Quantitative fetal MRI with diffusion tensor imaging in cases with ‘short’ corpus callosum

It has been suggested previously that the presence of Probst bundles (PB) in cases with a short corpus callosum (SCC) on diffusion tensor imaging (DTI) may help to differentiate between corpus callosal (CC) dysplasia and a variant of normal CC development. The objectives of this study were to compare DTI parameters between cases of SCC vs normal CC and between cases of SCC with PB (SCC‐PB+) vs SCC without PB (SCC‐PB–).

apparently isolated SCC detected by sonography between November 2016 and December 2022 (IRB: 00011928).MRI was performed using a 1.5-Tesla Signa system.T2-weighted axial and sagittal sequences of the fetal brain were used to measure the length and thickness of the CC.16-direction DTI axial brain sequences were performed to identify the presence of PB and to generate quantitative imaging parameters (fractional anisotropy (FA) and apparent diffusion coefficient (ADC)) of the entire CC, genu, body and splenium.Cases in which other associated brain abnormalities were detected on MRI were excluded.Cases were matched for fetal gender and gestational age with controls in a 1:3 ratio.Control cases were normal fetuses included in the LUMIERE on the FETUS trial (NCT04142606) that underwent the same DTI evaluation of the brain.Comparisons between SCC and normal CC cases, and between SCC-PB+ and SCC-PB-cases were performed using ANOVA and adjusted for potential confounders using ANCOVA.
In SCC-PB+ cases, all FA values were significantly lower, and ADC values in the CC body were significantly higher compared with normal CC cases (all P < 0.05).In SCC-PB-cases, there was no significant difference in FA and ADC compared with normal CC cases (all P > 0.05).
Conclusions Fetal DTI evaluation of the CC showed that FA values were significantly lower and ADC values tended to be significantly higher in SCC-PB+ compared with normal CC cases.This may highlight alterations of the white matter microstructure in SCC-PB+.In contrast, isolated SCC-PB-did not demonstrate significant changes in DTI parameters, strengthening the possibility that this is a normal CC variant.© 2023 International Society of Ultrasound in Obstetrics and Gynecology.

INTRODUCTION
The corpus callosum (CC) is the main cerebral commissure, which enables the integration of motor, sensory and cognitive activities between the two hemispheres of the brain 1 .CC anomalies are one of the most common cerebral malformations, with a reported incidence ranging between 1.8 and 7 per 10 000 births, and may be seen in up to 2-3% of pediatric neurology patients 2 .CC anomalies may occur as an isolated anomaly or as a component of a complex disorder.Consensus appears to exist regarding the constituents of a diagnosis of complete agenesis of the CC.However, the definitions of other CC abnormalities, including dysplasia and hypoplasia with or without dysplasia, remain unclear, making the diagnosis and prognosis of CC dysgenesis difficult 3 .
A short CC (SCC) is poorly defined within the existing literature and may reflect generalized hypoplasia with intact morphology 3 .In the absence of a consensus definition, the diagnosis of SCC remains challenging 4 .Tepper et al. 5 have proposed employing a high occipitofrontal dimension/CC length ratio to identify SCC.Other studies defined SCC based on the anteroposterior diameter ≤ 1 st , < 5 th or < 10 th percentile, using various reference charts created for ultrasound or magnetic resonance imaging (MRI), which have been shown to have significant discrepancies, potentially confounding accurate diagnosis [6][7][8][9] .Despite the discrepancies in the definition, SCC represents the most frequent CC anomaly 10 .The finding of SCC is a source of anxiety for parents and complicates counseling for practitioners managing these cases, as data regarding this condition, including neurodevelopmental outcome, are scarce, and its significance is not well established 7,11,12 .
Fetal diffusion tensor imaging (DTI) offers a unique tool that enables the evaluation of white matter microstructure by probing the characteristics of water diffusion in tissues 13 .Millischer et al. 10 have shown that, in cases of isolated SCC, the presence of Probst bundles (PB), visualized by DTI, was more likely to be associated with CC dysplasia than with a variation of normal CC development.However, little is known about the integrity of the white matter microstructure in cases of SCC.
We hypothesized that quantitative DTI evaluation of the white matter microstructure of CC fibers may help to better characterize SCC.This information may be beneficial in parental counseling when SCC is suspected during pregnancy.Therefore, this study aimed to compare DTI parameters between cases of SCC vs normal CC, and between cases of SCC with PB (SCC-PB+) vs SCC without PB (SCC-PB-).

Study design and population
This was a retrospective case-control study of data collected between November 2016 and December 2022 at the department of Fetal Medicine and the LUMIERE Platform in Necker Hospital, Paris, France.
Inclusion criteria were fetuses referred for SCC with a CC length (anteroposterior diameter) below the 5 th percentile in the midsagittal plane on ultrasound, according to Achiron & Achiron 14 , with no other anomaly of CC morphology detected, that also underwent DTI.Cases with complete or partial agenesis of the CC, associated brain anomaly, genetic anomaly or incomplete data were excluded.All subjects provided written informed consent approved by the institutional review board (IRB: 00011928; CERAPH 2020-12-06).
For the control group with a normal CC, the inclusion criteria were women with a pregnancy between 26 and 34 weeks' gestation, with no known fetal anomaly or condition, who had consented to participate in our research program on fetal MRI (LUMIERE on the FETUS trial (NCT04142606)).Controls were matched to SCC fetuses based on fetal gender and gestational age (GA) at the time of MRI in a 3:1 ratio.All subjects provided written informed consent approved by the institutional review board (2019-A01498-49).

MRI protocol
Women underwent MRI without sedation or breathholding.Subjects were scanned using a 1.5-Tesla 450 W MRI scanner (GE Healthcare, Chicago, IL, USA) with a 48-channel body array coil.All subjects underwent a full scan protocol, consisting of initial axial and sagittal T2-weighted gradient-echo scout acquisitions to identify true axial and sagittal sections of the fetal brain.Axial diffusion-weighted imaging of the brain was then performed.Imaging parameters of the diffusion-weighted acquisition for each slice were as follows: 15 non-collinear diffusion-weighted pulsed magnetic field gradients with a b-value of 700 s/mm 2 and one b0 image without diffusion-weighting, field of view (FOV) of 36 mm, phase FOV of 1.00, slice thickness of 3 mm, voxel size of 4.5 x 2.8 x 3 mm, 7-10 slices, repetition time of 2800 ms and minimum echo time and acquisition time of 2.3 min.There was no change in protocol acquisition parameters throughout the duration of the study.Movements of the fetus, including movement of the head, body or extremities, were quantified subjectively during acquisition.Acquisition was repeated in cases of fetal movements (three attempts maximum).

Anatomical measurements
Anatomical measurements of the CC were performed on T2-weighted MRI (Figure 1).The anteroposterior diameter of the CC was measured in the midsagittal plane from the most anterior aspect of the genu to the most posterior aspect of the splenium.Curvilinear length of the CC and CC thickness at the level of the genu, body and splenium were also measured in the midsagittal plane.Fronto-occipital and biparietal cerebral diameters were measured in the axial plane.

DTI processing
Postprocessing of DTI data was performed in MRtrix3 (www.mrtrix.org) 15and NiftyMIC (https://github.com/gift-surg/NiftyMIC) 16 .First, raw DICOM data were converted to NIFTI format and to MRtrix format.Preprocessing steps included denoising 17 , Gibbs-ringing artifact correction 15 , motion and eddy correction 18 and bias removal.Images were then corrected for interslice motion and reconstructed using NiftyMIC 16 .A constrained spherical harmonics convolution model was applied to estimate diffusion tensors (Figure 2) 15 .Regions of interest were drawn manually in the midsagittal plane of the CC, encompassing the entire CC, genu, body and splenium, to obtain DTI parameters, including fractional anisotropy (FA) and apparent diffusion coefficient (ADC), using MRtrix3 15 .
The presence of PB was assessed by two independent reviewers (R.C. and D.G.).PB were defined as present when fibers positioned in a rostrocaudal direction, creating the roof of the lateral ventricles, were clearly identified in the color-coded FA maps 10,19,20 .
Quantitative variables with a non-normal distribution are expressed as median (range) and were compared using ANOVA.Multivariate comparison of DTI parameters was performed and adjusted for potential confounders using ANCOVA.Z-scores of DTI parameters in cases of SCC-PB+ and SCC-PB-compared with normal CC were calculated.P < 0.05 was considered to indicate statistical significance.

RESULTS
A total of 22 fetuses with isolated SCC and a median GA of 31.6 (range, 26.5-34.0)weeks were matched to 66 control fetuses with a normal CC.PB were present in 10 (45.5%) and absent in 12 (54.5%)SCC cases.In the control group, there was no relationship between DTI parameters and GA (Figures S1 and S2).Baseline characteristics, anatomical measurements and DTI parameters of control fetuses and SCC fetuses, overall and according to the presence of PB, are presented in Table 1.

Short vs normal corpus callosum
As expected, all measurements of CC length and thickness were significantly reduced in SCC compared with normal CC cases.When evaluating the entire SCC cohort, including SCC-PB+ and SCC-PB-cases, FA of the entire CC, genu, body and splenium was significantly lower in SCC compared with normal CC cases after adjusting for the anteroposterior diameter of the CC (Figure S3).There was no significant difference in ADC values between the two groups (Figure S4).
There was no significant difference in the CC anteroposterior diameter, curvilinear length, body thickness or splenium thickness between SCC-PB+ and SCC-PBcases.However, genu thickness was significantly reduced in SCC-PB+ compared with SCC-PB-cases (P = 0.02).The FA of the entire CC, genu, body and splenium was significantly lower in SCC-PB+ compared with SCC-PBcases after adjusting for genu thickness (Figure S3).The ADC of the entire CC, genu and body was significantly higher in SCC-PB+ compared with SCC-PB-cases after adjusting for genu thickness (Figure S4).
All measurements of CC length and thickness were significantly reduced in SCC-PB+ compared with normal CC cases.The FA of the entire CC, genu, body and splenium was significantly lower in SCC-PB+ compared with normal CC cases after adjusting for the anteroposterior diameter of the CC (Figure S3).The ADC of the CC body was significantly higher in SCC-PB+ compared with normal CC cases after adjusting for the anteroposterior diameter of the CC (Figure S4).Compared with normal CC cases, Z-scores of FA ranged between -3.8 and -2.14, and those of ADC ranged between 0.79 and 1.44 in SCC-PB+ cases (Table S1).
All measurements of CC length and thickness were significantly reduced in SCC-PB-compared with normal CC cases.There was no significant difference in FA and ADC parameters between SCC-PB-and normal CC cases after adjusting for the anteroposterior diameter of the CC (Figures S3 and S4).Compared with normal CC cases, Z-scores of FA ranged between -1.15 and -1.02, and those of ADC ranged between -0.32 and -0.20 in SCC-PB-cases (Table S1).

DISCUSSION
To our knowledge, this is the first study to compare quantitative fetal DTI findings between SCC and normal CC cases.The use of DTI allows differentiation between two subgroups of SCC, which cannot be distinguished on the basis of CC biometry alone.Indeed, we found that FA is significantly lower and ADC tends to be significantly higher in SCC-PB+ compared with SCC-PB-and normal CC cases.The observed differences in DTI parameters may reflect an alteration of the white matter microstructure in SCC-PB+ cases, which may be related to an underlying developmental dysplasia of the CC.
The findings of this study are important for the management of patients referred for suspected SCC on ultrasound.Parental counseling is challenging in such cases; thus, reliable information permitting better characterization of the anomaly may improve counseling.
In this study, SCC was defined as a morphologically intact CC but with a reduced anteroposterior diameter, as this is one of the main indications for further brain evaluation.The definition of SCC is controversial, as there are discrepancies between the different reference range charts, and SCC may represent a broad spectrum of different anomalies 4,9 .In the literature, the term 'short  corpus callosum' has been used to refer to a reduction in the anteroposterior diameter [6][7][8]11,12 , an increase in the CC occipitofrontal dimension/length ratio 10 , a reduction in thickness 12 , a morphologically incomplete CC 11 or a combination of these features. Intrestingly, CC measurements are not reported in the postnatal description of CC abnormalities 3 .In studies 21,22 reporting the neurodevelopmental outcome of SCC cases, adverse outcome was more likely to be related to the associated brain malformations or genetic syndrome rather than the anatomical description of the CC abnormality.It would be particularly useful to be able to redefine CC anomalies based on a functional description rather than an anatomical one.
In a study of 33 isolated SCC cases undergoing DTI tractography evaluation, Millischer et al. 10 reported that visualization of PB may help differentiate between SCC due to CC dysplasia and SCC as a variant of normal development of the CC.However, they did not report quantitative DTI parameters of SCC cases or compare SCC with normal CC controls.Similar to other studies 23,24 , they reported that genetic anomalies were more frequent in SCC-PB+ compared with SCC-PBcases.However, they found that the presence of PB was not associated with a poor prognosis (although outcome data were available for only three infants).In another DTI study, of 20 fetuses with a CC abnormality (four with partial agenesis and 16 with complete agenesis) by Kasprian et al. 25 , PB was detected in 75% of cases.In adults with complete agenesis of the CC, PB can be considered as functional tissue, with electrophysiological properties similar to those of neural tissue of the intact CC 26 , and are thought to contribute to the cognitive process.The absence of PB in cases of complete agenesis would, therefore, suggest a lack of these alternative pathways and function.However, little is known about the implications of PB in the neurodevelopment of fetuses with SCC, and whether they have the same significance as in complete agenesis of the CC.
The pathophysiology of a SCC anomaly is currently unclear.At 18 weeks, all components of the CC have developed, and CC growth is then linear throughout gestation 14,27,28 .Disruption of an essential early event may, therefore, lead to hypoplasia.SCC with dysplasia may be attributed to multiple mechanisms, including midline axonal misguidance, reduction in the number of cortical neurons and proportional reduction in the number of axons crossing to the opposite hemisphere, reduced ability of small neurons to produce long-range interhemispheric axons, modification of the number of CC axons by noxious factors during the dynamic process of prenatal increase and postnatal decrease in the number of axons, or a combination of these mechanisms 29 .The reduction in the number of CC axons may provide one explanation as to why DTI parameters are modified in SCC-PB+ cases.However, we do not have detailed pathology data to suggest a cause for the difference in DTI parameters between the SCC-PB+ and PB-groups.It is possible that SCC-PB-is a normal variant of CC development, with anatomical measurements outside the normal range, in contrast to SCC-PB+, which seems to have a background of axonal pathfinding.
Limitations of this study include its retrospective design, the small number of cases and the lack of data on the neurodevelopmental outcome of these patients.The small size of the SCC-PB-group may explain the lack of a significant difference between the SCC-PB-and normal CC groups.Additionally, the definition of SCC remains controversial, and the findings of our study may not translate to cases with a thin or morphologically incomplete CC.However, all cases included in this study fulfilled the anatomical landmark criteria described by Raybaud 30 to define a completely developed CC (one-third anterior to the mammillary body and the splenium reaching to the level of the quadrigeminal plate).
The rationale behind our study was to investigate whether fetal DTI evaluation of the CC could help determine if abnormal CC biometry (short length) is attributable to a variant of normal development (or even biological variability, as observed for fetal growth) or CC dysplasia.We hypothesize that, when SCC is a variant of normal development, fulfilling the functional needs of the developing central nervous system, and interhemispheric connections are functionally complete, PB do not develop.However, in cases in which SCC is dysplastic and has altered quantitative DTI parameters, such as those observed in our cohort, the interhemispheric connections are not complete and functional needs are not met, which leads to the development of PB, as in cases of complete agenesis discussed above.Based on our findings, we believe that fetal DTI evaluation of the CC may help to redefine and characterize CC abnormalities.

Conclusions
Fetal DTI evaluation of SCC may highlight alterations of the white matter microstructure in cases of SCC-PB+.SCC-PB-cases did not present with DTI parameters that differed from those of normal CC cases.The findings of this study show that fetal DTI may help to redefine CC abnormalities and improve parental counseling.Long-term neurodevelopmental outcome should be investigated in future studies to evaluate whether DTI could aid in the prediction of outcome in cases of isolated SCC.

Figure 2
Figure 2 Fractional anisotropy color map of normal corpus callosum in sagittal plane.

Table 1
Characteristics of fetuses with normal corpus callosum (CC) and those with short CC (SCC), overall and according to whether Probst bundles were present (PB+) or absent (PB-)