Anatomical Studies of Executive Dysfunction in Acute Stroke

Executive dysfunction is a common, disabling consequence of stroke, frequently undetected and poorly characterized at the level of its neural substrate. Against a network-level understanding in which executive control depends on a distributed circuit of cortical hubs, subcortical relays, and connecting white-matter tracts, this thesis characterizes its structural signature, asking which components of that substrate, when damaged, predict the deficit, and whether the deficit is carried by the location of damage independently of its magnitude and premorbid reserve. Sixty patients with acute stroke (mean age 63 years; median 6 days post-stroke) underwent neuropsychological assessment with the Frontal Assessment Battery (FAB) and the Trail Making Test, including the derived TMT B-A set-shifting cost, interpreted against Italian normative cutoffs; language was assessed with the Esame Neuropsicologico per l'Afasia (ENPA) to test whether the executive findings were confounded by aphasia. Lesions were manually segmented on FLAIR and normalized to standard space, then characterized through both focal tissue damage and structural disconnection across cortical (Schaefer 100-region), subcortical (Tian 16-region), and white-matter (Yeh 77-tract) atlases and combined whole-grey-matter representations, yielding seven predictor sets per outcome. Lesion-symptom mapping used ridge regression with leave-one-out cross-validation, adjusting in every model for age, education, lesion volume, and stroke severity. The unique contribution of lesion location was isolated by variance decomposition, tested by permutation, and corrected across the full family of tests by the Benjamini-Hochberg procedure. The FAB’s latent structure was examined by principal component analysis, and its subtests analyzed as supplementary outcomes. FAB performance was robustly predicted by lesion location beyond all covariates, surviving correction in six of seven anatomical representations, whereas timed trail-making was not. Processing speed showed no topographic signal and set-shifting only trend-level signals that did not survive correction. The FAB signature was distributed rather than focal, converging on a single left-lateralized circuit: left peri-Sylvian cortex with salience-network opercular and insular nodes, the left striatal-pallidal-thalamic grey matter, and the left frontal tracts connecting them. Structural disconnection predicted impairment at least as strongly as focal damage, and subcortically survived correction where focal damage did not, indicating that the deficit follows the severing of connections as much as the destruction of tissue. The signature was not attributable to aphasia, in that after adjustment for language, it persisted on the same circuit. Salience-network involvement was consistent with that network's role in executive control, though the structural design cannot test the underlying dynamics and the cortical findings are bounded by the predominance of anterior-circulation stroke; the convergent subcortical and white-matter evidence grounds the distributed-circuit conclusion more securely than the cortical signal alone. Localizable FAB subtests converged on this shared anatomy rather than dissociable substrates, expressing the unity of executive function; the isolated set-shifting cost showed a trend-level, disconnection-weighted signal consistent with, though insufficient to confirm, a disconnection account of shifting deficits. Post-stroke executive dysfunction has a definable structural signature: a left-lateralized frontal-executive circuit, distributed across cortical, subcortical, and white-matter substrates, disrupted through both destruction of its nodes and disconnection of its pathways, and carried by location of damage independently of its magnitude. These findings establish a structural foundation for understanding the disorder as one of distributed network disruption,