591 - Using in vivo MRI imaging as a Biomarker for the Expanded Neonatal Subventricular Zone

Publication Number: 3572.591

Aunoy Poddar, UCSF, San Francisco, CA, United States; Melissa Coloma, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Quanhao Sun, Keck School of Medicine of the University of Southern California, Los Angeles, CA, United States; Hannah Glass, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Yvonne Wu, University of California, San Francisco, School of Medicine, San Francisco, CA, United States; Mercedes Paredes, University of California, San Francisco, School of Medicine, SF, CA, United States; Yi Li, Department of Radiology, University of California, San Francisco, San Francisco, CA, United States

Background: Early life brain injury can have lifelong neurological consequences. We recently characterized the perinatal developmental brain structure, the Arc -an expansion of the neonatal subventricular zone that harbors migrating neurons and persists into the first year of infancy. Late neuronal migration may support cognition and plasticity but is vulnerable to perinatal injury.

Objective: Here, we describe the radiological features used to identify the Arc structure from in vivo neonatal brain imaging and compare Arc volumes between normotypic and injured infants.

Design/Methods: 10 high-resolution neonatal T2w sequences (isotropic 0.8 mm2 voxels) from the Developing Human Connectome Project (dHCP) were selected to represent the normotypic subset. 67 T2w sequences (1 mm2 - 2 mm2 voxels) from the HEAL study were used to construct the encephalopathy cohort (determined by Sarnat Criteria) (2). Criteria for Arc segmentation included (Fig 1): 1) T2 hypointensity on at least the axial plane in the region of histologically-defined Arc 2) T2 hypointensity was bounded posteriorly in the sagittal plane based by on the foramen of Monro 3) medial extension of T2 hypointensity bounded by the corpus callosum. Three readers manually segmented 17 HEAL MRIs, which were used to train an algorithm (nnU-Net) to segment arcs on 50 additional MRIs. To assess if the Arc volume could be compared across patients, we compared segmented volumes from the dHCP cohort to those in the HEAL cohort with moderate encephalopathy.

Results: We found that overall volumes did not significantly differ despite differences in voxel size (median dHCP vol 417 mm3 vs. median HEAL vol in moderate encephalopathy 308 mm3, p =0.051). To test if Arc volume could be a biomarker for brain injury, we compared the segmented Arc volumes from infants with severe encephalopathy to the infants with moderate encephalopathy and found that there was a significant decrease in volume (median vol in moderate encephalopathy 417 mm^3 vs. median vol in severe encephalopathy 197 mm^3, p= 0.01). We corroborated these results by comparing histology of normotypic infants to infants diagnosed with HIE (Fig 2), observing a decrease in migratory neurons, thus suggesting a relationship between the T2 MRI and the underlying cytoarchitecture.

Conclusion(s): Our findings suggest that a developmental brain structure known as the Arc has an imaging correlate on T2w MRI and that the volume of this structure is associated with severity of clinically-diagnosed brain injury. We envision that this structure may be a biomarker for neurodevelopmental disability in the setting of perinatal injury.

Figure 1. Arc segmentation of dHCP Images.
A) Unsegmented and segmented axial T2-weighted images at the level of the left frontal horn demonstrates curvilinear T2 hypointensity corresponding to the Arc (blue segmentation overlay). In the axial plane and coronal, the arc creates an "eyebrow" shape around the lateral ventricle. For consistency, the medial segmentation of the arc did not extend into the corpus callosum. B) Unsegmented and segmented sagittal T2-weighted images demonstrate how the Arc parallels the frontal horn. The yellow dashed line indicates vertical rostro-caudal boundary, anatomically defined by the foramen of Monroe. For consistency, the posterior segmentation of the arc did not extend beyond this line.

Figure 2. Histological-MRI Correlates of Arc region.
A) Post-mortem Nissl stain of a sagittal section of a GW39 fetus (3) compared to in vivo T2w MRI from neonatal dHCP scan. Red and Purple outlines outline cytoarchitectural structure of Arc region (from Paredes et al. 2016). B) Post-mortem immunohistochemistry of the Arc in a 3 day old neurotypical infant versus 3 day