Abnormal development of transient fetal zones in mild isolated fetal ventriculomegaly

Abstract

Mild isolated fetal ventriculomegaly (iFVM) is the most common abnormality of the fetal central nervous system. It is characterized by enlargement of one or both of the lateral ventricles (defined as ventricular width greater than 10 mm, but less than 12 mm). Despite its high prevalence, the pathophysiology of iFVM during fetal brain development and the neurobiological substrate beyond ventricular enlargement remain unexplored.

In this work, we aimed to establish the relationships between the structural development of transient fetal brain zones/compartments and increased cerebrospinal fluid volume.

For this purpose, we used in vivo structural T2-weighted magnetic resonance imaging of 89 fetuses (48 controls and 41 cases with iFVM). Our results indicate abnormal development of transient zones/compartments belonging to both hemispheres (i.e. on the side with and also on the contralateral side without a dilated ventricle) in fetuses with iFVM. Specifically, compared to controls, we observed enlargement of proliferative zones and overgrowth of the cortical plate in iFVM with associated reduction of volumes of central structures, subplate, and fetal white matter. These results indicate that enlarged lateral ventricles might be linked to the development of transient fetal zones and that global brain development should be taken into consideration when evaluating iFVM.

Introduction

Fetal ventriculomegaly (FVM), involving enlargement of one or both of the lateral ventricles, is the most common abnormality of the fetal central nervous system (CNS), with a prevalence ranging between 0.3 and 1.5 per 1,000 live births (Myrianthopoulos 1977; Pilu and Hobbins 2002). The majority of FVM is initially diagnosed on screening fetal sonography at the time of a routine second trimester fetal survey, warranting further classification (based on the size of ventricles) and clinical investigation [screening for congenital infection, karyotyping, and magnetic resonance imaging (MRI) to exclude associated brain abnormalities]. In approximately half of FVM cases, no other identifiable CNS or systemic structural abnormalities are seen, constituting FVM subgroup classified as “isolated fetal ventriculomegaly” (iFVM) (Mahony et al. 1988; Goldstein et al. 1990; Bromley et al. 1991; Tomlinson et al. 1997; Den Hollander et al. 1998; Lipitz et al. 1998; Robson et al. 1998; Vergani et al. 1998; Pilu et al. 1999; Senat et al. 1999; Kelly et al. 2001).

In contrast to the higher incidence of adverse neurodevelopmental outcomes in FVM that is associated with other brain abnormalities (Pier et al. 2011), the incidence of abnormal long-term outcomes among children with a history of iFVM varies substantially and ranges from 3% to 64% (Bloom et al. 1997; Vergani et al. 1998; Pilu et al. 1999; Gaglioti et al. 2005; Ouahba et al. 2006; Falip et al. 2007; Sadan et al. 2007; Leitner et al. 2009; Beeghly et al. 2010; Pier et al. 2011; Gómez-Arriaga et al. 2012; Tatlı et al. 2012; Chu et al. 2016; Fox et al. 2018). However, only certain neurodevelopmental outcomes in iFVM are explained by the size of the ventricles (Ouahba et al. 2006; Falip et al. 2007; Beeghly et al. 2010), whose proportions are roughly estimated by serial measurements of the atrial diameter (borderline iFVM = 10 mm; mild iFVM = 10–12 mm; moderate iFVM = 13–15 mm; severe iFVM > 15 mm) on ultrasound or MRI (Malinger et al. 2020). Thus, in the absence of other brain abnormalities and with standard ventricular measurements that are of limited clinical significance, the diagnosis of fetal iFVM remains a source of considerable anxiety for prospective parents and poses a significant challenge to clinicians during fetal counseling (Melchiorre et al. 2009).

Identifying biomarkers of outcome with high diagnostic performance (e.g. volumes of transient fetal zones) in the classification of iFVM and prediction of neurodevelopmental outcomes is central for parental counseling. iFVM biomarkers of neurodevelopmental outcome might help with the future development of clinical risk stratification tools for early interventions, treatments, and clinical counseling in this distinct population of pregnant women and their offspring. However, a lack of knowledge about the pathophysiology of ventriculomegaly and mechanisms underlying the increase in ventricular cerebrospinal fluid volume poses a limitation in hypothesis-driven research. Finally, whether the presence of iFVM is associated with altered fetal brain development and whether this altered brain development underlies poor neurodevelopmental outcome remains unknown. Our work is a first step towards identifying these associations.

Because of continued advances in acquisition and processing techniques, MRI offers unprecedented ways to study fetal brain development with higher resolution and improved conspicuity of the developing fetal cerebral mantle as compared to sonography (Gholipour et al. 2017; Vasung et al. 2019, 2020). Thus, fetal MRI can be used to characterize both normal and abnormal brain development (Rollins et al. 2020).

The aim of our study was to explore the anatomical substrate beyond the ventricular enlargement and to characterize the differences in spatiotemporal brain maturation between iFVM and controls that could serve as potential biomarkers of the neurodevelopmental outcome.

Material and methods

Subjects

This study was a retrospective observational study of fetuses with iFVM and controls. We retrospectively screened all pregnant women who had a fetal MRI at our institution between July 2013 and January 2019 (N = 2,623) and identified 136 cases of iFVM. For controls, we used fetal MRIs from pregnant women who were prospectively recruited as controls in previous studies (Gholipour et al. 2017; Rollins et al. 2020; Vasung et al. 2021). Eligibility criteria for iFVM cases and controls are shown in Supplementary Table S1.

This study was carried out under an Institutional Review Boards (IRB) approved research protocol. Since this study used retrospective review of records (clinical chart review was conducted for all pre- and post-natal clinical data), was of minimal risk to the participants, and did not involve the collection of identifiable private information the IRB waived the requirement for informed consent.

Magnetic resonance imaging

All retrospectively collected fetal MRIs were acquired on the same MRI scanner (3T Siemens Skyra, Siemens Healthcare, Erlangen, Germany). T2-weighted half-Fourier single shot turbo spin echo (HASTE) acquisitions of the fetal brain were acquired using the following image acquisition parameters; field of view = 300–400 mm, time echo = ~120 ms, time repetition = ~1.6 s, matrix size 256 × 256, slice thickness = 2–3 mm, and 3–14 sets of triplane HASTE acquisitions. Three to 5 sets of triplane HASTE acquisitions (each set consisting of runs obtained in the 3 orthogonal planes) were needed to reliably perform image processing and analysis. Control subjects had a mean of 12.3 (min–max No.: 7–8) T2 HASTE acquisitions per scan with the mean duration of the T2 portion of the scan of 14.9 min per subject (min–max: 7.5–21 min). Clinically acquired iFVM subjects had a mean of 6.4 (min–max No.: 3–14) T2 HASTE acquisitions per scan with the mean duration of the T2 portion of the scan of 8.5 min per subject (min–max: 4–18.6 min).

MRI preprocessing

Fetal MRI scans acquired during routine clinical examinations that met our eligibility criteria (Supplementary Table S1) were identified using a search engine. To ensure accuracy of clinical findings and absence of additional abnormal brain or body findings, all fetal brain and body MRI scans were read by both a pediatric neuroradiologist (EY, PEG) and a pediatric body radiologist (JE, SC, or CB), all with extensive experience in reading fetal MRIs. After the successful download, a visual quality check (QC) was performed (CV-A) on the multiple series of HASTE acquisitions and 3D reconstructions (LV). For each MRI, 4–14 of the best available T2 stacks were chosen as input images for reconstruction. Stacks that had persistent fetal motion, oblique angles, signal artifacts that obscured the brain, or an ROI which did not contain the entire brain were excluded. All HASTE acquisitions which passed the QC were combined and reconstructed into a motion-corrected high-resolution MRI volume. These MRI volumes were preprocessed with an in-house pipeline (Gholipour et al. 2017) composed of the following steps: (i) motion correction and super-resolution 3D reconstruction, (ii) brain masking (with occasional manual edits), (iii) N4 bias field correction with intensity normalization, and (iv) rigid registration to a spatiotemporal fetal brain MRI atlas. The resulting outputs were visually inspected, and poor reconstructions were attempted again with adjusted sets of input stacks until an acceptable reconstruction was produced. All the pipeline steps have been validated (Gholipour et al. 2017), and the pipeline was used in several previous studies (Vasung et al. 2019, 2020). Finally, we included only the data of subjects with excellent preprocessing QC and no need for manual correction of automatic brain segmentation (QC by LV and CV-A).

MRI segmentation

Using the previously published spatiotemporal atlas (Gholipour et al. 2017) and PSTAPLE (Akhondi-Asl and Warfield 2013), we automatically generated labels on reconstructed 3D images to achieve efficient and accurate fetal brain tissue segmentation (Supplementary Fig. 1).

The spatiotemporal atlas of fetuses younger than 32 gestational weeks (GW) includes the segmentation of transient fetal compartments (cortical plate, proliferative, intermediate, and subplate zone; Supplementary Fig. 1A and B). The window during which one can identify transient fetal brain structures on MRI (15-32GW) has been studied extensively and is reported in previous work (Vasung et al. 2016, 2019, 2020; Diogo et al. 2019; Kostović 2020). In fetuses older than 32 GW, the segmented brain structures are permanent (i.e. adult-like, Supplementary Fig. 1C and D) with less visible border between transient fetal compartments (i.e. subplate zone, intermediate zone). Thus, after 32 GW the subplate zone and intermediate zone were combined into one structure called the fetal white matter.

Next, we conducted a QC for each reconstructed volume and segmented MRI to ensure the accuracy of segmentation.

Finally, segmented brain structures were clustered together according to their proximity and visibility on MRI (Diogo et al. 2019; Vasung et al. 2019, 2021). Before 32 GW, the tissues were combined into the following clusters: cortical plate, limbic structures, subplate, intermediate zone, proliferative compartments, ganglionic eminence, and central structures (combined basal ganglia, amygdala, and thalamus). After 32 GW, the tissues were combined into the following clusters: cortical plate, limbic structures, fetal white matter, proliferative compartments, ganglionic eminence, and central structures (combined basal ganglia, amygdala, and thalamus).

The volume of each cluster was measured by summing all the voxel volumes of structures belonging to a specific cluster.

Statistical analyses

Demographic and clinical information were reported using the count, percentage, mean, and standard deviation.

To characterize the differences in the volume of the measured brain clusters between iFVM and controls, we employed a mixed-effects linear model with hemisphere as the unit of analysis. A model was constructed separately for each cluster and for the younger and the older group of fetuses. In the older group, due to the loss of continuity of the border between the intermediate zone and subplate zone, those 2 structures were combined into one cluster termed “fetal white matter.”

The dependent variable was the relative volume of the cluster, measured in percentage of the total hemispheric volume. Based on the literature review, we hypothesized that the relative volume of the cluster is explained by the following factors or covariates (M1): condition (iFVM or control), hemispheric abnormality [normal or abnormal (iFVM) hemisphere], gestational age (in GW), hemisphere (left or right), sex (male or female), and amount of total intracranial subarachnoidal volume (in cm³). A random effect was included in the model to account for correlation between the 2 hemispheres from a given subject.

The use of the parametric model was confirmed by examining and verifying a quasi-Gaussian distribution of residuals. We employed robust regression with bisquare weighting (Huber 1981) to detect and reduce the influence of outliers.

From parameters of the fitted model for each cluster, we constructed estimates for the pairwise differences in relative volume among the 3 classes of hemisphere (control, iFVM normal, and iFVM abnormal), with standard errors derived from the pooled variance.

To protect against inflation of type I error from multiple comparisons, we designated the 2 age groups as “families,” with the goal of keeping the familywise type I error rate at 5%. For each cluster within the family, we conducted a 2-df test of the null hypothesis that relative volume was equal for the 3 classes of the hemisphere. We adjusted the resulting P-values by the step-down procedure of Holm (Holm 1979) to limit the familywise type I error rate to 5%. When the 2-df test proved significant, the principle of closed testing (Bender and Lange 2001) allowed us to conduct all 3 pairwise tests for that cluster using a critical threshold of P < 0.05 without further inflating the type I error rate. We used the Benjamini–Hochberg (Benjamini and Hochberg 1995) procedure to estimate the false discovery rate for those differences declared significant.

Results

Subjects

Of 136 retrospectively identified iFVM cases, 30% (N = 41) had a technically satisfactory MRI and passed all steps of the QC. All of the 41 cases had either borderline or mild ventriculomegaly (i.e. atrial diameter 10–12 mm).

Based on gestational age determining the MRI visibility of the transient fetal compartments (Diogo et al. 2019; Vasung et al. 2021), the fetuses were assigned to younger (subplate and intermediate zone can be distinguished in continuity on MRI scans) or older groups (subplate and intermediate zone cannot be distinguished in continuity on MRI scans and are combined into the fetal white matter).

The younger group of fetuses (N = 54, age range = 18–32 GW) included 30 control subjects (age range = 18–31 GW; 8 females) and 24 fetuses with iFVM (age range = 18.71–32 GW; 12 females). Approximately half of the iFVM cases (58%) in the younger group had unilateral iFVM (Table 1). Twenty-seven subjects with iFVM had postnatal medical records that were available for this study. Four subjects with iFVM had abnormal postnatal clinical signs or symptoms (conductive hearing loss, proximal muscle weakness, absence seizures, or low axial tone) but did not have evidence for the overt neurodevelopmental delay in the available medical records during the follow-up period [from 1–7 years for subjects (N = 27) that continued their clinical follow-up at our institution]. There was no significant difference in age between controls and iFVM subjects in the younger group.

Table 1.

Demographic and clinical information of fetuses included in the study.

Group	Younger fetuses		Older fetuses
	Control, N = 30	iFVM, N = 24	Control, N = 18	iFVM, N = 17
Abnormal hemisphere
Healthy	60/60 (100%)	14/48 (29%)	36/36 (100%)	15/34 (44%)
Dilated ventricle	0/60 (0%)	34/48 (71%)	0/36 (0%)	19/34 (56%)
iFVM type
Bilateral	n/a	10/24 (42%)	n/a	2/17 (12%)
Unilateral	n/a	14/24 (58%)	n/a	15/17 (88%)
Age in GW (mean ± SD)	26.67 (3.89)	26.98 (3.57)	34.17 (1.54)	34.18 (2.49)
Sex
Female	8/30 (27%)	12/24 (50%)	9/18 (50%)	7/17 (41%)
Male	22/30 (73%)	12/24 (50%)	9/18 (50%)	10/17 (59%)
Abnormal postnatal clinical signs	0/30 (0%)	4/24 (17%)	0/18 (0%)	5/17 (30%)

SD, standard deviation; n/a, not applicable.

The older group of fetuses (N = 35, age range = 30–38 GW) was composed of 18 control subjects (age range = 32.14–36.86 GW; 9 females) and 17 fetuses with iFVM (age range = 30–38 GW; 7 females). The majority of iFVM cases (88%) in the older group had unilateral iFVM (Table 1). Five iFVM cases had an abnormal postnatal clinical signs or symptoms (hypotonia, hearing loss, mild neck flexion weakness, GM hemorrhage detected 10 days after birth, increased head circumference) but did not have evidence for overt neurodevelopmental delay in the available medical records during the follow-up. There was no significant difference in age between controls and iFVM subjects in the older group.

Younger

Between-group analysis

We tested if there were significant differences in the relative volume of clusters between iFVM and control subjects after adjusting for the age, hemispheric differences, sex, and absolute volume of the subarachnoid cerebrospinal fluid (CSF). Our analysis showed that, compared to the controls, fetuses with iFVM had significantly larger relative volumes of the (in the order of the magnitude; mean difference ± standard error): lateral ventricle (6.16% ± 0.53), cortical plate (2.96% ± 0.79), proliferative zones (1.57% ± 0.3), and ganglionic eminence (0.18% ± 0.07) belonging to the abnormal hemisphere (left hemisphere in Fig. 1A and C, Table 2).

Fig. 1.

An illustration summarizing significant differences (color-coded; scale bar in the middle) in the relative volume of the tissue clusters between controls (A, B) and iFVM cases (C, D) before (first column) and after (second column) 32 GW. Note that we compared with controls: (i) the volumes of the hemisphere with the dilated ventricle in iFVM (left hemisphere in C and D) and (ii) the healthy hemisphere in iFVM (right hemisphere in C and D).

Table 2.

The relative volume of segmented clusters in the younger group of fetuses (18–32 GW), comparing normal and abnormal hemispheres in iFVM subjects and corresponding hemispheres in control subjects.

Anatomical structure	Contrast	Difference, % of hemispheric volume	SE	P a
Limbic	iFVM Normal – Control	−0.18	0.06	0.005
	iFVM Abnormal – Control	−0.16	0.05	0.005
	iFVM Abnormal – iFVM Normal	0.02	0.04	0.61
Central	iFVM Normal – Control	−0.35	0.16	0.03
	iFVM Abnormal – Control	−0.44	0.15	0.005
	iFVM Abnormal – iFVM Normal	−0.09	0.06	0.15
Lateral ventricle	iFVM Normal – Control	0.89	0.57	0.13
	iFVM Abnormal – Control	6.16	0.53	<0.0001
	iFVM Abnormal – iFVM Normal	5.27	0.37	<0.0001
Cortical plate	iFVM Normal – Control	2.55	0.81	0.003
	iFVM Abnormal – Control	2.96	0.79	0.0004
	iFVM Abnormal – iFVM Normal	0.42	0.35	0.24
Intermediate zone	iFVM Normal – Control	−0.87	0.87	0.32
	iFVM Abnormal – Control	−2.04	0.85	0.02
	iFVM Abnormal – iFVM Normal	−1.16	0.26	<0.0001
Subplate zone	iFVM Normal – Control	−1.45	0.61	0.02
	iFVM Abnormal – Control	−2.25	0.56	0.0002
	iFVM Abnormal – iFVM Normal	−0.80	0.40	0.047
Proliferative zone	iFVM Normal – Control	0.08	0.31	0.80
	iFVM Abnormal – Control	1.57	0.30	<0.0001
	iFVM Abnormal – iFVM Normal	1.49	0.14	<0.0001
Ganglionic eminence	iFVM Normal – Control	0.10	0.07	0.19
	iFVM Abnormal – Control	0.18	0.07	0.015
	iFVM Abnormal – iFVM Normal	0.08	0.03	0.004
Cerebellum	iFVM Normal – Control	−0.03	0.15	0.83
	iFVM Abnormal – Control	−0.19	0.14	0.20
	iFVM Abnormal – iFVM Normal	−0.16	0.04	0.0004

^aPairwise differences were evaluated only after rejecting H₀: iFVM Abnormal = iFVM Normal = Control within each anatomical structure. ^bSignificant differences in bold type, with an estimated false discovery rate of 1.1%.

However, compared to the controls, iFVM subjects also had significantly smaller volumes (in the order of the magnitude) of: the subplate zone (−2.25% ± 0.56), intermediate zone (−2.04% ± 0.85), central structures (−0.44% ± 0.15), and limbic structures (−0.16% ± 0.05) belonging to the abnormal hemisphere (Table 2, left hemisphere in Fig. 1A and C). We did not find significant differences between control subjects and iFVM in the relative volume of the cerebellum.

Besides the significant differences between controls and iFVM in the relative volume of the clusters belonging to the abnormal hemisphere, we also found significant differences between iFVM and control subjects in the relative volume of the clusters belonging to the healthy iFVM hemisphere. More specifically, the healthy hemisphere of the iFVM subjects compared to the controls showed significantly larger volumes of the cortical plate (2.55% ± 0.81) and smaller volumes of the limbic structures (−0.18% ± 0.06), central structures (−0.35% ± 0.16), and the subplate zone (−1.45% ± 0.61) (Table 2, right hemisphere in Fig. 1A and C).

Within-group analysis

We also tested if there were significant differences in the relative volume of the clusters between abnormal and normal hemispheres in iFVM subjects after adjusting for the age, hemispheric differences, sex, and absolute volume of the subarachnoid CSF. Our results showed that in fetuses with iFVM, the abnormal hemisphere, compared to the normal hemisphere, had similar relative cortical plate volumes but significantly larger relative volumes of the lateral ventricles (5.27% ± 0.37), larger volumes of the proliferative zones (1.49 ± 0.14), and ganglionic eminence (0.08% ± 0.03). In contrast, the abnormal hemisphere had significantly smaller volumes of the intermediate zone (−1.16 ± 0.26), subplate zone (−0.8 ± 0.4), and cerebellum (−0.16 ± 0.04) (Table 2).

Older

Between-group analysis

As in the younger group, we tested if there were significant differences in the relative volume of the clusters between iFVM and control subjects in the older group after adjusting for the age, hemispheric differences, sex, and absolute volume of the subarachnoid CSF.

Our analyses showed that, compared to the controls, fetuses with iFVM had significantly larger relative volumes (in the order of the magnitude; mean difference ± standard error) of: the lateral ventricle (3.22% ± 0.35), cortical plate (2.95% ± 1.02), proliferative zones (0.71% ± 0.15), and ganglionic eminence (0.21% ± 0.03) belonging to the abnormal hemisphere (left hemisphere in Fig. 1B and D, Table 3). Furthermore, compared to the controls, iFVM subjects also had significantly smaller volumes (in the order of the magnitude) of: the fetal white matter (−3.17% ± 1.05) and central structures (−0.63% ± 0.13) belonging to the abnormal hemisphere (Table 3, left hemisphere in Fig. 1B and D). We did not find significant differences between control subjects and iFVM in the relative volume of the cerebellum.

Table 3.

The relative volume of segmented clusters in the older group of fetuses (30–38 GW), comparing normal and abnormal hemispheres in iFVM subjects and corresponding hemispheres in control subjects.

Anatomical structure	Contrast	Difference, % of hemispheric volume	SE	Pa
Limbic	iFVM Normal − Control	−0.07	0.05	—b
	iFVM Abnormal − Control	−0.09	0.05
	iFVM Abnormal − iFVM Normal	−0.02	0.03
Central	iFVM Normal − Control	−0.46	0.13	0.001
	iFVM Abnormal − Control	−0.63	0.13	<0.0001
	iFVM Abnormal − iFVM Normal	−0.18	0.05	0.0009
Lateral ventricle	iFVM Normal − Control	0.44	0.35	0.21
	iFVM Abnormal − Control	3.22	0.35	<0.0001
	iFVM Abnormal − iFVM Normal	2.78	0.24	<0.0001
Cortical plate	iFVM Normal − Control	1.92	1.02	0.07
	iFVM Abnormal − Control	2.95	1.02	0.007
	iFVM Abnormal − iFVM Normal	1.03	0.51	0.055
Fetal white matter	iFVM Normal − Control	−1.76	1.05	0.10
	iFVM Abnormal − Control	−3.17	1.05	0.005
	iFVM Abnormal − iFVM Normal	−1.41	0.53	0.012
Proliferative zone	iFVM Normal − Control	0.22	0.15	0.15
	iFVM Abnormal − Control	0.71	0.15	<0.0001
	iFVM Abnormal − iFVM Normal	0.49	0.08	<0.0001
Ganglionic eminence	iFVM Normal − Control	0.13	0.03	0.0003
	iFVM Abnormal − Control	0.21	0.03	<0.0001
	iFVM Abnormal − iFVM Normal	0.08	0.02	<0.0001
Cerebellum	iFVM Normal − Control	−0.49	0.23	—b
	iFVM Abnormal − Control	−0.39	0.23
	iFVM Abnormal − iFVM Normal	0.10	0.06

^aPairwise differences were evaluated only after rejecting H₀: iFVM Abnormal=iFVM Normal=Control within each anatomical structure.

^bH₀: iFVM Abnormal=iFVM Normal=Control not rejected, pairwise differences not tested. Significant differences in bold type, with an estimated false discovery rate of 1.8%.

Furthermore, similar to the younger group, in the older group, we also found significant differences between controls and iFVM in the relative volume of the structures belonging to the normal hemisphere. Compared to the controls, the normal hemisphere of the iFVM subjects showed significantly larger volumes of the ganglionic eminence (0.13% ± 0.03) and smaller volumes of the central structures (−0.46% ± 0.13), (Table 3, right hemisphere in Fig. 1B and D). Similar to the younger group, the cortical plate was not significantly different.

Within-group analysis

Compared to the normal hemisphere in iFVM, the abnormal hemisphere in iFVM fetuses has significantly larger relative volumes of the lateral ventricles (2.78% ± 0.24), proliferative zones (0.49% ± 0.08), and ganglionic eminence (0.08% ± 0.02) after adjusting for the age, hemispheric differences, sex, differences between controls and iFVM subjects, and absolute volume of the subarachnoid CSF. In addition, compared to the normal hemisphere, the abnormal hemisphere in iFVM fetuses has significantly smaller volumes of the fetal white matter (−1.41% ± 0.53) and central structures (−0.18% ± 0.05) (Table 3). The cortical plate was not significantly different.

Discussion

To our knowledge, this is the first study to identify structural differences of majority of transient fetal brain parenchymal compartments between iFVM cases and controls. Our results indicate abnormal development of transient compartments belonging to both hemispheres (i.e. with and without dilated ventricles) in fetuses with iFVM. We observed enlargement of proliferative zones and overgrowth of the cortical plate in iFVM with associated reduction of volumes of central structures, subplate, and fetal white matter.

Compared to controls, the hemisphere with the dilated ventricle in iFVM fetuses had a significantly larger volume of the proliferative zones, cortical plate, and ganglionic eminence, regardless of the age group. These results indicate an enlargement of proliferative pools (proliferative zones and ganglionic eminence), which might, in turn, result in an increase in the production of neuronal progenitors and overgrowth of the cortical plate. Our results are in agreement with previous studies showing cortical overgrowth in ventriculomegaly (Gilmore et al. 2008; Lyall et al. 2012; Kyriakopoulou et al. 2014; Lockwood Estrin et al. 2016), but here we add evidence of proliferative pool enlargement.

Neuronal precursors are born in proliferative zones (ventricular and subventricular zones and ganglionic eminence (Petanjek et al. 2009), and the majority migrate to the cortical plate, reaching their final destination around 20 GW (Bystron et al. 2008). These neurogenic processes (cellular proliferation, migration, and differentiation) are regulated by extracellular molecules (Clowry et al. 2010; Bakken et al. 2016; Kostović 2020), whose concentration is tightly linked to the osmolarity of interstitial fluid.

The effect of enlarged ventricles on the neuronal proliferative pool and cortical growth needs to be also viewed through the prism of hypotheses addressing fetal CSF physiology. According to a classical hypothesis (Dandy 1929), the choroid plexuses (ChPs) of ventricles produce CSF in the postnatal brain. From ventricles, CSF circulates unidirectionally to the subarachnoid space (Dandy 1929; Spector et al. 2015). In the subarachnoid space, it is passively absorbed into the intracranial venous system. However, animal models have challenged this classical hypothesis (Bulat and Klarica 2011), showing that capillaries in both the intraparenchymal and ChPs are simultaneously the primary generators and absorbers of CSF (constituted 99% of water (Bulat and Klarica 2011)) and interstitial fluid. Hydrostatic and osmotic pressures within intraparenchymal capillaries determine the net pressure that will drive the filtration or reabsorption of CSF and interstitial fluid (Bulat and Klarica 2011). Also, in the embryonic and fetal brain, both ChPs and capillaries are thought to contribute significantly to CSF production. Before 8 GW, when ChPs are formed, the CSF is believed to be trapped amniotic fluid (Johansson et al. 2008). From 8 to 10 GW, the ChPs increase in volume (occupying almost 75% volume of the ventricles; Boassa and Yool 2005) and start to produce CSF and CSF-borne signaling factors required for normal brain development (Lun et al. 2015). During this period, the first perforating telencephalic vessels can be observed (Kuban and Gilles 1985; Marín-Padilla 2012), and both classical and alternative hypotheses most likely explain CSF production and elimination. Finally, from 10 GW, the growth of ChPs, relative to the brain and ventricles, slows down. By 20 GW, it is assumed that ChPs have reached an adult-like appearance (Boassa and Yool 2005). Around 20 GW, most neurons reach their final destination in the cortex (Bystron et al. 2008) and contribute to the exponential increase in the brain volume, amount of pial capillary plexus, and the number of perforating medullary vessels (Marín-Padilla 2012). Thus, after 20 GW, as the relative size of ChPs in comparative relation to the brain decreases dramatically (to 0.5% at birth), the perforating capillary vessels, whose surface is 5,000 times larger compared to the surface of the ChP, likely become instrumental in CSF production and absorption as proposed by this newly developed model of CSF physiology (Bulat and Klarica 2011).

Animal models in chick embryos show that ventricular pressures gradually increase during normal embryonic development, which is crucial for an increase in mitotic activity of neuroepithelial cells (Desmond and Jacobson 1977). Thus, early during embryonic development and before 32 GW, the increase in intraventricular pressures seen in iFVM may lead to stimulation of cell proliferation and subsequent cortical overgrowth as suggested by Kyriakopoulou et al. 2014, a phenomenon that has also been observed in chick embryos (Gato and Desmond 2009). Our findings of larger volumes of proliferative compartments and ganglionic eminence belonging to the hemisphere with enlarged ventricles in iFVM, compared to the hemisphere with normal size ventricles, corroborate these findings and are in agreement with cortical overgrowth interpretations suggested by Kyriakopoulou et al. 2014.

After 32 GW, that is after neuronal proliferation subsides and after the majority of neurons reach their final cortical destination, the enlargement of proliferative compartments and cortical plate could be explained by the density of capillary networks within these compartments. (Gato et al. 1993) showed that the accumulation of osmotic molecules regulates the CSF pressure in chick embryos. Furthermore, the new CSF hypothesis argues that an increase in osmolarity of interstitial fluid and an increase in the contact area between interstitial fluid and capillaries (seen in tissues with the dense capillary network) result in an increase in CSF filtration and an increase in CSF volume (Maraković et al. 2010). Thus, an increase in tissue osmolarity in compartments with high capillary density might be a result of a higher concentration of osmotic and extracellular matrix molecules leading to an increase in the volume of these structures and a compensatory decrease in volume of structures with less dense capillary networks. However, little is known about intraparenchymal vessel architecture remodeling during prenatal human fetal development, and further studies are needed to elucidate the roles of parenchymal vessels in the creation of CSF.

Finally, we also provide evidence that in addition to abnormalities in the hemisphere with dilated ventricles in iFVM, the development of the “healthy” hemisphere (the hemisphere with normal-sized ventricles in iFVM) is also altered. More specifically, we also found structural differences between hemispheres with normal-sized ventricles in iFVM and controls (larger volumes of cortical plate and smaller volumes of limbic structures, central structures, and subplate before 32 GW; a larger volume of ganglionic eminence and smaller volume of central structures after 32 GW). In the absence of experimental models, the pathophysiological mechanism and the cause for these alterations remain unknown. However, studies like Gilmore et al. (2008), which suggest that ventricular size is tightly linked to changes in CSF flow, and from (Sawamoto et al. 2006), which provide evidence that in the adult brain the newly created neurons follow the cerebrospinal flow during their migration, suggest that an increase in CSF volume observed in iFVM most likely affects global brain development.

Despite several aforementioned attractive theories about the pathophysiology of mild ventriculomegaly, we have to highlight the fact that our work addresses statistical dependence on which our interpretation relies. However, one should mention that mild ventriculomegaly is an anatomical finding that might have several causes, which remain to be determined. Recent work by Jin et al. (2020) suggests that genetic disruption of neurogliogenesis underlies the early appearance of ventriculomegaly that evolves to sporadic congenital hydrocephalus. These results strongly suggest that prenatal ventriculomegaly might be a neuroradiological vestige of abnormal brain development (Duy et al. 2022). Similarly, approximately 5% of fetuses with apparently mild-to-moderate iFVM have an abnormal karyotype or abnormal findings on the chromosomal microarray (Donnelly et al. 2014). Furthermore, several reports suggest that up to 5% of cases of iFVM result from congenital infections (Jamieson et al. 2006; Devaseelan et al. 2010), which include CMV, toxoplasmosis, Zika virus, sporadic mumps, enterovirus, parainfluenza virus type 3, parvovirus B19, and lymphocytic choriomeningitis virus. Finally, iFVM can also result from early fetal damage or loss of brain tissue that was not captured or documented during the routine clinical ultrasound. Thus, more systematic prospective neuroimaging, genetic, and histological studies of truly iFVM, as well as animal models, are needed to elucidate the pathophysiology and identify underlying causes of this frequent imaging finding and help in predicting the postnatal course which might lead to clinically overt disease.

Conclusion

Our results offer evidence of cortical overgrowth and enlargement of proliferative compartments in the brain of fetuses with iFVM, which are associated with a decrease of volumes belonging to central structures, subplate zone, intermediate zone, and fetal white matter. These results indicate that iFVM might be tightly linked to the development of transient fetal zones. In conclusion, we suggest that global brain development should be taken into consideration when examining patients with iFVM.

Limitations

There are several limitations of our study. Due to the retrospective nature of the study, we were not able to collect all relevant postnatal clinical data (i.e. development of hydrocephalus, adverse long-term neurodevelopmental outcomes). Furthermore, MRIs of 30% of iFVM cases (all with mild or borderline ventriculomegaly) were of excellent quality for structural analysis. Given that these were retrospectively analyzed MRIs that were acquired for clinical purposes, we cannot exclude potential bias in the selection of iFVM cases. Due to this difficulty in distinction of the zones on in utero MRI, the proliferative zones in this manuscript and our previous work (Vasung et al. 2016, Vasung et al. 2019, Rollins et al. 2020, Vasung et al. 2020) included the ventricular and inner subventricular zones while the outer subventricular zone (less cell dense) was most likely captured within the intermediate zone label. In addition, the differences in the relative hemispheric percentage of certain structures were small with a very small standard error of the mean (e.g. ganglionic eminence). Given that these differences reached statistical significance, these results suggest relatively high reliability of findings (Tables 2 and 3) but with a small size effect. Therefore, they should be interpreted with caution. Finally, due to the retrospective nature of the study, only a very limited number of postnatal MRI scans were available for evaluation (N = 2) and, therefore, we cannot exclude the possibility that some of the subjects had additional brain abnormalities (e.g. polymicrogyria) and do not represent a truly iFVM.

Funding

Ralph Schlaeger Foundation in Neuroradiology; National Institute of Neurological Disorders and Stroke of the National Institutes of Health (award number K23NS101120); American Academy of Neurology Clinical Research Training Fellowship; National Alliance for Research on Schizophrenia & Depression Young Investigator Award from the Brain and Behavior Foundation; National Institutes of Health (grants R01EB018988, R01NS106030, and R01EB013248); Technological Innovations in Neuroscience Award from the McKnight Foundation. Conflict of interest statement. None declared.