Increased blood-brain barrier permeability (BBBP) is frequently observed after ischemic stroke and can be a sign of acute injury. In the subacute setting, increased BBBP may be a result of recovery mechanisms including neuroprotective inflammation and angiogenesis. Bernardo-Castro et al. and Yang et al. previously summarized the main pathophysiological processes likely underlying the evolution of BBBP in subacute stages post-stroke (1, 2). Directly following an acute vessel occlusion, oxidative stress and subsequent inflammation lead to early impairment of the blood-brain barrier (BBB). Subsequent reperfusion and endothelial damage lead to further BBB impairment (3) and can last up to weeks following the index event (4). The underlying pathophysiology of increased BBBP is highly complex (5, 6) and may play a crucial role in tissue recovery and outcome post-stroke. However, the exact time course and role of selected biomarkers in increased BBBP following ischemia remain poorly understood.

In the acute phase, both the ischemic event and subsequent reperfusion can result in oxidative stress and neuro-inflammation of the brain (1). The inflammatory processes result in increased cytokine levels such as interleukin-6 (IL-6), tumor necrosis factor-alpha (TNFα), and the acute phase reactant C-reactive protein (CRP) (7–9). Preclinical studies suggest that selected pro-inflammatory proteins and proteolytic enzymes may play a causal role in increased BBBP by negatively modifying tissue recovery following cerebral ischemia (10–13). Clinical imaging studies in stroke patients found high matrix metallopeptidases-9 (MMP-9) blood levels to be associated with secondary brain damage in the hyper-acute and acute stage after an ischemic stroke (14, 15). Animal studies with histological assessment of rat brains suggest that high MMP-9 expression in histological samples have a negative effect on BBB function and ultimately post-stroke recovery (16). In the acute setting of an ischemic stroke, the upregulation of vascular endothelial growth factor (VEGF) was observed to influence the BBB integrity by increasing para-cellular permeability (17, 18). Moreover, using magnetic resonance imaging (MRI) of rodent brains, Zhang et al. could show that intravenous application of VEGF may enhance angiogenesis in the ischemic penumbra during the subacute phase of an IS and hence potentially modify outcome (19).

Stroke rehabilitation studies hypothesize that through the reduction of oxidative stress and anti-inflammatory processes, aerobic fitness training may improve BBB integrity (20). In line with this hypothesis, previous studies suggest that physical training post-stroke may promote synaptic plasticity and enhance neurogenesis and angiogenesis, leading to better functional recovery (21–23).

Both the underlying mechanisms of increased vascular remodeling and angiogenesis were observed to be associated with better functional recovery following ischemia in rodent studies (24). Previously published clinical studies using imaging techniques (including MRI and single photon emission computed tomography) found that increased BBBP in hyperacute and acute phases of ischemic stroke was associated with poor functional outcome (14, 25, 26). There are still no clinical studies published investigating potential associations of increased BBBP and functional recovery in subacute ischemic stroke patients. However, a comprehensive analysis of how selected blood serum biomarkers and aerobic physical training may affect BBBP post-stroke in the clinical setting is still lacking.

In this study, we aimed to assess the evolution of BBBP in subacute ischemic stroke assessed via contrast-enhanced MRI using the prospective BAPTISe study (“Biomarkers and Perfusion—Training-Induced changes after Stroke”) that accompanied the randomized controlled stroke rehabilitation trial PHYS-STROKE (“Physical Fitness Training in Patients with Subacute Stroke”). In this exploratory analysis, we evaluated underlying associations of increased BBBP with selected blood biomarkers, exposure to early aerobic fitness training as well as long-term functional recovery after stroke.

In this exploratory analysis of the BAPTISe study, we observed that increased BBBP evaluated on contrast-enhanced MRI following moderate-to-severe ischemic stroke was detectable in approximately three out of four cases in early subacute phase of stroke. The presence and intensity of BBBP decreased over time; however, it remained detectable up to 2 months reflecting persisting BBB changes throughout recovery processes in the subacute phase post-stroke. Furthermore, we could demonstrate an association of an increased BBBP with larger lesion volumes and more severe strokes both before and after the trial intervention. To the best of our knowledge, this is the first clinical study to investigate the effect of increased BBBP on long-term outcome in subacute stroke patients.

Although increased BBBP assessed via contrast-enhanced MRI is frequently observed in the hyper-acute and acute phase following ischemia (3, 4, 33–36), the dynamic in later stages is far less understood. Several animal studies could demonstrate that an increased BBBP is still detectable up to 3 weeks following an ischemic event (3, 4). Previous clinical studies support the theory that BBBP may persist into subacute stages in some cases (33, 35, 36). In this study, we were able to observe that most stroke patients still demonstrated an increased BBBP of the ischemic lesion visible on contrast-enhanced MRI at 2 months after the acute event.

BBB dynamics are believed to be diverse and likely influenced by stroke severity (34). In our analyses, stroke severity defined by NIHSS was associated with the presence of an increased BBBP. This has not yet been observed in previous clinical BBBP studies (16–18); albeit BBBP was assessed using differing methodologies including CSF fluid-serum albumin ratios or CT-perfusion imaging in these studies. Moreover, in the current study, serially performed MRI allowed an evaluation of BBBP evolution over time. We observed that BBBP tended to decrease over time in this patient cohort. MR-based detection of increased BBBP might help to understand underlying regeneration mechanisms such as vascular remodeling and possible associations with functional outcome (24).

Whereas, previous studies have suggested that selected cytokines such as TNFα and IL-6 may serve as surrogate markers of increased BBBP (7, 8), we found no robust associations between selected pro-inflammatory biomarkers and BBB integrity in this patient cohort derived from a randomized-controlled stroke rehabilitation trial. Moreover, the levels of selected biomarkers representing pathophysiological mechanisms such as enzymatic proteolysis (MMP-9) and vascular remodeling (VEGF) did not differ between patients with and without increased BBBP.

One of our primary aims was to determine whether blood biomarkers could serve as surrogate markers of the presence and evolution of increased BBBP in subacute stroke. A handful of previously published studies described underlying associations between selected serum biomarkers and BBBP. For example, it has been shown that IL-6 increases BBB dysfunction in rodent models (7). Interestingly, it has been demonstrated that the inhibition of VEGF signaling diminishes the BBB impairment (37). This phenomenon was only observed in mice with pre-existing diabetes. However, in the current exploratory study, we found no robust associations between TNFα, hsCRP, IL-6, VEGF, and MMP-9 levels at v1 and the prevalence or evolution of BBBP in this cohort of subacute stroke patients.

We observed a higher rate of increased BBBP in patients with high LDL-cholesterol levels at inclusion (Table 5 and Supplementary Table 4). Our findings support the observation that elevated LDL levels influence the occurrence of an increased BBBP in the subacute stages of ischemic stroke. Cholesterol levels might influence BBB integrity through activation of inflammation and oxidative stress as described previously (38). Interestingly, previous clinical studies found that low LDL levels and low total cholesterol levels to be associated with higher rates of hemorrhagic transformation (39–41). Whether there is a causal connection between the effects of cholesterol levels on BBB integrity and increased risk of hemorrhagic transformation post-stroke should be investigated in detail in future, independent analyses.

Of note, preclinical studies have reported contradicting results. For example, Kalayci et al. (42) suggested that hypercholesterolemia might have a positive effect on the BBB by increasing the expression of tight junction proteins and thereby possibly decreasing paracellular permeability in rodents. Since the increased BBBP in the subacute stage of the ischemic event also represents regenerative processes, further studies on the lipoprotein's influence in the subacute setting may be of interest. Nevertheless, a comprehensive analysis in a larger independent cohort of stroke patients is warranted to support or refute the observations reported here.

Of note, due to the advantages and disadvantages of single biomarkers, there may be additional value of a combination of selected biomarkers for the prognostic value of increased BBBP. In their clinical study, Tu et al. (43) proclaimed that a panel of neuroendocrine biomarkers predicts functional outcome more efficiently than the NIHSS or single biomarkers alone, suggesting that certain biomarker panels may help in the early evaluation of stroke.

Previous studies suggest that the application of intravenous thrombolysis in the setting of acute stroke may contribute to BBB alteration due to its potential neurovascular toxicity and effect on neuro-inflammation following ischemia (44). However, we found no correlation between increased BBBP and intravenous thrombolysis in the current analysis.

The effect of training on the BBB is still under investigation; previous studies have implied that physical training might reduce oxidative stress and inflammatory processes and thereby strengthen the BBB integrity (20). In our study, a 4-week aerobic fitness training intervention was not identified as a modifier of BBBP over time (Table 2). These results are in line with the co-primary efficacy endpoints of the PHYS-STROKE trial. Here, no associations of the maximum walking speed and Barthel Index at 3 months post-stroke and the intervention were observed (29).

Assuming that BBBP affects the regenerative processes following ischemic tissue damage on a molecular basis, one might expect that long-lasting BBBP post-stroke may modify functional recovery (1). Previous studies that used imaging techniques as well as cerebrospinal fluid/serum albumin ratios to quantify the BBBP found that BBB alteration in the acute setting influenced patients' long-term outcome (14, 25, 26). Although in the current analysis, increased BBBP in subacute stages post-stroke were not associated with a favorable functional long-term outcome (mRS <3), we observed an association between a persisting BBBP and worse functional outcome (mRS >3). However, these findings should be validated in larger, independent cohort analyses. Detection of persisting BBBP in subacute stages could, for example, guide individual rehabilitation strategies (45, 46) if the studies find prognostic value in prolonged BBBP. The identification of a surrogate marker of BBBP (i.e., via increase in selected serum biomarkers) would provide an accessible tool to easily assess BBBP in the clinical setting and subsequently guide further therapeutic procedures.

In summary, this is the first study to show increased BBBP on contrast-enhanced MRI through early and late subacute stages of stroke, which takes us closer to understanding the complex regeneration and recovery processes taking place after ischemia (24, 32). A deeper knowledge of the prognostic value of patient-individual BBBP dynamics in subacute stages may guide therapeutic and rehabilitation strategies in the future.

This analysis has several limitations that warrant discussion. First, this is an exploratory analysis as part of a randomized-controlled stroke rehabilitation trial. Therefore, the existing MRI protocol was not explicitly designed for BBB evaluation. However, dynamic contrast enhanced-MRI is the most well-established method for the visualization and quantification of BBBP (30, 47, 48) and contrast-enhanced MRI is a reliable method to assess BBB integrity and has been successfully applied in previous clinical studies (5, 31). Of note, the time point of MR-imaging at v1 and v2 was inhomogeneous, causing overlap, and limiting comparability across individuals with differing imaging time points. Furthermore, we used subtraction maps to show increased BBBP changes within one subject over time (49, 50); the limitations of this technique include complex post-processing steps and the final visual evaluation of the subtraction images, which is subject to rater-bias. Lastly, the number of participants who received a contrast agent application at both v1 and v2 MRI time points was low, minimizing the sample size considerably and increasing the risk of Type II errors. Concerning the blood-based biomarkers, the total numbers of MMP9 and VEGF measurements were likewise considerably low, which also limits the statistical power of this analysis. Of note, there are several well-known biomarkers of inflammation influencing the BBBP in subacute stroke such as IL-1beta and IL-8 (51, 52). Although these cytokines were not included in the current analysis, they could add diagnostic value in further analyses.

The permeability of the blood-brain barrier assessed on contrast-enhanced MRI decreases during the subacute phase of ischemic stroke but remains detectable for up to 2 months in three out of four patients after moderate-to-severe ischemic stroke. The presence of increased BBBP is associated with more severe strokes and larger infarct volumes. We did not observe a relation between the presence of increased BBBP and exposure to aerobic fitness training. In our cohort, we could not detect an association between the selected serum biomarkers (hs-CRP, TNFα, IL-6, MMP-9, and VEGF) and the phenomenon of an increased BBBP. No clear associations were observed between an increased BBBP and functional outcome at 6 months post-stroke; hence, the long-term prognostic value of the phenomenon in subacute phases of an ischemic stroke remains unclear and warrants further investigation in independent cohorts.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional review board of Charité-Universitätsmedizin Berlin (EA1/137/13 and EA1/138/13). The patients/participants provided their written informed consent to participate in this study.

AF, MEb, AN, and TR conceived or designed the study and supervised the study. SM was responsible for analyzing the data and drafted the manuscript. AK and SP contributed in statistical analysis. AF, MEb, and MEn obtained funding. AN attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. All authors critically revised the manuscript for important intellectual content and gave final approval of the version to be published.

This trial was supported by the German Ministry for Health and Education (01EO0801) through Center for Stroke Research Berlin grant G.2.15. The funder had no role in study design; data collection, analysis, or interpretation; or writing the manuscript. AN was participant in the BIH-Charité Clinician Scientist Program funded by the Charité –Universitätsmedizin Berlin and the Berlin Institute of Health. MEn received funding from DFG under Germany's Excellence Strategy—EXC-2049–390688087, BMBF, DZNE, DZHK, EU, Corona Foundation, and Fondation Leducq.

MEn reports grants from Bayer and fees paid to the Charité from AstraZeneca, Bayer, Boehringer Ingelheim, BMS, Daiichi Sankyo, Amgen, GSK, Sanofi, Covidien, Novartis, and Pfizer, all outside the submitted work. JF reports consulting and advisory board fees from Abbvie, AC Immune, Artemida, BioClinica, Biogen, BMS, Brainomix, Cerevast, Daiichi-Sankyo, EISAI, F. Hoffmann-La Roche AG, Eli Lilly, Guerbet, Ionis Pharmaceuticals, IQVIA, Janssen, Julius Clinical, jung diagnostics, Lysogene, Merck, Nicolab, Premier Research, and Tau Rx, outside the submitted work. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.