Tract-based disconnection induced by WMH in patients with minor stroke is related to long-term post-stroke cognition

Background Over a third of minor stroke patients will suffer from post-stroke cognitive impairment (PSCI) but there are no validated tools to clearly identify at-risk patients in the early phase. We aimed to investigate the short and long-term cognitive decline using disconnection features from infarct and white matter hyperintensities (WMH) in first-ever minor ischemic stroke patients. Methods First-ever minor ischemic stroke patients (NIHSS≤7) were prospectively followed-up at 72-hour, 6- and 36-month post-stroke with cognitive tests and conventional brain MRI. Infarct and WMH volumes were semi-automatically evaluated on DWI and FLAIR sequences respectively. Bayesian models using tract-based structural disconnection were used to estimate the remote pathological effects of cerebral infarct and/or WMH. The disconnection approach was compared to features extracted from cerebral infarct and WMH volumes and locations. All lesion-based features were compared between patients with and without cognitive impairment at 6- and 36-month post-stroke. The potential association between the features and cognitive domains alterations was assessed by canonical correlation analyses. All statistical analyses were corrected for age, education and multiple comparisons. Results 105 patients (female, 31%) with a mean (± SD) age of 63 ± 12 years were enrolled. Infarct volume was 10.28 ± 17.10 cm3 and involved the middle cerebral artery territory in 83% of patients. The burden of WMH was higher within frontal periventricular white matter. Infarct-based features showed no significant relationship with 6 and 36-month PSCI. However, a WMH disconnection factor involving the commissural and frontal tracts was associated with 6- and 36-month PSCI, particularly in executive/attention, language and visuospatial domains. Memory domain alterations were associated with higher WMH burden in right temporal regions. Conclusions WMH-induced disconnectivity may predict short and long-term PSCI in minor ischemic stroke. These neuroimaging features extracted from routine MR sequences could help identifying at-risk patients to test future rehabilitation interventions.