Poststroke apathy: Major role of cognitive, depressive and neurological disorders over imaging determinants

Abstract

Apathy occurs in approximately one third of people after stroke. Despite its frequency and functional consequences, the determinants of apathy have only been partially defined. The major difficulty lies in disentangling the reduction in activity due to apathy itself from those secondary to comorbidities, such as depression, sensorimotor deficits, and cognitive impairment. Here, we aimed to examine the prevalence of apathy, identify confounding sources of hypoactivity, and define its neuroimaging determinants using multivariate voxel lesion symptom-mapping (mVLSM) analyses. We assessed apathy in a subgroup (n ¼ 325, mean age: 63.8 ± 10.5 years, 91.1% ischemic stroke) of the GRECogVASC cohort using the validated Behavioral Dysexecutive Syndrome Inventory, interpreted using GREFEX criteria, as well as confounding factors (depression, anxiety, severity of the neurological deficit, and gait disorders). mVLSM analysis was used to define neuroimagingdeterminants andwas repeated after controlling for confounding factors. Apathy was present for 120 patients (36.9%, 95% CI: 31.7-42.2). Stepwise linear regression identified three factors associated with apathy: depressive symptoms (R2 = .3, p < .0001), cognitive impairment (R2 = .015, p < .02), and neurological deficit (R2 = .110, p < .0001). Accordingly, only 9 (7.5%) patients had apathy without a confounding factor, i.e., isolated apathy. In conventional VLSM analysis, apathy was associated with a large number of subcortical lesions that were no longer considered after controlling for confounding factors. Strategic site analysis identified five regions associated with isolated apathy: the F3 orbitalis pars, left amygdala, left thalamus, left pallidum, and mesencephalon. mVLSM analysis identified four strategic sites associated with apathy: the right corticospinal tract (R2 = .11; p < .0001), left frontostriatal tract (R2 0 .11; p < .0001), left thalamus (R2 = .04; p < .0001), and left amygdala (R2 = .01; p < .013). These regions remained significant after controlling for confounding factors but explained a lower amount of variance.