Blood-Brain Barrier Leakage in the Penumbra Is Associated With Infarction on Follow-Up Imaging in Acute Ischemic Stroke

Abstract

Background: Blood-brain barrier (BBB) leakage measured with dynamic susceptibility contrast-enhanced magnetic resonance imaging (MRI) has been associated with hemorrhagic transformation in acute ischemic stroke. However, the influence of prethrombolysis BBB leakage on infarct growth has not been studied. Therefore, we aimed to characterize BBB integrity according to tissue state at admission and tissue fate on follow-up MRI.

Methods: This is a post hoc analysis of the WAKE-UP trial (Efficacy and Safety of MRI-Based Thrombolysis in Wake-Up Stroke). Ischemic cores were segmented on diffusion-weighted imaging at baseline and on fluid-attenuated inversion recovery images at follow-up (22-36 hours). Dynamic susceptibility contrast-enhanced-MRI provided penumbra masks (time to maximum of the tissue residue function >6 s minus ischemic core) and BBB leakage (extraction fraction [EF], Z scored) maps via automated analysis. EF was averaged within the ischemic core, total penumbra, 2 penumbra subtypes (salvaged/infarcted penumbra), and normal tissue. Adjusted linear mixed-effects models tested for differences between tissue types and associations of EF with clinical/imaging outcomes. Complementary voxel-wise analyses were performed.

Results: Of 503 patients enrolled in the trial, 165 with suitable dynamic susceptibility contrast-enhanced-MRI data were included in this analysis (mean age 66 years, 38% women, median National Institutes of Health Stroke Scale score of 6; 53% receiving alteplase). EF was significantly increased in the ischemic core and penumbra relative to normally perfused tissue, while differences between total penumbra and ischemic core were statistically nonsignificant. Infarcted penumbra exhibited higher EF than salvaged penumbra, even after adjusting for hypoperfusion severity (P<0.001, n=79 with baseline penumbral tissue and follow-up MRI). Voxel-wise analyses showed a significant association between EF and voxel-level infarction in the placebo group only. EF did not predict hemorrhagic transformation or functional outcomes.

Conclusions: Penumbral BBB leakage may identify tissue at increased risk of infarction. Larger, prospective studies are needed to determine the clinical relevance of BBB leakage as an imaging marker of tissue fate.

Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT01525290. URL: https://eudract.ema.europa.eu/; Unique identifier: 2011-005906-32.

Keywords: blood-brain barrier; cerebral hemorrhage; infarction; ischemic stroke; magnetic resonance imaging; plasminogen activators.

Figure 1.

Methodological overview of the imaging pipeline. A, First, ischemic cores were segmented semiautomatically on baseline apparent diffusion coefficient (ADC≤620 mm²/s) and follow-up fluid-attenuated inversion recovery (FLAIR) images using the SONIA software. Then, pharmacokinetic modeling was performed in 2 different software suites (B). (1) NordicIce was used to estimate extraction fraction (EF) as a marker of blood-brain barrier (BBB) leakage, based on delay-insensitive singular value deconvolution and residue function–based leakage correction using an automatically defined arterial input function. (2) RAPID was used for current state-of-the-art estimation of time to maximum of the tissue residue function (T_max) maps and perfusion deficit masks. C, After coregistration, different tissue types of interest were defined using the perfusion deficit, baseline, and follow-up ischemic core masks. D, Different analytical approaches. We performed both tissue type of interest analyses using averaged EF/T_max values and voxel-wise analyses. For simplicity, only the salvaged vs infarcted penumbra contrast is shown here.

Figure 2.

Box plots and strip plots showing the results of the linear mixed-effects models and post hoc Tukey tests. Here, differences in extraction fraction (EF) between different tissue types of interest are illustrated with box plots and strip plots. Bars above boxplots indicate the level of significance of pairwise post hoc Tukey tests following adjusted linear mixed-effects models (Tables 2 through 4). Next to each part label, schematics of the corresponding tissue type of interest from Figure 1 are displayed. A, Comparison of normal tissue, penumbra, and ischemic core. B, Differences in EF between varying time to maximum of the tissue residue function (T_max) thresholds. C, Differences in EF between the infarcted and salvaged penumbra stratified by treatment status (placebo light colors, alteplase dark colors). NS indicates nonsignificant. *P<0.05; **P<0.01, ***P<0.001.