Long-term stability of CVR and its lag response presents patterns that are equally explained by vascular anatomy, neural activity, and anatomy.

Background

Cerebrovascular Reactivity (CVR) can be measured with BOLD functional MRI and induced with Breath-Hold (BH) in a highly reliable way. However, the reliability of BH-induced CVR, frequently expressed using the Intraclass Correlation Coefficient (ICC), is not constant across the whole brain, possibly suggesting regional state-like or trait-like driving factors. While it is possible that the neural activations elicited by the BH task are responsible for regional differences in CVR reliability, we hypothesize that the observed spatial patterns of long-term stability are associated with the vascular architecture of the brain.

Methods

After computing ICC maps of CVR and its lag response, we compute 1000 whole-brain surrogate datasets of both. We then use three different atlases (vascular, functional, and subcortical) to assess the probability of the ICC variance being lower in the original data than in the surrogate null data - if they are, we can assume that the parcellation matches the intrinsic organisation of the data.

Results

Parcels from all atlases show significant specificity, indicating that the three architectures taken into account here could explain equally well the observed patterns in ICC_CVR and ICC_lag. This could indicate that CVR and its lag could share specific individual characteristics with other cerebral features (e.g. functional networks). Moreover, given that functional areas are vascularised by sections of the carotid arteries, our observation that both atlases present similarly specific reliability corroborates the presence of an important link between vascular physiology and functional networks.

Full Text

Background

The estimation of cerebrovascular reactivity (CVR) can be a valuable assessment in the clinical routine. The blood oxygenation level-dependent (BOLD) based functional magnetic resonance imaging (fMRI) contrast is commonly used for CVR mapping studies, and a non-invasive way to implement BOLD-CVR measurements is to adopt voluntary respiratory challenges, such as breath-hold (BH) tasks [1]⁠. A BH task induces the subject in a state of hypercapnia that causes a transient increase in blood flow and the BOLD signal, similar to CO2 gas challenges [2]⁠. BH-induced CVR can be assessed despite inconsistent subject performance [3]⁠, and it can be successfully implemented in many types of populations [4,5]⁠. More importantly, it was shown that measuring BH-induced CVR with BOLD-fMRI is highly reliable [3,6–10⁠]. However, the reliability of BH-induced CVR, frequently expressed using the Intraclass Correlation Coefficient (ICC), is not constant across the whole brain, possibly suggesting regional state-like or trait-like driving factors. Local variability has been observed not only across different brain tissues, but also across hemispheres and between cortical and subcortical areas [8,10⁠]. The same observation can be extended to the CVR response lag [10⁠]. Despite these observations, little is known about which factors induce this regionally-specific reliability. Therefore, in this study we inspect how different architectures of the brain (e.g. a vascular architecture vs a functional network architecture) can specifically explain local differences in the long-term stability of CVR (i.e. be more homogeneous). We use three different atlases, one based on functional networks11⁠⁠, one based on vascular anatomy (arterial flow territories) [12]⁠, and one based on anatomical subcortical and cerebellar parcellation [13]⁠, to investigate if local CVR reliability is specific to certain regions of interests (ROI). While it is possible that the neural activations elicited by the BH task are responsible for regional differences in CVR reliability, we hypothesize that the observed spatial patterns of long-term stability are associated with the vascular architecture of the brain. As such, we expect that ROIs circumscribing vascular territories will demonstrate more specific (i.e. homogeneous) reliability as compared to functional or anatomical ROIs.

Methods

We leveraged a precision functional mapping dataset of BH tasks [3]⁠ (Figure 1B), acquired in seven subjects over ten sessions, using optimally-combined multi-echo BOLD fMRI14⁠. Subjects were instructed through visual cues. Further details about the breath-hold paradigm, data acquisition and preprocessing can be found in [10⁠] (see Figure 1). For the current study, we used the maps of the voxelwise reliability of CVR and lag metrics (ICC_CVR and ICC_lag, respectively) of the optimally-combined analysis (see Figure 2 adapted from [10] ). Then, we generated 1000 whole-brain surrogate datasets of the voxelwise ICC_CVR and ICC_lag maps, maintaining spatial autocorrelation properties [15]⁠. We used each atlas (vascular, functional, and subcortical) to extract within-parcel variance of the ICC values from the original ICC maps and from all surrogates maps, scaling them by the variance of all brain voxels. We used a modified version of the atlas described in [12⁠] to take into account inter-hemispheric differences. Finally, for each ROI we assessed the probability of the ICC variance being lower in the original data than in the surrogate null data, that would indicate a better match with the intrinsic organisation of the data. Otherwise, the underlying data might be better described with a different organization than that delimited by the region.

Results

Figures 2 and 3 show the probability of ICC_CVR and ICC_lag maps being more homogeneous in the true data than in surrogate data, respectively. For both ICC_CVR and ICC_lag maps, the areas vascularised by the left intermediate and distal anterior, medial, and posterior carotid artery, as well as by the right intermediate anterior carotid artery, have high probability of presenting homogeneous reliability beyond that expected due to spatial autocorrelation. Contrary to our hypothesis, for both ICC_CVR and ICC_lag maps, the left somatomotor, dorsal attention, salience, control, and default mode networks, as well as the right somatomotor, salience, and default mode network (plus the dorsal attention network for ICC_CVR maps), also exhibit high probability of presenting homogeneous reliability. Similarly, most of the subcortical areas, as well as the left cerebellum in the case of ICC_CVR maps, feature high homogeneity. Parcels from all atlases show significant specificity, indicating that the three architectures taken into account here could explain equally well the observed patterns in ICC_CVR and ICC_lag. This could indicate that CVR and its lag could share specific individual characteristics with other cerebral features (e.g. functional networks [16,17]). Moreover, given that functional areas are vascularised by sections of the carotid arteries, our observation that both atlases present similarly specific reliability corroborates the presence of an important link between vascular physiology and functional networks [18]⁠.

References

  1. Kastrup A, Li T, Takahashi A, Glover GH, Moseley ME. Functional Magnetic Resonance Imaging of Regional Cerebral Blood Oxygenation Changes During Breath Holding. Stroke. 1998;29(12):2641-2645. doi:doi: 10.1161/01.STR.29.12.2641
  2. Kastrup A, Krüger G, Neumann-Haefelin T, Moseley ME. Assessment of cerebrovascular reactivity with functional magnetic resonance imaging: Comparison of CO2 and breath holding. Magn Reson Imaging. 2001;19(1):13-20. doi:10.1016/S0730-725X(01)00227-2
  3. Bright MG, Murphy K. Reliable quantification of BOLD fMRI cerebrovascular reactivity despite poor breath-hold performance. Neuroimage. 2013;83:559-568. doi:10.1016/j.neuroimage.2013.07.007
  4. Thomason ME, Burrows BE, Gabrieli JDE, Glover GH. Breath holding reveals differences in fMRI BOLD signal in children and adults. Neuroimage. 2005;25(3):824-837. doi:10.1016/j.neuroimage.2004.12.026
  5. Handwerker DA, Gazzaley A, Inglis BA, D’Esposito M. Reducing vascular variability of fMRI data across aging populations using a breathholding task. Hum Brain Mapp. 2007;28(9):846-859. doi:10.1002/hbm.20307
  6. Peng S-L, Yang H-C, Chen C-M, Shih C-T. Short- and long-term reproducibility of BOLD signal change induced by breath-holding at 1.5 and 3 T. NMR Biomed. Published online 2019. doi:10.1002/nbm.4195
  7. Magon S, Basso G, Farace P, Ricciardi GK, Beltramello A, Sbarbati A. Reproducibility of BOLD signal change induced by breath holding. Neuroimage. 2009;45:702-712. doi:10.1016/j.neuroimage.2008.12.059
  8. Lipp I, Murphy K, Caseras X, Wise RG. Agreement and repeatability of vascular reactivity estimates based on a breath-hold task and a resting state scan. Neuroimage. 2015;113:387-396. doi:10.1016/j.neuroimage.2015.03.004
  9. Cohen AD, Wang Y. Improving the Assessment of Breath-Holding Induced Cerebral Vascular Reactivity Using a Multiband Multi-echo ASL/BOLD Sequence. Sci Rep. 2019;9(1):1-12. doi:10.1038/s41598-019-41199-w
  10. Moia S, Termenon M, Uruñuela E, et al. ICA-based Denoising Strategies in Breath-Hold Induced Cerebrovascular Reactivity Mapping with Multi Echo BOLD fMRI. bioRxiv. Published online 2020:2020.08.18.256479. doi:10.1101/2020.08.18.256479
  11. Yeo BTT, Krienen FM, Sepulcre J, et al. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J Neurophysiol. 2011;106:1125-1165. doi:10.1152/jn.00338.2011.
  12. Mutsaerts HJMM, Van Dalen J, Heijtel DFR, et al. Cerebral perfusion measurements in elderly with hypertension using arterial spin labeling. PLoS One. 2015;10(8):1-13. doi:10.1371/journal.pone.0133717
  13. Fischl B, Salat DH, Busa E, et al. Whole Brain Segmentation. Neuron. 2002;33(3):341-355. doi:10.1016/s0896-6273(02)00569-x
  14. Moia S, Uruñuela E, Ferrer V, Caballero-Gaudes C. EuskalIBUR. Published online 2020. doi:10.18112/openneuro.ds003192.v1.0.1
  15. Burt JB, Helmer M, Shinn M, Anticevic A, Murray JD. Generative modeling of brain maps with spatial autocorrelation. Neuroimage. 2020;220(February):117038. doi:10.1016/j.neuroimage.2020.117038
  16. Finn ES, Shen X, Scheinost D, et al. Functional connectome fingerprinting: Identifying individuals using patterns of brain connectivity. Nat Neurosci. 2015;18(11):1664-1671. doi:10.1038/nn.4135
  17. Gratton C, Laumann TO, Nielsen AN, et al. Functional Brain Networks Are Dominated by Stable Group and Individual Factors, Not Cognitive or Daily Variation. Neuron. Published online 2018:439-452. doi:10.1016/j.neuron.2018.03.035
  18. Bright MG, Whittaker JR, Driver ID, Murphy K. Vascular physiology drives functional brain networks. Neuroimage. 2020;217. doi:10.1101/475491

Keep following to stay updated!

You can also download the poster!