Ultrasound fetal weight estimation at term may do more harm than good

Fetal size assessment is an established part of antenatal care, be it by manual palpation, symphysis–fundus height measurement or ultrasound biometry. So longstanding is this practice, that the identification of a small-for-gestational-age (SGA) baby has become inadvertently adopted into clinical practice as an important goal. However, it is vital to acknowledge that the practice of screening for fetal growth restriction (FGR) was introduced because of the association of SGA with increased risk of stillbirth, neonatal death and neurological impairment1, 2. The presumed biological mechanism for these associations is that placental insufficiency leads to poor fetal growth, fetal hypoxemia and related adverse outcomes. The ongoing controversy regarding the correct choice of fetal biometry chart does not seem likely to settle down soon. There are many concerns regarding the appropriateness of using one of the 100 or so published fetal growth charts. Proponents of implementing 'national' charts for individual assessment rationalize their views on the completely unfounded basis that a baby's weight is determined by the passport held by its mother. Others highlight that the majority of fetal biometry reference charts were developed using infants born preterm – which are dubiously considered to be 'normal' and likely to skew the distribution of expected fetal weight downwards because of undiagnosed pathology. To overcome this concern, in this issue of the Journal, Nicolaides et al. present data that develop and validate the use of in-utero fetuses for the construction of fetal growth and birth-weight reference charts3, 4. The authors quite correctly suggest that variation between published charts is likely to be the consequence of underlying differences in the study populations from which they were derived. However, they add to the debate by proposing that a single international standard for all countries is not appropriate as it would underestimate SGA in countries with normal big babies, such as Norway, and overestimate SGA in countries with normal small babies, such as India. In making this statement, the authors clearly insinuate that a baby's weight is determined by the ethnicity of its parents and that the proportion of SGA births should be the same (10%) in all countries. It is often muted that 'Asian women have smaller babies'. The perennial question is whether Asian women have smaller babies because they are themselves smaller or whether the case is that Asian and Caucasian women of the same stature and in good health would have similar sized babies. The INTERGROWTH-21st (IG-21) collaboration conducted a well-planned and monitored study to answer this question and concluded that ethnicity per se did not influence fetal growth, but rather that ethnicity is a biological marker of malnutrition and poor health – both of which contribute to poor fetal growth5. This would support the natural intuition of clinicians who observe smaller infants in developing countries and recent economic migrants/refugees to high-income ones. These findings would also counter the assertion that the SGA rate should be identical in all countries. A well-developed growth standard such as IG-21 does not underestimate growth restriction in women from high-income countries such as Norway. These countries have obesity rates of 20–30% – even in childhood – and therefore, it is logical that fetal assessment will 'overdiagnose' fetal macrosomia and 'underdiagnose' SGA in this setting (Figure 1). Similarly, the IG-21 growth standard does not overestimate SGA when used in low-income countries such as India, where there is a 20–30% rate of stunting in infancy. The apparent 'overdiagnosis' of SGA in low-income countries is consistent with higher rates of stillbirth, perinatal death and neurological impairment from pregnancy and similar rates of failure-to-thrive in infancy. Other fetal growth reference standards, such as those from the World Health Organization collaboration6, have been developed using methodology similar to that of IG-21. However, comparatively less stringent entry criteria or study monitoring is likely to have resulted in the inclusion of women predisposed to either SGA or macrosomic birth and, as a consequence, influenced the size distribution in these charts. Customization of fetal growth for maternal characteristics has had a popular resurgence of late as a consequence of the debate over which fetal growth standards to use. The arguments presented above undermine the rationale for customization on the basis of maternal ethnicity. Can it possibly be correct to assume that the baby of an Asian/Afro-Caribbean or nulliparous mother is 'normally' small when these very women are two- to three-fold more likely to have a stillbirth or neonatal death? The principal observation behind the adoption of customized charts was the apparent reduction in stillbirths in units that adopted the customization program in England and Wales7. However, recent population-based linkage studies from Scotland have shown that customization of fetal size did not improve prediction of stillbirth and that the previously observed reduction in stillbirth rate was similar between units that did and those that did not adopt customization of fetal weight into clinical practice8. Currently, the risk of FGR in pregnancy is assessed on the basis of a checklist of maternal characteristics at the time of presenting for antenatal care, resulting in a relatively high false-positive rate and poor sensitivity for FGR. In this issue of the Journal, data from the SPREE and ASPRE studies demonstrate that use of a first-trimester algorithm of maternal factors, mean arterial pressure, uterine artery pulsatility index and serum placental growth factor identified a high-risk group that contained about 46% of SGA neonates born preterm at a 12% screen-positive rate9. More effective early-pregnancy screening for SGA is clearly possible in the clinical setting and these data make a compelling case for the review of professional guidelines and implementation of this protocol into routine clinical practice. The recent DELPHI consensus on FGR provides clinical guidance for the identification of early-onset FGR10. The concomitant publication of the TRUFFLE series data provides robust evidence that early-onset FGR requires monitoring with both computerized cardiotocography/non-stress test as well as ductus venosus Doppler assessment11. The TRUFFLE data provide a pragmatic framework for monitoring and thresholds for scheduling birth – both related to excellent clinical outcomes with > 90% intact survival in this high-risk, very preterm FGR cohort. The clinical focus on fetal size at term is predicated on the association with stillbirth and the desire to deliver the pregnancy before this occurs. However, unlike preterm stillbirth, intrauterine demise at term occurs in appropriately grown fetuses (Figure 2) in 60–70% of cases12. Furthermore, it is likely that many of the SGA stillbirths were wrongly classified as SGA as a consequence of postmortem fetal weight loss through maceration and dehydration13. Sebire and colleagues, in a review of over 1000 postmortems, estimated that most fetuses lost over 20% of their body weight between intrauterine demise and being weighed after birth. There is undoubtedly an association between SGA and stillbirth, but a health policy focused on identification of SGA fetuses at term will not be able to prevent the majority of stillbirths. This is a situation analogous to risk stratification using maternal age for trisomy-21 screening in pregnancy13. The weakness of the association between fetal size and stillbirth at term is, at least unofficially, acknowledged in many screening programs that support the evaluation of fetal growth velocity in pregnancy by serial ultrasound assessment. However, growth velocity is a complex metric in which the ability of deviation in fetal growth to predict stillbirth and adverse outcome has yet to be established. Thresholds for deviation in fetal growth must surely depend on gestational age at onset of placental insufficiency, as well as the rate over which that growth deviation occurs and the ability of the fetus to endure such compromise. It would be churlish to assume that a single growth-deviation threshold could serve to identify and prevent stillbirth at any given gestation. Indeed, in the POP study in which women were randomized to routine pregnancy care or serial ultrasound scans, fetal growth velocity was significantly associated with adverse outcome, but only in the SGA group, not in appropriate-for-gestational-age births14. Our understanding of FGR is complicated by the incorrect assumption that the association between fetal size and adverse outcome is a causative one, that is to say that fetal smallness causes directly stillbirth. The placenta is a multifunctional organ responsible for nutrition, respiration and excretion amongst many other life processes. Placental dysfunction confers both nutritional and respiratory limits to the developing fetus, and, to understand its consequences, one must first evaluate fetal nutritional and respiratory needs. Fetal nutritional needs follow a logarithmic/power curve whilst respiratory demands show exponential growth (Figure 3). Early-onset placental dysfunction will affect fetal nutrition/growth far more than will respiratory compromise, explaining why early-onset FGR cases remain SGA for several weeks before requiring delivery. In contrast, onset of placental dysfunction at term will affect exponentially increasing fetal respiratory demands at a time when nutritional demands have plateaued. Thus, at term, a 3000-g fetus that is affected by placental failure may suffer demise from respiratory dysfunction before it has time to 'fail to grow'. Put a better way, SGA is a sign of placental dysfunction that is characteristic of early-onset disease, but is found only infrequently in late-onset disorder. Classically, FGR has been attributed to primary placental maldevelopment. However, large, prospective and blinded histological studies of placentas at term have shown a lack of any histological markers for either SGA or pre-eclampsia15, 16. The only lesion of note associated with SGA and PE is perivillous fibrin deposition, a defect on the maternal surface of the placenta attributed to maternal hypoperfusion of the placental bed. In support of these findings, Professor Khalil's research group has shown, in this issue of Ultrasound in Obstetrics & Gynecology, associations between new-onset impaired uterine artery perfusion in the third trimester (unrelated to impaired trophoblast development) and subsequent perinatal demise17, 18. These data are consistent with recent findings of impaired maternal cardiovascular function at term as a pathophysiological pathway to the development of placental dysfunction and pre-eclampsia. With fetal size established as a poor marker for adverse perinatal outcome at term, markers of fetal hypoxemia have come to the fore. Fetal arterial Doppler assessment, whether using cerebroplacental ratio (CPR) or umbilicocerebral ratio, has been shown to have stronger association with perinatal death than does fetal size19. Even studies that demonstrated poor predictive ability of CPR for stillbirth, demonstrated convincingly that CPR was two- to three-fold better than SGA at prediction of this particular adverse outcome20. A recent systematic review of 128 studies involving over 47 000 women, showed that CPR outperformed fetal size, and umbilical and middle cerebral artery Doppler assessment in the prediction of perinatal death and composite adverse outcome21, 22. Before fetal Doppler assessment is hastily implemented in clinical practice, properly conducted and powered trials of fetal wellbeing at term to predict perinatal death and severe adverse outcome are required. Placental dysfunction is associated with fetal smallness due to intrauterine malnutrition as well as fetal compromise from respiratory hypoxemia. Fetal size currently serves as a common clinical proxy for placental dysfunction, but the issue of which fetal biometry chart to use continues to be a vexatious topic. In my opinion, the rationale for use of a common international fetal growth reference standard makes a lot of sense. However, the differing fetal nutritional and respiratory demands with advancing gestation mean that the majority of stillbirths at term are not SGA, calling into question both health policies focused on the detection of SGA at term and the need for more 'accurate' fetal growth charts. Emerging data show consistently and convincingly that fetal arterial redistribution is associated more strongly than is fetal size with perinatal death at term. A greater emphasis on properly conducted and powered trials of fetal wellbeing at term are now desperately required.