In Utero MRI Identifies Impaired Second Trimester Subplate Growth in Fetuses with Congenital Heart Disease

Abstract

 

The subplate is a transient brain structure which plays a key role in the maturation of the cerebral cortex. Altered brain growth and cortical development have been suggested in fetuses with complex congenital heart disease (CHD) in the third trimester. However, at an earlier gestation, the putative role of the subplate in altered brain development in CHD fetuses is poorly understood. This study aims to examine subplate growth (i.e., volume and thickness) and its relationship to cortical sulcal development in CHD fetuses compared with healthy fetuses by using 3D reconstructed fetal magnetic resonance imaging. We studied 260 fetuses, including 100 CHD fetuses (22.3–32 gestational weeks) and 160 healthy fetuses (19.6–31.9 gestational weeks). Compared with healthy fetuses, CHD fetuses had 1) decreased global and regional subplate volumes and 2) decreased subplate thickness in the right hemisphere overall, in frontal and temporal lobes, and insula. Compared with fetuses with two-ventricle CHD, those with single-ventricle CHD had reduced subplate volume and thickness in right occipital and temporal lobes. Finally, impaired subplate growth was associated with disturbances in cortical sulcal development in CHD fetuses. These findings suggested a potential mechanistic pathway and early biomarker for the third-trimester failure of brain development in fetuses with complex CHD.

Significance Statement

Our findings provide an early biomarker for brain maturational failure in fetuses with congenital heart disease, which may guide the development of future prenatal interventions aimed at reducing neurological compromise of prenatal origin in this high-risk population.

Introduction

The subplate zone is a transient structure situated between the cortical plate and the intermediate zone of the developing cerebral mantle. Subplate neurons receive thalamic inputs and project into the developing cortical plate, where they play key regulatory role in cortical development and plasticity (Kanold and Luhmann 2010). In the human brain, subplate neurons are generated around 6 gestational weeks and start to disappear in the third trimester, primarily through developmental apoptosis but also through migration into the cortical plate, with some surviving and remaining functional throughout life (Rakic 1976; Kostovic and Goldman-Rakic 1983; Kostovic and Rakic 1990; Chan et al. 2002; Judaš et al. 2010; Kanold and Luhmann 2010; Corbett-Detig et al. 2011). Using advanced in vivo fetal brain magnetic resonance imaging (MRI) techniques, the subplate zone is traceable on structural MRI images from 17 gestational weeks until the early third trimester of pregnancy (Corbett-Detig et al. 2011; Vasung, Rollins, Yun, et al. 2020). Visualization and measurement of the subplate by MRI is enhanced by its prominent hydrophilic extracellular matrix (Widjaja et al. 2010; Kostović et al. 2014; Milos et al. 2020). Within this period, the subplate develops into the most prominent zone of the developing brain becoming, at its peak, around four times thicker than the overlying cortical plate (Kostovic and Rakic 1990; Kostović et al. 2002; Perkins et al. 2008; Widjaja et al. 2010; Corbett-Detig et al. 2011). Subplate neurons are vulnerable to injury during the fetal period and may provide a crucial link between early alterations in brain development and later impairments in cortical function leading to neurodevelopmental dysfunction (Kanold and Luhmann 2010).

Fetuses with congenital heart disease (CHD) are at increased risk for impaired brain development in utero and postnatally (Limperopoulos et al. 2010; Ortinau et al. 2013; Rollins et al. 2021). We and others have previously reported that CHD is associated with altered fetal brain development, including impaired brain volumetric growth, increased cerebrospinal fluid, altered cortical folding, abnormal brain metabolism, and reduced cerebral oxygen delivery (Limperopoulos et al. 2010; Clouchoux et al. 2013; Schellen et al. 2015; Sun et al. 2015; Masoller et al. 2016; Ortinau et al. 2019; Rollins et al. 2021) and that certain types of CHD may be at heightened risk (Donofrio et al. 2003; Awate et al. 2010; Rollins et al. 2021). Animal and neuropathological studies of CHD suggest that immature cortical development is secondary to white matter abnormalities, indicating developmental vulnerability of the developing subplate in CHD (Morton et al. 2017; Stinnett et al. 2017; Gertsvolf et al. 2018). Hypoxia/ischemia, a condition related to CHD, has been associated with disrupted subplate neuron maturation and connectivity as well as subplate neuron loss in animal models (McQuillen et al. 2003; McClendon et al. 2017; Mikhailova et al. 2017). Recently, a clinical study reported that fetuses with hypoplastic left heart syndrome (HLHS) or transposition of the great arteries (TGA) had lower total subplate volume compared with healthy fetuses (Rollins et al. 2021). Despite the growing body of evidence to support fetal onset of impaired brain growth and development in CHD, the subplate growth in different lobes and its association with cortical development in CHD fetuses remains unknown.

The objectives of this study were 3-fold: 1) to compare subplate growth (i.e., volume and thickness) and cortical development (i.e., sulcal depth) in the frontal, parietal, temporal, occipital lobes, and insula between CHD and healthy fetuses; 2) to compare subplate growth and cortical development among different CHD diagnostic groups; and 3) to determine the relationship between subplate growth and cortical sulcal development in CHD and healthy fetuses by using advanced 3D reconstructed MRI. We hypothesize that CHD fetuses will demonstrate altered global and/or regional subplate growth compared with healthy fetuses and that subtypes of CHD will be associated with different patterns of subplate development. We also hypothesize that fetal subplate growth will be associated with cortical sulcal development in healthy and CHD fetuses.

Materials and Methods

Study Design

We prospectively recruited women with normal pregnancies and those pregnancies complicated by a fetal CHD diagnosis between December 2013 and July 2019. Exclusion criteria included: 1) fetuses with extracardiac anomalies noted on antenatal ultrasound or identified chromosomal abnormalities and 2) pregnant women with i) pregnancy-related complications (e.g., preeclampsia, diabetes), ii) known medical disorders (e.g., psychiatric, genetic, metabolic, etc.), iii) medication or illicit drug use, iv) smoking or alcohol use, v) contraindications to MRI (e.g., metal implants or claustrophobia), and vi) multiple pregnancy. Eligible participants were scheduled to undergo the fetal MRI study between 17 and 32 weeks of gestational age (GA) because the anatomic boundaries of the subplate are difficult to discern reliably on structural MRI after 32 weeks of GA (Vasung, Rollins, Yun, et al. 2020). Informed written consent was obtained from all participants, following a protocol approved by the institutional review board at Children’s National Hospital.

Baseline Characteristics of the Study Cohort

A diagram illustrating participant recruitment is shown in Supplemental Figure S1. We consecutively enrolled 342 pregnant women (135 fetal CHD and 207 fetal controls). We excluded 18 CHD fetuses that were subsequently diagnosed with genetic syndromes, 1 CHD with a previously undiagnosed brain malformation (lissencephaly), 57 subjects (16 CHD and 41 controls) with excessive fetal motion on MRI, and 6 controls that developed other complications (e.g., pregnancy-induced hypertension and hypothyroidism). There were 56 subjects (30 CHD and 26 controls) with an untraceable subplate region on 3D reconstructed MRI scans due to brain maturation and were excluded for subplate volume and thickness measurements. However, these 56 subjects were included in our study for cortical sulcal depth measurement. Our final sample consisted of 260 fetuses, including 100 CHD fetuses (57 males, 43 females) and 160 healthy fetuses (82 males, 78 females). The GA at MRI was 28.2 ± 2.3 (range: 22.3–32) weeks for CHD and 27.1 ± 2.7 (range: 19.6–31.9) weeks for controls. The subplate was measured on 204 fetuses, including 70 CHD fetuses (GA: 27.1 ± 2.0; range: 22.3–30.1 weeks) and 134 healthy fetuses (GA: 26.4 ± 2.3; range: 19.6–30.1 weeks). Mean maternal age was 31.9 ± 5.6 (range: 19–43) years and 31.7 ± 6.6 (range: 18–49) years for CHD and controls, respectively. Characteristics of the cohort are shown in Table 1.

Table 1.

Characteristics of the cohort

Characteristic	CHD (n = 100)	Controls (n = 160)	P value
Male, n (%)	57 (57)	82 (51)	0.37
GA at MRI, mean ± SD (range), week	28.2 ± 2.3 (22.3–32)	27.1 ± 2.7 (19.6–31.9)	0.001
Maternal age, mean ± SD (range), year	31.9 ± 5.6 (19–43)	31.7 ± 6.6 (18–49)	0.87
Maternal education, n (%)   ≤ High school   Partial college   College graduate   Graduate degree   Unknown	14 (14) 18 (18) 33 (33) 33 (33) 2 (2)	29 (18) 22 (14) 37 (23) 70 (44) 2 (1)	0.15
Maternal employment, n (%)   Professional   Skilled, clerical, or sales   Unskilled laborer   Unemployed or homemaker   Unknown	69 (69) 8 (8) 3 (3) 18 (18) 2 (2)	103 (64) 14 (9) 5 (3) 35 (22) 3 (2)	0.88
Primigravida, n (%)	36 (36)	50 (31)	0.49
Primipara, n (%)	44 (44)	72 (45)	0.79
Type of CHD, n (%) SV CHD   HLHS (including MS or MA and AS or AA)   HLHS variants with DORV   Unbalanced AVSD with DORV   Pulmonary stenosis or atresia with intact septum   Tricuspid atresia with HRH (all variants)   Heterotaxy with SV 2V CHD   Tetralogy of Fallot   TGA   Ventricular septal defect with aortic stenosis or atresia and coarctation   AVSD (complex with coarctation or heterotaxy)   Truncus arteriosus   DORV with pulmonary stenosis or atresia   Coarctation of the aorta   Tetralogy with absent pulmonary valve   Total anomalous pulmonary venous return   AS	41 (41) 24 (24) 4 (4) 4 (4) 3 (3) 3 (3) 3 (3) 59 (59) 17 (17) 14 (14) 8 (8) 6 (6) 4 (4) 4 (4) 3 (3) 1 (1) 1 (1) 1 (1)

Note:  P value for difference between CHD and controls based on two-tailed, unpaired t-test for continuous variables and Fisher exact test for categorical variables. SD, standard deviation; DORV, double outlet right ventricle; AVSD, atrioventricular septal defect; HRH, hypoplastic right heart; MS, mitral stenosis; MA, mitral atresia; AS, aortic stenosis; AA, aortic atresia.

MRI Acquisition Protocol

Fetal brain MRI images were acquired on a 1.5 Tesla GE DISCOVERY MR450 scanner with an eight-channel receiver coil. The scanning protocol included multiplanar single shot fast spin echo acquisitions (repetition time: 1100 ms; echo time: 160 ms; flip angle: 90°; field-of-view: 32 cm; matrix: 256 × 192; in-plane resolution: 1.25 × 1.66 mm²; 2-mm slice thickness). The acquisition time per plane was 2–3 min and participants were free-breathing during the scanning. All fetal brain MRI studies were reviewed by an experienced fetal neuroradiologist who was blinded to CHD versus control status.

Fetal Brain Reconstruction

Motion-corrupted images of axial, sagittal, and coronal planes were reconstructed into a high-resolution 3D volumetric image (Fig. 1) by using a validated pipeline (Kainz et al. 2015), a parallel slice-to-volume reconstruction method using evaluated point-spread functions for the image reconstruction from motion corrupted stacks of 2D slices. After reconstruction, all images were reoriented to fetal brain atlas (Serag et al. 2012) with the image resolution of 0.86 × 0.86 × 0.86 mm³. Subjects with image evidence of structural brain injury or severe motion artifact were excluded.

Figure 1.

An example of fetal brain reconstruction from 2D single shot fast spin echo MRI slices of axial, sagittal, and coronal planes (left) to a single 3D volumetric image (right).

Fetal Subplate and Cortical Plate Segmentation

The subplate and cortical plate were automatically segmented by combining the prior segmentation information from CRL fetal brain atlases (Gholipour et al. 2017) and SyN registration in ANTs toolbox (Tustison et al. 2014). An example of the automatic segmentation results of a fetal brain T2-weighted MR image is shown in Figure 2. The automatic segmentations were further manually corrected using ITK-SNAP software by a rater with >5 years of experience on fetal brain MR image segmentation. Thirty percent of scans were randomly selected and were corrected by the same rater and by another rater. Intra- and inter-rater reliabilities using intraclass correlation coefficient were >0.98 for both subplate and cortical plate. Raters were blinded to CHD versus control status for all subjects.

Figure 2.

Automatic segmentation results of a 3D reconstructed fetal brain MR image at 24 gestational weeks. The subplate regions (see red arrows) are shown in yellow and gray colors.

Fetal Brain Parcellation

Five regions of interest for each of left and right hemispheres were obtained, including frontal lobe, parietal lobe, temporal lobe, occipital lobe, and insula (Supplemental Fig. S2). The parcellation for each region of interest was first obtained on CRL fetal brain atlases (Gholipour et al. 2017) by consolidating 50 parcellated brain regions from Draw-EM pipeline (Makropoulos et al. 2014) and was then mapped to our dataset by using SyN registration in ANTs toolbox. Finally, the parcellation results were manually corrected by the same rater using ITK-SNAP.

Subplate Volumetric, Thickness, and Cortical Sulcal Measures

After segmentation and parcellation, global and regional subplate volumes, thickness, and cortical sulcal depth were calculated. Subplate volume was calculated by multiplying the voxel number of the segmented subplate region and the image resolution. Subplate thickness was measured as the shortest distance between the inner and outer surfaces of the segmented subplate zone (Aganj et al. 2009). Cortical sulcal depth was calculated as the Euclidean distance from each vertex on the cortical inner surface to the nearest point on the cerebral hull surface (Yun et al. 2013).

Subplate Growth Using Shape Analysis

Subplate growth was shown on a template surface measured by subplate thickness change across GA (Supplemental Fig. S3). First, the thickness was calculated at each voxel of the segmented subplate region (Aganj et al. 2009). A template was chosen from CRL fetal brain atlases (Gholipour et al. 2017) based on the mean GA of our study cohort. Thereafter, T2-weighted images of fetal brains were registered to the template by using SyN in ANTs toolbox. Subsequently, the subplate thickness of each individual brain was mapped to the template according to the deformation field obtained via SyN in the previous step. Finally, a linear regression model was performed at each vertex on the subplate surface of the template to calculate the subplate thickness change across GA. Note, the registration was only performed to illustrate the subplate thickness change across GA on a template surface (Supplemental Fig. S3). The subplate and cortical measures in CHD versus controls were calculated from original (nonregistered) images.

CHD Diagnostic Categories

Fetal CHD was diagnosed using fetal echocardiography with standard imaging protocols. All diagnoses were reviewed by a fetal cardiologist. The CHD cohort was classified into single-ventricle (SV) and two-ventricle (2V) CHD (Clancy et al. 2000), and the CHD diagnoses of our cohort are summarized in Table 1. Of the 100 fetuses with CHD, 41 had SV and 59 had 2V physiologies. The three most common CHD types were HLHS (28 subjects), tetralogy of Fallot (TOF, 17 subjects), and TGA (14 subjects) (Table 1).

Statistical Analysis

Statistical analyses were performed using SAS University Edition and MATLAB 2019a. ANCOVA was used to measure the subplate volume, thickness, and cortical sulcal depth in CHD versus controls and different types of CHD, controlling for GA at MRI and fetal gender. Generalized linear models were used to evaluate the association between subplate volume/thickness and cortical sulcal depth, adjusting for GA at MRI and fetal gender; group by subplate volume/thickness interaction term was added to models to evaluate any differential association between subplate volume/thickness and cortical sulcal depth in CHD versus controls. For shape analysis, generalized linear models were used to measure the subplate thickness changes across GA at each vertex on the subplate surface of the template. P values were adjusted for multiple testing by controlling the false discovery rate based on the number of outcomes (Benjamini and Hochberg 1995), and two-tailed adjusted P values ≤ 0.05 were considered to be significant.

Results

Global and Regional Subplate Volumes

In both CHD and control cohorts, fetal subplate volumes significantly increased with advancing GA (Supplemental Fig. S4). The comparison of global and regional subplate volumes in CHD and controls is shown in Table 2. Fetuses with CHD had smaller global subplate volumes for both left (11.33 vs. 12.37 cm³, P < 0.0001) and right (11.61 vs. 12.63 cm³, P < 0.0001) hemispheres compared with controls. All five parcellated regions in each hemisphere were smaller in CHD fetuses versus controls (all adjusted P < 0.05, Table 2). After accounting for the total cerebral volume, the relative subplate volume (subplate volume/cerebrum volume) in the right hemisphere overall (0.257 vs. 0.262; P = 0.01) and right frontal lobe (0.094 vs. 0.097; P = 0.003) remained smaller in CHD fetuses versus controls (Supplemental Table S1). In CHD fetuses, the subplate volume was smaller in SV versus 2V CHD for the right occipital (1.25 vs. 1.44 cm³, P = 0.007) and temporal (2.36 vs. 2.55 cm³, P = 0.008) lobes (Supplemental Table S2). Considering HLHS and TGA have been suggested as high-risk CHD for impaired brain development, we further performed subanalyses on comparing global and regional subplate volumes in fetuses with HLHS versus other SV CHD versus TGA versus other 2V CHD (Supplemental Table S3). Fetuses with HLHS had significantly smaller subplate volumes in right occipital (1.22 vs. 1.47 cm³, P = 0.001) and temporal (2.35 vs. 2.57 cm³, P = 0.007) lobes compared with fetuses with other 2V CHD types (i.e., 2V CHD excluding TGA).

Table 2.

Global and regional subplate volume in CHD and controls

Volume (cm3)	Left subplate			Right subplate
	CHD	Control	P value	CHD	Control	P value
Frontal lobe	3.93	4.34	<0.0001*	4.22	4.63	<0.0001*
Parietal lobe	3.42	3.65	0.001*	3.30	3.54	0.0006*
Occipital lobe	1.32	1.47	0.0005*	1.25	1.36	0.007*
Temporal lobe	2.20	2.39	0.0002*	2.32	2.53	<0.0001*
Insula	0.22	0.24	0.0002*	0.24	0.27	0.002*
Global	11.33	12.37	<0.0001*	11.61	12.63	<0.0001*

Note: Results of least squares means from ANCOVA, controlling for GA at MRI and gender.

^*Significant after adjusting for multiple testing.

Global and Regional Subplate Thickness

Supplemental Figure S5 shows the subplate shape and thickness of five healthy fetuses at 20, 22, 25, 28, and 30 gestational weeks, where red color indicates larger thickness and blue color indicates smaller thickness. Subplate thickness change across GA in healthy fetuses is shown in Supplemental Figure S3. Subplate thickness of red regions increased faster than other regions. In both hemispheres, the posterior regions had the most significant increases in subplate thickness between 19.6 and 30.1 gestational weeks (Supplemental Fig. S3). The subplate mean thickness for each hemisphere increased linearly as a function of GA in both CHD and controls (Supplemental Fig. S6). Subplate mean thickness in global and five parcellated regions of left and right hemispheres are shown in Table 3. Fetuses with CHD had smaller subplate thickness in the left and right frontal (left: 2.76 vs. 2.84 mm, P = 0.005; right: 2.96 vs. 3.06 mm, P = 0.0004) and temporal (left: 2.60 vs. 2.70 mm, P = 0.002; right: 2.67 vs. 2.75 mm, P = 0.009) lobes, the insula (left: 1.79 vs. 1.92 mm, P < 0.0001; right: 1.92 vs. 2.04 mm, P = 0.002) as well as the right global hemisphere (3.19 vs. 3.27 mm, P = 0.01). Moreover, subplate thickness was reduced in SV compared with 2V CHD in the right occipital (2.52 vs. 2.81 mm, P = 0.001) and temporal (2.64 vs. 2.78 mm, P = 0.006) lobes (Supplemental Table S4). Fetuses with HLHS had significantly reduced subplate thickness in the right occipital (2.49 vs. 2.83 mm, P = 0.0007) and temporal (2.61 vs. 2.79 mm, P = 0.002) lobes compared with fetuses with other 2V CHD types (i.e., 2V CHD excluding TGA) (Supplemental Table S5).

Table 3.

Global and regional subplate thickness in CHD and controls

Mean Thickness (mm)	Left subplate			Right subplate
	CHD	Control	P value	CHD	Control	P value
Frontal lobe	2.76	2.84	0.005*	2.96	3.06	0.0004*
Parietal lobe	4.0	4.02	0.72	4.02	4.07	0.30
Occipital lobe	2.72	2.81	0.05	2.60	2.66	0.22
Temporal lobe	2.60	2.70	0.002*	2.67	2.75	0.009*
Insula	1.79	1.92	<0.0001*	1.92	2.04	0.002*
Global	3.13	3.19	0.04	3.19	3.27	0.01*

Note: Results of least squares means from ANCOVA, controlling for GA at MRI and gender.

^*Significant after adjusting for multiple testing.

Global and Regional Cortical Sulcal Development

Supplemental Figure S7 shows the cortical sulcal depth of left and right hemispheres in CHD and healthy fetuses, where the sulcal depth increased significantly with advancing GA for both CHD and controls. Global and regional differences in cortical sulcal depth in CHD versus controls are shown in Supplemental Table S6. CHD fetuses showed reduced cortical sulcal depth in the right occipital lobe (0.64 vs. 0.71 mm, P = 0.03) (Supplemental Table S6). In CHD fetuses, cortical sulcal depth was reduced in SV compared with 2V CHD in the right temporal lobe (1.34 vs. 1.43 mm, P = 0.05) (Supplemental Table S7). However, these differences were no longer significant after adjusting for multiple testing. The cortical sulcal depth did not significantly differ in fetuses with HLHS versus other SV versus TGA versus other 2V CHD types (Supplemental Table S8).

Association between Subplate Volume/Thickness and Cortical Sulcal Development

The associations between subplate volume and cortical sulcal depth are shown in Supplemental Table S9. The subplate volume in each hemisphere was positively associated with cortical sulcal depth in both CHD (left: 0.06 mm/cm³; 95% confidence interval [CI]: 0.03–0.08 mm/cm³; right: 0.05 mm/cm³; 95% CI: 0.02–0.07 mm/cm³) and controls (left: 0.04 mm/cm³; 95% CI: 0.02–0.05 mm/cm³; right: 0.04 mm/cm³; 95% CI: 0.03–0.06 mm/cm³). Notably, the associations between subplate volume and cortical sulcal depth in left frontal lobe and right occipital lobe were significantly different in CHD versus controls, where the association between subplate volume and sulcal depth was greater in CHD compared with controls (left frontal lobe: 0.05 vs. 0.04 mm/cm³, P = 0.006; right occipital lobe: 0.06 vs. 0.05 mm/cm³, P = 0.008). We further showed the association between subplate thickness and cortical sulcal depth (Supplemental Table S10). The association between subplate thickness and cortical sulcal depth in frontal and occipital lobes were significantly greater in CHD versus healthy fetuses (left frontal lobe: 0.21 vs. 0.10 mm/mm; P = 0.007; right frontal lobe: 0.25 vs. 0.05 mm/mm; P = 0.003; left occipital lobe: 0.22 vs. 0.07 mm/mm; P = 0.005; right occipital lobe: 0.28 vs. 0.09 mm/mm; P = 0.003).

Discussion

In this study, we show evidence of impaired second trimester brain growth in fetuses with CHD using the largest known cohort. Specifically, we report decreased global and regional subplate volume and thickness (adjusted for GA and gender) compared with healthy fetuses. Notably, a selective reduction in subplate volume and thickness was observed in the right occipital and temporal lobe of SV CHD compared with 2V CHD fetuses. Moreover, fetuses with HLHS (the most severe form of CHD) showed the greatest impairment in subplate growth in the right occipital and temporal lobes. Interestingly, subplate volume was positively associated with emerging cortical sulcal depth in controls and CHD. In fetuses with CHD, reduced subplate thickness was associated with decreased sulcal depth in the frontal and occipital lobes.

It is increasingly recognized that prenatal factors contribute to neurodevelopmental disabilities in infants with CHD. Although the prenatal mechanisms of injury are complex and likely multifactorial, abnormal neurological development in fetuses with CHD likely includes both genetic and environmental factors (Andelfinger 2008; Limperopoulos et al. 2010; Wu et al. 2020). Critical events in fetal brain development occur in the second and third trimesters of pregnancy (Limperopoulos et al. 2010) and require the appropriate delivery of oxygen and nutrients by the fetal cardiovascular system. Cerebral blood flow may be abnormal in fetuses with certain forms of CHD (Donofrio et al. 2003; Kaltman et al. 2005), which may impair myelination and cortical organization. Additionally, hypoxia related to reduced blood flow may impair proliferation and neurogenesis in the subventricular zone (Morton et al. 2017) and chronic fetal brain hypoxia may slow brain oxygen metabolism and result in altered cellular development (Lim et al. 2016), which is accompanied by impaired subplate development (Morton et al. 2017; Gertsvolf et al. 2018). Environmental factors, such as impaired maternal mental health, the most common complication of pregnancy, may be especially important when fetal CHD has been diagnosed (Wu et al. 2020). Elevated levels of maternal mental distress have been associated with increased uterine artery resistance and decreased placental expression of neurotropic precursor monoamine oxidase A, which may limit blood flow to the fetal brain and affect cell neurogenesis, migration, and differentiation in fetal brain, respectively (Fisk et al. 1999; Blakeley et al. 2013).

Our data showed that the posterior regions of the subplate in healthy fetuses had the most significant growth measured by the thickness increase across 19.6–30.1 gestational weeks, corroborating the previously reported regional variation in subplate growth rate (Corbett-Detig et al. 2011). An earlier study on 21 healthy fetuses between 20.6–25.9 gestational weeks showed the greatest increase in subplate thickness, which occurred in the temporal and occipital lobes (Corbett-Detig et al. 2011). We also found the anatomical border between the subplate and intermediate zone became less distinguishable, with the subplate becoming confined to certain gyral crests on structural MRI for most of subjects >29 weeks GA. This is consistent with the findings in other fetal MRI studies (Perkins et al. 2008; Corbett-Detig et al. 2011). These MRI features of subplate maturation are thought to reflect the relocation of ascending thalamocortical axons into the overlying cortex (Kostović and Judaš 2002; Kostović et al. 2002; Perkins et al. 2008) and shrinking of the hydrophilic extracellular matrix (Milos et al. 2020).

Our previous study showed that fetal global brain volumes in CHD and controls were comparable at the end of the second trimester but that there was a progressive fall off in brain growth over the third trimester (Limperopoulos et al. 2010). In the current study, we demonstrate that the subplate volume and thickness impairments in fetuses with CHD were already evident in the second trimester, which may reflect an earlier biomarker of brain maturational failure in this high-risk fetal population. We found that global and regional subplate volumes were decreased in CHD versus healthy fetuses; the difference in frontal lobe remained significant even after accounting for the cerebral volume. Subplate thickness in the CHD group was decreased bilaterally in the frontal and temporal lobes as well as in the insula. The functional topography of these impairments in subplate growth corresponds to the spectrum of neurodevelopmental deficits described in survivors of CHD. Notably, the frontal lobes control important cognitive skills and executive functions, while the temporal lobes play vital roles in emotion, memory, sensory input, speech, and comprehension (Fuster 2002; Hickok and Poeppel 2007), and the insula is involved in sensory and affective processing and cognition (Uddin et al. 2017). These regions of impairments in fetuses with CHD may play a role in cognitive, language, and behavior dysfunction reported in CHD infants, years following surgical repair (Majnemer et al. 2006, 2008).

We also reported that CHD fetuses showed reduced cortical sulcal depth in the right occipital lobe; however, this difference was no longer significant after adjusting for multiple testing. Altered cortical folding and sulcal depth have been suggested in CHD fetuses at later GAs and postnatally before cardiac surgery (Awate et al. 2010; Clouchoux et al. 2013; Ortinau et al. 2013). The late second trimester GA of our cohort, that is, prior to the major acceleration of cortical sulcation that takes place in the third trimester, likely under-represents the impairment of cortical development in CHD. We also reported positive associations between subplate volume and cortical sulcal depth for both hemispheres in controls and CHD, supporting the important role of the subplate in cortical maturation. Notably, we found that the association between subplate thickness and cortical sulcal depth in frontal and occipital lobes were significantly greater in CHD fetuses. Given that our CHD fetuses showed impaired subplate thickness in the frontal lobe compared with controls (Table 3), we posit that there may be enhanced susceptibility to alterations in frontal cerebral cortical development for CHD infants later in life, which needs to be confirmed in follow-up studies.

Our subanalyses on SV versus 2V CHD physiology suggested smaller regional subplate volume and thickness in SV fetuses. In CHD fetuses with SV physiology, reduced fetal oxygen and substrate delivery results in hypoxia (Sun et al. 2015). Fetal lamb studies have shown that chronic hypoxia is associated with white matter hypervascularity, decreased neuronal density, impaired myelination, and altered microglial morphology, suggesting fetal hypoxia may delay cellular maturation and result in brain dysmaturation (Lawrence et al. 2019, 2020; McGovern et al. 2020). Importantly, animal studies have demonstrated that the subplate neurons are vulnerable to hypoxia-ischemia (McQuillen et al. 2003; Mikhailova et al. 2017) and that transient hypoxemia disrupts subplate neuron arborization and functional maturation (McClendon et al. 2017). These studies lend credence to the impairments in subplate volume and thickness in SV versus 2V CHD fetuses in the current study. HLHS and associated variants represent one of the most severe forms of SV CHD. In our subanalyses, we found fetuses with HLHS had smaller subplate volumes and thickness in the right occipital and temporal lobes compared with 2V CHD excluding TGA. We and others have previously reported impaired global subplate volume and delayed cortical development in certain complex CHD types (HLHS and/or TGA) compared with healthy fetuses (Clouchoux et al. 2013; Rollins et al. 2021). This study suggested a selective vulnerability of the subplate in the setting of HLHS in the occipital and temporal lobes compared with other CHD types. A recent MRI-histological correlative study suggested that the MR-signal intensity and thickness of the subplate varied in limbic and neocortical regions and that this regional difference already exists at 15 postconceptional weeks (Bobić-Rasonja et al. 2021). Changes in the MR-signal intensity have been suggested to reflect the developmental changes in the amount and composition of the extracellular matrix, which largely exists in subplate zone (Kostović et al. 2002; Corbett-Detig et al. 2011; Milošević et al. 2014; Bobić-Rasonja et al. 2021). The subplate remnants were suggested to disappear first in the central and occipital regions and then in the prefrontal region in infants (Kostović et al. 2014), indicating regional difference in the timing of the subplate expansion in prenatal stage. In addition, the occipital and temporal lobes were suggested to have the most significant increases in subplate thickness by a previous MRI study (Corbett-Detig et al. 2011). These fastest subplate growing regions may be more vulnerable in severe forms of CHD compared with slower growing regions. Furthermore, regional differences in subplate gene expression, neuronal maturation, dendritic branching, and density and time of the arrival of thalamocortical afferents in the subplate may contribute to this regional difference in subplate impairment in CHD types (Kostovic and Rakic 1990; Corbett-Detig et al. 2011; Kim et al. 2013; Kostović et al. 2014; Vasung, Rollins, Velasco-Annis, et al. 2020). Finally, fetuses with HLHS have been reported to have reduced pulsatility indices in the main cerebral arteries, where the occipital and temporal lobes are supplied by the posterior and middle cerebral arteries (Peng et al. 2018). Whether the difference in regional cerebral blood flow impairments in HLHS (Peng et al. 2018) is associated with regional subplate alterations is an intriguing question which requires further study.

Although our study has a number of strengths, the limitations deserve mention. First, the current paper did not correlate subplate findings with postnatal brain growth and neurodevelopment in infancy and childhood. Our longitudinal follow-up study is currently underway. Also, we found regional vulnerabilities that need to be further explored. In addition, the subplate region was not prominent in T2-weighted MRI scans for most subjects with GA > 29 weeks due to brain maturation. In our study, we only chose subjects with clear subplate boundaries and our intra- and inter-rater reliabilities were >0.98. Moreover, since we approached CHD subjects after their confirmed diagnoses (mean GA at CHD diagnoses: 21.9 gestational weeks), our CHD subjects had a minimum GA at MRI of 22.3 weeks which could not match to that of our healthy cohort (i.e., minimum GA at MRI of 19.6 weeks). In the statistical models, we controlled for GA at MRI to account for this age difference. Finally, the subplate development may also be abnormal in fetuses with certain types of CHD, as suggested by our subanalysis of HLHS. Our sample size for other CHD diagnostic subgroups (e.g., TGA and pulmonary atresia) was too small to examine whether subplate alterations were present. Ongoing work is needed and currently underway to increase our sample size for more specific CHD types and to explore the relationship between early fetal subplate growth impairments and later brain development as well as child long-term neurodevelopment in CHD.

Conclusion

We provide evidence for impaired second trimester subplate growth and development in a large cohort of CHD fetuses compared with healthy fetuses. Disturbances in subplate development were more pronounced among severe forms of fetal CHD. We also report that a direct relationship between subplate growth and emerging cortical sulcal development, suggesting that third trimester delayed cerebral cortical maturation may have its origins in second trimester subplate growth failure in CHD. This may, in turn, represent an earlier biomarker of elevated risk for altered later brain development in CHD population and potential therapeutic target.

Funding

National Institutes of Health (R01 HL116585-01); Thrasher Research Fund (14764).

Notes

We thank the pregnant participants and their families in the current study. We also thank our laboratory members for their contributions. Conflict of Interest: None declared.