Association of Neuroimaging Markers on Clinical CT Scans With Domain-Specific Cognitive Impairment in the Early and Later Post-Stroke Stages.

BACKGROUND: Post-stroke cognitive impairment (PSCI) is associated with neuroimaging markers, including cortical atrophy and white matter lesions (WMLs), on clinically acquired CT neuroimaging. OBJECTIVES: To investigate the association between cortical atrophy/WMLs and PSCI in specific cognitive domains in the acute/subacute and chronic stages after stroke, in order to provide clarity on the relationship between these neuroimaging markers and the temporal evolution of PSCI. METHODS: We visually assessed cortical atrophy using the Global Cortical Atrophy (GCA) scale and WMLs using the Fazekas scale. Oxford Cognitive Screen (OCS) or Birmingham Cognitive Screen (BCoS) assessed PSCI at two timepoints (acute/subacute and chronic) in six domains (language, memory, number processing, executive function, attention, praxis). We binarised domain-specific performance as impaired/unimpaired using normative cut-offs. Multivariable linear and logistic regression analyses evaluated associations between GCA/Fazekas scores with acute/subacute/chronic global and domain-specific PSCI, and ANCOVAs examined whether these scores were significantly different in patients with recovered versus persistent PSCI. Age, sex, education, NIHSS, lesion volume, and recurrent stroke were covariates in these analyses. RESULTS: Among 411 stroke patients (Mdn/IQR age= 76.16/16.64; 193 female; 346 ischaemic stroke; 107 recurrent stroke), GCA and Fazekas scores were not associated with global cognitive impairment in the acute/subacute stage after stroke, but GCA score was associated with chronic global PSCI (B = 0.01, p < .001, 95% CI [0.00 0.01]). In domain-specific analyses, GCA score was associated with chronic impairment in the memory (B = 0.06, p < .001, 95% CI [0.03 0.10]) and attention (B = 0.05, p = .003, 95% CI [0.02 0.09]) domains, and patients with persistent PSCI these domains showed significantly higher GCA scores than patients who had recovered (Memory: F(1,157) = 6.63, p = .01, η 2 G = .04; Attention: F(1,268) = 10.66, p = .001, η 2 G = .04). CONCLUSIONS: The present study highlights the potential impact of cortical atrophy on the cognitive recovery process after stroke and demonstrates the prognostic utility of CT neuroimaging for post-stroke cognitive outcomes. Clinical neuroimaging could help identify patients at long-term risk of PSCI during acute hospitalisation.