Introduction

Resting-state functional MRI (rs-fMRI) reveals spontaneous fluctuations of the BOLD signal, which depends upon cerebral blood flow, volume, and oxygenation¹. During early brain development, it has been shown that preterm infants’ brain expresses regional and age-specific rs-fMRI activity patterns². Given that the intrinsic fluctuations measured by rs-fMRI are a consequence of oxygenated blood supplying active neurons, the BOLD signal variability (BOLD SD) may serve as a physiological signal related to an individual’s cerebrovascular status³ and neural flexibility/excitability⁴. We hypothesized that rs-BOLD variability during early development may reflect neuronal and cerebrovascular maturity (optimal brain function⁵). To this end, we aimed to investigate, for the first time, changes in BOLD variability across two time-points in the developing preterm infant brain.

Methods

Cohort: Twenty-two (n=22) preterm infants were recruited at Geneva University Hospital neonatal units. All infants underwent magnetic resonance imaging at 33 weeks (mean±SD: 29.23±2.31, range: 24.57-32.57 and at term-equivalent-age (TEA; 40 post-conception weeks; PCWs) Thus, the neuroimaging dataset consists of longitudinal imaging scans.
MRI acquisition: An EPI sequence (TR=700ms, Siemens 3T-MAGNETOM) was used resulting in 590 volumes/subject/time-point. No sedation was used and the infants were scanned while resting quietly in the scanner with MR-compatible earmuffs.
Preprocessing: For each subject and time-point, the functional data were realigned and co-registered using SPM12. Volumes with a frame-wise displacement (FD>0.5mm) were removed, along with the previous and two subsequent images. Nine subjects with more than 1/3 of motion-affected volumes were excluded. The UNC neonatal atlas⁶ was registered to each subject’s native space using Advanced-Normalization-Tools⁷. The deformation field was applied to the atlas Gray/White/CSF tissues (TMPs) and to the atlas image to bring them into the native space. Using these TPMs, the subject’s T2 is segmented and the subject-specific probability maps were obtained. GM voxel signals were extracted by masking the atlas using the GM TPM obtained by the segmentation, and by reslicing it to the functional space. The regions of the UNC atlas were further grouped into 11 primary regions for a more global brain coverage: Orbito-frontal cortex (OFC), Dorsolateral prefrontal cortex (DFC), Ventrolateral prefrontal cortex (VFC), Medial prefrontal cortex (MFC), Primary motor cortex (M1C), Primary somatosensory cortex (S1C), Posterior inferior parietal cortex (IPC), Primary auditory cortex (A1C), Posterior superior temporal cortex (STC), Inferior temporal cortex (ITC), Primary visual cortex (V1C). Average regional BOLD time-courses were extracted for each subject and finally, the BOLD variability (BOLD SD) was estimated as the standard deviation of the signal change across time, after bandpass filtering [0.01Hz-0.1Hz] (Figure 1).
Analysis: For each subject and time-point, we extracted the BOLD SD measures and we compared them across time-points (33 vs 40 weeks). Paired statistical tests were used to assess the statistical significance of our results.

Results

Regional and temporal differences in the evolution of the BOLD variability (SD) were found between 33 and 40 weeks time-points for the preterm infants. When comparing the BOLD variability evolution across the two time-point, the regions exhibiting the highest increase in BOLD SD (delta: TEA > birth) were found to be the primary visual (V1C), primary somatosensory (S1C), primary auditory (A1C), primary motor (M1C) and posterior inferior parietal cortex (Figure 2 & 4). These results were significant after applying FDR correction for multiple comparisons (α=0.05) (Figure 4). On the other hand, in the frontal and temporal brain areas such as the orbito-frontal cortex (OFC), dorsolateral prefrontal cortex (DFC), ventrolateral prefrontal cortex (VFC), medial prefrontal cortex (MFC), and inferior temporal cortex (ITC), the BOLD variability was not significantly increased or decreased (Figure 3 & 4).

Discussion & Conclusion

In the present study, we explored the longitudinal evolution of the BOLD variability between birth and TEA, a time window of rapid brain development. Previous research has shown that the brain is a variable, dynamic system that fluctuates naturally from moment-to-moment⁴. Moreover, it has been reported that the BOLD variability of neuronal responses may reflect neural flexibility and maturity⁵. In particular, increased variability in neural systems is reported to be reflective of functional maturation including flexibility, adaptability, and higher dynamic range⁴.
Our resting-state BOLD variability results highlight the increasing role of the primary visual, somatosensory, and motor cortices during this early extrauterine development of preterm infants. The BOLD SD was significantly increased in V1C, S1C, A1C, which are primary sensory areas, as well as in M1C, the primary motor cortex, and the IPC, which is known to be implicated in the integration of sensory information. These findings indicate that major hemodynamic changes occur in these brain regions between birth (33 weeks) and term-equivalent-age (40 weeks), providing evidence in favor of the rapid and early functional (linked to neuronal flexibility) and cerebrovascular maturity of the primary sensory and motor cortices during early development. However, these two are hard to disentangle. On the other hand, the lack of BOLD SD change in the higher-order cortical regions (i.e., frontal/prefrontal areas) suggests that these regions remain immature in terms of functional maturation. This is reasonable given that the frontal regions are late-developing regions of the neocortex⁸. BOLD variability could, therefore, be a valuable and meaningful biomarker of functional neuronal maturity.

Acknowledgements

This study was supported by the Swiss National Science Foundation n°32473B_135817/1 and the foundation Prim’ enfance. We thank the Plateforme de Recherche de Pédiatrie and the Centre for Biomedical Imaging (CIBM) of the University Hospital of Geneva for their support.