Objective
Prenatal imaging has identified alterations of brain growth in fetuses with congenital heart disease. However, little is known about the timing of altered brain development and its occurrence in specific congenital heart disease subgroups. This magnetic resonance imaging study aimed to identify early (median, 25 gestational weeks [GW]) changes in fetal total brain (TBV), gray matter (GMV), and subcortical brain (SBV) volumes in Tetralogy of Fallot (TOF) cases in utero.
Study Design
Fetal magnetic resonance imaging (1.5 Tesla) was performed in 24 fetuses who were diagnosed with TOF and 24 normal age-matched control fetuses (20-34 GW). TBV, GMV, SBV, intracranial cavity, cerebellar, ventricular, and external cerebrospinal fluid volumes were quantified by manual segmentation based on coronal T2-weighted sequences. Mixed model analyses of variance and t -tests were conducted to compare cases and control fetuses.
Results
TBV was significantly lower ( P < .001) in early (<25 GW) and late TOF cases. Both GMV ( P = .003) and SBV ( P = .001) were affected. The GMV-to-SBV ratio declined in fetuses with TOF ( P = .026). Compared with normal fetuses, ventricular volume was increased ( P = .0048). External cerebrospinal fluid was enlarged in relation to head size ( P < .001). Intracranial cavity volume ( P = .314) and cerebellar volume ( P = .074) were not significantly reduced in fetuses with TOF.
Conclusion
TOF is associated with smaller volumes of gray and white matter and enlarged cerebrospinal fluid spaces. These changes are present at ≤25 GW and indicate altered fetal brain growth in this pathophysiologic entity during early stages of human brain development.
Children born with congenital heart disease (CHD) frequently show structural and metabolic brain alterations that are associated with neurobehavioral impairment during later life. Because these aberrations are already detectable prenatally, postnatal surgical repair is limited in its potential to improve the neurodevelopmental outcome of these cases.
Tetralogy of Fallot (TOF) is a severe cyanotic heart disease with a right-to-left shunt that comprises right ventricular outflow obstruction (RVOTO), outlet ventricular septal defect, aorta dextroposition (override of the ventricular septum with biventricular aortic connection), and right ventricular hypertrophy. It has an estimated prevalence of 2.8-3.5 per 10,000 live births and, thus, represents approximately 2.8-3.5% of all CHDs and approximately 5% of severe CHD cases. In fetuses with TOF, a right-to-left shunt and increased left ventricular output result in consecutively low oxygen saturation in cerebral arteries and inadequately high cerebral vascular resistance.
Previously, abnormalities of fetal brain development have been identified at later gestational stages (from 25-28 to 37 gestational weeks [GW]) in mixed CHD populations. A third-trimester drop-off in global brain growth and delayed cortical folding have been reported in a more specific left ventricular pathologic condition: the hypoplastic left heart syndrome.
In fetuses with hypoplastic left heart syndrome, severely compromised cerebral perfusion is maintained by the right ventricle and achieved by forward flow in the arterial duct and retrograde flow in the aortic arch. Intracardiac mixing of oxygenated blood and deoxygenated venous return contributes to an adverse outcome. Hypoplastic left heart syndrome has been assumed to have a greater impact on cerebral perfusion and brain development than RVOTO, such as in TOF.
The aim of this single-center, retrospective, case-control study was to identify alterations of fetal brain and cerebrospinal fluid (CSF) volumes in cases of TOF, with the use of 3-dimensional volumetric fetal brain magnetic resonance imaging (MRI). The hypothesis was that TOF impacts brain development at a very early stage that results in decreased parenchymal volumes and enlarged CSF spaces.
In contrast to previous works, this study was intended to detect brain volume abnormalities in a clinically and pathophysiologically distinct type of CHD at gestational stages before the completion of neuronal migration (25 GW). Consequently, we aimed to provide further evidence that altered brain development is not limited to CHD with left ventricular dysfunction.
Materials and Methods
Test subjects
A total of 58 fetal MRI datasets that were acquired between March 2004 and April 2014 were studied retrospectively. The data were obtained from pregnant women who had been informed about the procedure and the possible risks of the examination and who had given written, informed consent for a prenatal MRI study before the examination. MRI was performed at a median gestational age (GA) of 25 GW (20-34 GW).
The retrospective study was approved by the local ethics committee (registration no. 1367/2014), and all image data were pseudonymized before further analysis. Thirty fetuses with confirmed TOF were recruited as cases, and 28 normal control fetuses were recruited from pregnancies with a normal fetal echocardiogram. The diagnosis was made during routine obstetric ultrasound screening and confirmed by 2 expert fetal cardiologists (U.S.M., E.M.).
Exclusion criteria were multiple gestation pregnancy and prenatally documented abnormal karyotype (22q11 microdeletion; trisomy 13, 21, and 18). Two expert fetal neuroradiologists (D.P., G.K.) screened the existing image data for signs of intraventricular hemorrhage, infections, or any gross structural brain or spinal cord abnormalities.
Fetuses with TOF
A retrospective investigation of medical patient data indicated a normal karyotype without numeric aberrations or 22q11 microdeletion in 16 of 30 (53%) fetuses with TOF. In 11 of 30 fetuses (37%), no detailed genetic workup was available because of the parents’ preference. A PraenaTest (LifeCodexx AG, Konstanz, Germany) for trisomy 13, 18, and 21 was performed on 1 of the fetuses, with negative results. A 22q11 microdeletion was found in 1 of 30 fetuses (3%), and the case was excluded from the study. In 1 of 30 fetuses (3%) with TOF, an inversion on chromosome 14 (inv[14][q13q24]) was present; in another 1 of 30 fetuses (3%), a rearrangement on the p arm of chromosome 5 was found. Both fetuses had no apparent brain malformation and were included in the study. A bilateral superior vena cava was observed in 1 of 30 fetuses (3%) with TOF. No further malformations were observed in the remaining 29 of 30 cases (97%). Five of 30 fetuses (17%) with TOF were excluded from the study because of excessive fetal motion imaging artifacts.
Normal control fetuses
Twenty-eight fetuses were entered into the study as normal control fetuses after 3 fetal MRI experts (D.P., P.C.B, G.K.) performed a detailed retrospective review of control examination image data.
In 19 of 28 fetuses (68%), no structural abnormalities were observed on fetal MRI. MRI findings in the control group included premature rupture of membranes (6/28 fetuses; 21%) and fetal body malformations without a known association with structural brain abnormalities or heart defects (9/28 subjects; 32%), such as malformations of the genitourinary system (3/28 subjects; 11%), congenital cystic adenomatoid malformation (2/28 subjects; 7%), isolated cleft lip without genetic abnormality (2/28 subjects; 7%), scoliosis (1/28 subjects; 4%), and a suspected omental cyst (1/28 subjects; 4%). Two of 28 cases (7%) were imaged because of maternal indications.
Four of 28 normal control fetuses (14%) were excluded from the study because of excessive fetal motion during image acquisition.
Age matching
Twenty-four fetuses with TOF and 24 normal control fetuses were included in the final study cohort (n = 48).
Mean GA was 179.7 ± 25.3 days (n = 24), with a median age of 176 days (25 + 1 GW) among fetuses with TOF and 179.3 ± 24.9 days (n = 24), with a median age of 175 days (25 + 0 GW) in the control group. Mean maternal age was 34.7 ± 5.2 years (n = 24) among cases, and 36.4 ± 6.1 years (n = 24) in control fetuses.
Twelve of 24 fetuses (50%) with TOF and 12 of 24 (50%) control fetuses were at <25 GW.
Cases and control fetuses were matched based on GA, and 24 matched pairs were obtained with a GA difference of no >4 days between the matched subjects and an average difference of 1.00 ± 1.02 days.
Data acquisition and processing
Ultrafast 1.5 Tesla, T2-weighted MRI image sequences of the fetal brain (echo time, 100-140 msec; repetition time, 9000-19,000 msec; acquisition time, <25 seconds) were acquired in the coronal, sagittal, and axial orientations without the administration of contrast agents or sedatives. The resulting anisotropic image data had a 256 × 256 pixel matrix, an in-plane resolution of 0.72-1.17 mm in both dimensions, and a slice thickness of 3.0-4.4 mm.
The acquired gray-scale image data were converted from DICOM (Digital Imaging and Communications in Medicine) to NIfTI (Neuroimaging Informatics Technology Initiative) format with DCM2NII software (version 12.12.2012; https://www.nitrc.org/frs/?group_id=556 ) and edited with ITK-SNAP (version 2.4.0; www.itksnap.org ).
A single expert (C.S.) used MRI image data in the coronal orientation to perform a manual segmentation of cortical gray matter, subcortical brain parenchyma, the cerebellum, the ventricular system, and the external CSF spaces ( Figure 1 ).
The segmentation of case and control group data was conducted in random order. The rater was blinded to the information about the gray-scale image data, GA, and the individual indication for the MRI examination before the segmentation.
The segmentation results were then transferred to and verified in 3D Slicer software (version 4.3.1; www.slicer.org ), and the width of the left and right ventricular atrium was measured manually on a coronal image by the same rater who had performed the manual segmentations.
Based on the created label-maps, gray matter (GMV), subcortical brain (SBV), cerebellar (CBV), ventricular (VV), and external CSF (eCSFV) volumes were determined, and total brain (TBV = GMV + SBV) and intracranial cavity (ICV = TBV + CBV + VV + eCSFV) volumes were calculated.
Atrium widths of the left and right lateral ventricles (LVW) were measured and examined for mild (10mm ≤ atrium width <13mm), moderate (13mm ≤ atrium width <15mm), or severe (atrium width ≥15mm) ventriculomegaly.
Statistics
Statistical analyses were conducted with IBM SPSS Statistics software (version 21.0; IBM, Armonk, NY). The metric data were calculated and expressed in means and standard deviation of means.
Mixed-model analyses of variance were performed on the 24 age-matched pairs. The samples were subdivided based on GA, with subjects aged <25 GW (cases, 12; control fetuses, 12), as opposed to those ≥25 GW (cases, 12, control fetuses, 12). Case vs control group affiliation served as a within-subjects variable (group); affiliation to either age group was used as a between-subjects variable (age group). The means of GMV, SBV, TBV, cerebellum, VV, eCSFV, ICV, and LVW for the narrower (nLVW) and the wider (wLVW) of the 2 lateral ventricles were compared ( Table ). To obtain more detailed results, post hoc t -tests were conducted.
| Variable | Difference of estimated means | P value | |||
|---|---|---|---|---|---|
| Tetralogy of Fallot–normal | Older-younger | Group (A) | Age group (B) | A × B | |
| Total brain volume, mm 3 | −13,874 | 69,961 | < .001 | < .001 | .446 |
| Gray matter volume, mm 3 | −4242 | 16,061 | .003 | < .001 | .056 |
| Subcortical volume, mm 3 | −9633 | 53,899 | .001 | < .001 | .976 |
| Gray matter volume/subcortical brain volume | 0.001 | −0.015 | .887 | .354 | .015 |
| Cerebellar volume, mm 3 | −426 | 3712 | .074 | < .001 | .740 |
| Intracranial cavity volume, mm 3 | −7528 | 115,873 | .314 | < .001 | .896 |
| Ventricular volume, mm 3 | 1962 | 2421 | .005 | .004 | .329 |
| External cerebrospinal fluid spaces volume/intracranial cavity volume | 0.05 | −0.02 | < .001 | .105 | .238 |
| Atrium width, narrower lateral ventricle, mm | 1.29 | 0.21 | .012 | .711 | .719 |
| Atrium width, wider lateral ventricle, mm | 1.33 | 0.65 | .023 | .280 | .466 |
To avoid an increasing error of the second kind, no multiplicity correction was performed. A significance level of α = .05 was applied.
Results
Mean TBV in fetuses with TOF was significantly lower ( P < .001) than in normal control fetuses, with reduced GMV ( P = .003) and SBV ( P = .001). TOF fetuses and normal control fetuses at ≥25 GW had larger TBV ( P < .001) than those at <25 GW, with a significant increase of GMV ( P < .001) and SBV ( P < .001). Neither GMV ( P = .056) nor SBV ( P = .976) showed significant interaction between the within-subjects (group) or between-subjects (age group) variables ( Figures 2 and 3 ). In contrast, significant interaction was observed for the mean ratio of GMV to SBV ( P = .015). A Student t test showed that GMV declined in proportion to TBV ( P = .026) in fetuses with TOF. Consequently, the mean ratio of GMV to SBV dropped significantly in cases ( P = .026), which resulted in a higher GMV-to-SBV ratio in normal control fetuses than in fetuses with TOF who were older than 25 GW ( Figure 4 ). Mean ICV was not significantly lower in cases ( P = .314). It increased with age ( P < .001) with a steady (ie, no interaction; P = .896) nonsignificant difference between cases and control fetuses. Mean VV was significantly higher ( P = .005) in fetuses with TOF; the lateral ventricles were, on average, significantly wider than that in normal control fetuses (nLVW, P = .012’ wLVW, P = .023). Compared with normal control fetuses, fetuses with TOF exhibited greater eCSFV ( P < .001) in relation to ICV. A slight asymmetry of LVW was observed in all fetuses. The mean difference between nLVW and wLVW was 0.82 ± 0.70 mm in the control group and 0.86 ± 0.87 mm in fetuses with TOF, with no significant variation between cases and control fetuses ( P = .850; paired samples t -test). Ventriculomegaly was observed in 5 of 24 fetuses (21%) with TOF: bilateral mild ventriculomegaly was found in 3 of 24 fetuses (13%); unilateral mild ventriculomegaly was found in 1 of 24 fetuses (4%); and a combination of mild and moderate ventriculomegaly was found in another 1 of 24 fetuses (4%). Ventriculomegaly occurred in none of the 24 control fetuses who were studied. LVW showed no significant increase (nLVW, P = .711; wLVW, P = .280) with age. Mean CBV was not significantly reduced ( P = .074) in fetuses with TOF.
