Structural connectome changes as biomarkers of stroke recovery: a longitudinal 7T MRI study.

BACKGROUND

Stroke leads to widespread brain connectivity changes, impacting areas both close and remote to the lesion. Post-stroke recovery dynamics are not fully understood. Investigating structural network reorganization over time can thus provide valuable information on adaptive and maladaptive neural plasticity changes on a subject-specific level.

METHODS

Four first-time ischemic stroke patients (3 M, aged 50-69 years) with upper-extremity motor impairment were examined using ultra-high field 7T MRI structural imaging protocols. For each patient, we performed longitudinal lesion quantification and white matter connectivity analysis at three critical timepoints associated with post-stroke recovery: within 1 week, at ~ 1 month, and at ~ 3 months. Using the structural MRI images, we generated patient-specific weighted structural connectivity matrices at each timepoint. We utilized the Schaefer-Yeo and Automated Anatomical Labeling atlases to derive both anatomical regions and resting-state networks based on pre-defined parcel assignments. We examined lesion evolution and white matter connectivity changes as disconnections, re-emerging connections, and existing connections with an increase in estimated connectivity strength over time. We further conducted exploratory edge-level analyses to examine the connectivity strength changes for each patient.

RESULTS

Across all patients and timepoints, an increase in estimated connectivity strength of pre-existing connections dominated structural reorganization. Temporally, the four patients revealed distinct neural reorganization patterns. Patient 1 exhibited robust structural changes in the late ~1 to ~ 3 months stage, whereas Patient 2 in the early < 1 week to ~ 1 month stage. Patient 3 had continuous network growth, while Patient 4 demonstrated stable network reorganization. In our sample, the somatomotor and attention networks underwent the most dynamic reorganization. Somatomotor and salience/ventral attention regions exhibited increased connectivity strength, and in cortical stroke cases, dorsal attention regions demonstrated decreased connectivity strength.

CONCLUSIONS

In this longitudinal case series, post-stroke neural network reorganization appears to be driven by an increase in estimated connectivity strength of surviving white matter connections, suggesting compensatory neuroplasticity. Adaptive changes were most evident in the somatomotor and salience/ventral attention networks within this sample, while the dorsal attention network suggested a more limited contribution to adaptive network changes. Individual differences in the timing and pattern or reorganization highlight the potential need for further research into personalized treatment approaches to promote adaptive recovery.