Antonio Maria Chiarelli1, Eleonora Piccirilli1, Carlo Sestieri1, Daniele Mascali1, Emma Biondetti1, Antonio Ferretti1, Richard Wise1, and Massimo Caulo1,2
1Department of Neuroscience, Imaging and Clinical Sciences, University G. D'Annunzio of Chieti Pescara, Chieti, Italy, 2Department of Radiology, SS. Annunziata University Hospital, Chieti, Italy
Synopsis
During the
perinatal period the brain undergoes extensive development, including
increasing brain perfusion and metabolism. More rapidly modifying brain areas
may be more sensitive to insults and transient hypoxic states. We performed
PCASL on 115 infants at term equivalent age (TEA) with variable gestational age
at birth. Insular-subcortical and somato-motor regions exhibited high grey
matter CBF(GM) in both term and preterm newborns. However, premature
neonates showed a redistribution of perfusion compared to term infants, with
somato-motor regions showing even higher CBFGM. These results
suggest that insular-subcortical and somato-motor regions may reflect brain
health and development at TEA.
Introduction
The third trimester of gestation and the first months of life
are critical periods for brain development1,2. Premature birth can be associated
with poor neurodevelopmental outcome even without evident radiological
alterations3. Indeed, rapidly developing brain areas
with a higher need for oxygen may be more sensitive to perinatal suffering and
transient hypoxic states. Cerebral blood flow (CBF) tends to be coupled to
brain oxygen consumption4. Hence, regions with high CBF may be
the focus of perinatal investigation to probe brain health and infer long-term
neurodevelopmental outcome5.
Here we report a pseudo-continuous
arterial spin labelling (PCASL) study, performed at term-equivalent age (TEA), to
evaluate CBF patterns in term and preterm infants with variable gestational age
at birth (GAB). Methods
After ethical committee approval, preterm
infants were recruited from the Neonatology Unit of the University Hospital of
Chieti based on the following inclusion criteria: absent neurological
abnormalities (e.g. stroke, germinal matrix hemorrhage with grade >2) or
congenital infection. Infants at term were selected from a group of neonates
without asphyxia. The selection resulted in a group of 115 infants (67 males)
between 25 and 40 weeks of GAB (mean = 34 weeks, SD = 4.7 weeks), 45 were born at
term (>37 weeks of GAB) .
MRI was
performed at TEA with a 3T scanner using a 32-channel head array coil (Ingenia
Cx, Philips, Best, the Netherlands). Patients were fed and then sedated with
0.05 mg of oral Midazolam/kg6. pCASL was performed acquiring 4
paired tag-control volumes with a multi-shot 3D GRaSE readout (t=1.8s, PLD=2s, TE=14ms, TR=4.4s, res.=2.5×2.5x7mm3 , FOV=200×200x91mm3)
and background suppression.
2 images were acquired without background suppression to measure M0.
The imaging-labeling planes gap ranged from 10mm to 20mm. An axial T2-w volume
was acquired for segmentation and coregistration purposes (res.=0.4×0.4x3.5mm3 , FOV=204×204x98mm3).
CBF was
obtained by averaging the 4 tag-control difference images, using voxelwise M0
normalization and applying the single-compartment kinetic model with a
brain/blood partition coefficient of λ=0.9ml/g, labelling efficiency α=0.85, labelling efficiency reduction
due to background suppression αINV=0.88 and T1blood=1.67s7. T2-w images were segmented using dHCP
pipeline8 and registered to the UNC Infant
Atlas9,10. Original T2-w images, their tissue
segmentations and the UNC Atlas were transformed into the PCASL space. Grey
Matter CBF(GM) was extracted from the 90 regions of interest (ROIs).
Due to T1blood
variability caused by significant differences in hemoglobin concentration
within the population, to evaluate modifications of the CBFGM
pattern across subjects we normalized the CBF maps through z-scoring. The normalized
(n)CBFGM for each ROI represented the distance, in units of standard
deviation, from the subject global CBFGM. A hierarchical clustering11 of ROIs (ward metric, preceded by
spatial principal component analysis retaining 95% of the variance) was
performed on nCBFGM to reduce the number of regions and increase the
study power and interpretability. The effect of prematurity on nCBFGM
was also evaluated. Results
Figure 1
shows a CBF map and the UNC Infant Atlas registered to the T2-w image as well
as the tissue segmentation of the T2-w image of the same subject.
Figure 2
shows the group average CBFGM in the 90 ROIs. Regions of higher
perfusion (above 16 ml/100g/min) such as subcortical and somato-motor regions
and lower perfusion (~8 ml/100g/min) such as ventral-frontotemporal regions are
visible.
Figure 3
reports the hierarchical clustering
outcome. 5 main clusters were identified retaining more than 70% of the data variance.
The 5 clusters were named: ventral-frontotemporal, medial-occipitotemporal,
fronto-parietal, insular-subcortical and somato-motor clusters. Ventral-frontotemporal and fronto-parietal
clusters had a nCBFGM below 0 (t=-23.1, df=114, p~0; t=-12.7, df=114, p~0) and insular-subcortical
and somato-motor clusters had a nCBF above 0 (cluster 1: t=21.1, df=114, p~0;
t=15, df=114, p~0).
Figure 4
reports t-score maps (p<0.05, uncorrected) evaluating the difference in nCBFGM
between premature and term infants. Significant (Bonferroni corrected)
differences were obtained for somato-motor ROIs, with higher relative perfusion,
and for left temporal pole and orbitofrontal regions, with lower relative
perfusion in premature neonates.
Figure 5
reports the boxplots comparing premature and term infants in the 5 clusters
identified. nCBFGM was higher in premature neonates in the
somato-motor cluster (t=3.3, df=113, p=1.4∙10-3) and lower in the ventral-frontotemporal
(t=-3.0, df=113, p=2.8∙10-3) and fronto-parietal (t=-3.4, df=113, p=7.8∙10-4) clusters.Discussion and Conclusion
A data-driven
clustering of ROIs, using nCBFGM, identified 5 main clusters, two
with lower perfusion (ventral-frontotemporal and fronto-parietal clusters), one
with intermediate perfusion (medial-occipitotemporal cluster), and two with higher perfusion (insular-subcortical
and somato-motor clusters), Premature infants had a redistribution of CBF
compared to term infants, with higher nCBFGM in the somato-motor
cluster and lower nCBFGM in ventral-frontotemporal and
fronto-parietal clusters. The high perfusion in the somato-motor regions in
premature neonates might be associated with the longer extrauterine exposure.
We suggest
that insular-subcortical and somato-motor regions should be a focus of
investigation after perinatal suffering for assessing brain health and
predicting neurodevelopmental outcome at TEA. Moreover, accurate evaluation of
T1blood and labelling efficiency would allow the cross-sectional
results to be reinterpreted in absolute CBF units rather than being limited to
consideration of the spatial distribution of CBF.Acknowledgements
This work was
partially conducted under the framework of the Departments of Excellence
2018–2022 initiative of the Italian Ministry of Education, University and
Research for the Department of Neuroscience, Imaging and Clinical Sciences
(DNISC) of the University of Chieti-Pescara, Italy.References
1. Herschkowitz
N. Brain Development in the Fetus, Neonate and Infant. Neonatology 1988;
54: 1–19.
2. Zun Z, Kapse K, Jacobs M, Basu S, Said
M, Andersen N et al. Longitudinal Trajectories of Regional Cerebral
Blood Flow in Very Preterm Infants during Third Trimester Ex Utero Development
Assessed with MRI. Radiology 2021; 299: 691–702.
3. Anderson PJ, Cheong JLY, Thompson DK.
The predictive validity of neonatal MRI for neurodevelopmental outcome in very
preterm children. Semin Perinatol 2015; 39: 147–158.
4. Payne SJ. Cerebral Blood Flow and
Metabolism: A Quantitative Approach. WORLD SCIENTIFIC, 2017
doi:10.1142/10463.
5. Tortora D, Mattei PA, Navarra R, Panara
V, Salomone R, Rossi A et al. Prematurity and brain perfusion: Arterial
spin labeling MRI. NeuroImage Clin 2017; 15: 401–407.
6. Chiarelli AM, Sestieri C, Navarra R,
Wise RG, Caulo M. Distinct effects of prematurity on MRI metrics of brain
functional connectivity, activity, and structure: Univariate and multivariate
analyses. Hum Brain Mapp 2021; 42: 3593–3607.
7. Alsop DC, Detre JA, Golay X, Günther M,
Hendrikse J, Hernandez-Garcia L et al. Recommended implementation of
arterial spin-labeled perfusion MRI for clinical applications: A consensus of
the ISMRM perfusion study group and the European consortium for ASL in
dementia. Magn Reson Med 2015; 73: 102–116.
8. Makropoulos A, Robinson EC, Schuh A,
Wright R, Fitzgibbon S, Bozek J et al. The Developing Human Connectome
Project: a Minimal Processing Pipeline for Neonatal Cortical Surface
Reconstruction. NeuroImage 2018; 173: 88–112.
9. Shi F, Yap P-T, Wu G, Jia H, Gilmore
JH, Lin W et al. Infant Brain Atlases from Neonates to 1- and
2-Year-Olds. PLOS ONE 2011; 6: e18746.
10. Avants BB, Tustison NJ, Song G, Cook PA,
Klein A, Gee JC. A reproducible evaluation of ANTs similarity metric
performance in brain image registration. NeuroImage 2011; 54:
2033–2044.
11. Murtagh F, Contreras P. Algorithms for
hierarchical clustering: an overview. WIREs Data Min Knowl Discov 2012; 2:
86–97.