TL;DR
This study reveals that Moyamoya disease (MMD), a cerebrovascular disorder, disrupts the brain's "critical dynamics" – a state of optimal information processing found in healthy brains. Both hemorrhagic (HMMD) and ischemic (IMMD) MMD subtypes exhibit "faded criticality," particularly during rest, with IMMD showing some recovery during cognitive tasks. These findings offer new insights into MMD's neural underpinnings and suggest potential biomarkers for diagnosis.
What Problem Does This Paper Address?
Healthy brain activity is thought to operate near a state of "criticality," a dynamic balance that optimizes information processing and transmission. This critical state is characterized by scale-free neural avalanches (short-term dynamics) and long-range temporal correlations (LRTC, long-term dynamics). However, it's unclear how this critical dynamic is affected in Moyamoya disease (MMD), a chronic cerebrovascular disorder characterized by progressive stenosis of the internal carotid arteries, leading to reduced blood flow and potential neurological deficits. MMD has two main clinical outcomes: hemorrhagic (HMMD) and ischemic (IMMD). While both share a common pathophysiology, their specific neural dynamics and how they differ, especially in resting and task states, remain poorly understood. This paper aims to investigate these altered dynamics using a multimodal approach (EEG and fMRI) and to explore if these differences can help distinguish between the two MMD subtypes.
Methods
The researchers employed a combination of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) to assess brain dynamics in MMD patients and healthy controls.
EEG Analysis: * Neural Avalanches: They analyzed EEG data to quantify neural avalanches, which are brief, cascading bursts of neural activity. Four metrics were used: avalanche size (number of events), duration (time bins), the kappa ($\kappa$) value (deviation from criticality), and the branching parameter ($\sigma$) (average number of subsequent events triggered by a single event). * Long-Range Temporal Correlation (LRTC): Detrended Fluctuation Analysis (DFA) was used to calculate the Hurst exponent, a measure of LRTC in the alpha band of EEG signals. This assesses how correlated brain activity is over longer time scales.
fMRI Analysis: * LRTC: Voxel-wise Rescaled Range (R/S) analysis was applied to fMRI data to quantify LRTC in brain regions.
Experimental Conditions: * EEG data was collected during three states: eyes-closed (EC) resting state, eyes-open (EO) resting state, and a working memory (WM) task. * fMRI data was collected during a resting state.
Statistical Analysis: * One-way ANOVA was used to compare groups, with Tukey-Kramer's post-hoc tests for pairwise comparisons.
Datasets
The study utilized data from:
- 50 adult patients with Moyamoya disease (16 HMMD and 34 IMMD).
- 25 healthy control subjects.
EEG data from 24 patients (11 HMMD, 13 IMMD) and 21 controls were used for analysis due to quality. fMRI data was collected from 16 HMMD, 34 IMMD, and 25 controls. No new benchmarks were introduced; existing methods for avalanche and LRTC analysis were applied.
Key Insights and Ideas
- MMD Disrupts Brain Criticality: The core idea is that MMD significantly impairs the brain's ability to maintain a critical state, which is crucial for efficient information processing. This "faded criticality" is observed across different timescales and states.
- Subcriticality in Resting State: Both HMMD and IMMD patients exhibit "subcritical" dynamics in the resting state (eyes-closed), meaning their brain activity is less dynamic and potentially less efficient than healthy controls.
- Task-Dependent Dynamics: While HMMD remains largely subcritical even during a working memory task, IMMD patients show a remarkable shift towards "supercritical" dynamics, even surpassing healthy controls. This suggests a different compensatory mechanism or a more pronounced disruption in HMMD.
- LRTC as a Biomarker: Long-range temporal correlations (LRTC) are significantly reduced in MMD patients compared to controls, making their brain activity appear more random. This reduction is observed in both EEG (temporal) and fMRI (spatial) data.
- Spatial Patterns of Disruption: The study identified specific brain regions where LRTC is reduced in MMD. Interestingly, these regions show partially overlapping but also distinct patterns between HMMD and IMMD, hinting at subtype-specific neural network alterations.
Key Results
- Neural Avalanches:
- In the eyes-closed (EC) state, both HMMD and IMMD showed subcritical dynamics (lower avalanche size and duration exponents, $\kappa$ values > 1).
- During the working memory (WM) task, HMMD remained subcritical, while IMMD became supercritical (higher avalanche size and duration exponents, $\kappa$ values < 1, p < 0.05), surpassing controls.
- Long-Range Temporal Correlation (LRTC):
- In the EC state, MMD patients' EEG amplitude envelopes were more random (lower Hurst exponent) than controls.
- During state transitions (EC to EO/WM), LRTC sharply decreased in controls but remained chaotic in MMD individuals (p < 0.05).
- fMRI analysis revealed significantly lower Hurst exponents in MMD groups compared to controls in specific brain regions, including the left SMA for HMMD and right pre-/postcentral gyrus and right inferior temporal gyrus for IMMD.
- Subtype Differences: While both subtypes showed faded criticality, IMMD demonstrated a greater ability to transition to a more active, supercritical state during cognitive tasks compared to HMMD.
Limitations and Future Work
- Simultaneity of Data Acquisition: The EEG and fMRI data were not acquired simultaneously. The authors suggest that future studies should use simultaneous EEG-fMRI technology for more stable and reliable results.
- Sample Size: The study acknowledges a relatively small sample size, particularly for the HMMD group. They recommend increasing the sample size and including patients with lower Suzuki grades (I-II and V-VI) to gain broader insights into disease progression and variability.
- Executive Dysfunction: The study notes that patients with executive dysfunction might require more patients to fully characterize their brain dynamics.
- Future Directions: The findings suggest that the identified EEG and fMRI features could be valuable for rapid, bedside identification of acute cerebral ischemia and hemorrhage, which is a critical clinical need. Further research could explore these features for diagnostic purposes and to understand the underlying neurophysiological processes of cognitive impairment in MMD subtypes.
Summary updated