International Journal of Technology Assessment in Health Care, 8:Suppl. 1 (1992), 160-169. Copyright © 1992 Cambridge University Press. Printed in the U.S.A.
ANTENATAL DIAGNOSIS OF INTRAUTERINE GROWTH RETARDATION BY ULTRASOUND Karel Marsal University of Lund
Abstract Ultrasound estimation of fetal weight or ultrasound measurement of fetal abdomen alone enables identification of small-for-gestational-age fetuses. A prerequisite for this is a reliable dating of pregnancy, which is provided by a routine ultrasound screening in the first half of gestation. The fetal growth can be followed by serial fetometric measurements. As a standard, charts of intrauterine growth based on the ultrasonic measurement can be used. As a secondary diagnostic test for monitoring fetal health in pregnancies suspected of intrauterine growth retardation, Doppler ultrasound evaluation of fetal and uteroplacental hemodynamics provided useful for early detection of imminent fetal distress.
Fetal growth is measured as the change in fetal weight and size in relation to age (1). Recognition of impaired intrauterine growth, that is, intrauterine growth retardation (IUGR), is of great clinical importance, as the growth-retarded fetuses are at risk of developing asphyxia and they show increased perinatal mortality and morbidity. It is thus desirable to establish the size and growth of the fetus in utero by use of a noninvasive method, enabling direct measurements of fetal dimensions. Besides, to further evaluate the well-being of fetuses identified as growth-retarded, a method for dynamic studies of various fetal functions is required. Real-time ultrasound fulfills both above requirements and is therefore widely used for fetometry and study of fetal movements, fetal breathing, evaluation of the amniotic fluid, etc. More recently, another modality of ultrasound—Doppler ultrasound — is increasingly used for studies of fetal circulation. SAFETY OF THE USE OF ULTRASOUND IN PREGNANCY
The safety of the antenatal application of diagnostic ultrasound is continuously supervised by researchers and international organizations as ultrasound can undoubtedly produce biological effects. However, there is no evidence of adverse bio-effects from diagnostic ultrasound on mammalian tissues. No adverse effects have been found in follow-up studies of children exposed for diagnostic ultrasound in utero (26). The American Institute of Ultrasound in Medicine Bio-Effects Committee stated The Malmo studies reported below are based on the work by Drs. S. Eik-Nes, J. Laurin, P.-H. Persson, and research midwife B.-M. Weldner. The fruitful cooperation with them throughout the years is gratefully acknowledged.
160
Ultrasound diagnosis of IUGR
in 1983 that current data indicate that the benefits to patients of the prudent use of diagnostic ultrasound far outweigh the eventual risk. The European Federation for Ultrasound in Medicine and Biology: Biological Effects Committee published in May 1984 in Strasbourg a document stating that "Routine ultrasound scanning of every woman during pregnancy is not contra-indicated by the evidence currently available from biological investigations and its performance should be left to clinical judgement." ULTRASOUND VARIABLES
Modern real-time ultrasound scanners produce generally good quality images with high resolution. This enables a detailed study of fetal anatomy and accurate measurement of fetal size already in early pregnancy. The following variables are commonly used for estimation of fetal size: crown-rump length (CRL), biparietal diameter (BPD), femur length (FL), and abdominal circumference, area or diameter (AD: mean of two orthogonal abdominal diameters). For definitions and measurement techniques see standard textbooks of obstetrical ultrasound (2;9). To lessen the measurement error, a mean of measurements of each variable in three subsequent images should be used. Fetal weight can be estimated from fetal dimensions obtained by ultrasound. Many formulas have been published for this purpose. For the Scandinavian population, EikNes et al. (12) designed a formula based on BPD and AD. This formula was further developed by Persson and Weldner (33) to include also FL: log fetal weight = 0.972 x log BPD + 1.743 x log AD + 0.367 x log FL -2.646.
(1)
The prediction error (standard deviation; SD) is 7.1%. Using pulsed wave Doppler ultrasound, signals of blood velocity can be recorded from a number of fetal vessels (descending aorta, cerebral arteries, renal arteries, etc.). Blood velocity signals from umbilical artery are very easily obtained even by continuous wave Doppler ultrasound. The velocity waveform contains important information on fetal circulation, which can be used to indicate fetal well-being. Figure 1 presents indices most often used to characterize the waveform. Of them, pulsatility index (PI) is the most robust one. A decrease in the diastolic flow gives an increased PI. A semiquantitative evaluation of waveform classes (blood flow class; BFC) (22) or even just discrimination between the waveforms with end-diastolic flow velocity present and absent was shown to have high predictive capacity as to the occurrence of IUGR and fetal distress (20;27). ULTRASOUND ASSESSMENT OF GESTATIONAL AGE
Knowledge of gestational age is a prerequisite for evaluation of fetal growth. The gestational age can be estimated accurately if the date of conception is known. This is, however, not the case in most of instances. Calculation of dates from the last menstrual period is uncertain in 20-30% of all pregnant women (6;18). Ultrasound measurement of fetal dimensions in early pregnancy has proved to be the most accurate assessment of gestational age (1), superior even to a reliable menstrual history (8). The most accurate estimation of gestational age is achieved by measurements of CRL between 6 and 12 weeks and by measurements of BPD between 13 and 18 weeks (8). In Malmo, a formula combining BPD and FL used at 15-18 weeks estimated the gestational age with SD of 2.7 days (32): INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL. 1, 1992
161
Marsal
50 -
B 0 L
A-B
PI=
-rr~ V
RI =
AB
A / B ratio (S / D ratio) Figure 1. Indices used to characterize the blood velocity waveform recorded by Doppler ultrasound from fetal arteries. A = peak systolic velocity (S); B = the least diastolic velocity (D); V = mean velocity over the heart cycle.
gestational age (days) = BPD x 1.2 + FL x 1.0 + 49.
(2)
The routine dating of pregnancies by ultrasound influences profoundly the clinical obstetrical practice by reducing the number of suspected postmaturity, facilitating the decision making in preterm pregnancies, enabling the evaluation of intrauterine growth, etc. Thus, it can be concluded that the reliability of the estimation of gestational age in itself justifies a screening ultrasound program in early gestation. INTRAUTERINE GROWTH CURVES AND DIAGNOSIS OF IUGR
In clinical practice, the intrauterine growth is usually judged by its endpoint, that is, the birthweight. The birthweight is compared to standard charts of the growth and, to define IUGR, certain centiles are used. Most of the published growth charts show a flattening of the curve toward and beyond term. This was interpreted as an expression of a placental constrain of normal fetal growth (19). In the studies constituting the basis of those charts, the gestational age of pregnancies was most often not reliably known. Two independent Scandinavian studies have shown that the growth curves based on ultrasonically dated pregnancies have an appearance different from that of the curves of pregnancies where the age was not ascertained (15;34). The former curves have a linear course even at and beyond term, and their distribution is normal and narrow with SD of 11% of the mean birthweight (15). The linear relationship between age and growth continues even postpartum, indicating that there is a uniform incremental weight 162
INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL. 1, 1992
Ultrasound diagnosis of IUGR
26
28
30
32
34
36
38
40
42
completed weeks 6ESTATI0NAL AGE Figure 2. Intrauterine growth chart of Malmo population based on the ultrasonically estimated fetal weights. Mean ± 2 SD.
gain during the last trimester of pregnancy and during the first 4 months of postnatal life (11). The growth charts based on birthweights are not necessarily representative for the intrauterine population, especially not in the preterm period. A preterm delivery is abnormal by definition and is often associated with complications of pregnancy. The growth charts based on birthweights of ultrasonically dated neonates are usually weakly sigmoid-shaped with negatively skewed distribution in the preterm and postterm period when compared to the intrauterine weight curves (31;35). An intrauterine weight curve shows a linear course with normal distribution and SD of 11% after 32 weeks of gestation (Figure 2) (33). To establish the diagnosis of IUGR is not easy. Already the definitions of IUGR differ considerably between countries and researchers, the cutoff points being defined as the 10th, 5th, or 3rd centile, and in Scandinavia as the mean birthweight - 2 SD. Different results will be obtained using growth charts based on birthweights or intrauterine weights. At 33 weeks, the 5th centile of birthweight corresponds to 4.5 SD below the mean of intrauterine weights and at 39 weeks, the 5th centile corresponds to 1.8 SD below the mean (31). Among preterm neonates, a high frequency of small-forgestational-age (SGA) infants is found if the intrauterine weight standards are used (Table 1) (25). Thus, it is not obvious which growth curves and which cutoff points are preferable for IUGR diagnosis. Many attempts were made to use various fetometric variables or combination and ratios of them to diagnose IUGR. Of single ultrasound measurements, the measures of fetal abdomen showed the best diagnostic efficacy. A ratio between the fetal head and abdomen is used to differentiate between the symmetrical and asymmetrical types INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL. 1, 1992
163
Marsal Table 1. Frequency of IUGR (Birthweight < Mean - 2 SD) According to the Intrauterine Weight Chart
Gestational age at delivery (weeks)
No. of IUGR
% IUGR of the total no. of deliveries
«28 29 30 31
1 1 5 8
3 10 29 22
32 33 34 35 36
10 11 13 9 17
23 18 14 6 5
37 38
7 20
2
39 40 >41
26 31 22
2
2
")
Preterm ' 10.1%
1 1 Term
f 2-3%
33 J
Source: Laurin and Persson (24).
of IUGR, the latter being the predominating one (7). From the above-described experience on intrauterine weight curves, it seems reasonable to use the ultrasound estimation of fetal weight to diagnose IUGR. By predicting deviation from the expected birthweight, it should be possible to characterize the individual growth and to identify in this way a group of SGA fetuses. Based on the finding of normal distribution of estimated intrauterine weights, Eik-Nes et al. (13) developed a method for predicting the birthweight deviation (%) from the deviations in BPD and AD. An advantage of the method is that it is not related to a certain gestational week and can be used from 32 weeks until term. Later, FL deviation was added to the formula by Laurin and Persson (24): Predicted birthweight deviation (%) = BPD dev. x 0.86 + AD dev. x 1.59 + FL dev. x 0.91 -0.11.
(3)
When tested on the Malmo population, the mean of the predicted birthweight deviation was 0% and SD 11%. The weight deviation can be also calculated from the ultrasonically estimated intrauterine weight (Equation [1]) and the expected weight at that gestational age (24): intrauterine weight deviation (%) = (estimated weight - expected weight) x 100/expected weight.
(4)
Similar approaches to the IUGR diagnosis were applied later by others, for example, by calculating individual growth curve standards (10) or by estimating the ideal neonatal weight (29). To guarantee high reliability of the ultrasonic method, both when applied for dating of gestations and for fetal weight estimation, it is very important to continuously monitor
164
INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL 1, 1992
Ultrasound diagnosis of IUGR Table 2. Ultrasound Screening for IUGR Neilson et al. (28) Method
CRL, trunk area
IUGR definition IUGR prevalence (%) Sensitivity (%) Specificity (
ANTENATAL DIAGNOSIS OF INTRAUTERINE GROWTH RETARDATION BY ULTRASOUND Karel Marsal University of Lund
Abstract Ultrasound estimation of fetal weight or ultrasound measurement of fetal abdomen alone enables identification of small-for-gestational-age fetuses. A prerequisite for this is a reliable dating of pregnancy, which is provided by a routine ultrasound screening in the first half of gestation. The fetal growth can be followed by serial fetometric measurements. As a standard, charts of intrauterine growth based on the ultrasonic measurement can be used. As a secondary diagnostic test for monitoring fetal health in pregnancies suspected of intrauterine growth retardation, Doppler ultrasound evaluation of fetal and uteroplacental hemodynamics provided useful for early detection of imminent fetal distress.
Fetal growth is measured as the change in fetal weight and size in relation to age (1). Recognition of impaired intrauterine growth, that is, intrauterine growth retardation (IUGR), is of great clinical importance, as the growth-retarded fetuses are at risk of developing asphyxia and they show increased perinatal mortality and morbidity. It is thus desirable to establish the size and growth of the fetus in utero by use of a noninvasive method, enabling direct measurements of fetal dimensions. Besides, to further evaluate the well-being of fetuses identified as growth-retarded, a method for dynamic studies of various fetal functions is required. Real-time ultrasound fulfills both above requirements and is therefore widely used for fetometry and study of fetal movements, fetal breathing, evaluation of the amniotic fluid, etc. More recently, another modality of ultrasound—Doppler ultrasound — is increasingly used for studies of fetal circulation. SAFETY OF THE USE OF ULTRASOUND IN PREGNANCY
The safety of the antenatal application of diagnostic ultrasound is continuously supervised by researchers and international organizations as ultrasound can undoubtedly produce biological effects. However, there is no evidence of adverse bio-effects from diagnostic ultrasound on mammalian tissues. No adverse effects have been found in follow-up studies of children exposed for diagnostic ultrasound in utero (26). The American Institute of Ultrasound in Medicine Bio-Effects Committee stated The Malmo studies reported below are based on the work by Drs. S. Eik-Nes, J. Laurin, P.-H. Persson, and research midwife B.-M. Weldner. The fruitful cooperation with them throughout the years is gratefully acknowledged.
160
Ultrasound diagnosis of IUGR
in 1983 that current data indicate that the benefits to patients of the prudent use of diagnostic ultrasound far outweigh the eventual risk. The European Federation for Ultrasound in Medicine and Biology: Biological Effects Committee published in May 1984 in Strasbourg a document stating that "Routine ultrasound scanning of every woman during pregnancy is not contra-indicated by the evidence currently available from biological investigations and its performance should be left to clinical judgement." ULTRASOUND VARIABLES
Modern real-time ultrasound scanners produce generally good quality images with high resolution. This enables a detailed study of fetal anatomy and accurate measurement of fetal size already in early pregnancy. The following variables are commonly used for estimation of fetal size: crown-rump length (CRL), biparietal diameter (BPD), femur length (FL), and abdominal circumference, area or diameter (AD: mean of two orthogonal abdominal diameters). For definitions and measurement techniques see standard textbooks of obstetrical ultrasound (2;9). To lessen the measurement error, a mean of measurements of each variable in three subsequent images should be used. Fetal weight can be estimated from fetal dimensions obtained by ultrasound. Many formulas have been published for this purpose. For the Scandinavian population, EikNes et al. (12) designed a formula based on BPD and AD. This formula was further developed by Persson and Weldner (33) to include also FL: log fetal weight = 0.972 x log BPD + 1.743 x log AD + 0.367 x log FL -2.646.
(1)
The prediction error (standard deviation; SD) is 7.1%. Using pulsed wave Doppler ultrasound, signals of blood velocity can be recorded from a number of fetal vessels (descending aorta, cerebral arteries, renal arteries, etc.). Blood velocity signals from umbilical artery are very easily obtained even by continuous wave Doppler ultrasound. The velocity waveform contains important information on fetal circulation, which can be used to indicate fetal well-being. Figure 1 presents indices most often used to characterize the waveform. Of them, pulsatility index (PI) is the most robust one. A decrease in the diastolic flow gives an increased PI. A semiquantitative evaluation of waveform classes (blood flow class; BFC) (22) or even just discrimination between the waveforms with end-diastolic flow velocity present and absent was shown to have high predictive capacity as to the occurrence of IUGR and fetal distress (20;27). ULTRASOUND ASSESSMENT OF GESTATIONAL AGE
Knowledge of gestational age is a prerequisite for evaluation of fetal growth. The gestational age can be estimated accurately if the date of conception is known. This is, however, not the case in most of instances. Calculation of dates from the last menstrual period is uncertain in 20-30% of all pregnant women (6;18). Ultrasound measurement of fetal dimensions in early pregnancy has proved to be the most accurate assessment of gestational age (1), superior even to a reliable menstrual history (8). The most accurate estimation of gestational age is achieved by measurements of CRL between 6 and 12 weeks and by measurements of BPD between 13 and 18 weeks (8). In Malmo, a formula combining BPD and FL used at 15-18 weeks estimated the gestational age with SD of 2.7 days (32): INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL. 1, 1992
161
Marsal
50 -
B 0 L
A-B
PI=
-rr~ V
RI =
AB
A / B ratio (S / D ratio) Figure 1. Indices used to characterize the blood velocity waveform recorded by Doppler ultrasound from fetal arteries. A = peak systolic velocity (S); B = the least diastolic velocity (D); V = mean velocity over the heart cycle.
gestational age (days) = BPD x 1.2 + FL x 1.0 + 49.
(2)
The routine dating of pregnancies by ultrasound influences profoundly the clinical obstetrical practice by reducing the number of suspected postmaturity, facilitating the decision making in preterm pregnancies, enabling the evaluation of intrauterine growth, etc. Thus, it can be concluded that the reliability of the estimation of gestational age in itself justifies a screening ultrasound program in early gestation. INTRAUTERINE GROWTH CURVES AND DIAGNOSIS OF IUGR
In clinical practice, the intrauterine growth is usually judged by its endpoint, that is, the birthweight. The birthweight is compared to standard charts of the growth and, to define IUGR, certain centiles are used. Most of the published growth charts show a flattening of the curve toward and beyond term. This was interpreted as an expression of a placental constrain of normal fetal growth (19). In the studies constituting the basis of those charts, the gestational age of pregnancies was most often not reliably known. Two independent Scandinavian studies have shown that the growth curves based on ultrasonically dated pregnancies have an appearance different from that of the curves of pregnancies where the age was not ascertained (15;34). The former curves have a linear course even at and beyond term, and their distribution is normal and narrow with SD of 11% of the mean birthweight (15). The linear relationship between age and growth continues even postpartum, indicating that there is a uniform incremental weight 162
INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL. 1, 1992
Ultrasound diagnosis of IUGR
26
28
30
32
34
36
38
40
42
completed weeks 6ESTATI0NAL AGE Figure 2. Intrauterine growth chart of Malmo population based on the ultrasonically estimated fetal weights. Mean ± 2 SD.
gain during the last trimester of pregnancy and during the first 4 months of postnatal life (11). The growth charts based on birthweights are not necessarily representative for the intrauterine population, especially not in the preterm period. A preterm delivery is abnormal by definition and is often associated with complications of pregnancy. The growth charts based on birthweights of ultrasonically dated neonates are usually weakly sigmoid-shaped with negatively skewed distribution in the preterm and postterm period when compared to the intrauterine weight curves (31;35). An intrauterine weight curve shows a linear course with normal distribution and SD of 11% after 32 weeks of gestation (Figure 2) (33). To establish the diagnosis of IUGR is not easy. Already the definitions of IUGR differ considerably between countries and researchers, the cutoff points being defined as the 10th, 5th, or 3rd centile, and in Scandinavia as the mean birthweight - 2 SD. Different results will be obtained using growth charts based on birthweights or intrauterine weights. At 33 weeks, the 5th centile of birthweight corresponds to 4.5 SD below the mean of intrauterine weights and at 39 weeks, the 5th centile corresponds to 1.8 SD below the mean (31). Among preterm neonates, a high frequency of small-forgestational-age (SGA) infants is found if the intrauterine weight standards are used (Table 1) (25). Thus, it is not obvious which growth curves and which cutoff points are preferable for IUGR diagnosis. Many attempts were made to use various fetometric variables or combination and ratios of them to diagnose IUGR. Of single ultrasound measurements, the measures of fetal abdomen showed the best diagnostic efficacy. A ratio between the fetal head and abdomen is used to differentiate between the symmetrical and asymmetrical types INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL. 1, 1992
163
Marsal Table 1. Frequency of IUGR (Birthweight < Mean - 2 SD) According to the Intrauterine Weight Chart
Gestational age at delivery (weeks)
No. of IUGR
% IUGR of the total no. of deliveries
«28 29 30 31
1 1 5 8
3 10 29 22
32 33 34 35 36
10 11 13 9 17
23 18 14 6 5
37 38
7 20
2
39 40 >41
26 31 22
2
2
")
Preterm ' 10.1%
1 1 Term
f 2-3%
33 J
Source: Laurin and Persson (24).
of IUGR, the latter being the predominating one (7). From the above-described experience on intrauterine weight curves, it seems reasonable to use the ultrasound estimation of fetal weight to diagnose IUGR. By predicting deviation from the expected birthweight, it should be possible to characterize the individual growth and to identify in this way a group of SGA fetuses. Based on the finding of normal distribution of estimated intrauterine weights, Eik-Nes et al. (13) developed a method for predicting the birthweight deviation (%) from the deviations in BPD and AD. An advantage of the method is that it is not related to a certain gestational week and can be used from 32 weeks until term. Later, FL deviation was added to the formula by Laurin and Persson (24): Predicted birthweight deviation (%) = BPD dev. x 0.86 + AD dev. x 1.59 + FL dev. x 0.91 -0.11.
(3)
When tested on the Malmo population, the mean of the predicted birthweight deviation was 0% and SD 11%. The weight deviation can be also calculated from the ultrasonically estimated intrauterine weight (Equation [1]) and the expected weight at that gestational age (24): intrauterine weight deviation (%) = (estimated weight - expected weight) x 100/expected weight.
(4)
Similar approaches to the IUGR diagnosis were applied later by others, for example, by calculating individual growth curve standards (10) or by estimating the ideal neonatal weight (29). To guarantee high reliability of the ultrasonic method, both when applied for dating of gestations and for fetal weight estimation, it is very important to continuously monitor
164
INTL. J. OF TECHNOLOGY ASSESSMENT IN HEALTH CARE 8:SUPPL 1, 1992
Ultrasound diagnosis of IUGR Table 2. Ultrasound Screening for IUGR Neilson et al. (28) Method
CRL, trunk area
IUGR definition IUGR prevalence (%) Sensitivity (%) Specificity (
