LAUSR

Background: Stroke survivors are at an increased risk of developing post-stroke cognitive impairment and post-stroke dementia; those at risk could be identified by brain imaging routinely performed at stroke onset. Aim: This systematic review aimed to identify features which are associated with post-stroke cognitive impairment (including dementia) on magnetic resonance imaging (MRI) performed at stroke diagnosis. Summary of review: We searched the literature from inception to January 2022 and identified 10,284 records. We included studies that performed MRI at the time of stroke (0–30 days after a stroke) and assessed cognitive outcome at least 3 months after stroke. We synthesized findings from 26 papers, comprising 27 stroke-populations (N = 13,114, average age range = 40–80 years, 19–62% female). When data were available, we pooled unadjusted (ORu) and adjusted (ORa) odds ratios. We found associations between cognitive outcomes and presence of cerebral atrophy (three studies, N = 453, ORu = 2.48, 95% CI = 1.15–4.62), presence of microbleeds (two studies, N = 9151, ORa = 1.36, 95% CI = 1.08–1.70), and increasing severity of white matter hyperintensities (three studies, N = 704, ORa = 1.26, 95% CI = 1.06–1.49). Increasing cerebral small vessel disease score was associated with cognitive outcome following unadjusted analysis only (two studies, N = 499, ORu = 1.34, 95%CI = 1.12–1.61; three studies, N = 950, ORa = 1.23, 95% CI = 0.96–1.57). Associations remained after controlling for pre-stroke cognitive impairment. We did not find associations between other stroke features and cognitive outcome, or there were insufficient data. Conclusion: Acute stroke MRI features may enable healthcare professionals to identify patients at risk of post-stroke cognitive problems. However, there is still substantial uncertainty about the prognostic utility of acute MRI for this.