Connectome reorganization and functional recovery after subarachnoid hemorrhage

Background and Objectives

Magnetic resonance diffusion tensor imaging (DTI) allows inferences on brain connectivity via quantitative mapping of white matter structures. Since white matter tracts may be damaged following aneurysmal subarachnoid hemorrhage (aSAH) and given a critical need for better prognostication in aSAH, we evaluated the association between DTI connectivity measures and functional outcome in this population.

Methods

Patients with suspected aSAH enrolled in a prospective observational cohort underwent DTI during their acute hospitalization. A structural connectome was created then sorted into four canonical large-scale networks: default mode network (DMN), executive control network (ECN), salience network (SAL), and whole brain (WB). Clinical and graph features were used separately and in combination to train random forest (RF) and logistic regression classifiers to predict modified Rankin Score (mRS) at discharge and 6 months after discharge (favorable outcome mRS 0-2, unfavorable outcome mRS 3-6).

Results

A total of 56 suspected aSAH patients underwent DTI a median of 7 (IQR 3.8-12.3) days after admission. The best performing model for predicting 6-month mRS combined clinical and DTI graph features specific to the SAL network; mean±SEM area under the receiver operator characteristic curve (AUROC) and area under the precision recall curve (AUPRC) were, respectively, 0.94 ± 0.004 and 0.95 ± 0.004 (vs 0.91±0.004 and 0.94±0.004 for clinical only, respectively). Results for clinical+ECN were AUROC 0.92±0.004 and AUPRC 0.94±0.004 and for clinical+DMN AUROC 0.93±0.004 and AUPRC 0.95±0.004.

Discussion

Accuracy of prognostication in patients with SAH can be significantly improved by integrating connectivity measures derived from DTI. The highly predictive DTI graph features suggest a dynamic process of structural reorganization occurring in the early phase after aSAH.