
presentation, or failure to progress in labour.
Statistical analysis
Between two and five consecutive sets of Doppler
observations at least two weeks apart were obtained for
each fetus. The average of three individual measurements performed at each sitting was recorded. The measurements from each were expected to follow a
curvilinear trend as the fetus developed. In fetuses with
a normal outcome this trend was expected to conform to
the course of the normal centiles for Doppler measurement~’~.
For fetuses with an abnormal outcome, it was hypothesised that either the average of the measurements
would be abnormal or that the trend would deviate
sharply from normal. Such a divergence from the normal trend might occur at any stage beyond 26 weeks of
gestation in the weeks immediately prior to delivery,
which is often premature when the outcome is not normal. This makes the analysis of the raw observations
difficult because the normal trends of fetal Doppler
measurements are generally curvilinear, with a standard
deviation and non-normal distribution, which change
with gestational age. In addition, the variation between
subjects at a given age was likely to be much greater
than that of repeated observations on the same subject,
after eliminating the influences of time trends in the
observations.
The analysis was therefore simplified by modelling
the centiles for gestational age of the observations on
fetuses with normal outcome using Grostat”, and then
using the programme to convert all this data to normal z
scores, as described previo~sly’~.
If the Grostat models
fit the data perfectly for fetuses with a normal outcome,
z scores corresponding to each Doppler measurement
should be normally distributed with zero mean and unit
standard deviation at each gestational age. After this
transformation a sequence of observations at different
gestational ages, but on the same centile become a
sequence of identical z scores. Thus all observations on
the 3rd centile transform to z = 1-88,on the 16th centile
to z = 1, on the 50th centile to z = 0, etc.
Thus, for each Doppler measurement we had a set of z
scores, repeated a variable number of times on each
baby in the study. Such data have been described as
0 RCOG 1999 Br J Obstet Gynaecol 106,453-466
455
‘multi-level’ with two levels of variation: level 1, with
subject variation, and level 2 the between subject variation. Both levels of variation” are regarded as random
and independent.
The analysis of such data is implemented in MLWin’”.
Using ML3, an earlier version of MLWin, the z scores for
each Doppler variable were modelled for the four groups
of pregnancies, normal, PPIH, SGA and PPIH + SGA.
The trends of the group mean z scores were modelled as
a linear or quadratic curve as required. The fitted model
also represented the pattern of observation on each individual subject by a constant difference from the group
mean plus, if necessary, a linear trend towards or away
from the group mean. These two terms summarised the
inter-subject variation at level 2. Quadratic components
of individual trends were not significant. The variation of
the individual observations about the fitted trend for that
subject, when pooled over all subjects, provided the estimate of level 1 intra-subject variation.
The first models of the fetal Doppler measurements
included gestational age alone. However, when gestational age and gestational age at birth were both
included in the statistical model of abdominal circumference z scores, the analysis showed that the key predictor was the time interval (in weeks) between
obtaining measurements and the birth of the baby. The
inclusion of this variable eliminated the need to include
gestational age or gestational age at birth. This finding
led to the replacement of the gestational age at which
measurements were made and the gestational age at
birth by a single variable: the interval before birth in
weeks at which the measurements were made. This substitution was made in all analyses of trends in the
Doppler measurements.
Finally, to collate all the data on repeated measurements, ML3 was used to derive summary statistics of
the intercept and trend of each subject’s measurements
according to each Doppler z score at two and a half
weeks prior to birth. Since these statistics could be
assumed to be approximately normally distributed, discriminant analysis was used to determine the extent to
which the data showed that the groups differed significantly in their development, as well as the key measurements for such separation.
RESULTS
The intra-operator variability for the various measurements is presented in Table 1. Figure 1 shows the trend of
mean z scores for the abdominal circumference and
Doppler measurements in pregnancies complicated by
PPIH and SGA. The divergence of the Doppler ratios of
these scores is shown in Figure 2. The abdominal circumference deviates from normal at an early stage in the process, as does elevation of the umbilical artery pulsatility
456 K . H A R R I N G T O N ET A L .
3
UAPl
2
1
u)
2
8
0
u)
h(
-1
-2
-3
I
-12
I
-11
I
I
I
-10
-9
-8
I
I
I
I
-6
-5
Time to delivery (weeks)
-7
-4
I
-3
1
I
I
-2
-1
0
Fig. 1. Trends before birth of mean z scores of the abdominal circumference and Doppler PI values ( z scores) for PPIH + SGA patients.
index. Changes in the middle cerebral artery pulsatility
index and aorta pulsatility index become clear about three
weeks before delivery. Figure 2 shows the pattern of
change in the Doppler ratios in the same group of patients.
Figures 3-8 illustrate the slope for the abdominal circumference and each Doppler parameter. Figures 9-1 1
illustrate the slope of the ratios for each of the Doppler
ratios. Table 2 shows the combinations of variables that
best predicted each complicated outcome.
The difference between the abdominal circumference
measurements for appropriately grown infants (AGA)
and those who were small for gestational age only
(without PPIH) grew wider nearing delivery, so that the
abdominal circumferences of SGA babies were significantly less than those in AGA babies two weeks prior to
delivery (Fig. 3). This was shown by the significant negative slope for the SGA group. The abdominal circumference z score for the SGA group was 1.064 less than
that for AGA babies by two weeks before birth and
-1.20 at half a week before birth.
Doppler measurements
The three groups of complicated pregnancies (SGA,
PPIH, and PPIH + SGA) provided an opportunity to
examine the haemodynamic responses of the fetus to
conditions that frequently share a common aetiology.
There were differences in the umbilical artery pulsatility
index between AGA and SGA babies, with or without
PPIH (Fig. 4).In the last 3 and a half weeks before birth
there was also an increase in the difference between
measurements from the AGA normal babies and those in
the PPIH group. As a result, the umbilical artery pulsatility index was found to demonstrate good discriminant ability in the separation of normal from the
abnormal groups.
The middle cerebral artery pulsatility index means
differed significantly from normal in those in the PPIH,
SGA and PPIH + SGA groups at two weeks before birth
(Fig. 5). The difference increased with proximity to
birth for SGA fetuses (with or without PPIH). The individuals differed from their respective group means, the
variation being constant over time (i.e. unaffected by
gestational age). This variable was one of two recorded
that showed very clear differences between the normal
group and the three other groups (PPIH, SGA and PPIH
+ SGA). There was little difference in middle cerebral
artery time-averaged velocity between the normal group
and those with complications, although blood flow
velocities appeared higher in the PPIH + SGA group at
0 RCOG 1999 Br J Obstet Gynaecol 106,453466
L O N G I T U D I N A L A N A L Y S I S O F DOPPLER FETAL CIRCULATION
5
UAPVMCAPI
4
3
2
m 1
e!
8
u)
‘ v o
-1
-2
-3
I
I
-4
-12
-11
I
I
I
-10
-9
-8
I
I
I
-6
-5
Time to delivery (weeks)
-7
I
I
I
I
I
-4
-3
-2
-1
0
Fig. 2. Trends before birth of mean z scores of ratios of Doppler measurementsfor the PPM + SGA group.
0.5
Normal
0
-0.5
m
?!
PPlH
8ln -1.0
ru
-1 -5
-2.0
-2.5
I
-12
1
-11
I
I
I
-10
-9
-8
I
I
Fig. 3. Abdominal circumference mean z scores, described by weeks before birth.
0 RCOG 1999 Br J Obstet Gynaecol 106,453466
I
-6
-5
Time to delivery (weeks)
-7
I
I
I
I
I
-4
-3
-2
-1
0
457
458 K. HARRJNGTON E T At.
PPIH SGA
2.5
2.0
1.5
1.o
v)
?!
8
u)
N
0.5
0
Normal
-0.5
-1
.o
-12
-11
-10
-9
-8
-7
-6
-5
Time to delivery (weeks)
4
-3
-2
-1
0
Fig. 4. Umbilical artery pulsatility index mean z scores, described by weeks before birth.
two weeks to birth (Fig. 6). The individual readings varied about the group means both in intercept (at two
weeks to birth), and in slope, (i.e. the differences from
the group means increased over time). The upward trend
of the mean z scores of the normal group shows that the
gestational age related increase in normal middle cerebral artery time-averaged velocity had not been fully
captured by the fitted centiles.
The average thoracic artery pulsatility index was
higher than normal for those in the SGA and PPIH +
SGA groups and there was a tendency towards higher
values among those in the PPIH group (Fig. 7). This difference widened increasingly until birth. The significance of the coefficients for PPIH and PPIH + SGA
showed that measurements of the thoracic artery pulsatility index were of value in differentiating between
the normal and abnormal groups. Individual values varied significantly from the group mean at two weeks
before birth but there was a common slope. Aorta timeaveraged velocity measurements showed differences
between the normal and the other three important
groups studied (Fig. 8). There were differences evident
in flow velocity at 2 and one-half weeks before birth
between the normal group and the PPIH, SGA and PPIH
+ SGA groups. The separation increased with age for
those in the PPIH and PPIH + SGA groups. Individuals
varied from their group means but had a common slope.
An examination of the ratios of the Doppler parameters such as the umbilical artery pulsatility index: middle cerebral artery pulsatility index ratio revealed that
some means (specifically those associated with SGA)
exhibited an even greater deviation from the normal
than the individual measurements (Figs 9-1 1).
Potential value of the data for diagnosis of PPIH
+ SGA
The longitudinal analyses suggested substantial differences between pregnancies with a normal and those
with a complicated outcome, in particular the PPIH +
SGA group, which most closely correlates with true
intrauterine growth restriction. Using ML3, an estimate
was made of the predicted value of every measurement
on each subject at 2 and one-half weeks-before birth.
This estimate was based on the group mean and the
deviation from the mean of the trend line fitted to the
subjects’ measurements. A series of two group stepwise
discriminant analyses was applied to these data as
described in the methods section. These analyses suggested that only four of the seven sets of measurements
0 RCOG 1999 Br J Obstet Gynaecol 106,453466
L O N G I T U D I N A L ANALYSIS OF D O P P L E R FETAL CIRCULATION
459
0.5
0
-0.5
fn
e!
8
-1.0
fn
h(
-1.5
-2.0
\
PPIH SGA
-2.5
I
I
-12
-11
I
-10
I
I
I
-9
-8
-7
1
I
-6
-5
Time to delivery (weeks)
I
I
I
I
I
-4
-3
-2
-1
0
Fig. 5. Middle cerebral artery pulsatility index mean z scores, described by weeks before birth.
provided significant independent information on the
diagnosis of these conditions (Table 2). Moreover, the
percentage of variation exhibited showed that the
Table 2. The variables selected by stepwise discriminant analysis as
best predictors of the conditions listed, compared with ‘normals’,
from estimated z scores 2 and a half weeks before birth. PPIH =
pre-eclampsidproteinuricpregnancy induced hypertension, with an
appropriately grown baby; SGA = delivery of a small for
gestational age (i
10th centile) baby, with no evideece of PPIH;
PPIH + SGA = development of PPIH and the delivery of an SGA
baby; AC = abdominal circumference; UA = umbilical artery;
PI = pulsatility index; MCA = middle cerebral artery; TAV = time
averaged mean velocity.
Condition
PPIH
SGA
PPIH + SGA
Variables selected
MCA PI
AC
Aorta TAV
AC
MCA PI
Aorta TAV
UA PI
AC
MCA PI
UA PI
Percent
variation explained
complicated groups are almost completely identifiable
as abnormal at least 2 and one-half weeks before birth
by using these scores. By constructing a score from the
discriminant function for each subject, the extent of separation of the groups was evident (Fig. 12).
The precise depths of the pockets of amniotic fluid
measured by ultrasound were simply recorded as normal
(when the measurement of the deepest vertical pool of
liquor seen was 2 2 cm), or abnormal (if c 2 cm). The
incorporation of amniotic fluid volume data in the eventual analysis would have required continuous measurements. However, no pregnancy in this study exhibited
an abnormal amniotic fluid volume (< 2 cm) prior to two
weeks before delivery. It was also apparent that the clinical decision-taking in some cases would be influenced
if oligohydramnios were reported during ancillary
investigations.
79.3
DISCUSSION
88-8
87.5
0 RCOG 1999 Br J Obstet Gynaecol 106,453466
In this study we conducted a follow up of pregnancies
from the second trimester, before the onset of complications such as IUGR or pre-eclampsia. We have demonstrated the chronological changes that occur in growth
and associated fetal Doppler parameters in pregnancies
460 K. HARRINGTON ET A L .
3.0
PPlH
/
2-5
2.0
s
1 *5
1.o
0.5
0
-0.5
-12
-11
-10
-9
-8
-7
-6
-5
Time to delivery (weeks)
-4
-3
-2
-1
0
Fig. 6. Middle cerebral artery time-averaged velocity mean z scores, described by weeks before birth.
2.5
PPlH SGA
2.0
1.5
g
1.0
8u1
h(
h
$ 0.5
0
-0.5
-1
.o
-12
-11
-10
-9
-8
-7
-6
-5
Time to delivery (weeks)
4
-3
-2
-1
0
Fig. 7. Thoracic aorta pulsatility index mean z scores, described by weeks before birth.
0 RCOG 1999 Br J Obstet Gynaecol 106,453-466
LONGITUDINAL ANALYSIS O F DOPPLER FETAL CIRCULATION
0.4
r
0.2
'
-0.2
461
'
$ -0.4
L
8
-0.6
>
-0.8
-1
.o
-1-2
'
-1.4
-1.6
I
I
I
I
I
I
-12
-11
-10
-9
-8
I
I
I
-7
-6
-5
Time to delivery (weeks)
I
I
I
4
-3
-2
I
I
0
-1
Fig. 8. Thoracic aorta time-averagedvelocity mean z scores, described by weeks before birth.
that developed pre-eclampsia and fetal growth restrictions, and compared these findings with those in pregnancies with a normal outcome.
To simplify the analysis all the measurements have
been transformed to z scores. To determine z scores, the
centile values of measurement by gestational age are
constructed for babies with normal outcome. Then the z
score of any observation is the standard Normal Deviate
corresponding to the centile on which the measurement
lies. The z scores of the normal ultrasound observations
are therefore expected to lie in a broad band between +2
and - 2 from 22 to 42 weeks of gestation. The drift of the
normal means z scores from zero (e.g. in Fig. 6), shows
that the fitted centiles did not fully take account of the
5r
-12
PPlH SGA
-11
-10
-9
-8
-7
-6
5
Time to delivery (weeks)
-4
-3
Fig. 9. Umbilical artery: middle cerebral artery pulsatility index ratio, described by weeks before birth.
0 RCOG 1999 Br J Obstet Gynaecol 106,453-466
-2
-1
0
462 K . H A R R I N G T O N ET A L .
3
2
1
0
In
2 -1
8
In
h(
-2
-3
I
-4
-12
I
I
I
I
-11
-10
-9
-8
I
I
I
-7
-6
-5
Time to delivery (weeks)
I
I
I
I
I
-4
-3
-2
-1
0
Fig. 10. Middle cerebral artery: thoracic aorta pulsatility index ratio, described by weeks before birth.
2.0
-
1.5 1.0
-
0.5
-
0-
8
-0.5
-
h(
-1.0
-
8
-1.5 -2.0
-
-2.5
-
-3.0
-3.5
-
'
I
-12
I
I
I
I
-11
-10
-9
-8
I
I
1
-7
-6
-5
Time to delivery (weeks)
I
I
I
I
I
-4
-3
-2
-1
0
Fig. 11. Middle cerebral pulsatility index: thoracic aorta time-averaged velocity ratio, described by weeks before birth.
0 RCOG 1999 Br J Obstet Gynaecol 106,453-466
LONGITUDINAL ANALYSIS O F DOPPLER FETAL CIRCULATION
40
”
r
16 18 20 22
28 30
36 38 40 42 44
z scores
Fig. 12. Discriminant scores at 2 and one-half weeks before delivery
from a comparison of PPIH + SGA and ‘normal’ babies, using
longitudinal fetal biometry and Doppler observations.
curvature in the normal centiles at the end of this age
range. But in almost all the other figures the normal
means show little deviation from zero.
PPIH may develop, and IUGR become apparent, at
various stages in the third trimester. Analysis of the
abdominal circumference data showed that the changes
were best described in terms of weeks before delivery
rather than gestational age, and consequently all the
changes have been described in terms of this parameter.
Figures 1 to 11 show that differences between the group
means for the various measurements tend to increase
progressively and are generally significantly different
from zero immediately prior to delivery. These are average values, and are regardless of gestational age at
delivery. It is possible that for any Doppler measurement the deviations may differ from the average according to gestational age at birth, as suggested by the
placental studies of Krebs et al.”. This would imply an
interaction of the changes with gestational age.
Our previous paperI5indicates the possibility of such
an effect, particularly in umbilical artery measurements,
but we did not attempt to examine this here. Statistically
one looks first for main effects and if these are present,
one may look for interaction. In this first study of serial
observations of Doppler measurement the number of
cases in the non-normal groups is scarcely sufficient to
go further.
We have not presented patterns of the individual
observations on babies with normal and abnormal outcomes in Figs 1 to 11, because the overlap of different
groups of observations would make the graphs meaningless. It will be remembered that the normal range is f 2.
Only when individual observationsare used to determine
predicted values for the individual at 2 and one-half
weeks before birth and the results for three or four
Doppler parameters are combined into a discriminant
0 RCOG 1999 Br J Obstet Gynaecol 106,453-466
463
score do clear distinctions emerge between the groups,
as in Fig. 12. This establishes that Doppler measurements contain information that distinguishes these
groups from normal more than two weeks prior to delivery, as currently practised. Computer programmes have
yet to be developed to extract and synthesise such data
on a routine basis.
A reduction in the abdominal circumference mirrored a
rise in the umbilical artery pulsatility index, and preceded
flow velocity waveform changes in the middle cerebral
artery and thoracic aorta. At 2 and one-half weeks before
delivery, the abdominal circumference in the SGA group
was 1.045 standard deviations lower (SD 0-140),and that
in the PPIH + SGA group was 1.2 standard deviations
(SD 0.151), lower than the normal group. The AC
remains a useful test for identifying small
Although identifying the SGA fetus for the purpose of
closer surveillance during pregnancy may result in the
inclusion of a number of normal small fetuses, it remains
an important finding. This is because of the association
demonstrated between an estimated fetal weight, below
the 10th or 5th centile for gestation, and the subsequent
development of perinatal complications.
The umbilical arteries were selected for study because
they reflect maldevelopment of the placental tertiary
stem villous tree”. Histopathological embolisation studies have correlated an increase in umbilical artery resistance with a reduction in the functional area of the
placenta’. While the origin or precise aetiology of raised
umbilical artery resistance is contentious, the clinical use
of umbilical artery is well established, An increased
resistance to flow or a decrease in flow in the umbilical
arteries may represent a deficiency of forward flow, such
as in severe fetal cardiac compromise. In the absence of
severe fetal cardiac problems, the resistance in the
umbilical arteries is a good indicator of placental resistance and flow. The mean pulsatility index in the umbilical artery was 1.210 standard deviations (SD 0.205)
higher in the SGA group than in the normal group, 2 and
one-half weeks before delivery. This measurement
achieved the greatest separation between the SGA and
normal babies at four to eight weeks preceding delivery.
Doppler ultrasound has enabled the non-invasive confirmation of the brain-sparing effect in IUGR in human
f e t u s e ~ ~ and
~ . ~ correlation
~,
between acidaemia and
absent end diastolic frequencies in the fetal aorta has
been established. In this longitudinal analysis the greatest degree of change in the fetal circulation occurred
during the last three weeks preceding the birth of a
growth restricted baby, and the ratios of the Doppler measurements obtained from the fetal and umbilical circulation demonstrated the greatest deviation from normal.
Earlier studies have demonstrated that in the fetal
aorta significant changes occur later, and can be independent of changes in the umbilical artery”. Also, the
464 K .
HARRINGTON ET AL.
umbilical artery pulsatility index is typically elevated if
IUGR and hypoxia occur before 34 weeks, but may
remain normal if changes occur near or at
Behavioural states in the fetus may affect Doppler
but this effect is of more significance in the fetal venous circulation than in the fetal
arterial circulation. In order to minimise the influence
of fetal behaviour, the measurements in this study were
obtained in the absence of fetal movement and breathing movements.
The pregnancies used to define a normal outcome
were a mixture of women from the routine obstetric
population, with no known complications in the current
pregnancy. Some had expressed concern in the clinic
because of a previous poor pregnancy outcome, and
some were without complaints but happy to act as controls for the research project. What is important in relation to this study is that the pregnancy outcome in this
group was normal (i.e. delivered with birthweights
appropriate for gestational age). While they represented
a ‘normal’ group in this study, they do not necessarily
represent the norm for every other population. We therefore cannot assume that all obstetric populations would
respond in an identical manner.
Historically, many investigators have sought to
understand the variation in haemodynamics of the uteroplacental and fetal circulations, in a bid to increase our
current understanding of the physiologic basis of health
and disease in the fetoplacental unit during pregn a n ~ y ~ ~ -Various
’~.
epidemiological, clinical, and
immunological studies indicate that a failure of the placenta to create an adequate uteroplacental circulation is
the major underlying factor in the pathogenesis of fetal
growth restriction and pre-eclarnp~ia~”.’~.
Several animal
and biochemical studies have been conducted to try and
unravel the hormonal and biochemical changes in the
fetus that accompany growth deprivation, by examining
the effects of altering maternal or uteroplacental blood
and nutrient supply on fetal growth35-”.
This study differs from earlier animal experiments
because the placental complications in the latter were
induced artificially (by uterine artery occlusion techniques, microsphere embolisation of placental segments or the surgical reduction of placental mass) in
essentially normal pregnancies with normal fetuses.
Previous human research also involved evaluation of
the fetal circulation after complications were already
diagnosed’4*39*40.
Ultrasound assessment of amniotic fluid volume, and
electronic fetal heart rate monitoring (CTG) are also
established tests of fetal wellbeing. Amniotic fluid volume was measured at each visit during this study, and
classed as either normal or abnormal according to
whether the deepest vertical pool measured was 2 2 cm
or < 2 cm4’. The readings were normal, except when
measured within two weeks of delivery. CTGs were not
performed as part of the study but were used in the clinical management of the pregnancy. A CTG was not
requested if the fetus showed evidence of normal activity and there was a normal amount of amniotic fluid
found at the ultrasound scan. Our experience in clinical
practice indicates that changes in fetal growth and
Dopplers of the fetal arterial circulation precede abnormalities in fetal heart rate pattern^'^.
Changes in Doppler indices of the venous circulation,
and ultimately ominous fetal heart rate patterns, suggest
severe asphyxia and imminent death if delivery is not
expeditedl2.I4.This implies that significant abnormalities in the CTG are comparatively late signs of fetal
compromise. Longitudinal, computer analysed fetal
heart rate pattern studies also support the suggestion that
ominous fetal heart rate changes typically occur in the
presence of acidaemia4*.This study suggests that fetal
heart rate changes are a late sign, and using CTGs as a
first line of investigation may result in a compromised
fetus being missed by antenatal assessment.
Examination of the regression analyses of the data
revealed that the groups in this study could seldom be
fully identified by a single variable, as was apparent
from the univariate analysis, but they were identified by
a simple linear combination of the repeated measures. It
should also be noted that discrimination was independent of the background variables considered, namely
race and panty. Measurement ratios were not included
in the analysis of trends. But apart from this, the discriminant scores combined the observations on those
measurements with the largest F ratios in the univariate
analysis of last observations.
The results of discriminant analysis of the various
parameters revealed that it should be possible to differentiate, using a quantitative measure, between pregnancies with normal and complicated outcomes, 2 and
one-half weeks before delivery (Fig. 12). As the data
involves relatively small numbers, and used the time of
delivery in calculating the discriminant scores, we cannot assume that this analysis could be used for prospective diagnosis of IUGR. However, it does suggest that
such a goal is realistic, with the possibility that in the
future, we may be able to utilise the information derived
from fetal Doppler studies directly in the diagnosis and
management of pregnancies complicated by preeclampsia or restricted fetal growth.
The clinical presentation of uteroplacental insufficiency may be maternal, fetal or a combination of
both. This suggests that our understanding of the true
pathophysiological processes responsible for a failure
of placental development and function requires further
evaluation. The sequence of events associated with
fetal deterioration and demise are important, because
of the current lack of consensus regarding the most
0 RCOG 1999 Br J Obstet Gynaecol 106,453-466
LONGITUDINAL ANALYSIS OF DOPPLER FETAL CIRCULATION
accurate means of detecting the fetus at risk of IUGR,
the most informative modes of monitoring, and the
timing of delivery.
Analysis of fetal haemodynamics therefore has the
potential to play an important role in the assessment of
fetal wellbeing. It complements existing techniques,
such as the ultrasound measurement of growth and
amniotic fluid volume, and exhibits changes, which typically coincide with the reduction in fetal growth, but
appear before the development of an abnormal CTG or
biophysical profile. Because the changes occur in the
fetal circulation at a time when other tests are still normal, they should be helpful in identifying the fetus that
is truly hypoxic. In addition, because fetal Doppler
waveform measurements reflect the ability (or inability)
of the fetus to cope with hypoxia, a deterioration in the
values obtained may be useful in determining the optimum time for delivery, to minimise perinatal morbidity
and any subsequent complications.
We have described longitudinal changes in fetal
biometry and various Doppler ultrasound indices and
ratios in complicated and uncomplicated pregnancies.
Longitudinal data pose unique problems in analysis.
Nevertheless, a definite sequence of events within the
fetus is proposed in pregnancies that develop chronic
uteroplacental complications such as pre-eclampsia and
IUGR. Knowledge of this sequence of events may be
useful in assessing the condition of an individual fetus
under investigation: we have demonstrated the potential
of this data in the management of the at risk fetus. Further studies are needed to determine the optimum use of
this data for everyday clinical practice.
References
1 Danielian PJ, Allman ACJ, Steer PJ. Is obstetric and neonatal outcome
worse in fetuses who fail to reach their own growth potential? Br J
Obstet Gynaecol 1992; 99: 452454.
2 Department of Health. Report on confidential enquiries into maternal
deaths in the United Kingdom: HMSO, London, 1991-1993.
3 Gardosi J, Mongelli M, Wilcox M, Chang A. Screening and assessment of fetal growth. Amsterdam: Elsevier. 1992.
4 Campbell S, Wilkin D. Ultrasonic measurement of fetal abdominal
circumference in the estimation of fetal weight. Br J Obster Gynaecol
1975; 82: 689497.
5 Chard T, Yoong A, Macintosh M. The myth of fetal growth retardation
at term. Br J Obsrer Gynaecol1993; 100: 1076-1081.
6 Harrington K, Campbell S. Fetal size and growth. Curr Opinion
Obsrer Gynecol 1993; 5: 186-194.
7 Abitbol MM. Ober WB. Gallo GR. Experimental toxaemia of pregnancy in the monkey with a preliminary report on renin and aldosterone. Am J Pathol 1977; 86: 573-575.
8 Combs CS, Katz MA, Kitzmiller JL, Brescia RJ. Experimental preeclampsia produced by chronic constriction of the lower aorta: validation with longitudinal blood pressure measurements in conscious
rhesus monkeys. Am J Obsret Gynecol 1993; 169: 215-223.
9 Trudinger BJ, Stevens D, Connelly A. Umbilical artery flow velocity
waveforms and placental resistance: the effect of embolization of the
umbilical circulation. Am J Obsrer Gynecol 1987; 157: 143-1448.
10 Deane C. Harrington K. A practical approach to obtaining optimum
Doppler signals. In: Harrington K,Campbell S. editors. A Colour
Atlas of Doppler Ultrasonography in Obstetrics. London: Arnold,
1995: 3 5 4 6 .
0 RCOG 1999 Br J Obstet Gynaecol 106,453466
465
11 Fitzgerald DE. Drumm JE. Non-invasive measurement of human
fetal circulation using ultrasound: a new method. BMJ 1977; 2:
1450-1451.
12 Rizzo G, Capponi A, Arduini D, Romanini C. The value of fetal arterial, cardiac and venous flows in predicting pH and blood gases measured in umbilical blwd at cordocentesis in growth retarded fetuses.
Br J Obstet Gynaecoll995; 102: 963-969.
13 Wladimiroff JW.Wijngaard JAGW. Degani S, Noordam J. van Eyck
J. Tonge HM. Cerebral and umbilical artery arterial flow velocity
waveforms in normal and growth retarded pregnancies. Obsrer
Gynecol 1987; 69: 705-709.
14 Hecher K, Campbell S , Doyle P, Harrington K. Assessment of fetal
compromise by Doppler ultrasound investigation of the fetal circulation. Circulation 1995; 91: 129-138.
15 Harrington K. Carpenter RG. Nguyen M, Campbell S. Changes
observed in Doppler studies of the fetal circulation in pregnancies
complicated by pre-eclampsia or the delivery of a small-for-gestational age baby: I. Cross-sectional analysis. Ultrasound Obster
Gynaecoll996; 6: 19-28.
16 Chamberlain PF, Manning FA, Morrison I. Harman CR, Lange IR.
Ultrasound evaluation of amniotic fluid volume. I. The relationship of
marginal and decreased amniotic fluid volumes to perinatal outcome.
Am J Obstet Gynecoll984; 150: 245-249.
17 Davey DA, MacGillivray I. The classification and definition of the
hypertensive disorders of pregnancy. Am J Obster Gynecol 1988;
158: 892-898.
18 Yudkin PL. Aboualfa M. Eyre JA, Redman CW, Wilkinson AR. New
birth weight and head circumference centiles for gestational ages
24-42 weeks. Early Hum Dev 1987; 15: 45-52.
19 Goldstein H. Multilevel statistical models. London: Edward Arnold,
1995.
20 Goldstein H, Rasbash J, Plewis I, Draper D, Browne. Young M.
Woodhouse G, Healy M. A users guide to Mlwin. Multilevel Models
Project, Institute of Education, University of London. 1998.
21 Krebs C. Macara LM. Leiser R, Bowman AW. Greer IA. Kingdom
JCP. Intra-uterine growth restriction with absent end-diastolic flow
velocity in the umbilical artery is associated with maldevelopment
of the placental terminal villous tree. Am J Obster Gynecol 1996;
175: 15361542.
22 Chang TC, Robson SC, Boys RJ, Spencer JA. Prediction of the small
for gestational age infant: which ultrasonic measurement is best?
Obstet Gynecoll992; 80: 103CL1038.
23 Hedriana HL, Moore TR. A comparison of single versus multiple
growth ultrasonographic examinations in predicting birth weight. Am
J Obstet Gynecol 1994; 170: 160&1606.
24 Vyas S. Nicolaides KH, Bower S, Campbell S. Middle cerebral artery
flow velocity waveforms in fetal hypoxaemia. Br J Obsret Gynaecol
1990; 97:797-803.
25 Akalin-Sel T, Nicolaides KH, Campbell S. Understanding the
pathophysiology of intrauterine growth retardation: the role of the
lower limb reflex in redistribution of blood flow. Eur J Oh Gyn
1992; 46: 2 4 ,
26 Chang TC, Robson SC. Spencer JAD, Gallivan S. Prediction of
perinatal morbidity at term in small fetuses: comparison of fetal
growth and Doppler ultrasound. Br J Obstet Gynaecol 1994; 101:
422427.
27 Spencer JAD, Price J, Lee A. Influence of fetal breathing and movements on variability of umbilical Doppler indices using different numbers of waveforms. J UltrasoundMed 1991; 10: 3 7 4 1 .
28 Van Eyck J, Wladimiroff JW. Noordam MJ. Tonge HM, Prechtl HIT.
The blood flow velocity waveform in the fetal descending aorta: its
relationship to fetal behavioural states in normal pregnancy at 37-38
weeks gestation. Early Hum Dev 1985; 12: 137-143.
29 Frusca T, Morassi L. Percorelli S, Grifolato P, Gastaldi A. Histological features of uteroplacental vessels in normal and hypertensive patients in relation to birth weight. Br J Obstet Gynaecol 1989;
96: 835-839.
30 Khong Ty, De Wolf E Robertson WB. Brosens I. Inadequate maternal
vascular response to placentation in pregnancies complicated by preeclampsia and by small-for-gestational age infants. Br J Obster
Gynaecoll986; 93: 1049-1059.
3 1 Brosens I, Robertson WB, Dixon HG. The physiological response of
the vessels of the placental bed to normal pregnancy. J Pathol Bacreriol 1967; 93:569-579.
466 K . H A R R I N G T O N E T A L .
32 Brosens I, Robertson WB, Dixon HG.The role of the spiral arteries
in the pathogenesis of pre-eclampsia. Obster Gynaecol Annu 1972;
1: 177-191.
33 Kingdom JCP, Burrell SJ, Kaufmann. Pathology and clinical implications of abnormal umbilical artery Doppler waveforms. Ultrasound
Obstet Gynecoll997;9 271-286.
34 Redman CWG, Sargent IL, Sutton L. Immunological aspects of
human pregnancy and its disorders. London: Buttenvorths, 1984.
35 Mass6 J, Forest JC, Moutquin JM, Marcoux S, Brideau NA, BClanger
M. A prospective study of several potential biologic markers for early
prediction of the development of pre-eclampsia. Am J Obsrer Gynecol
1993;169 501-508.
36 Bewley S,Chard T,Grudzinskas G, Campbell S . The relationship of
uterine and umbilical Doppler resistance to fetal and placental protein
synthesis in the second trimester. Placenta 1993;1 4 663470.
37 De long CLD, Dekker GA, Sibai BM. The renin angiotensin aldosterone system in pre-eclampsia; a review. Clin Perinatol 1991;18:
683-71 1.
38 Oian P, Monrad-Hansen H, Maltau JM. Serum uric acid correlates
39
40
41
42
with bZ-microglobulin in pre-eclampsia. Acfa ObsfefGynecol Scand
1986;65: 103-106.
Weiner 2,Farmakides G, Schulman H, Penny 6.Central and peripheral hemodynamicchanges in fetuses with absent end diastolic velocity in umbilical artery: correlation with computerized fetal heart rate
pattern. Am J Obsrer Gynecoll994; 170:509-515.
Rizzo G,Arduini D. Fetal cardiac function in intra-uterine growth
retardation. Am J Obsrer Gynecol1991; 165:876-882.
Manning F, Platt L, Sipos L. Antepartum fetal evaluation: development of a fetal biophysical score. Am J Obstet Gynecol 1980;136
787-795.
Guzman ER,Vintzeilos AM, Martins M, Benito C, Houlihan C, Hanley M. The efficacy of individual computer heart rate indices in
detecting acidemia at birth in growth-restricted fetuses. Obstet
Gynecol, 1996;87:969-974.
Received 23 December 1997
Returned for revision 7 May 1998
Accepted 22 September 1998
0 RCOG 1999 Br J Obstet Gynaecol 106,453466