Early fetal growth: concepts and pitfalls
Normal prenatal growth is one of the best indicators of good health and nutrition, while abnormal growth (macrosomia or growth restriction) is associated with an increased likelihood of perinatal complications1, 2. Thus, screening for and management of abnormal fetal growth are important aspects of prenatal care. Growth-restricted fetuses are at an increased risk of perinatal and infant death3, 4. Fetal macrosomia, an increasing clinical challenge, is associated with a risk of prolonged labor, perinatal mortality, asphyxial injuries, meconium aspiration, shoulder dystocia, soft-tissue trauma, humeral and clavicular fractures, brachial plexus and facial palsies5–8. Evidence is emerging that macrosomic birth is also associated with future health risks9, 10. Unfortunately, sonographic fetal weight estimation at term is not perfectly accurate, particularly in cases of macrosomia11–13, and efforts to improve screening for fetuses at risk of abnormal growth should therefore be encouraged.
In this issue of the Journal, Thorsell et al. analyze the risk of macrosomia with respect to a proxy for fetal growth during the first trimester14. They used the discrepancy between the expected date of delivery based on ultrasound examination at 16–20 weeks and that derived from the last menstrual period (LMP) to classify the study population into three groups: those with fetal size in keeping with LMP (discrepancy ⩽ 6 days), those with expected date of delivery moved forward ≥ 7 days (smaller than expected) and those with expected date moved backward ≥ 7 days (larger than expected). Based on a retrospective cohort of 19 377 women, they demonstrated that in the larger-than-expected group, there was a 59% increase in the risk of a birth weight ≥ 4500 g and a 145% increase in the risk of a birth weight of ≥ 5000 g (odds ratio (OR), 1.59 (95% CI, 1.12–2.24) and 2.45 (95% CI, 1.22–4.90), respectively). In the smaller-than-expected group, the risk of being born macrosomic was reduced by around 20% (OR, 0.81 (95% CI, 0.73–0.90) for birth weight ≥ 4000 g; OR, 0.70 (95% CI, 0.55–0.87) for birth weight ≥ 4500 g; and OR 0.66 (95% CI, 0.35–1.28) for birth weight ≥ 5000 g). Accordingly, they concluded that accelerated fetal growth can be detected as early as at the beginning of the second trimester.
This interesting report adds to the increasingly complex picture of fetal growth. Fetal growth used to be evaluated solely from size at birth. However, size at birth does not correspond just to fetal growth rate, also depending on the length of gestation. Birth size must therefore be considered with respect to gestational age. The most widely used size indictor is birth weight. However, a birth weight of 2300 g does not have the same significance at 35 weeks (normal preterm birth weight) as it does at 38 weeks (small, term fetus)15. Most diagnostic classifications of fetal growth, for both individuals and populations, are based on birth weight for gestational age16. Birth weight, indirectly reflecting fetal growth, is also strongly associated with fetal, neonatal and post neonatal morbidity and mortality, and with morbidity during infancy and childhood17–19. However, one should always keep in mind the difference between fetal size, which represents a single measurement point, and fetal growth, which is a dynamic process. The two concepts contain completely different information, as we learned for adults during our medical studies: a person's weight and his or her weight variations over the past few months provide different information. The same applies to fetuses. Fetal size can be normal despite abnormal growth, while normal fetal growth can be associated with abnormal size.
Gestational age, fetal size and fetal growth
Gestational age may be easy to determine in assisted reproductive technology (ART) pregnancies and when the date of the LMP is known with certainty and the menstrual cycle is regular. However, most patients conceive spontaneously and LMP is certain in only a fraction of cases20. Fetal size nomograms have been constructed from pregnancies in which the age was certain. Pregnancy dating by transabdominal sonographic measurement of crown–rump length (CRL) was first described by Robinson21, 22 and subsequent studies using high-frequency transvaginal sonography showed similar growth patterns in early pregnancy23, 24. These studies assumed uniform growth across the first half of pregnancy, regardless of maternal or fetal characteristics. Indeed, because measurement error and individual variations in fetal size are small in the first half of pregnancy, ultrasound proved much more reliable than did LMP for estimating gestational age. Therefore, fetal size up to mid-pregnancy was considered equivalent to gestational age, and this appeared to be generally, if not absolutely, correct25. In those days things were relatively simple: LMP was used only to calculate the approximate date at which the first ultrasound scan should be booked. At the time of the first scan, ultrasound measurements of fetal size, based for example on CRL and/or biparietal diameter (BPD), together with validated algorithms for deriving GA, were considered more reliable than LMP for dating. Subsequent scans were used to assess fetal growth, most of which was considered to occur during the second half of pregnancy, with physiological variations in fetal size emerging during this half of pregnancy26. Accordingly, it was claimed a few years ago that ‘it is time to forget the LMP’27.
Early fetal growth: concepts
Prenatal evaluation of fetal size and growth has changed in recent years. It is likely that, as gestational age advances, the biological variability of fetal size increases, reflecting variable fetal growth, and that such variability could emerge earlier than previously thought; it is now clear that fetal growth is not uniform even during the first trimester28, 29. An increasing number of studies are examining possible relationships between early fetal size and subsequent outcome. We have learned that fetuses can show tremendous measurement differences as early as the 11–14-week scan, owing to early expression of the genetic or chromosomal background, as well as have markers for possible malformations30. It is now clear that the vast majority of major fetal abnormalities can be diagnosed prenatally by ultrasound examination and that most of these abnormalities can be detected in the first trimester31; it would be surprising, therefore, if growth was uniform across fetuses during this trimester. Yet, during the first half of pregnancy, prenatal evaluation is usually limited to a single ultrasound examination, fetal size being used to determine gestational age, based on the assumption of uniform growth. Early markers of fetal growth are clearly required to move beyond this concept.
Twin pregnancies offer an interesting model for evaluating early fetal growth, since each twin can serve as both subject and control. Many studies have evaluated the impact of early intertwin growth discrepancies on pregnancy outcome. Intertwin discrepancy may be an ominous finding with respect to adverse perinatal outcome, including chromosomal abnormalities, structural anomalies, intrauterine growth restriction and fetal death. However, results are controversial32–39. Another interesting population for studying early fetal growth consists of ART pregnancies40, in which gestational age is considered certain. However, neither twin nor ART pregnancies can be considered ‘normal’ in this setting, and it may not be appropriate to extrapolate findings in these populations to normal singleton pregnancies.
A third way of evaluating early fetal growth consists of comparing the actual size of the embryo or fetus with the size predicted using standard growth charts and LMP41. The use of large databases has made it possible to better understand the impact and significance of slight deviations of CRL from the expected value. Using this approach, first-trimester growth restriction has been linked to an increased risk of low birth weight, a low birth-weight percentile for gestational age, and extremely preterm birth. In 1998, Smith et al.42 published a large study in the New England Journal of Medicine. In a cohort of around 4000 women with certain LMP, they found a clear association between a smaller-than-expected CRL and both growth restriction and extreme preterm birth. This was confirmed in a large cohort of ART pregnancies40. In a previous study by Thorsell et al.43, it was found that pregnancies in which the expected date of delivery was moved forward 7 days or more at the time of dating ultrasound faced a two-fold increase in the risk of the fetus being small-for-gestational age at birth. It is also known that a smaller-than-expected first-trimester CRL is associated with an increased likelihood of miscarriage44.
The paper of Thorsell et al.14 in this issue of the Journal is in line with a report from Hackmon et al.45, who demonstrated that severe macrosomia can manifest as early as 11–14 weeks. They used the same approach, comparing the actual size of the embryo or fetus with its expected size based on standard growth charts and LMP.
Finally, early fetal growth can be assessed by using a previous ultrasound measurement as reference. An association between early BPD growth and birth weight was found in such a study. The assessment of fetal growth therefore becomes independent of the gestational-age error46.
These observations add to a growing body of evidence suggesting that the duration of pregnancy and the complications of late pregnancy (preterm birth, growth restriction or macrosomia) may be the ultimate consequence of conditions that have their origins in the very earliest weeks of gestation, preceding the first prenatal visit35, 47. The relationship between early fetal growth and later pregnancy outcome is difficult to interpret, however. It may be that the same factors that determine nutrition in later pregnancy also do so in the first trimester. Intrauterine factors, such as abnormal placentation or unequal distribution of uterine blood flow in twin pregnancies, may affect growth earlier than previously thought. Alternatively, a favorable or unfavorable environment (e.g. nutritional, hormonal) in the first trimester might permanently affect the fetus and its growth potential41, 47, 48. A recent study also showed that maternal physical characteristics and lifestyle habits are associated with first-trimester fetal growth based on CRL. Higher diastolic blood pressure and a higher hematocrit were both associated with a shorter CRL, as were maternal smoking and a lack of folic acid supplementation49. Shorter first-trimester CRL was associated with preterm birth, small size for gestational age at birth and low birth weight, which are associated with increased perinatal mortality and morbidity49. Interestingly, first-trimester growth variations were also associated with growth in early childhood49. The paper of Thorsell et al.14 underlines that larger-than-expected ‘incorrectly dated’ fetuses may in fact simply be large fetuses.
Early fetal growth: pitfalls
Discrepancies between observed and expected fetal size at the first ultrasound examination may, however, principally reflect features of the menstrual history, particularly cycle length. This is important, as cycle length is positively associated with the risk of delivering a low-birth-weight or growth-restricted baby50, and a long menstruation–conception interval is associated with adverse outcome51–54. Studies of ART pregnancies, however, do not support this hypothesis.
Whatever the physiological reason for the relationship between early growth and pregnancy outcome, such findings may challenge our current practices. However, the approach of interpreting early fetal growth based on a discrepancy between expected and observed measurements should not be misunderstood and misused. Current practice is to simply re-evaluate gestational age if the measured CRL does not correlate with the value predicted from the LMP, even when the discrepancy is small20. Although this does not take into account individual variations resulting from pathological or physiological processes and may at first sight contradict the concept of early fetal growth, ultrasound is more accurate than is LMP for dating, and its use reduces the number of ‘post-term’ pregnancies55. A single fetal size measurement, of the CRL for example, cannot be used simultaneously to establish gestational age and to assess early fetal growth. In practice, sonographers should keep in mind that the actual size of the embryo or fetus in the first trimester is more likely to differ from the expected size because of variations in the timing of ovulation than because of abnormal early growth. They should avoid the pitfall of thinking first of abnormal early growth rather than of misdating when they find a discrepancy between the observed and expected CRL on the first scan.
However, such discrepancies should be recorded in order to assist clinical research into the obvious variability of early fetal growth. This could be particularly relevant in women with a certain date of LMP and a regular cycle, and in women who conceived through ART. Such information, together with other potential early markers of fetal growth, such as pregnancy-associated plasma protein-A, could help to predict abnormal subsequent fetal growth56, 57, allowing closer fetal assessment and surveillance, and early delivery when necessary. It could also help to plan rigorous interventional studies which may prove useful if offered earlier on in pregnancy58. However, the predictive power of early fetal growth is not sufficient for stand-alone screening in routine practice. This is most likely because there are various causes of abnormal fetal growth, some of which may manifest early and others later, and also because an embryo may be larger or smaller than expected for a variety of reasons. Thorsell et al.14 found that only a small proportion of macrosomic babies were larger than expected at the time of the dating ultrasound scan.
Another potential pitfall in the study of early fetal growth is related directly to the methodology used. If the reason for a discrepancy between observed and expected fetal size is misdating rather than abnormal fetal growth, an association between early and late growth might be deduced in error. A fetus conceived earlier than assumed will be larger than expected on ultrasound examination, and gestational age will be underestimated at birth. Such a baby could therefore be considered as large for gestational age at birth, leading to a spurious association between faster-than-expected growth in early pregnancy and the risk of macrosomia. Thorsell et al.14 avoid this pitfall by using a definition of large-for-gestational age that is independent of gestational age itself, namely, a birth weight of ≥ 4000 or 4500 g. Variations in the risk of delivering a large infant in relation to early growth cannot therefore be explained by gestational age error41. Another way to deal with the issue is to quote all gestational age-related outcomes for both LMP- and CRL-based gestational age. Because CRL is subject to measurement error59, larger CRL values will, on average, have a positive measurement error and therefore overestimate GA, while smaller values will, on average, have a negative measurement error and therefore underestimate GA. These measurement errors would ‘protect’ against spurious statistical associations.
The 11–14-week examination is already being seen as the most pertinent medical risk assessment for pregnancies at high medical risk that would benefit from specific management60. Most karyotype anomalies as well as fetal malformations can now be suspected as early as in the first trimester. Variations in early fetal growth, assessed by various means, could soon help to better identify cases at risk of abnormal growth—both growth restriction and macrosomia—later in pregnancy.