Markers of Endothelial and Hemostatic Activation and Progression of Cerebral White Matter Hyperintensities
Longitudinal Results of the Austrian Stroke Prevention Study
Background and Purpose— The pathogenesis of cerebral small vessel disease (SVD) is poorly understood, but endothelial activation and dysfunction may play a causal role. Cross-sectional studies have found increased circulating markers of endothelial activation, but this study design cannot exclude causality from secondary elevations. Confluent white matter hyperintensities (WMHs) on magnetic resonance imaging (MRI) appear to represent asymptomatic cerebral SVD. In a prospective study, we determined whether circulating markers of endothelial activation predicted progression of WMH.
Methods— In the community-based Austrian Stroke Prevention Study, MRI was performed at baseline in 296 subjects and repeated at 3 and 6 years. The following were measured on baseline plasma samples: intercellular adhesion molecule (ICAM), thrombomodulin, tissue factor plasma inhibitor, prothrombin fragments 1 and 2, and D-dimers.
Results— ICAM was associated with age- and gender-adjusted WMH lesion progression at both 3 and 6 years, respectively; (odds ratio [OR], 1.007; 95% confidence interval [CI], 1.002 to 1.012; P=0.004; and OR, 1.004; 95% CI, 1.000 to 1.009 per ng/mL; P=0.057). After multivariate analysis controlling for other cardiovascular risk factors and C-reactive protein, 3-year OR was 1.010 (95% CI, 1.004 to 1.017; P=0.001) and 6-year OR was 1.008 (1.002 to 1.014 per ng/mL; P=0.006). Baseline log lesion volume was a strong independent predictor of progression but associations remained after controlling for this (3-year OR, 1.011; 95% CI, 1.002 to 1.020; P=0.013; and 6-year OR, 1.009; 95% CI, 1.000 to 1.017; P=0.039 per ng/mL). There was no association between WMH progression and other markers.
Conclusions— ICAM levels are related to progression of WMH on MRI. The prospective study design increases the likelihood that this association is causal and supports a role of endothelial cell activation in disease pathogenesis. In contrast, we found no evidence for coagulation activation being important.
Cerebral small vessel disease (SVD) causes one quarter of all ischemic strokes, a large proportion of cases of vascular dementia, and contributes to age-related cognitive decline. It results in lacunar infarction, resulting from ischemia in the territory of a single perforating artery, as well as in more diffuse ischemic changes referred to as leukoaraiosis. Leukoaraiosis, which is seen as low signal on computed tomography imaging or as high signal on T2 or FLAIR magnetic resonance imaging (MRI), is most prominent in the periventricular and deep white matter regions. At autopsy in patients with symptomatic cerebral SVD, thickening and hyaline deposition of the small perforating end arterioles supplying the white matter can be seen. In some cases of larger symptomatic lacunar infarction, localized microatheroma was found at the origin of the deep perforating arterioles.1 The neuropathological appearances corresponding to leukoaraiosis are neuronal loss, ischemic demyelination, and gliosis.2
White matter high-signal changes can seen on MRI of normal subjects, particularly with increasing age. Such white matter hyperintensities (WMHs) may be punctate or confluent. Correlations of MRI appearances with neuropathology have shown that punctate WMHs often represent widened perivascular spaces without substantial ischemic damage.3 In contrast, confluent lesions appear to be ischemic and result from similar vascular changes to those seen in patients with symptomatic SVD.3 Imaging a community population to identify the presence of confluent WMH may therefore provide a surrogate marker to investigate the pathogenesis of SVD. A particularly powerful approach may be to study relationships between risk factors and progression of WMH; recent longitudinal studies have shown that confluent lesions progress relatively rapidly.4
The pathogenesis of SVD is not fully understood. Hypertension is the major risk factor but fails to account for much of the risk.5 Several lines of evidence suggest that chronic endothelial dysfunction plays a pivotal role.6 This may be responsible for breakdown of the blood–brain barrier,7 and impaired cerebral reactivity and autoregulation,8 both of which have been observed in SVD. Endothelial dysfunction can be assessed in vivo by measuring soluble plasma markers. These are released into the circulation in response to endothelial perturbation by a number of different stimuli. Expression of intercellular adhesion molecule 1(ICAM) is a precondition for the adhesion and transendothelial migration of lymphocytes; increased blood levels of soluble ICAM reflect endothelial cell activation. Thrombomodulin (TM) is normally expressed on the endothelial cell surface where, once activated by thrombin, it regulates the activity of protein C. Increased plasma levels are thought to reflect endothelial damage. Levels of both ICAM and TM are elevated in patients with symptomatic SVD.6,9,10⇓⇓ However, such changes could be merely secondary to systemic vascular or cerebral parenchymal damage. Prospective studies offer a more robust method to determine whether such associations are causal or secondary. Therefore, we determined whether plasma markers of endothelial function predicted disease progression, determined by WMH on MRI, in a community population. Previous work has also suggested a prothrombotic state may be important in symptomatic SVD.11 Therefore, we also determined levels of 2 markers of hemostatic turnover, D-dimer, and prothrombin fragments 1 and 2.
Subjects and Methods
The Austrian Stroke Prevention Study is a single-center prospective follow-up study on the cerebral effects of vascular risk factors in the normal elderly population of Graz, Austria.12 Individuals aged 50 to 75 years, stratified by gender and 5-year age groups, were randomly selected from the official community register. Between September 1991 and March 1994, 8193 were invited to participate and 2794 agreed; the first 2007 eligible subjects were enrolled. All were examined by a trained neurologist. Exclusion criteria included neuropsychiatric disease, including stroke and dementia, with the latter screened for by the Mini Mental State Examination.
Every fourth study participant or, in case of refusal, the next participant was invited for an imaging and neuropsychology substudy, including MRI; 509 subjects were included in this part. MRI was repeated at 3 or 6 years or both. The current article uses data from those 296 individuals who had repeat MRI at either 3- or 6-year follow-up; 271, 204, and 191 had repeat MRI at the 3-year, 6-year, and both 3-year and 6-year time points, respectively. Of the 509, 42 had no baseline (MRI contraindications in 3 and claustrophobia in 39). Follow-up MRI was not possible in 171 (death 12, stroke 13, subject refusal 82, and no response to recontact 23). In 41 subjects, we were unable to retrieve the electronically stored data.
Blood was obtained in sodium citrate, centrifuged at 3000g for 10 minutes, and plasma separated and stored at −70°C. Levels of circulating markers were measured using sandwich enzyme-linked immunosorbent assay-based assays for the following: ICAM (R&D Systems, UK); TM and tissue factor pathway inhibitor (TFPI), (Diagnostica Stago and American Diagnostica, Axis Shield, Scotland); prothrombin fragments 1 and 2 (Dade Behring, Sysmex UK Ltd); and D-Dimers (Biopool, Alpha Laboratories UK). Intra-assay (inter-assay) coefficients of variation for ICAM1, TM, TFPI, prothrombin fragments 1 and 2, and D-Dimers were 3.6% (7.4%), 4.8% (4.0%), 7.2% (7.4%), 6.0% (9.0%), and 3.6% (4.8%), respectively. High-sensitivity C-reactive protein was assayed with an enzyme-linked immunosorbent assay.13
MRI was performed on 1.5-T systems (Gyroscan S 15 and ACS; Philips Medical Systems) using proton density and T2-weighted (TR/TE 2000 to 2500/30 to 90 ms) sequences in the transverse orientation. Additionally, sagittal T1-weighted images (TR/TE 580 to 620/15 to 20 ms) were acquired. Baseline and follow-up MRI protocols were identical. All slice thicknesses were 5 mm and acquisition matrices were 205×256.12
Image analysis was performed in a 2-stage process. First, white matter lesions were identified on the hard copies by an experienced investigator. Lesion volume was then calculated semi-automatically on the original data. This allowed the same lesions to be identified on repeat scans and lesion volume quantification to be performed blind to patient identity and time of scan. White matter lesions were defined as focal areas hyperintense on proton density and T2-weighted images, and graded into punctate, early confluent, or confluent lesions, as previously described.12 Caps and pencil-thin lining were disregarded because these changes represent normal anatomical variants.12 The reader marked and roughly outlined on an overlaid transparency all white matter lesions on the baseline MRI initially, and then on the follow-up scans from the same individual. The identified lesions then had their volumes measured on T2 images on an UltraSPARC workstation (Sun Microsystems) using DISPunc (David Plummer, UCL, London, UK). The hard copy with lesions outlined on the overlaid transparency was placed alongside the corresponding images displayed on the workstation. Computer-based volumetric measurements were performed blinded to subject identity and order of scans. Total lesion volume (cm3) was calculated by multiplying the total lesion area by slice thickness. The operator had a proven maximum median intrarater coefficient of variation of 6.4%.
Scans of 50 randomly selected participants were analyzed on 2 occasions, separated by a few months, to establish the error in volumetric assessments. We set the error range as the 95% confidence interval (CI) of the most pronounced difference between repeated measurements (−1.59 to 1.81 cm3). Lesion progression was defined as an increase in WMH volume that exceeded measurement error (1.81 cm3) at 3- and 6-year follow-up. All decreases in lesion volume were within the range of measurement error. The WMH lesion load results from the first and second repeatability studies were highly correlated. The intraclass correlation coefficients ranged from 0.96 at baseline to 0.98 at 6-year follow-up.
Distributions of thrombomodulin, TFPI, and D-dimer were skewed and required logarithmic transformation to obtain normal distributions. We determined associations between markers and lesion progression, already defined, using logistic regression. Because of strong associations between age and lesion volume and progression, we initially determined age- and gender-adjusted relationships and then for significant associations, adjusted for other cardiovascular risk factors (hypertension, diabetes, smoking (current, ex-smoker, nonsmoker), body mass index, and serum cholesterol, triglyceride, and fibrinogen. We also adjusted for C-reactive protein. We also determined associations with absolute change in lesion volume at 3 and 6 years, using multivariate regression.
Because progression was related to baseline WMH lesion load, we controlled for this in multivariate analysis. Baseline values of zero were replaced with 0.1 to allow logarithmic transformation that best fit the distribution of lesion volume.
Seventeen samples were visually lipidemic and excluded before data analysis, leaving 267 samples. Baseline characteristics are shown in Table 1. Baseline lesion volume was median 0.3, range 0 to 35.4, mean (standard deviation [SD]) 1.32 (3.49)cm3, and grading of WMHs was none in 82 (30.7%), punctate in 148 (55.4%), early confluent in 26 (9.7%), and confluent in 11 (4.1%).
In 259 subjects rescanned at 3 years, WMH load was median (range) 0.3 (0 to 42.6), mean (SD) 1.88 (4.83) cm3, and in 196 rescanned at 6 years WMH load was median 0.45 (0 to 65.9) and mean (SD) 2.74 (7.35) cm3. At 3 years, 24 (9.3%) showed disease progression, whereas at 6 years 34 (17.3%) showed disease progression.
Relationships between conventional cardiovascular risk factors and WMH progression are shown in Table 2. Progression at both time points was associated with age and hypertension, but not with other risk factors. Baseline log lesion volume was a strong independent predictor of progression at 3-year (odds ratio [OR], 24.466 log cm3; 95% CI, 6.939 to 86.263; P<0.0001) and 6-year (OR, 24.264 log cm3; 95% CI 7.023 to 83.836; P<0.0001) values after adjustment for age, gender, and conventional risk factors.
ICAM was associated with age- and gender-adjusted WMH lesion progression at both 3 and 6 years; 3-years OR was 1.007 (95% CI, 1.002 to 1.012) per ng/mL (P=0.004) and 6-year OR was 1.004 (95% CI, 1.000 to 1.009) per ng/mL (P=0.057). Unadjusted mean [SD] ICAM levels were higher in subjects who showed lesion progression at both 3 (372.60 [86.49] versus 312.75 [88.42] ng/mL; P=0.003) and 6 years (347.44 [83.29] versus 309.01 [92.55] ng/mL). After multivariate analysis controlling for conventional risk factors and also C-reactive protein, ICAM remained associated with WMH lesion progression at both 3 and 6 years (3-year OR, 1.010 per ng/mL; 95% CI, 1.004 to 1.017; P=0.001; and 6-year OR, 1.008 per ng/mL; 95% CI, 1.002 to 1.014; P=0.006). This association remained after controlling for baseline WMH lesion load (3-year OR, 1.011 per ng/mL; 95% CI, 1.002 to 1.020; P=0.013; and 6-year OR, 1.009 per ng/mL; 95% CI, 1.000 to 1.017; P=0.039).
ICAM was also positively correlated with age- and gender-adjusted change in absolute lesion volume at 3 years (β=0.163, P=0.012) and 6 years (β=0.175, P=0.020). These associations remained significant after controlling for these cardiovascular risk factors at 3 years (β=0.176, P=0.001) and 6 years (β=0.219, P=0.005).
There was no association between WMH progression at 3 or 6 years and other markers (Table 3 for age- and gender-adjusted values).
This is the first study to investigate the relationship between plasma markers of endothelial and hemostatic activation in relation to progression of WMH. ICAM was significantly related to progression. The prospective study design increases the likelihood that this association is causal. Baseline lesion load was a strong independent predictor of progression, as we have previously published.4 However, relationships with ICAM remained after also controlling for this, supporting a causal association between ICAM and progression. This is in contrast to cross-sectional studies that are of less value because they cannot exclude the possibility that a particular measure is elevated because of, rather than caused by, the disease. Increased blood levels of soluble ICAM reflect endothelial cell activation. Therefore, our results suggest that this process is important in the pathogenesis of WMH.
Endothelial cell activation could play a role in the pathogenesis of cerebral SVD in a number of ways, including effects on cerebrovascular hemodynamics, atherosclerosis, and blood–brain barrier function. Endothelial nitric oxide release plays a crucial role in regulation of cerebral blood flow14 and in autoregulation15 in humans. Endothelial dysfunction results in impaired nitric oxide release and bioavailability. Reductions in both resting cerebral blood flow16 and cerebral autoregulation8 have been reported in cerebral SVD. Adhesion of circulating leukocytes to endothelial cells and subsequent transendothelial migration is thought to be an important step in the initiation of atherosclerosis. In part, this process is mediated by cellular adhesion molecules, including ICAM, expressed on the endothelial membrane. Pathological studies have shown increased ICAM expression in several components of the atherosclerotic plaque.17 Prospective studies have also shown associations between soluble ICAM and risk of future myocardial infarction,18,19⇓ supporting a role in atherosclerosis, although this may be partly caused by confounding by conventional risk factors and socioeconomic status.20 Increased blood–brain barrier permeability has been suggested as playing a role in the pathogenesis of cerebral SVD.21 Expression of ICAM on brain microvascular endothelium may allow activated leukocytes to enter the central nervous system via binding to leukocyte ligand on ICAM.22
Therefore, although our data provide evidence that endothelial cell activation plays a role in WMH progression, further studies are required to determine how this endothelial activation and likely consequent dysfunction mediate its effect. Furthermore, the source of ICAM in our population is uncertain. Circulating forms of adhesion molecules may be derived from vascular wall components, including endothelial and smooth muscle cells, although we feel it is most likely derived from endothelial cell activation in this population. However, its origin may not be confined to the brain vasculature. It is likely that cerebral SVD represents a systemic arteriopathy; this is supported by studies showing abnormalities in other vascular beds.23
Some potential limitations of the data merit consideration. First, because they are based on the storage of samples at −70°C for up to 7 years, the possibility of protein degradation cannot be excluded. However, this seems unlikely because concentrations of ICAM were similar to those reported in fresh plasma. Second, previous studies have found ICAM is elevated in patients with acute and chronic inflammatory disorders and unstable coronary symptoms. However, participants in this study were healthy when blood samples were obtained.
In contrast to the associations with ICAM, we found no associations with other plasma markers of endothelial and hemostatic activation. Previous studies have found an association between plasma TM and advanced symptomatic cerebral SVD,6,9⇓ although a prospective study found no association with 87 cases of all subtypes of ischemic stroke.24 Our results failed to show an association with progression of WMH, not supporting a role for TM in the pathogenesis of early cerebral SVD. TM expression in cerebral small vessels is low or absent in many areas;25 this may be relevant to the lack of association. It is possible that our sample size is too small to detect associations with these markers. We performed power calculations, based on the means and SDs of measurements in those who progressed and did not at 3 years; if such differences were significant, sample sizes required for detection would be: thrombomodulin, 37492; TFPI, 1034; D-dimers, 1152; and prothrombin fragments 1 and 2, 3742. Therefore, such differences would only be detected by a much larger study and would be an order of magnitude lower than those with ICAM.
TFPI is the major physiological inhibitor of the effect of tissue factor. It binds deactivated factor Xa within the tissue factor VIIa/Xa complex, preventing thrombin formation. The relationship between TFPI and stroke has been little studied. Decreased plasma levels have been found in adult patients with ischemic stroke, including atherothrombotic and lacunar infarction.26 In a cross-sectional study of advanced symptomatic cerebral SVD, patients with widespread leukoariosis had lower TFPI levels and a higher TF/ TFPI ratio than those with isolated lacunar infarction,6 but no prospective studies have investigated these relationships. Ours is the first prospective study to our knowledge, and it failed to show a relationship.
It has been suggested that a prothrombotic state plays a role in cerebral SVD.13 Prothrombin fragments 1 and 2 are markers of thrombin generation, which arises from conversion of prothrombin to thrombin. D-dimer is a marker of fibrinolytic turnover and has been elevated in studies assessing those at risk for future atherothrombotic events. We found no relationship between prothrombin fragments 1 and 2 or D-dimer and progression of WMH. This does not support a role of activation of coagulation in progression of early cerebral SVD. However, it could be important in pathogenesis of acute ischemic events resulting in lacunar stroke in patients with established SVD; this would not have been detected in our study of asymptomatic individuals.
In summary, this study shows ICAM levels are significantly related to WMH progression. This provides support for endothelial activation being involved in early pathogenesis. In contrast, our results do not support a role for hemostatic changes in early disease, and this would be consistent with local thrombosis playing a secondary role of endothelial activation. Our results, together with previous studies implicating a role of endothelial activation and dysfunction in the disease, would suggest that therapies that stabilize the endothelium may have a role in preventing disease progression.
This study was supported by a project grant from The Stroke Association of the UK.
- Received December 20, 2004.
- Revision received March 17, 2005.
- Accepted April 5, 2005.
Pantoni L, Garcia JH. Pathogenesis of leukoaraiosis: a review. Stroke. 1997; 28: 652–659.
Fazekas F, Kleinert R, Offenbacher H, Schmidt R, Kleinert G, Payer F, Radner H, Lechner H. Pathologic correlates of incidental MRI white matter signal hyperintensities. Neurology. 1993; 43: 1683–1689.
You R, McNeil JJ, O’Malley HM, Davis SM, Donnan GA. Risk factors for lacunar infarction syndromes. Neurology. 1995; 45: 483–487.
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.
Tomimoto H, Akiguchi I, Suenaga T, Nishimura M, Wakita H, Nakamura S, Kimura J. Alterations of the blood–brain barrier and glial cells in white-matter lesions in cerebrovascular and Alzheimer’s disease patients. Stroke. 1996; 27: 2069–2074.
Bakker SL, de Leeuw FE, de Groot JC, Hofman A, Koudstaal PJ, Breteler MM. Cerebral vasomotor reactivity and cerebral white matter lesions in the elderly. Neurology. 1999; 52: 578–583.
Fassbender K, Bertsch T, Mielke O, Muhlhauser F, Hennerici M. Adhesion molecules in cerebrovascular diseases. Evidence for an inflammatory endothelial activation in cerebral large- and small-vessel disease. Stroke. 1999; 30: 1647–1650.
Kario K, Matsuo T, Kobayashi H, Asada R, Matsuo M. ‘Silent’ cerebral infarction is associated with hypercoagulability, endothelial cell damage, and high Lp(a) levels in elderly Japanese. Arterioscler Thromb Vasc Biol. 1996; 16: 734–741.
Tomimoto H, Akiguchi I, Wakita H, Osaki A, Hayashi M, Yamamoto Y. Coagulation activation in patients with Binswanger disease. Arch Neurol. 1999; 56: 104–108.
Schmidt R, Fazekas F, Kapeller P, Schmidt H, Hartung H-P, MRI white matter hyperintensities: three-year follow-up of the Austrian Stroke Prevention Study. Neurology. 53; 1999: 53: 132–139.
Macy E, Hayes T, Tracy R. Variability in the measurement of C-reactive protein in healthy subjects: implications for reference intervals and epidemiological applications. Clin Chem. 1997; 43: 52–58.
Markus HS, Lythgoe DJ, Ostergaard L, O’Sullivan M, Williams SCR. Reduced cerebral blood flow in white matter in ischaemic leukoaraiosis demonstrated using quantitative exogenous contrast based perfusion MRI. J Neurol Neurosurg Psychiatry. 2000; 69: 48–53.
Hwang SJ, Ballantyne CM, Sharrett AR, Smith LC, Davis CE, Gotto AM, Boerwinkle E. Circulating adhesion molecules VCAM-1, ICAM and E-selectin in carotid atherosclerosis and incident coronary heart disease cases: the Atherosclerosis Risk in Communities (ARIC) study. Circulation. 1997; 96: 4219–4225.
Wardlaw JM, Sandercock PA, Dennis MS, Starr J. Is breakdown of the blood–brain barrier responsible for lacunar stroke, leukoaraiosis, and dementia? Stroke. 2003; 34: 806–812.
Kwa VIH, van der Sande JJ, Stam J, Tijmes N, Vrooland JL, for the Amsterdam Vascular Medicine Group. Retinal arterial changes correlate with cerebral small-vessel disease. Neurology. 2002; 59: 1536–1540.
Wang L, Tran ND, Kittaka M, Fisher MJ, Scheeiber SS, Zlokovic BV. Thrombomodulin expression in bovine brain capillaries. Anticoagulant function of the blood–brain barrier, regional differences, and regulation mechanisms. Arterioscler Thromb Vasc Biol. 1997; 17: 3139–3146.