Changes in Background Blood–Brain Barrier Integrity Between Lacunar and Cortical Ischemic Stroke Subtypes
Background and Purpose— Lacunar stroke is associated with endothelial dysfunction and histologically with intrinsic cerebral microvascular disease of unknown cause. Endothelial dysfunction could impair blood–brain barrier integrity. We assessed background blood–brain barrier leakage in patients with lacunar ischemic stroke compared with cortical stroke controls.
Methods— We recruited patients with lacunar or mild cortical ischemic stroke and assessed generalized cerebral blood–brain barrier leak with MRI and intravenous gadolinium at least 1 month after stroke. We used detailed image processing to compare signal change before and for 30 minutes postcontrast throughout gray matter, white matter, and cerebrospinal fluid with summary analyses and general linear modeling.
Results— Among 48 patients (29 lacunar, 19 cortical), postcontrast enhancement was significantly higher in cerebrospinal fluid (P=0.04, Mann-Whitney U), and nonsignificantly higher in white matter, in lacunar than in cortical strokes, with no difference in gray matter. General linear modeling confirmed significantly greater postcontrast enhancement in cerebrospinal fluid in lacunar patients than in cortical controls (t=3.37, P<0.0008).
Conclusion— These preliminary data suggest that the blood–brain barrier may be dysfunctional throughout subcortical white matter (white matter drains via interstitial spaces to cerebrospinal fluid) in patients with lacunar stroke. Further studies are required to confirm these findings and determine whether abnormal blood–brain barrier might predate development of lacunar disease. Blood–brain barrier dysfunction may be an important mechanism for brain damage in cerebral microvascular disease.
- blood-brain barrier
- cerebral infarction
- cerebral small vessel disease
- lacunar stroke
- magnetic resonance imaging
The cause of lacunar ischemic stroke is debated.1–3 Emboli can cause lacunar ischemic stroke but probably not most lacunar strokes.3 Instead, a substantial proportion seem to be attributable to an intrinsic cerebral small vessel disease of unknown cause described as “segmental arteriolar disorganization,” lipohyalinosis, or fibrinoid necrosis.4 The fact that it is a generalized rather than focal small vessel condition is supported by associations with white matter lesions (WMLs)5 and microhemorrhages.6
Theories on what causes the cerebral microvascular abnormality include microatheroma, endothelial dysfunction, or inflammation.3 The primary event could be alteration of cerebral microvascular endothelial (ie, blood-brain barrier [BBB]) function,7 with extravasation of plasma components into the arteriolar wall (causing the wall thickening, cellular infiltration, disintegration, perivascular damage observed pathologically8), and then leakage into the adjacent brain causing “perivascular edema-related lesions.”9,10 This observation is supported by the systemic11,12 and cerebral13 endothelial dysfunction found in lacunar stroke.
We sought evidence of generalized background BBB leakiness in patients with lacunar stroke using MRI to examine the signal change in gray matter, white matter, and cerebrospinal fluid (CSF) over the course of 30 minutes after intravenous gadolinium contrast administration. The controls were patients with mild cortical ischemic stroke to account for age, having any stroke, vascular risk factors, and secondary prevention treatments that might influence the vascular endothelium. Normal older subjects would not be appropriate controls as any associations found could just be with any stroke, not with lacunar stroke per se.
Patients and Methods
We recruited patients prospectively with clinical lacunar or mild cortical stroke from our hospital stroke service. All patients were examined by a stroke physician and classified into lacunar or cortical stroke clinical syndromes according to the Oxfordshire Community Stroke Project classification.14 All patients underwent usual investigations for stroke (carotid Doppler ultrasound, ECG, blood tests, and other tests). We recorded history of diabetes and hypertension. Patients with abnormal urea and creatinine were excluded to avoid nephrogenic systemic sclerosis, which can occur with Gd-DTPA. Mild cortical stroke was defined as 2 of hemiparesis, hemi-sensory loss, or loss of higher cerebral dysfunction (eg, dysphasia or neglect), equivalent to a partial anterior circulation stroke syndrome.14 The study was approved by the Local Research Ethics Committee and all patients gave written informed consent.
Patients had diagnostic MRI at presentation to identify the site of the recent infarct and quantify WMLs. All scanning was performed on a 1.5-T MR scanner (Signa LX; General Electric) with 22 mT m−1 maximum strength gradients. Diagnostic MRI included axial diffusion-weighted, T2-weighted, fluid-attenuated inversion recovery, and gradient echo sequences (details available on request).
The BBB MRI was performed between 1 (to avoid any acute focal effects of the stroke on the BBB) and 3 months after the index stroke. All patients were well enough to attend as outpatients. The BBB imaging technique was similar to that used previously in the brain,15 retina,16 and breast.17 We tested 2 temporal modifications of the technique in 2 consecutive cohorts, both including patients with lacunar and cortical ischemic strokes. The BBB imaging consisted of T1-weighted whole brain volume acquisitions before, and then repeated sequentially up to 30 minutes after, intravenous injection of 40 mL (ie, ≈0.2 mg/kg) of gadolinium diethyltriamine penta-acetic acid (Gd-DTPA) as gadodiamide 0.5 mmol/mL (Omniscan; GE Healthcare AS). The first 21 patients (first cohort) were imaged with a spoiled gradient echo recalled sequence with TR 29 ms, flip angle 24°, 8 acquisitions, slice thickness 3 mm, and acquisition time of 3 minutes 46 seconds. Scanner modifications enabled use of a higher temporal resolution sequence in the 27 subsequent patients (second cohort) consisting of a fast SPRG with TR 8.1 ms, flip angle 12°, 26 acquisitions, slice thickness 4 mm, and acquisition time 1 minute 9 seconds. Both sequences used a TE of 3 ms, matrix 256×256, and field of view 240. We used Gd-DTPA, a small molecule (molecular weight, 590 Daltons), because it crosses the cerebral microvascular endothelium passively (ie, independent of any specific transport mechanism).15 In the retina (developmentally related to the brain), Gd-DTPA provided a sensitive, noninvasive, and linear assay of passive blood–retinal barrier leak.16
We aligned the postcontrast to the precontrast T1-weighted volume acquisitions to remove bulk patient motion (www.fmrib.ox.ac.uk/fsl); placed multiple small regions of interest using a standard template (total 145; Figure 1) on the precontrast scan in white matter (84), gray matter (56), ventricular CSF (10), and internal carotid/basilar arteries (5); sampled the signal change from baseline to 30 minutes postcontrast injection; and calculated the percentage signal enhancement attributable to contrast in each region of interest. One observer placed all the regions of interest. We coded the WMLs using the Fazekas scale18 on the fluid-attenuated inversion recovery and T2-weighted images and brain tissue loss (enlargement of ventricles and major fissures) using the Wahlund scale.19 All image analysis was performed blind to all patient details.
The final stroke subtype was determined from the Oxfordshire Community Stroke Project clinical subtype and the recent infarct site (lacunar or cortical) on diagnostic MRI. We compared lacunar and cortical patient demographics using t tests and χ2 tests. We produced signal enhancement values per time point per tissue type in the whole brain and in the asymptomatic hemisphere (to avoid residual effects of the incident stroke on BBB leakiness) per patient and plotted enhancement-time curves. We first analyzed change in enhancement with time using repeated measures mixed modeling keeping the 2 cohorts separate because of the different temporal resolution. In the second cohort, the higher temporal resolution enabled us also to perform more detailed linear mixed modeling of temporal evolution of enhancement by tissue type and stroke type to examine whether the parenchymal and CSF signals were independent of intravascular signal. Finally, with both cohorts combined, we calculated the area under the signal–time curves20 to examine differences between lacunar and cortical stroke patients using Mann Whitney U tests (data not normally distributed). All analyses were performed in SPSS for windows (version 13.0; SPSS Inc) and SAS (version 9.1; SAS Institute Inc).
Twenty-one patients (14 lacunar, 7 cortical final stroke subtype) were imaged with the lower temporal resolution sequence (first cohort) and 27 patients (15 lacunar, 12 cortical final stroke subtype) were imaged with the higher temporal resolution sequence (second cohort), for a total of 48 patients. There was no difference in the median time from stroke to BBB permeability imaging (lacunar 52 vs cortical 45 days; difference of 2 days; 99% CI, −20 to 23 days), age (mean 67 and 66 years, respectively; P=not significant), the proportion with diabetes (lacunar 4 of 30 [13%] vs cortical 1 of 19 [5%]; P=not significant), or hypertension (lacunar 12 of 30 [40%] vs cortical 11 of 19 [58%]; P=not significant) between lacunar and cortical groups. None of the lacunar stroke patients had a definite atherothromboembolic source (eg, >50% carotid stenosis; atrial fibrillation). Both groups had low WML scores in deep (median score lacunar and cortical both 1; χ2=2.2; P=0.5) and periventricular (mean score lacunar 2; cortical 1; χ2=1.6; P=0.6) white matter. There were no microhemorrhages. Brain volume loss was similar between the 2 groups, with 10 of 30 lacunar and 8 of 19 cortical stroke patients having a Wahlund score of 2 or 3 for ventricles or fissures (33% vs 42%; P=NS).
After intravenous Gd-DTPA injection, the enhancement in brain parenchyma and blood peaked early and declined slowly thereafter (Figure 2). The absolute maximum enhancement was ≈2% in white matter and 6% to 7% in deep and superficial gray matter. In the ventricular CSF, the enhancement increased steadily throughout the sampling period and was still increasing at 30 minutes.
The signal change from baseline to 30 minutes, keeping the 2 cohorts separate, did not show any significant difference between lacunar or cortical patients for any tissue (but the separate cohorts were too small to expect significance at this stage; Figure 2). Note the blood signal in lacunar patients appeared higher in the first cohort (Figure 2a) but lower in the second cohort (Figure 2b) compared with cortical patients, despite which the CSF and white matter signals tended to be higher in lacunar patients in both cohorts.
To test whether the parenchymal signal might simply reflect intravascular Gd-DTPA, we modeled the high temporal resolution sequence data (more time points, more patients). The white matter signal–time curve gradient was not different to that of gray matter (estimate of difference 0.38; standard error, 0.76; P=0.62), but declined faster than in blood (estimate of difference, −4.96; standard error, 0.76; P<0.001), and was different from CSF (estimate of difference, 3.40; standard error, 0.76; P<0.001). Thus, the brain parenchymal signal did not simply reflect intravascular Gd-DTPA and CSF regions of interest do not contain any intravascular tissue. General linear modeling also showed significantly greater accumulation of contrast in CSF in lacunar patients than in cortical controls (estimate of difference, 3.37; P<0.0008).
Combining the 2 cohorts to achieve a larger sample size using summary data (area under curve) showed greater overall enhancement in lacunar than in cortical stroke in CSF (area under curve Mann-Whitney U, P=0.04) and a trend toward greater enhancement in white matter (Figure 3). The gray matter and blood signal were similar in lacunar and cortical stroke (Figure 3c). The results did not change if analysis was restricted to just the asymptomatic hemisphere, ie, the difference in enhancement between lacunar and cortical stroke was not simply attributable to any residual BBB leakage at the site of the index infarct.
There appear to be small but potentially important differences in the passage of Gd-DTPA across the BBB in patients with lacunar compared with cortical stroke syndromes. This would be consistent with the hypothesis that a major contributing factor to the pathogenesis of “lacunar disease” is BBB leak secondary to cerebral microvascular endothelial dysfunction. Whereas it is possible for small emboli to enter the lenticulostriate arteries,1 accumulating data suggest that the majority (80%) of lacunar strokes are secondary to an intrinsic cerebral small vessel abnormality. It is the nature of that abnormality that has caused much debate3 and that is the subject of the present study.
The patients in the present study were at an early stage in their disease, so if BBB dysfunction is part of the pathogenesis, then we would expect any Gd-DTPA leak at this stage to be small. Contrast reaches CSF from choroid plexus and via the cerebral perivascular spaces.21 Greater passage of contrast from the capillary microvasculature into the perivascular space and thence into the ventricles therefore could first manifest as greater difference in CSF enhancement between patient groups, rather than in white matter where the absolute signal change is much lower. Other factors that might affect BBB leak, such as blood pressure, drugs used in secondary stroke prevention, and diabetes, require evaluation in a larger study. There was no difference in the proportion with hypertension or diabetes in the present study. We minimized the effect of these and secondary prevention treatments by using cortical stroke controls.
Neither lacunar nor cortical strokes are “pure” diseases. Patients with cortical strokes can have lacunar strokes (and vice versa) and both atherothromboembolic sources and features suggesting intrinsic small vessel disease (white matter lesions, microhemorrhages) may be present in the same patients. Risk factors (atheroma, intrinsic small vessel disease) are likely to have been developing for some time before the first symptoms appear. In some patients, symptoms might never appear. Therefore, we identified patients using a clear cut-point at which there was a definite clinical manifestation of cortical or lacunar stroke (ie, the first presentation with stroke symptoms) to be certain that the patient had the disease sufficiently badly to cause the stroke symptoms (ie, not just asymptomatic white matter lesions) and to sort the patients into definite stroke subtypes (this could not be performed on the basis of white matter lesions because cortical stroke patients have these also). To detect a significant difference between cortical and lacunar subtypes in the white matter after Gd-DTPA, and to adjust for WML load, atrophy, and other key factors, with 80% power, (α=0.05) will require at least 200 subjects per group. Furthermore, a cross-sectional study would only demonstrate an association between BBB dysfunction and lacunar stroke. To demonstrate that the increased dysfunction might be causative (or at least a key manifestation of the pathogenesis) would require a longitudinal study to demonstrate that BBB leak predated disease progression and that the most abnormal BBB dysfunction at baseline predicted the highest risk of recurrent lacunar stroke and progressive WMLs during follow-up.
Small but measurable quantities of Gd-DTPA passively cross the normal BBB and blood–retinal barrier (developmentally and functionally related).16,22 The passage is linear and changes associated with retinopathy,16,22 development of WMLs, and subcortical stroke23 can be detected with Gd-DTPA MRI in experimental models. More work is required to characterize the relationship between change in signal and change in blood or tissue Gd-DTPA. For this reason, we used mixed modeling to compare the enhancement in the different regions of interest with the baseline signal intensities, which allows for differences of concentration–signal curves between tissues to be examined, but not for direct inference of concentrations from signal intensity data. Therefore, we cannot, for example, state that there is more enhancement or contrast entering the CSF than white matter. Rather we have focused on differences within each tissue between cortical and lacunar stroke subtypes. It is unlikely that the parenchymal enhancement was only attributable to intravascular Gd-DTPA as the statistical model shows that enhancement declined significantly more rapidly in blood than in brain and the CSF enhancement (uncontaminated by intravascular Gd-DTPA) was quite different to that of blood. Although a difference in blood flow could account for the observed tissue differences, it is unlikely that the difference between lacunar and cortical patients is attributable to any difference in blood signal; the blood signal was higher in cortical than lacunar patients in the second (larger) cohort (Figure 2b), despite which the white and gray matter and CSF signals was higher in lacunar than cortical patients. In both cohorts combined there was no overall difference in blood signal. A larger sample will enable further detailed modeling of the influence of vascular, brain tissue, and patient factors on signal change in brain and CSF after Gd-DTPA. Newer contrast agents may help target specific BBB transport mechanisms and be useful in the study of drug delivery to the brain and the role of BBB in disease pathogenesis.
Lacunar stroke is associated with elevated plasma markers of endothelial activation and inflammation,12 impaired cerebrovascular reactivity,13 and abnormal flow-mediated dilatation,11,13 although many of these and similar studies (with one exception11) used age-matched normal rather than nonlacunar stroke controls. Thus, some of these endothelial abnormalities might reflect the presence of any stroke and might not be specific to lacunar stroke. There are no other studies of background BBB leakiness in lacunar stroke. A systematic review of BBB changes with normal aging, Alzheimer disease, vascular dementia, and WMLs, including >3000 patients, found that BBB permeability increased with increasing age; compared with age-matched normal controls, patients with Alzheimer disease and vascular dementia had more permeable BBB (worst in vascular dementia), and BBB permeability increased with increasing WMLs.24 The spontaneously hypertensive stroke-prone rat will have cerebral microvascular abnormality develop that pathologically resembles human cerebral microvascular disease.23 In this rat, the onset of cerebral disease is characterized by BBB leak in white matter seen on Gd-DTPA MRI, followed by increasing WMLs and discrete subcortical “infarcts.”23 The changes in apparent diffusion coefficient on MR diffusion imaging25 and in tissue T226 at disease onset indicate accumulating extracellular fluid, quite different to the intracellular edema of large artery occlusive ischemic stroke. Thus, the mechanism of brain damage in small vessel stroke may not be primarily ischemic, but endothelial/BBB dysfunction with microvascular leak may be the major mechanism.
The present work is preliminary but suggests that subtle, abnormal background BBB dysfunction may be present early in patients with lacunar stroke syndrome. Further studies are needed to determine whether BBB dysfunction worsens as the clinical and imaging manifestations of the disease progress, and whether in humans worse BBB dysfunction predates the development of the clinically apparent disease state.
Sources of Funding
The study was funded by the Chief Scientist Office of the Scottish Executive (CZB/4/281), the Wellcome Trust (075611), Chest Heart Stroke Scotland (ResFell04), the Row Fogo Charitable Trust, the Cohen Charitable Trust, and the UK Stroke Association (TSA02/01). The work was conducted in the SFC Brain Imaging Research Centre at the University of Edinburgh (www.sbirc.ed.ac.uk).
- Received July 25, 2007.
- Accepted August 15, 2007.
Futrell N. Lacunar infarction. Embolism is the key. Stroke. 2004; 35: 1778–1779.
Norrving B. Lacunar infarction. Embolism is the key: Against. Stroke. 2004; 35: 1778–1779.
Davis SM, Donnan GA. Why lacunar syndromes are different and important. Stroke. 2004; 35: 1779–1780.
Lammie GA. Pathology of small vessel stroke. Br Med Bull. 2000; 56: 296–306.
Wardlaw JM, Lewis SC, Keir SL, Dennis MS, Shenkin S. Cerebral microbleeds are associated with lacunar stroke defined clinically and radiologically, independently of white matter lesions. Stroke. 2006; 37: 2633–2636.
Wardlaw JM, Sandercock PA, Dennis MS, Starr J. Is breakdown of the blood-brain barrier responsible for lacunar stroke, leukoariosis, and dementia? Stroke. 2003; 34: 806–812.
Chen PL, Wang PY, Sheu WH, Chen YT, Ho YP, Hu HH, Hsu HY. Changes of brachial flow-mediated vasodilation in different ischemic stroke subtypes. Neurology. 2006; 67: 1056–1058.
Hassan A, Hunt BJ, O’Sullivan M, Parmar K, Bamford JM, Briley D, Brown MM, Thomas DJ, Markus HS. Markers of endothelial dysfunction in lacunar infarction and ischaemic leukoaraiosis. Brain. 2003; 126: 424–432.
Pretnar-Oblak J, Sabovic M, Sebestjen M, Pogacnik T, Zaletel M. Influence of atorvastatin treatment on L-arginine cerebrovascular reactivity and flow-mediated dilatation in patients with lacunar infarctions. Stroke. 2006; 37: 2540–2545.
Tofts PS, Kermode AG. Measurement of the blood-brain barrier permeability and leakage space using dynamic MRI. 1. Fundamental concepts. Mag Res Imaging. 1991; 17: 357–367.
Berkowitz BA, Roberts R, Luan H, Peysakhov J, Mao X, Thomas KA. Dynamic contrast-enhanced MRI measurements of passive permeability through blood retinal barrier in diabetic rats. Invest Ophthalmol Vis Sci. 2004; 45: 2391–2398.
Matthews JN, Altman DG, Campbell MJ, Royston P. Analysis of serial measurements in medical research. BMJ. 1990; 300: 230–235.
Sironi L, Guerrini U, Tremoli E, Miller I, Gelosa P, Lascialfari A, Zucca I, Eberini I, Gemeiner M, Paoletti R, Gianazza E. Analysis of pathological events at the onset of brain damage in stroke-prone rats: a proteomics and magnetic resonance imaging approach. J Neurosci Res. 2004; 78: 115–122.
Farrall AJ, Wardlaw JM. Blood brain barrier: ageing and microvascular disease—systemic review and meta-analysis. Neurobiol Aging. 2007;in press.
Guerrini U, Sironi L, Tremoli E, Cimino M, Pollo B, Calvio AM, Paoletti R, Asdente M. New insights into brain damage in stroke-prone rats. A nuclear magnetic imaging study. Stroke. 2002; 33: 825–830.