Good outcome associated with blood-brain barrier disruption and lower blood pressure after endovascular therapy

Abstract

Objectives:

To evaluate the association between post-endovascular thrombectomy (EVT) blood-brain barrier (BBB) disruption on MRI or CT and average systolic blood pressure (SBP) with favorable 90-day functional outcome. Observational studies have found elevated SBP associated with worse outcomes post-EVT, while recent randomized trials found no difference in targeted BP reduction. There may be a subgroup of patients who benefit from targeted BP reduction post-EVT.

Methods:

This is a single-center study of 1) anterior large vessel occlusion stroke patients treated with EVT from 2015 to 2021, 2) achieved mTICI grade 2b or 3. Hyperintense acute reperfusion marker (HARM), hemorrhagic transformation (HT), and midline shift at 3-hour post-EVT and 24-hour imaging were assessed independently by multiple raters. Binary logistic regression models were used to determine the association of post-EVT SBP with outcomes. BBB disruption was defined as HT or HARM on 3h post-EVT imaging.

Results:

Of 103 patients, those with SBP 100-129 versus SBP 130-160 found no significant difference in favorable 90-day outcome (64% vs. 46%, OR 2.11, 95% CI 0.78-5.76, p=0.143). However, among 71 patients with BBB disruption, a significant difference in favorable outcome of 64% in SBP 100-129 vs. 39% in SBP 130-160 group (OR 5.93, 95% CI 1.50-23.45, p=0.011) was found. There was no difference in symptomatic ICH, 90-day mortality, midline shift (≥5 mm), and hemicraniectomy, between BP or BBB groups.

Conclusions:

BBB disruption on 3h post-EVT imaging and lower SBP was associated with favorable outcome. This imaging finding may guide targeted BP therapy and suggests need for a randomized control trial.

Introduction

Observational studies have found associations between elevated systolic blood pressure (SBP) with symptomatic intracranial hemorrhage (sICH) and worse clinical outcomes post-endovascular thrombectomy (EVT).¹ Blood pressure (BP) variability^2,3, peak systolic⁴, and rebound elevation of SBP⁵ were associated with worse outcomes, while targets < 160 versus <180 have reported better outcomes.⁶ This effect is similarly seen in patients with incomplete recanalization.⁷ While reperfusion and collateral status have guided some centers in BP parameters, there have been no controlled trials to validate this. Milestone EVT trials used varying post-EVT SBP targets with current guidelines by the American Heart Association recommending <180 mm Hg - based on the majority of enrolled patients receiving IV alteplase.⁸ The optimal balance between perfusion and ischemia remains a challenge in BP management post-EVT.⁹ The only randomized trial to date, BP-TARGET, found no difference in intensive versus standard BP reduction for hemorrhagic transformation (HT) of infarct or functional outcome.⁹ This suggests that while elevated BP is associated with worse clinical outcomes, the selection of patients who would benefit from intensive therapy is unclear.

Following a stroke, there is a loss of blood-brain barrier (BBB) integrity leading to increased paracellular permeability.¹⁰ Increased BBB permeability facilitates the infiltration of immune cells into the parenchyma, triggering inflammation mediated by intracellular and extracellular signaling mechanisms that lead to neuronal death or dysfunction.¹¹ Studies have shown that BBB disruption is associated with poor clinical prognosis.^12-14

According to the Starling equation,¹⁵ the influx from the capillary to the parenchyma is determined by the osmotic pressure gradient, filtration coefficient, and hydrostatic pressure gradient. Ischemic changes impair autoregulation of the vessels and reduce arteriolar tone.¹⁶ Therefore, arterial hypertension may result in a dramatic increase of the hydrostatic pressure across the BBB of the ischemic area.¹⁶ If the increase in the hydrostatic pressure exceeds the osmotic pressure gradient, the cerebral water influx increases.¹⁶ The filtration coefficient of the BBB is normally very low, ¹⁶ thus the amount of influx would still be insignificant. When the filtration coefficient is elevated (BBB disruption), the amount of influx increase would be drastic and could lead to vasogenic edema and worse outcome.

We hypothesize that a lower SBP is associated with favorable clinical outcome if a patient has BBB disruption post-EVT, given impaired cerebral autoregulation and dependence on systemic blood pressure. The same association may not be observed if a patient does not have BBB disruption.

Methods

This is a single-center retrospective study of patients enrolled in the Institutional Review Board approved NIH Natural History of Stroke study (NCT00009243). Patients enrolled by the NIH stroke team at MedStar Washington Hospital Center in Washington, DC between January 1, 2015, to December 31, 2021, were included in this study if patients had 1) anterior large vessel occlusion stroke treated with EVT, 2) achieved modified treatment in cerebral ischemia (mTICI) grade 2b or 3, and 3) patient or surrogate provided written informed consent.

Patients were scanned pre-EVT with either MRI brain with perfusion imaging and MRA head or CT head with perfusion imaging (CTP) and CTA of head and neck. After the completion of the EVT, patients were scanned within 3 hours and after 24 hours with either MRI brain or CT head, as per the NIH Natural History of Stroke protocol. The ischemic core volume was calculated by NIH coretool software, based on an ADC threshold < 620x10⁻⁶mm²/s.¹⁷ When the baseline scan was CTP, CBF<30% volume calculated by RAPID software was used for ischemic core volume.¹⁸

Three-hour post-EVT and 24h post-EVT MRI GRE or CT were evaluated for HT. Three-hour post-EVT MRI pre- and post-contrast FLAIR was evaluated for assessment of hyperintense acute reperfusion injury marker (HARM).¹⁹ Twenty-four-hour post-EVT MRI FLAIR or CT were evaluated for midline shift.²⁰ Post-EVT BBB disruption is defined as any HT on 3h post-EVT MRI or CT, or HARM on 3h post-EVT MRI. Two trained readers (PU, YK) performed independent imaging assessments with discordance settled by a third reader (ML) as a tiebreaker. All readers were blinded to functional outcome and blood pressure.

At our center, post-EVT patients get vital signs every 15 minutes for the first 2 hours, then every 30 minutes for 6 hours, then every hour for the rest of the first 24 hours. Every BP measurement from the completion of EVT to 24h post-EVT imaging was abstracted from the electronic medical record. Averages of the 24-hour systolic blood pressures (24h-aSBP) and standard deviation were calculated. Demographic and clinical characteristics, pre-morbid modified Rankin Scale (mRS), admission NIH stroke scale (NIHSS), onset-to-recanalization time, 24-hour NIHSS, and 90-day mRS were also obtained.

All post-EVT patients at our center were treated with antihypertensives to keep SBP <160. 24h-aSBP ranges of 100-129 mmHg and 130-160 mmHg were compared to be consistent with BP-TARGET study.⁹ The primary outcome was a favorable outcome at 90 days defined as mRS 0-2 or equal to pre-morbid mRS. Secondary clinical outcomes were death at 90 days, sICH defined as any HT with an increase of NIHSS by 4 points or more, significant midline shift ≥5mm at 24h scan,²¹ and hemicraniectomy before discharge.²¹ Then, only the patients with post-EVT BBB disruption were selected for the outcome comparison between the two blood pressure groups.

Unadjusted analyses were performed on the ordinal or continuous variables using the Mann-Whitney U test. As appropriate based on sample size, either the Chi-squared test or Fisher’s exact test was used for categorical variables. Univariate analysis to compare the frequency of potential confounders between the patients with and without BBB disruption and between the patients with 24h-aSBP<130 and ≥ 130 was performed first. Two variables with p value < 0.1 were chosen to enter multiple regression analysis (core volume and pre-morbid mRS). Details are shown in the supplemental material (Table I). The standard deviation of 24h SBP was not significantly different between patients with 24h-aSBP <130 and ≥ 130, nor between the patients with and without BBB disruption. A logistic regression analysis was then used to adjust the effect of core volume and pre-morbid mRS on the outcome. Statistical analyses were performed with IBM SPSS software (version 1.0.0.1406).

The data that support the findings of this study are available on reasonable request.

Results

From January 1, 2015 to December 31, 2021, 103 patients were treated with EVT for large vessel occlusion of the anterior circulation. Baseline demographic and clinical characteristics are shown in Table 1. There was no significant difference in characteristics between the 44 (43%) patients with 24h-aSBP 100-129 (median 24h-aSBP 120, IQR 115-126) vs. 59 (57%) patients with 24h-aSBP 130-160 (median 24h-aSBP 144, IQR 137-148). There was no significant difference in proportion of tandem occlusions between BP groups. During EVT, balloon angioplasty was performed for concurrent cervical carotid stenosis. No carotid stents or dual antiplatelet therapy were used acutely. Subsequent carotid revascularization with either stent or endarterectomy was performed within 2 weeks from the EVT.

Table 1.

Baseline Characteristics

	All patients (n=103)	24h-aSBP<130 (n=44)	24h-aSBP>=130 (n=59)	p-value
Age, median (IQR), years	66 (57-75)	65 (55-72)	67 (57-76)	0.447
Female, n (%)	63 (61)	26 (59)	37 (63)	0.709
Black/African American, n (%)	64 (62)	25 (57)	39 (66)	0.337
Hypertension, n (%)	63 (61)	29 (66)	34 (59)	0.453
Diabetes Mellitus, n (%)	17 (17)	7 (16)	10 (18)	0.868
Atrial fibrillation, n (%)	31 (30)	14 (33)	17 (30)	0.710
Previous cerebral infarction or TIA, n (%)	13 (13)	6 (14)	7 (12)	0.805
Tobacco use, n (%)	17 (17)	6 (14)	11 (19)	0.513
Pre-morbid modified Rankin, median (IQR)	0 (0-0)	0 (0-0)	0 (0-0)	0.325
Admission NIHSS, median (IQR)	19 (13-22)	20 (14-22)	19 (13-22)	0.731
Ischemic core volume*, median (IQR), Ml	8 (2-19)	9 (1-25)	7 (2-16)	0.144
Clot location, n (%)				0.773
Intracranial ICA	16 (16)	8 (18)	8 (14)
Proximal M1	42 (41)	19 (43)	23 (39)
Distal M1	31 (30)	11 (25)	20 (34)
Proximal M2	14 (14)	6 (14)	8 (14)
Tandem cervical ICA occlusion, n (%)	8 (8)	2 (5)	6 (10)	0.461
IV thrombolysis before EVT, n (%)	62 (60)	27 (61)	35 (59)	0.834
Onset-to-recanalization time, median (IQR), min	296 (232-382)	267 (216-365)	301 (243-430)	0.225
Modified TICI score, n (%)				0.843
mTICI 2b	41 (40)	18 (41)	23 (39)
mTICI 3	62 (60)	26 (59)	36 (61)
Post-EVT BBB disruption†	71 (69)	33 (75)	38 (64)	0.145
Standard deviation of 24h SBP, median, IQR	14 (11-17)	13 (10-20)	14 (12-17)	0.779

IQR, interquartile range; 24h-aSBP, 24-hour average systolic blood pressure post-thrombectomy; TIA, transient ischemic attack; mRS, modified Rankin Scale; ICA, internal carotid artery; IV, intravenous; EVT, endovascular therapy; mTICI, modified thrombolysis in cerebral infarction; BBB, blood-brain-barrier

^*22 patients missing ischemic core volume data.

^†BBB disruption was defined as any hemorrhagic transformation or hyperintense acute reperfusion marker on the immediate post-EVT scan.

The median time from EVT completion to 3h post-EVT imaging was 73 minutes (IQR 57-161). Sixty-three (61%) patients had pre-EVT and 3h post-EVT MRI. Twenty-six (25%) patients had pre-EVT CT and 3h post-EVT MRI. Fourteen (14%) patients did not get 3h post-EVT MRI. (Three of them had 3h post-EVT CT). The median time from EVT completion to 24h post-EVT scan was 22 hours (IQR 20 – 25). Ninety-seven (94%) patients had 24h post-EVT MRI and four (4%) had 24h post-EVT CT.

Seventy-one (69%) patients demonstrated evidence of BBB disruption on the 3h post-EVT scan. Six (8%) with any grade HT without HARM, forty-three patients (48%) with HARM only, and twenty (28%) patients had HARM with HT. Of three patients with post-EVT CT two had HT (Supplemental Table II).

Fifty-five patients (53%) achieved favorable 90-day outcome. After adjustment for potential confounding factors, no significant difference in favorable 90-day outcome was observed between 24h-aSBP 100-129 group versus 24h-aSBP 130-160 group (64% vs. 46%, OR 2.11, 95% CI 0.78-5.76, p=0.143). When only the patients with post-EVT BBB disruption were selected, 24-aSBP 100-129 was significantly associated with favorable 90-day outcome (64% vs. 39%, OR 5.93, 95% CI 1.50 – 23.45, p =0.011). No differences were found in sICH, death at 90-days, significant midline shift (≥5 mm), or hemicraniectomy, even when the patients with BBB disruption were selected. Secondary outcomes are shown in Table 2.

Table 2.

Outcomes

	24h-aSBP<130	24h-aSBP>=130	Adjusted OR (95% CI)	p-value
All patients (n=103)	N=44	N=59
Favorable 90-day outcome, n (%)	28 (64)	27 (46)	2.114 (0.776-5.762)	0.143
ENI, n (%)	32 (73)	40 (68)	1.914 (0.683-5.366)	0.217
Death by 90-day, n (%)	5 (11)	9 (15)	1.122 (0.246-5.117)	0.882
Symptomatic ICH, n (%)	1 (2)	4 (7)	0.000 (0.000-)	0.998
Significant midline shift, n (%)	2 (5)	1 (2)	0.000 (0.000-)	0.998
Hemicraniectomy, n (%)	1 (2)	1 (2)	1.702 (0.094-30.831)	0.719
Patients with BBB disruption (n=71)	N=33	N=38
Favorable 90-day outcome, n (%)	21 (64)	15 (39)	5.926 (1.497-23.451)	0.011
ENI, n (%)	22 (50)	22 (37)	1.933 (0.593-6.296)	0.274
Death by 90-day, n (%)	5 (15)	7 (18)	0.699 (0.104-4.688)	0.712
Symptomatic ICH, n (%)	1 (3)	4 (11)	0.000 (0.000-)	0.998
Significant midline shift, n (%)	2 (6)	1 (3)	0.000 (0.000-)	0.998
Hemicraniectomy, n (%)	1 (3)	1 (3)	1.877 (0.098-35.953)	0.676

The favorable 90-day outcome was defined as a 90-day modified Rankin Scale (mRS) of 0-2 or as same as pre-thrombectomy mRS; ENI, Early Neurological Improvement, defined as a reduction of NIHSS by 4 points or more by 24h; HT, hemorrhagic transformation; symptomatic intracranial hemorrhage (ICH) was defined as any HT with an increase of NIHSS by 4 or more points at 24h post-thrombectomy; significant midline shift was defined as midline shift of 5mm or greater at 24h post-thrombectomy scan; BBB disruption was defined as any HT or hyperintense acute reperfusion injury marker on the immediate post-thrombectomy scan. The p values less than 0.05 are bolded.

Discussion

This study identified a significant association with lower average SBP and favorable 90-day outcome in patients with large vessel occlusion anterior circulation ischemic stroke with early BBB disruption demonstrated on imaging immediately after EVT. This association was not present without BBB disruption. Our finding suggests that a trial such as BP-TARGET⁹ may be able to demonstrate a difference in outcome between intensive vs standard BP treatment if patients are selected based on this post-EVT imaging biomarker.

Contrast enhancement of CSF space on FLAIR has been used to assess early BBB integrity in acute ischemic stroke,^19,22-24 transient ischemic attack²⁵, and may indicate early hemorrhage²⁶ or suggest worse early neurologic recovery after thrombectomy.²⁷ Previous observational post-EVT BP studies have used only non-contrast CT for the assessment of hemorrhage.¹

In our study, while 64% of patients with BBB disruption and lower average SBP had a favorable clinical outcome, there were no significant differences in sICH, suggesting alternative factors may be contributing to hemorrhage. Measures that place one at increased risk of hemorrhage such as age, admission NIHSS, ischemic core volume, IV thrombolysis, successful recanalization, and onset to recanalization time were similar between both groups.

The precise biologic mechanism of elevated systemic BP and BBB disruption leading to worse clinical outcome is unclear, however, there may be at least two factors at play. In the setting of ischemia, dysfunctional tight junctions and endothelial damage lead to increased BBB permeability of the affected vessels.²⁸ The increased BBB permeability leads to parenchymal vasogenic edema from vascular fluid shifts and eventually greater contribution of hydrostatic forces driven by systemic BP, leading to post-ischemia inflammation.²⁹ Ischemic changes also impair dynamic cerebral autoregulation, the ability of vasculature to maintain stable blood flow with changes in blood pressure, leading to dependence on systemic BP that can last up to seven days.^30-33 These changes may lead to increased hydrostatic pressure gradient contributing to more fluid influx toward parenchyma. The relative hypertension necessary to increase perfusion to the brain and save residual penumbra must balance against subsequent increases in edema by hydrostatic forces and risk of post-ischemia inflammation. Conversely, an intact BBB may be able to tolerate higher blood pressure targets with lower risk.

This study may guide future randomized trial design to consider stratifying patients based on the presence of HARM or HT on post-EVT imaging to effectively identify appropriate BP targets; for example, those with an intact BBB may tolerate higher BPs if partially recanalized. This acknowledges the heterogeneity of patients undergoing EVT, both with regards to baseline characteristics including timing, blood pressure, ischemic core size, and collateral circulation, as well as factors post-EVT including incomplete recanalization with residual perfusion deficit, HARM, and HT. A one-size-fits-all approach cannot account for all patient factors.

Our study has several limitations including retrospective design, small sample size, radiologic heterogeneity, and possible selection bias that may introduce potential clinical and imaging confounding factors. Systolic variations, dynamic peaks, and trends of blood pressure beyond 24 hours, along with the use of anti-hypertensives and other active interventions can influence the presence of HARM, hemorrhage, and outcome. Eleven patients did not get post-EVT MRI for assessment of HARM affecting statistical power by low sample size. Obtaining a post-EVT MRI within 3h is also a challenge for many centers, reflecting a need for additional methods of BBB assessment. As all patients at our center were treated to a SBP <160 mmHg post-EVT, varying degrees of reperfusion injury and impaired cerebral autoregulation may have influenced the ability to reach SBP targets. Thus this secondary analysis regarding BBB disruption and primary outcome could not reflect causation and should be interpreted as hypothesis-generating.

Conclusions

Patients with BBB disruption and lower average SBP post-EVT in anterior circulation large vessel occlusion acute ischemic stroke had more favorable functional outcome at 90 days. This may guide future clinical trials in determining a select group of patients for whom more targeted BP therapy post-EVT is beneficial.