Background
Late-onset intrauterine growth restriction (IUGR) results from a failure of the placenta to supply adequate nutrients and oxygen to the rapidly growing late-gestation fetus. Limitations in current monitoring methods present the need for additional techniques for more accurate diagnosis of IUGR in utero. New magnetic resonance imaging (MRI) technology now provides a noninvasive technique for fetal hemodynamic assessment, which could provide additional information over conventional Doppler methods.
Objective
The objective of the study was to use new MRI techniques to measure hemodynamic parameters and brain growth in late-onset IUGR fetuses.
Study Design
This was a prospective observational case control study to compare the flow and T2 of blood in the major fetal vessels and brain imaging findings using MRI. Indexed fetal oxygen delivery and consumption were calculated. Middle cerebral artery and umbilical artery pulsatility indexes and cerebroplacental ratio were acquired using ultrasound. A score of ≥ 2 of the 4 following parameters defined IUGR: (1) birthweight the third centile or less or 20% or greater drop in the centile in estimated fetal weight; (2) lowest cerebroplacental ratio after 30 weeks less than the fifth centile; (3) ponderal index < 2.2; and (4) placental histology meets predefined criteria for placental underperfusion. Measurements were compared between the 2 groups (Student t test) and correlations between parameters were analyzed (Pearson’s correlation). MRI measurements were compared with Doppler parameters for identifying IUGR defined by postnatal criteria (birthweight, placental histology, ponderal index) using receiver-operating characteristic curves.
Results
We studied 14 IUGR and 26 non-IUGR fetuses at 35 weeks’ gestation. IUGR fetuses had lower umbilical vein ( P = .004) and pulmonary blood flow ( P = .01) and higher superior vena caval flow ( P < .0001) by MRI. IUGR fetuses had asymmetric growth but smaller brains than normal fetuses ( P < .0001). Newborns with IUGR also had smaller brains with otherwise essentially normal findings on MRI. Vessel T2s, oxygen delivery, oxygen consumption, middle cerebral artery pulsatility index, and cerebroplacental ratio were all significantly lower in IUGR fetuses, whereas there was no significant difference in umbilical artery pulsatility index. IUGR score correlated positively with superior vena caval flow and inversely with oxygen delivery, oxygen consumption, umbilical vein T2, and cerebroplacental ratio. Receiver-operating characteristic curves revealed equivalent performance of MRI and Doppler techniques in identifying IUGR that was defined based on postnatal parameters with superior vena caval flow area under the curve of 0.94 (95% confidence interval, 0.87–1.00) vs a cerebroplacental ratio area under the curve of 0.80 (95% confidence interval, 0.64–0.97).
Conclusion
MRI revealed the expected circulatory redistribution in response to hypoxia in IUGR fetuses. The reduced oxygen delivery in IUGR fetuses indicated impaired placental oxygen transport, whereas reduced oxygen consumption presumably reflected metabolic adaptation to diminished substrate delivery, resulting in slower fetal growth. Despite brain sparing, placental insufficiency limits fetal brain growth. Superior vena caval flow and umbilical vein T2 by MRI may be useful new markers of late-onset IUGR.
Intrauterine growth restriction (IUGR) is associated with stillbirth and adverse perinatal outcomes. It is commonly the result of placental insufficiency. The conventional approach to identifying IUGR is through serial ultrasound-based measures of fetal growth and Doppler measurements in the umbilical arteries (UAs). However, ultrasound-based fetal biometry and umbilical artery Doppler perform poorly in identifying IUGR in late gestation, with sensitivities ranging from 15% to 50% and false-positive rates in excess of 30%.
Late-onset IUGR can be more accurately identified with cerebroplacental ratio (CPR), which can be especially helpful when IUGR occurs in fetuses with estimated fetal weights (EFW) above the 10th percentile. However, animal studies indicate that chronic fetal hypoxia ultimately reduces fetal oxygen consumption with normalization of fetal blood flow distribution and resolution of the Doppler changes seen in acute hypoxia. This adaptive phenomenon could be an important limitation of middle cerebral artery (MCA) Doppler for the detection of fetuses at risk of abnormal brain development because of late-onset IUGR and highlights the need for more accurate tools for diagnosis of IUGR in utero.
Invasive studies in animal models and human cordocentesis have demonstrated profound reductions in fetal blood oxygen saturation (SaO2) in IUGR fetuses. Fetal monitoring might therefore be improved by direct assessment of fetal oxygenation (ie, fetal oximetry). A new magnetic resonance imaging (MRI) technique for performing oximetry of fetal blood in vivo offers a safe alternative to invasive cordocentesis. Furthermore, the combination of magnetic resonance oximetry with magnetic resonance blood flow quantification could enhance the assessment of placental function by allowing the quantification of fetal oxygen delivery (DO2), whereas the calculation of oxygen consumption (VO2) and measurements of superior vena caval flow could help to characterize fetal metabolic and circulatory adaptations to hypoxia.
Because the ultimate goal of fetal health assessment in late-onset IUGR is to optimize perinatal brain development, a more direct approach to assessing brain oxygenation may confer considerable potential benefits by facilitating more judicious timing of delivery.
Materials and Methods
We conducted a prospective, cross-sectional, case-control study comparing MRI and Doppler ultrasound measurements in fetuses with and without IUGR. Research MRI and ultrasound examinations were performed on a group of normal and suspected late-onset IUGR fetuses in the final weeks of pregnancy. Placental histology, anthropometric measurements, and brain MRI were performed soon after birth.
Our recruitment included fetuses across a range of weight percentiles but was focused on small-for-gestational-age fetuses to provide a study group enriched for IUGR cases. A composite scoring system, based on both pre- and postnatal parameters, was used to define IUGR. We compared the performance of MRI and ultrasound parameters in terms of their concordance with postnatal evidence of IUGR.
Participants
This study was approved by the research ethics boards at the Hospital for Sick Children and Mount Sinai Hospital. Written consent was obtained from every subject. Pregnant women between 32+0 and 41+0 weeks’ gestation with singleton pregnancies were invited to participate through the Obstetric Outpatient Clinic at Mount Sinai Hospital in Toronto from May 2013 to February 2015. Gestational age (GA) was determined from first-trimester crown-rump length measurements. Pregnancies complicated by chronic maternal illnesses including diabetes, autoimmune disease, and hypertension were excluded. Fetuses with anemia, prenatally diagnosed congenital malformations, and genetic syndromes were also excluded.
Imaging protocol
MRI imaging protocol
Each subject was scanned using the same imaging protocol according to our previously published technique. The scans were performed on a clinical 1.5T MRI system (Siemens Avanto, Erlangen, Germany). Details of the MRI sequence parameters are given in Appendices 1 and 2 .
Fetal body weight and brain weight
Although there are no established reference ranges for fetal volume or fetal brain volume, these can be converted to body and brain weights and compared with GA-specific autopsy reference ranges. Figure 1 shows an example of fetal body ( Figure 1 A) and brain ( Figure 1 B) segmentation.
Blood flow quantification
Phase-contrast (PC) MRI with metric optimized gating was used for the quantification of blood flow in the major fetal vessels. We prescribed acquisitions aligned perpendicular to the long axis of the descending aorta (DAo), superior vena cava (SVC), ascending aorta (AAo), main pulmonary artery (MPA), ductus arteriosus, umbilical vein (UV), and branch pulmonary arteries and indexed the measurements to EFW.
Magnetic resonance oximetry
T2-based MR oximetry has been shown to be feasible in lamb fetuses and humans. In this study, we used a T2 preparation pulse sequence with an steady-state free precession readout and a nonrigid registration motion correction algorithm (Myomaps; Siemens Healthcare, Erlangen, Germany) to perform T2 mapping in the major fetal vessels. Examples of T2 maps are shown in Figure 1 , C and D. T2 relaxation time was measured from the T2 maps with a region of interest placed over the central 50% of the vessel in accordance with established criteria. We used a previously reported conversion from T2 to SaO2 (percentage O 2 ) for adult blood according to our previously published technique. We determined the oxygen content of blood based on a GA appropriate estimation of fetal hemoglobin concentration.
Combined ventricular output (CVO), DO2, and O 2 consumption (VO2) calculation
CVO was estimated from the sum of the AAo and MPA flows plus an estimated coronary blood flow of 3% of the CVO based on lamb studies. Fetal DO2 and VO2 were calculated using the flow and T2 measurements according to our previously published technique. Fetal oxygen extraction fraction was calculated as VO2/DO2.
Doppler (UA and MCA)
A research ultrasound was performed on the same day as the MRI, which included measurements of MCA and UA, pulsatility index, and CPR, and an estimation of fetal weight based on biparietal diameter, head circumference, abdominal circumference, and femur length. In addition to the research ultrasound, all clinical Doppler ultrasound measurements were gathered. CPR percentiles were determined based on reference data.
IUGR diagnosis
We collected calculated the ponderal index and birthweight Z score for each newborn. The lowest GA-appropriate percentile CPR after 30 weeks was recorded. Gross and histopathological examination of every placenta was performed by an expert in perinatal pathology (S.K.).
In our IUGR scoring system, 1 point was allocated for a positive result in each of 4 categories, and a defined IUGR score of ≥ 2:
- 1.
Birthweight third or less percentile or ≥ 20% drop in the percentile of ultrasound-based EFW over serial visits ≥ 2 weeks apart.
- 2.
Ponderal index < 2.2 (grams per cubic centimeter).
- 3.
Lowest CPR after 30 weeks less than the fifth percentile.
- 4.
Placental histology meets predefined criteria for placental underperfusion (ie, placental weight less than the 10 th percentile, multifocal infarction, or decidual vasculopathy).
When performing the receiver-operating characteristic (ROC) analysis for the MRI and Doppler parameters in identifying IUGR, only postnatal parameters (≥ 2 of categories 1, 2, and 4 listed previously) were used to define IUGR to avoid bias toward CPR.
Statistical methods
All measurements passed the omnibus normality test of D’Agostino et al except for the following: UA and pulmonary blood flow in the normal group and MPA T2 in the IUGR group. The initial analysis compared MRI and ultrasound parameters between IUGR and non-IUGR fetuses using a Student t test for normally distributed measurements or a Mann-Whitney test for measurements that were not normally distributed. We then used Pearson’s correlation to investigate the relationships between measurements. Linear regression and Bland-Altman plots were used to assess intra- and interobserver variability and reproducibility for MRI flow and T2 measurements. Finally, ROC curves were created to evaluate the performance of MRI and Doppler measurements in identifying IUGR. Statistical analysis was performed using GraphPad Prism 6.0e (GraphPad Inc, San Diego, CA). All values in the text are expressed as means ± SDs. The results are expressed as means with SD and values of P < .05 were considered statistically significant.
Results
Participants
There were 69 women in late gestation who participated in the study. Among these, 29 subjects were excluded, 28 because the data obtained did not include all scoring parameters required for group categorization and 1 because of unacceptable MRI image quality. The remaining 40 subjects were included in the analysis and all fetuses were born in good condition except 1 stillbirth of an IUGR fetus in the setting of preeclampsia.
Subjects were classified based on our IUGR scoring system into 14 IUGR and 26 non-IUGR (clinical information is summarized in Appendix 3 ). A comparison of clinical characteristics between the 2 groups is shown in the Table . No significant differences were seen in either mean GA (P = 0.6) or maternal age (P = 0.7) at the time of MRI between the 2 groups. IUGR fetuses were born 2.4 ± 0.8 weeks earlier than normal fetuses. The mean interval between MRI and birth was 11 ± 13 days for IUGR fetuses and 21 ± 10 days for normals ( P = .01).
| Characteristics | Normal (n = 26) | IUGR (n = 14) | P values |
|---|---|---|---|
| GA at MRI scan, wks | 35.9 ± 0.9 | 35.4 ± 2.4 | .500 |
| Maternal age, y | 33.8 ± 4.5 | 34 ± 4 | .700 |
| Days from MRI to birth | 11 ± 13 | 21 ± 10 | .010 a |
| EFW at MRI scan, kg | 2.8 ± 0.3 | 1.9 ± 0.6 | .0001 a |
| EBW at MRI scan, g | 300 ± 29 | 249 ± 55 | .005 a |
| EBW Z-score | 0.02 ± 0.86 | –1.36 ± 0.73 | < .0001 a |
| EBW over EFW, % | 10.9 ± 1.0 | 13.8 ± 2.3 | .0004 a |
| GA at birth, wks | 39.6 ± 1.1 | 37.0 ± 2.8 | .008 a |
| Birthweight, kg | 3.16 ± 0.41 | 1.95 ± 0.64 | < .0001 a |
| Birthweight percentile | 33 ± 23 | 2 ± 2 | < .0001 a |
Imaging results
MRI growth findings
MRI-based EFW, birthweight, and birthweight percentile were significantly lower in IUGR fetuses. EFW by ultrasound and MRI were closely correlated (R 2 = 0.9, P < .0001). IUGR fetuses had lower EBW and brain weight Z score. We observed a higher ratio of EBW over EFW in the IUGR fetuses ( P = .0004). This impression of asymmetric growth restriction was confirmed by a higher ratio of the birth head circumference/birthweight in IUGR newborns ( P = .0005).
MRI hemodynamic findings
Approximately 20% of PC and T2 measurements needed to be repeated because of motion artefact. However, acceptable image quality was obtained in all but 1 subject for all vessels with an average scan duration of 45 minutes. We required 2-3 hours of postprocessing for each case. A high degree of intra- and interobserver agreement and reproducibility was found for MRI measurements ( Appendice 4 ).
Figure 2 A shows a comparison of the MRI-measured major fetal vessel flows indexed to EFW. IUGR fetuses had significantly increased SVC flow ( P < .0001) and ductus arteriosus flow ( P = .02) but decreased pulmonary blood flow ( P = 0.01) and UV flow ( P = .004) compared with normal fetuses. Figure 2 B demonstrates significantly lower mean T2 values in all measured vessels in the IUGR fetuses. DO2 and VO2 were both lower in IUGR fetuses, as shown in Figure 3 . Consequently, IUGR fetuses had significantly higher oxygen extraction fraction (40% ± 10%) than normal fetuses (34% ± 8%, P = .03).
Doppler findings
IUGR and normal fetuses were at a similar GA when the lowest CPR was recorded. Although a trend toward higher UA PI was seen in the IUGR fetuses, the difference was not statistically significant ( P = .08). By contrast, MCA PI ( P = .04) and CPR ( P = .005) were significantly lower in the IUGR group ( Appendix 6 ).
Neonatal brain MRI
Nine of 14 IUGR patients and 24 of 26 normal subjects underwent neonatal brain MRI, all of which were grossly normal. Three of 5 IUGR newborns that did not have neonatal MRI had normal cranial ultrasound findings. IUGR newborns had lower brain weight Z scores ( P < .05). Four of the IUGR fetuses had diffuse excessive high signal (DEHSI) of the white matter, whereas none of the normal newborns had DEHSI. On magnetic resonance spectroscopy, approximately a third of the subjects in both groups had lactate in the basal ganglia. We found no difference in white matter apparent diffusion coefficient or N-acetyl aspartate to choline ratios between the 2 groups ( Appendix 7 ).
Placental histology
Twelve of 14 of the IUGR subjects met our histopathological criteria for IUGR. The 2 IUGR placentas that did not show the classical changes had mild placental abnormalities (1 had placental weight of the 10th to 25th percentile with mild dysmaturity of chorionic villi, and the other had increased villous vascularity with placental weight of the 25th to 50th percentile, whereas both had overcoiling of the umbilical cord). Of the 17 available non-IUGR placentas, 4 were below the 10th weight percentile, but none had multifocal infarction or decidual vasculopathy. Nearly all had subtle abnormal finding (eg, chorangiosis, villitis, villous dysmaturity, meconium staining, etc).
Correlations
The relationships between the IUGR score and MRI parameters are shown in Figure 4 . As IUGR score increased, MRI-measured UV flow, UV T2, and calculated DO2 and VO2 all decreased. Higher SVC flow was correlated with higher IUGR score (R 2 = 0.56, P < .0001). In addition, the score correlated inversely with the fetal brain weight Z score, and positively with the weight percentage of the brain over the body. EBW was negatively correlated with SVC flow (R 2 = 0.25, P = .0007) and positively correlated with UV T2 (R 2 = 0.22, P = .002) and DO2 (R 2 = 0.14, P = .01). EBW was also correlated with CPR (R 2 = 0.29, P = .0004) and inversely correlated with UA PI (R 2 = 0.22, P = .002). An inverse correlation was found between SVC flow and UV T2 (R 2 = 0.25, P = .0007) ( Figure 5 ).
When the lowest CPR (including clinical data) was compared with MRI, the following correlations were found: as CPR decreased, so did UV T2 (R 2 = 0.12, P = .03) and DO2 (R 2 = 0.31, P = .0003), whereas SVC flow increased (R 2 = 0.26, P = .0009). UA PI was positively correlated with SVC flow (R 2 = 0.18, P = .006) and inversely related to UV T2 (R 2 = 0.16, P = .01) and DO2 (R 2 = 0.29, P = .0003).
MCA PI showed significant correlation with SVC flow (R 2 = 0.16, P = .02). When MRI parameters were correlated with only the research Doppler parameters obtained on the same day as the MRI, the allocation of patients into IUGR or non-IUGR groups did not change. CPR correlated with UV flow but not with other MRI parameters. UA PI correlated with SVC flow (R 2 = 0.12, P = .03), DO2 (R 2 = 0.27, P = .001), UV T2 (R 2 = 0.11, P < .05), and UV flow (R 2 = 0.20, P = .006).
Performance of MRI and Doppler by ROC curves
When defining the IUGR patients solely based on postnatal evidence, there were 12 IUGR and 28 normal fetuses. Figure 6 illustrates the ROC plots for MRI and Doppler measurements for the identification of IUGR. Appendix 8 confirms that SVC flow had the highest area under the curve, although the differences between Doppler and MRI parameters were not significant. The lowest CPR performed better than the CPR on the day of MRI at identifying IUGR.
Results
Participants
There were 69 women in late gestation who participated in the study. Among these, 29 subjects were excluded, 28 because the data obtained did not include all scoring parameters required for group categorization and 1 because of unacceptable MRI image quality. The remaining 40 subjects were included in the analysis and all fetuses were born in good condition except 1 stillbirth of an IUGR fetus in the setting of preeclampsia.
Subjects were classified based on our IUGR scoring system into 14 IUGR and 26 non-IUGR (clinical information is summarized in Appendix 3 ). A comparison of clinical characteristics between the 2 groups is shown in the Table . No significant differences were seen in either mean GA (P = 0.6) or maternal age (P = 0.7) at the time of MRI between the 2 groups. IUGR fetuses were born 2.4 ± 0.8 weeks earlier than normal fetuses. The mean interval between MRI and birth was 11 ± 13 days for IUGR fetuses and 21 ± 10 days for normals ( P = .01).
| Characteristics | Normal (n = 26) | IUGR (n = 14) | P values |
|---|---|---|---|
| GA at MRI scan, wks | 35.9 ± 0.9 | 35.4 ± 2.4 | .500 |
| Maternal age, y | 33.8 ± 4.5 | 34 ± 4 | .700 |
| Days from MRI to birth | 11 ± 13 | 21 ± 10 | .010 a |
| EFW at MRI scan, kg | 2.8 ± 0.3 | 1.9 ± 0.6 | .0001 a |
| EBW at MRI scan, g | 300 ± 29 | 249 ± 55 | .005 a |
| EBW Z-score | 0.02 ± 0.86 | –1.36 ± 0.73 | < .0001 a |
| EBW over EFW, % | 10.9 ± 1.0 | 13.8 ± 2.3 | .0004 a |
| GA at birth, wks | 39.6 ± 1.1 | 37.0 ± 2.8 | .008 a |
| Birthweight, kg | 3.16 ± 0.41 | 1.95 ± 0.64 | < .0001 a |
| Birthweight percentile | 33 ± 23 | 2 ± 2 | < .0001 a |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
