Abstract WMP50: Lesion-symptom Mapping For Post-stroke Depression

Background: Post-stroke depression poses an important challenge for patients and delays reintegration to societal roles. Earlier identification of patients at higher risk for depression based on clinical imaging would allow a tailored treatment approach. Methods: We conducted machine-learning based lesion-symptom mapping (LSM) using a retrospective cohort of 477 patients with first-ever acute ischemic stroke (AIS) presenting to a large cerebrovascular center from 2013-2019 with mild motor impairment defined by mRS 0-2. Patient Health Questionnaire (PHQ-9) depression scale was collected within 60 days of index stroke. The location and volume of AIS lesions on brain MRI were analyzed using a machine-learning based algorithm coupled with the FreeSurfer parcellation package. LSM after random field theory-based multiple comparison correction was conducted to examine the association of AIS location and volume with PHQ-9 scale, adjusting for patient age, sex and NIHSS. Results: AIS in the right temporal, parietal, occipital lobes and the right basal ganglia were associated with depressive symptoms (Figure A). Sensitivity analysis excluding those with significant sleep disturbance (n=24) demonstrated that strokes in the right frontal and parietal lobes were associated with severe depression measured by PHQ-9>10 (Figure B). Conclusions: This hypothesis generating study was suggestive of a neuroanatomic basis for the development of post-stroke depression. Further studies in a larger dataset are needed to confirm these associations. The methods will help to predict those patients at higher risk for post-stroke depression, allowing for earlier intervention and recovery.