Fetal cerebral redistribution: a marker of compromise regardless of fetal size

Stillbirth is both a tragedy for parents and traumatic for the professionals involved in the care of the mother. Incidence increases in the final weeks of pregnancy, in parallel with that of late-onset fetal growth restriction (FGR)1, 2, suggesting that the events are likely to be related. Although many risk factors are associated with stillbirth3, FGR as a result of placental insufficiency accounts for half of all cases that occur in high-income countries4. As a result, screening for fetal compromise is performed with ultrasound, comparing individual growth with population centiles5. However, this approach is unlikely to prevent stillbirth occurring after 32 weeks' gestation, as more than half of these fetuses are not small in size6, 7. FGR is also associated with a risk of neurocognitive dysfunction2. It is likely that stillbirth represents the outcome of a pathological process, and that for every fetal demise many more fetuses might theoretically be affected by neurological impairment as a consequence of mild hypoxia and suboptimal growth at term8. Term fetuses with a birth weight < 10th centile are more likely to show brain abnormalities on magnetic resonance imaging (MRI), in the form of under-development of the corpus callosum and brain fissures9, 10. However, some studies report that fetuses with signs of cerebral redistribution, defined by a low middle cerebral artery (MCA) pulsatility index (PI), might have a higher risk for abnormalities such as lower communication and problem-solving scores11. In addition, term small-for-gestational-age (SGA) fetuses with normal MCA-PI have been shown to have similar neurodevelopmental outcome to that of appropriate-for-gestational-age (AGA) fetuses, proving the importance of fetal cerebral redistribution in the etiology of adverse neurodevelopmental outcome11. We recently published new evidence supporting the role of fetal cerebral redistribution in the identification of fetuses that have failed to reach their growth potential12. We, and others, have demonstrated that, regardless of fetal size, fetal redistribution is associated with a risk of adverse pregnancy outcome13-21. Here, we explain the rationale for our proposed novel approach to the evaluation of fetal wellbeing near term, in order to improve identification of adverse outcome at term. Failure to reach growth potential has traditionally been difficult to quantify, or even diagnose. Abnormal fetal growth has conventionally been evaluated using arbitrary thresholds of fetal size, commonly the 10th centile, using either population or customized centiles22, 23. However, this is a measure of abnormal fetal size rather than FGR or failure to reach growth potential. SGA is used commonly as a proxy to identify fetuses with abnormal growth24, 25. However, the majority of SGA fetuses are not pathologically growth restricted and do not demonstrate signs of placental insufficiency. This is further confused by the finding that a proportion of AGA infants also fail to meet their growth potential or, more correctly, suffer from occult placental insufficiency, and this population contains the majority of stillbirths at term6, 7, 12, 26. In fact, the proportion of AGA fetuses in pregnancies that result in stillbirth has increased since the 1960s, from 55% (1967–1976) to 77% (1999–2006), as demonstrated in a population-based cohort study using data from the Medical Birth Registry of Norway (1.9 million singleton births at, or beyond, 37 weeks' gestation during 1967 to 2006)27. This observation suggests clearly that efforts to reduce stillbirth have so far focused on SGA fetuses, ignoring the AGA fetuses with occult placental insufficiency. Knowing the third-trimester growth potential of an individual fetus would be useful to evaluate accurately the degree of growth restriction and stratify the risk of stillbirth and adverse perinatal outcome. However, previous attempts to define this growth potential have been based mainly on second-trimester biometry extrapolated to the third trimester, and did not gain popularity because of mathematical complexity28-30. Alternatively, growth customization has succeeded allegedly in reducing the stillbirth rate by adjusting centile curves for maternal characteristics31. However, this approach cannot define fully the genetic potential of the individual fetus, but merely places it on a centile that is more appropriate for its own customized population. Despite such mathematical reallocation, it remains that the majority of perinatal deaths and adverse outcomes at term, such as cerebral palsy, occur in fetuses whose weight is well above the 10th centile7, 32-35. Therefore, fetal weight > 10th centile, considered an appropriate weight at term, does not necessarily reflect fetal wellbeing and accordingly does not rule out the risk of stillbirth in a fetus that may decompensate once an apparently normal weight has been achieved. National guidance in the UK and the USA does not recommend the use of fetal Doppler as a screening tool for placental insufficiency, except when the fetus is already known to be SGA24, 25. Late-onset placental insult might occur in fetuses that have already gained enough weight to be considered AGA. However, the finding of fetal hemodynamic changes in such a fetus might indicate that it is suffering from FGR, despite the fact that its weight is still above the 10th centile. Although estimated fetal weight (EFW) tends to be below the 10th centile in the majority of cases of early-onset growth restriction, in fetuses with late-onset growth restriction, EFW appears to remain well within the limits for AGA fetuses. As a consequence, a small proportion of AGA fetuses, at any weight centile, are at risk of stillbirth because they are growth restricted (in the sense that their rate of growth is impaired, although their absolute weight is still in the ‘normal’ range). This concept is supported by the fact that the incidence of stillbirth and perinatal mortality, stratified according to birth-weight centiles, shows an increasing prevalence as the centile falls below the 80th centile (Figure 1)7, 33. This resembles the pattern and proportion of fetuses selected by our novel approach using fetal cerebral blood flow redistribution12. The antenatal diagnosis of FGR using fetal biometry alone has been challenged recently12, 36 and an alternative approach, based on fetal hemodynamics, to the weight centile-based model was proposed by our group12, 13, 17-20 (Figure 2). For this approach, fetal Doppler measurements were used as an indirect measure of fetal wellbeing, relying on the Doppler indices of blood flow redistribution in the fetus. Independently of fetal weight, we used the cerebroplacental ratio (CPR), corrected for gestational age and expressed in multiples of the median (MoM), to define fetal compromise. CPR, defined as the ratio between the MCA-PI and the umbilical artery PI (UA-PI), has been shown to correlate better with adverse outcome than do its individual components (MCA-PI or UA-PI)37, 38. Using established population patterns of perinatal risk, we applied a threshold at the 5th centile of CPR-MoM of fetuses least likely to suffer from the consequences of growth restriction (birth weight > 90th centile). Accordingly, regardless of the fetal-weight centile, we propose that fetuses with CPR-MoM values below this cut-off (< 5th centile) are considered at increased risk of adverse pregnancy outcomes, secondary to late-onset placental insufficiency or insult. Considering that the majority of adverse outcomes, including stillbirth at term, occur in fetuses of normal size, we studied AGA fetuses with abnormal CPR. We observed that these fetuses were, in fact, more prone to poor acid–base status at birth compared with those with normal CPR13, 17, 18. Furthermore, the frequency of operative delivery for presumed fetal compromise was higher in AGA fetuses with low CPR than in SGA fetuses with normal CPR, indicating that CPR was more strongly associated with fetal compromise as a result of placental insufficiency than was fetal size19. Moreover, low CPR was associated with a higher risk of admission to the neonatal unit (NNU) at term20. Similar results have been published by other groups, indicating that fetuses with low CPR more frequently develop fetal compromise in labor, are delivered by Cesarean section for abnormal fetal heart-rate recordings and have meconium-stained amniotic fluid14-16. Furthermore, an abnormal CPR at term predicts neurobehavioral problems at 18 months of age, according to Internalizing and Somatic Complaints scales, as demonstrated in the Generation R Study (Rotterdam, The Netherlands, 2003–2007)39. Our model identifies two groups of fetuses at potential risk: fetuses with low birth-weight centiles and fetuses with abnormal CPR-MoMs. In addition, we demonstrated that this observation (the link between cerebral redistribution and adverse fetal outcome) was not exclusive of the MCA but also extended to other cerebral vessels, such as the vertebral artery, which develops from an entirely different embryological origin18. This similarity supports our proposal described here. We have also demonstrated that EFW discordance and CPR discordance are independent predictors of the risk of perinatal loss in twin pregnancy, and that their combination could identify the majority of twin pregnancies at risk of perinatal loss40. We have demonstrated that, compared with birth weight, CPR is better associated with the need for operative delivery for presumed fetal compromise, umbilical cord venous and arterial pH at birth, and admission to the NNU17, 19, 20. It is important to emphasize that although these associations exist, their predictive value is unlikely to be high, probably because adverse pregnancy outcomes, such as intrapartum fetal compromise, aberrant umbilical cord venous and arterial pH and admission to the NNU are influenced largely by intrapartum confounders41. Whilst it might therefore be possible to postulate that the hemodynamic model is ineffective, one could argue that it represents a significant advantage over the reliance on fetal weight, which has poorer performance than does CPR. Until now, it has been primarily small fetuses that are considered at risk of adverse perinatal outcome2. However, as a large proportion of small fetuses do not suffer from placental insufficiency, and a proportion of AGA fetuses with low CPR may have chronic hypoxia, we propose a combined assessment approach for predicting adverse outcome using both EFW and CPR. This approach might enable identification of the fetuses with failure to reach their individual growth potential. In a recent study we were able to demonstrate that CPR combined with uterine artery Doppler and EFW, recorded in the third trimester, could identify the majority of pregnancies resulting in stillbirth and perinatal loss42. Furthermore, the model that combines hemodynamic and weight parameters might provide a useful tool for further research in order to evaluate possible markers of placental insufficiency, and perform randomized clinical trials to establish the appropriate management of term SGA and AGA fetuses with abnormal Doppler and probably placental insufficiency. Such a screening protocol has not been investigated before. It must be acknowledged that not all studies measuring CPR at term compare with our results. In a recent retrospective study43, CPR was found to be a poor predictor of adverse perinatal outcome. However, in that study, SGA fetuses with low CPR were delivered frequently by planned Cesarean section, leading to underestimation of the predictive accuracy of CPR for fetal distress in labor. Moreover, the prevalence of stillbirth and birth asphyxia with low Apgar score and low cord-blood pH may have decreased as a consequence of earlier deliveries and, finally, CPR may have been corrected for maternal and fetal characteristics, without establishing whether these factors were involved in the causal pathway for adverse outcome. These sources of bias could theoretically lead to an underestimation of the predictive accuracy of CPR for adverse perinatal outcome. Management of the term fetus with growth restriction is yet to be established as the pathophysiology of late-onset growth restriction and the long-term consequences of hypoxia have not yet been fully clarified44, 45. Although our proposal adds complexity, in that some AGA fetuses are growth restricted and at risk of adverse outcome, it may provide an explanation as to why many stillbirth cases at term are not small. Prospective studies are needed to elucidate the best markers for diagnosis of subtle hypoxia at term, the magnitude of the potential neurological damage in AGA fetuses with abnormal CPR and the best time to screen for adverse outcome in the third trimester. Once these questions are answered, randomized clinical trials will be required to determine the optimal management of SGA fetuses and AGA fetuses with growth restriction, once an appropriate diagnosis, according to any of the indicated tests, has been performed.