Doppler ultrasound of umbilical and middle cerebral artery in third trimester small‐for‐gestational age fetuses to decide on timing of delivery for suspected fetal growth restriction: A cohort with nested RCT (DRIGITAT)

To assess the association of the umbilicocerebral ratio (UCR) with adverse perinatal outcome in late preterm small‐for‐gestational age (SGA) fetuses and to investigate the effect on perinatal outcomes of immediate delivery.


| I N TRODUC TION
Fetal growth restriction (FGR) is a condition in which the fetus does not reach its intrinsic growth potential, mainly concentrated in the lowest weight centiles. 1The most frequent underlying pathophysiological mechanism is placental insufficiency resulting in malnutrition and hypoxia.3][4][5] The imminent hypoxia that poses a severe threat of stillbirth is only prevented by timely delivery. 5In its prevention, the (low) risk and high impact of fetal demise should be balanced against the (high) risk and relatively low impact of unmet fetal needs to mature. 6here is a long-standing search for methods to differentiate growth-restricted fetuses, who are at risk for reduced placental nutritional supply and hypoxia, from fetuses that are constitutionally small for their gestational age (SGA).Generally, the 10th centile of a reference chart for estimated fetal weight (EFW) and/or fetal abdominal circumference (FAC) is used to screen for FGR and to select women who need additional surveillance and possibly timely delivery. 7nfortunately, this method not only selects FGR fetuses, but also healthy SGA fetuses, who do not need additional surveillance and most likely will be harmed by early delivery. 8n 2016, a Delphi procedure led to a widely adopted international FGR consensus definition that incorporated Doppler measurements, including the cerebroplacental ratio (CPR), the inverse of the umbilicocerebral ratio (UCR), and decline in fetal growth centile as parameters of placental function. 9A high UCR reflects decreased resistance in the middle cerebral artery (MCA) and/or increased resistance in the umbilical artery (UmbA), indicating redistribution of the fetal circulation, an adaptation mechanism to scarcity of nutrients and oxygen, also called 'brain sparing'. 10he UCR has shown diagnostic accuracy for short-term adverse outcomes of early preterm FGR and short-term adverse outcomes of late preterm FGR, mostly in retrospective studies. 11,12Long-term associations are less established. 13,14he biomarkers placental growth factor (PlGF) and soluble fms-like tyrosine kinase-1 (sFlt-1) are proposed as markers for placental insufficiency and may play a role in the diagnosis of FGR. 15 If earlier delivery reduces the risk of hypoxia, better perinatal outcome can be hypothesised, but cannot be proven from observational studies.Therefore, we designed a randomised controlled trial (RCT) in a cohort of late preterm SGA pregnancies that randomised the cases with abnormal fetal Doppler indices between immediate delivery and expectant management. 16A second objective was to assess the association of criteria used for randomisation, with adverse perinatal outcomes in the cohort.In this paper, we describe short-term perinatal outcomes of this study.A video abstract (Video S1) has been created by the authors to summarise the paper's findings.

| Study setting
The DRIGITAT trial was a nationwide multicentre nested RCT in the Dutch Obstetric Consortium, a collaboration of perinatal secondary and tertiary care centres in the Netherlands.The study protocol has been previously published. 16Briefly, the trial design was a cohort study of SGA pregnancies of at least 32 +0 weeks of gestation with a nested RCT including fetuses with two consecutive abnormal UCRs >0.8, and/or UmbA pulsatility index (PI) above the 90th centile.Fetal Doppler index centiles were calculated according to Arduini and Rizzo. 17Patient recruitment, patient monitoring and data collection were performed by the treating obstetricians, residents, midwives and research staff of participating sites.Patient organisations Hellp stichting and Care4Neo (Vereniging van Ouders Couveusekinderen) had an advisory role in the development of the trial.

| Participants (cohort)
][20] Exclusion criteria were maternal age below 18 years, unwillingness or inability to give informed consent, uncertainty about the estimated due date (no first-trimester dating scan), chromosomal abnormalities or suspicion of congenital anomalies that could influence the prognosis of the pregnancy or health of the fetus independent of the growth problem.

| Routine protocol
Participants were requested to complete a baseline characteristics survey at inclusion.Routine care after recruitment was according to local hospital protocol.According to study protocol, a biometry scan was performed at least every 2 weeks and a Doppler ultrasound scan at least every week, including PI of the UmbA and MCA.The UCR was calculated by dividing the values of the UmbA PI by the MCA PI.In many publications, the ratio of the UmbA PI and MCA PI is presented as CPR; the mathematical inverse of the CPR is the UCR.In this study, we used the UCR, as we have previously demonstrated that there may be some advantages to the use of UCR, 21 which are of particular relevance when fetal Doppler changes become increasingly abnormal: with lower MCA and higher UmbA impedance CPR tends towards zero, whereas UCR tends towards infinity, thus allowing improved observation of differences.Cardiotocography was indicated on the discretion of the treating healthcare team.Guidance and safety net criteria were provided for timing of delivery in the cohort and in the expectant monitoring arm of the RCT.Delivery was indicated when:

| Blood sample storage and biomarker analysis
Maternal blood samples (serum and plasma) were drawn at inclusion and stored at −80°C for biomarker analysis of PlGF and sFlt-1 in the centres with facilities for biobanking.All blood samples were analysed for PlGF and sFlt-1, in batches after completion of the inclusions, at the University Medical Centre in Groningen using Roche Elecsys assays on the electrochemiluminescence immunoassay platform Cobas e601module (Roche Diagnostics, Basel, Switzerland).4][25] Abnormal values at inclusion were calculated.Evaluation of the use of biomarkers in predicting adverse outcomes will follow in a secondary analysis and is beyond the scope of this paper.

| Participants (RCT)
An abnormal fetal Doppler measurement was defined by a UCR >0.8 (equals a CPR of less than 1.25) based on a previous meta-analysis, 21 or an UmbA PI >90th centile on two consecutive occasions with an interval >15 hours.
Women were eligible for randomisation from 34 +0 until 36 +6 weeks if they had an abnormal fetal Doppler in combination with an EFW and/or FAC below the 3rd centile and from 36 +0 until 36 +6 weeks in case of an EFW and/or FAC below the 10th centile.Women who were RCT-eligible but declined RCT participation were managed according to local protocol.

| Randomisation and blinding
RCT-eligible women who provided informed consent were randomly allocated to immediate delivery or expectant monitoring.Randomisation was centrally organised using the online computerised randomisation service of Castor (last updated version 2023.4.0.1). 26Randomisation was done using variable block sizes of 4 and 6, stratified by gestational age (dichotomous; before or after 36 weeks) and participating hospital.Participating sites were able to access the randomisation service 24 hours per day.Blinding of participants, obstetricians or outcome assessors was not possible because of the nature of the intervention.Participants allocated to immediate delivery were planned for immediate induction of labour or caesarean section (CS), unless the intervention was refused by the participant.Participants allocated to the expectant management group were monitored with biometry at least every 2 weeks, at least twice weekly follow up by Doppler ultrasound scan and daily cardiotocography.Timing of birth was according to local hospital protocol and based on study safety net criteria.

| Outcomes
The primary outcome of the RCT was set at neurological development at 2 years of age, to be assessed by a Bayley-III test in all children born in the RCT and a subset of children born in the cohort (delivery >38 weeks, normal Doppler) randomly selected from Castor and will be published later. 27 short-term composite adverse perinatal outcome, based on the intention to treat, was the primary outcome for this report.It was defined by (1) adverse condition at birth and/or (2) major neonatal morbidity or mortality.
All other items of the relevant core outcome set were collected. 28

| Sample size RCT
The sample size was calculated based on a clinically relevant difference of seven points on the primary outcome of the Bayley-III assessment at 2 years of age. 27With a standard deviation of 15 points in both treatment groups, an α of 0.05 and a desired power of 80% for a two-sample t test, 74 participants had to be recruited in each arm of the nested RCT.Assuming a 20% loss to follow up of randomised participants (because of the long interval before assessment of the primary outcome), 185 women had to be included in the RCT.Assuming a 20% incidence of abnormal UCR within the SGA cohort and accounting for an RCT inclusion rate of 60%, we calculated the need for a cohort of 1542 patients.

| Statistical analysis
This paper presents short-term outcomes of the complete study cohort.For these analyses the cohort was divided into the following subgroups, based on Doppler classification: Categorical baseline characteristics were summarised by presenting number and column percentages and compared using chi-square test.Continuous variables were summarised as medians with interquartile ranges (IQR) and compared using non-parametric Kruskal-Wallis test.A multiple logistic regression analysis was performed with the composite adverse perinatal outcome as dependent variable and EFW multiple of the median (MoM) at inclusion, preeclampsia, gestational hypertension and the Doppler group classification. 29To assess the performance of this model, receiver operating characteristics (ROC) curves and area under the ROC curve (AUC) with 95% confidence interval (CI) were calculated.Differences with a p value less than 0.05 were considered statistically significant.Statistical analyses were performed using SPSS software (version 28.0; IBM, Armonk, NY, USA).

| R E SU LTS
Cohort recruitment ran from 1 January 2018 until 1 January 2022.RCT recruitment ran from the start and continued until 25 April 2022.In total, 19 hospitals participated, of which 12 were secondary care centres and seven were tertiary care centres.A total of 692 women were included in the cohort (Figure 1).Two cohort participants were excluded from analyses because of missing biometry and Doppler measurements, leaving 690 women for analyses.On 25 April 2022, the Data Safety Monitoring Board recommended that the study be halted completely because of low RCT-inclusion rates.At the time of halting the trial, 18 RCT participants had been randomly allocated to immediate delivery and 22 participants had been allocated to expectant management.After randomisation for immediate delivery, one participant declined the intervention.
Slow accrual to the RCT was related to slightly lower than anticipated number of patients with abnormal UCR and higher than anticipated number of patients with exclusion criteria such as language barriers, further worsened by the challenges of the COVID-19 period and its aftermath, when patient recruitment was less prioritised.Also, women frequently did not want to be randomised because early delivery was not an acceptable option for them.
Study population data of all subgroups, including both randomisation groups, are presented in Table 1.Notwithstanding the low number of inclusions in the RCT, there was a significantly lower gestational age at delivery (p < 0.01) in the immediate delivery group compared with the expectant management group, and a near significant lower birthweight (p = 0.05) and higher composite adverse perinatal outcome (p = 0.07).Baseline characteristics were comparable between the groups, except that pre-eclampsia was most frequent in the women who delivered before 34 weeks of gestation, and more frequent in the Abnormal Doppler twice consecutive subgroups than in those with never, or only once or intermittent abnormal Doppler.
There were six perinatal deaths in the entire cohort.Two deaths occurred in the group of women who delivered before 34 weeks, caused by congenital anomalies unknown at time of inclusion.One death occurred in the Abnormal Doppler twice consecutive group (not within the RCT group) because of placental abruption and emergency CS at 38 weeks.One late death occurred in the Abnormal Doppler once or intermittent group 138 days after delivery caused by a genetic disorder unknown at time of inclusion.Two deaths occurred in the Never Abnormal Doppler group, one caused by asphyxia during spontaneous labour at 40 weeks of gestation and one because of a congenital anomaly unknown at the time of inclusion.
Tables 2 and 3 compare baseline characteristics and perinatal outcome data of the different Doppler classification groups.The 30 women who delivered before 34 weeks and women with incomplete data were excluded from this comparison.The Abnormal Doppler twice consecutive group differed from the Abnormal Doppler once or intermittent and Never abnormal Doppler groups (Table 2): at inclusion, median EFW was significantly lower (1578 g, IQR 1351-1750 g), growth restriction was more severe, as shown by a significant lower EFW MoM (0.73, IQR 0.65-0.77)and pre-eclampsia was significantly more frequent (n = 27; 24%).PlGF at inclusion was more frequently below 99 and the sFlt-1:PlGF ratio was more frequently above 33 in the Abnormal Doppler twice consecutive group, compared with the two other groups.
At delivery, gestational age, birthweight and birthweight MoM were significantly lower in the Abnormal Doppler twice consecutive group, and delivery was more often by CS, compared with the other two groups (Table 3).The composite adverse perinatal outcomes were significantly worse in the Abnormal Doppler twice consecutive group compared with the other groups, with significantly longer hospitalisation after delivery for both infant and mother.These differences remained significant after a sensitivity analysis that excluded the immediate delivery group from the total Abnormal Doppler twice consecutive group, in which composite adverse perinatal outcome was more frequent than in the expectant management group (Table 3).
Figure 2 shows the forest plot of the odds ratios for the composite adverse perinatal outcome.The logistic regression analysis showed that EFW MoM at inclusion, pre-eclampsia, gestational hypertension and study group classification were significantly associated with the composite adverse perinatal outcome, with an AUC in a ROC graph of 0.71 (95% CI 0.67-0.76)(Figure 2).

| Main findings
In this SGA cohort, an abnormal UCR was significantly associated with a higher incidence of short-term perinatal adverse outcomes, meaning that this can be a useful tool to discriminate between constitutional SGA and FGR needing surveillance and timing of delivery.It is unlikely that imminent delivery based on an abnormal UCR results in better outcomes.However, the randomised trial was halted prematurely because of low inclusion rates in the RCT and therefore the study did not reach the required sample size, and also the study was not powered for this outcome.
Most eligible women who elected not to be randomised favoured expectant management.As a result of this, perinatal outcome was similar between women in the RCT, who were allocated to delayed delivery, and women who were eligible, but refused randomisation.
The expectant management group showed near significant better short-term outcomes compared with the immediate delivery group.The higher median birthweight in the expectant

| 1047
FETAL GROWTH RESTRICTION AMONG SMALL-GESTATIONAL AGE FETUSES management group showed that infants in this group gained on average 275 g in the 13 additional days of gestation (and hence maturation) compared with the immediate delivery group.Although obviously underpowered for definitive statements, in a post-hoc power calculation, there is more than a 90% chance that immediate delivery is associated with increased composite adverse perinatal outcome.This result resonates with previous calls to caution translation of associations into interventions, 30 and that earlier delivery to avoid hypoxia may be protective of hypoxia, but that this benefit may be outweighed by the burden of prematurity. 6Clinicians should refrain from immediate delivery unless in the context of clinical trials.The effects on longterm neurodevelopmental outcome (on which the study sample size was designed) will be reported in due time, although this analysis may be hampered by the small sample size.
The selection criteria based on EFW centiles with abnormal Doppler were effective in selecting those fetuses with the highest risk for composite adverse perinatal outcome.Our findings are in line with numerous observations of the associations between abnormal fetal MCA Doppler and adverse outcomes in large meta-analyses of mostly retrospective studies and recent (semi-) prospective studies. 12This result therefore supports the common conception, distilled in the Delphi consensus definition (adopted by the International Society of Ultrasound in Obstetrics and Gynecology and FIGO), that an abnormal UCR indicates the perinatal risks of abnormal placental function. 9etuses in the Abnormal Doppler twice consecutive group gained less weight after inclusion than in the other two groups.In this group the fetal growth restriction became worse, as shown by a lower birthweight MoM than the EFW MoM at inclusion, whereas there was no change of weight MoM in the group with never abnormal Doppler.This aligns with other recent observations that fetuses that appear to have decreasing growth velocity are at risk of adverse perinatal outcome. 31,32aseline characteristics of women in the cohort with an abnormal Doppler once, or intermittent, were more similar to those who never had an abnormal Doppler.This supports the strategy that Doppler results should be checked for consistency to confirm abnormality. 33Furthermore, it should be noted that also fetuses who never showed an abnormal Doppler may still develop perinatal complications, as in the case of serious asphyxia during labour at 40 weeks in this group.Although that outcome is not restricted to SGA fetuses and might occur in appropriate for gestational age (AGA) fetuses too, the fair amount of composite adverse perinatal outcome in this group suggests that watchful monitoring of women with a small fetus remains necessary, even when Doppler results are normal.Nonetheless, these cohort data suggest that provider-initiated delivery before 40 weeks may not be necessary, in contrast with the pragmatic conclusions of the previous DIGITAT trial. 34

| Strengths and limitations
Although this study was the first to report the UCR in thirdtrimester SGA fetuses in a randomised intervention trial, a  limitation was the study was halted before the planned sample size was reached because of low inclusion rates.First, preparedness among physicians, patients and their partners to participate proved to be limited.Moreover, this study was recruiting during the COVID-19 pandemic and its aftermath with restricted options for laboratory study sampling, staff shortages and the resulting increased workload, hampering inclusion rates.Because of the insufficient sample size, no definite conclusions on primary study question could be drawn.
Another limitation is that we focused on SGA (EFW or FAC below the 10th centile) and pragmatically defined suspected FGR by an abnormal UCR or UmbA PI >90th centile.This disregards the fact that with abnormal fetal Doppler measurements but an EFW or FAC above the 10th centile FGR may be present, as highlighted in the Delphi Consensus definition. 9,35Women with AGA fetuses with suspected FGR were not included and classification of FGR could be more accurate if AGA fetuses with a decline in fetal growth velocity and abnormal Doppler had also been included.Moreover, practice variation between different physicians in the performance of Doppler measurements might have occurred based on the experience of the physician performing the measurement.
The results might have been influenced by intervention bias, as the Doppler measurements were not blinded. 6hysicians might possibly have increased the monitoring frequency and adjusted the timing of delivery of participants with abnormal Doppler results.Also, outcomes may have shifted based on protocol management, also known as the Hawthorne effect. 36,37 strength of this prospective multicentre cohort study is the sample size with complete data for 690 late preterm SGA pregnancies that were prospectively monitored with repeated biometry and Doppler ultrasound scans during the course of pregnancy.
Future studies should focus on further improved diagnosis of FGR by the combination of serial ultrasound of fetal size and Doppler in combination with maternal serum biomarkers (identified or novel; single time-point and/or serial).In this paper, we only described abnormal biomarkers at baseline.The predictive value of biomarkers will be further studied in subsequent analyses.Additionally, placental pathology may reveal if differences between underlying placental lesions are reflected in different SGA and FGR phenotypes.Some of these questions will be answered in secondary analyses of the DRIGITAT dataset, exploring associations of biomarkers and placental pathology with clinical variables.In addition, the results of the TRUFFLE2 and RATIO37 RCT investigating similar hypotheses are eagerly awaited. 38,39The RCT data of DRIGITAT will also be made available for individual patient data meta-analysis with relevant other developing datasets. 38,39In those larger datasets, it can also be studied if subpopulations of gestational age categories or extent of fetal size abnormality have different benefit/harm ratios from early delivery.

| CONCLUSIONS
Selection criteria based on EFW centiles and abnormal Doppler were effective in selecting those with the highest risk for composite adverse perinatal outcome in a cohort with late preterm SGA.This and subsequent analyses of these data may allow guidance of the clinical application of these monitoring parameters.This study could not answer the research question of whether early delivery improves perinatal outcomes.Our interpretation is that it is unlikely that fetuses that are SGA and show haemodynamic redistribution will benefit from early delivery at 34 up to 36 +6 weeks.
F I G U R E 2 Forest plot of the odds ratios for a composite abnormal infant outcome, calculated by logistic regression analysis.The odds ratio for the EFW MoM is calculated per 0.1 MoM.Area under the curve of a receiver operating characteristic curve 0.71 (95% confidence interval 0.67-0.76).EFW, estimated fetal weight; MoM, multiple of the median.

F
I G U R E 1 Flowchart of the study population.Abnormal Doppler = umbilical artery pulsatility index above the 90th centile or umbilicocerebral ratio>0.8, and estimated fetal weight (EFW) below the 3rd centile at 32-35 weeks or EFW below the 10th centile at 36 weeks.*Including two women with absent or reverse end-diastolic flow in the umbilical artery, which precluded randomisation.

T A B L E 1
Study population (n = 690).Abnormal Doppler = Umbilical artery pulsatility index above the 90th centile or umbilicocerebral ratio >0.8, and EFW below the 3rd centile at 32-35 weeks or below the 10th centile at 36 weeks.Table shows median (interquartile range) or number (%).Abbreviations: ARED flow, absent or reversed end-diastolic flow in the umbilical artery; EFW, estimated fetal weight; MoM, multiple of the median.a Significant difference (p < 0.05) with the five other columns to the left.b Significant difference (p < 0.05) with all other columns.Differences between randomisation arms: ‡ p = 0.00; § p = 0.05; ¶ p = 0.07.
Data at and after inclusion.
T A B E 2 a Significant difference (p < 0.05) with other two columns.b Significant difference (p < 0.05) with [Never Abnormal Doppler].c 85 missing values.d 47 missing values.e 206 missing values.f 195 missing values.g Including two women with ARED flow.