Vascular and Neurodegenerative Markers for the Prediction of Post-Stroke Cognitive Impairment: Results from the TABASCO Study

Abstract

Background:

Stroke is a major cause of cognitive impairment and dementia. However, the underlying mechanisms beyond post-stroke cognitive impairment (PSCI) are not fully explained to date.

Objective:

We studied the contribution of vascular pathology measures to PSCI, separate from and in conjunction with pathologic markers associated with Alzheimer’s disease (AD).

Methods:

Data from 397 cognitively intact ischemic stroke patients were available. All patients underwent 3T MRI and evaluated for white matter hyperintensity volume (WMHV) and integrity, ischemic lesions, small vessel disease (SVD) markers and grey matter (GM), hippocampal and cerebrospinal fluid (CSF) volumes. Comprehensive cognitive tests were performed on admission and after two years. We used multiple regression to evaluate the contributions of vascular pathology measures (Framingham risk score, WMHV, and existence of SVD) and AD-associated markers (apolipoprotein E4 status and hippocampal volume).

Results:

During two years follow-up, 80 participants (20.2%) developed PSCI. Low GM and cortex volume and high WMHV and CSF volume, but not the new lesion volume, predicted the development of PSCI in a dose-dependent relationship (p = 0.001). Vascular related imaging markers and risk factors predicted PSCI better than AD related markers (p < 0.001).

Conclusions:

Brain structural measures, including total GM volume, WMHV, and CSF volume were independently associated with PSCI and may serve as early biomarkers for risk prediction. In our sample, vascular pathology measures contributed significantly better to PSCI prediction than markers associated with AD. The newly detected ischemic lesion has not emerged as biomarker for PSCI risk, thus maybe a part of the ongoing vascular pathology.

Keywords

Cognitive impairment neuroimaging predictors stroke

INTRODUCTION

Cerebrovascular disease is the second most frequent cause of dementia [1]. About 20% of patients suffering from stroke will develop dementia, and a much larger undetermined portion will develop mild cognitive impairment (MCI) [2, 3]. However, the underlying mechanisms beyond post-stroke cognitive impairment (PSCI) are not fully explained to date.

Although in some cases the cognitive impairment appears adjacent to the ischemic event, directly connecting the acute ischemic to the cognitive impairment, in others, the cognitive impairment develops months or even years after the event [4]. The question remains whether the ischemic event is a marker of ongoing cerebrovascular disease progressing to vascular dementia, or whether the cognitive impairment is a consequence of other neurodegenerative processes, unrelated to the ischemic event. A double hit mechanism may also be suggested [4].

The aim of this study was to contribute to the understanding of PSCI mechanism by evaluating the contribution of vascular pathology, separate from and in conjunction with Alzheimer’s disease (AD) related measures.

METHODS

Study population

Participants were enrolled from the Tel-Aviv Brain Acute Stroke Cohort (TABASCO) [5], which included 575 consecutive patients with mild to moderate first-ever acute ischemic stroke or transient ischemic attack (TIA). The present analysis included all patients who underwent MRI acquisition during admission and completed follow-up cognitive assessment two years after. MRI scans were acquired during admission within 7 days of stroke onset, unless patients were contraindicated or unable to perform the examination.

Volumetric analysis of brain structures

Volumetric analysis was performed on a 3D T1-WI axial fast SPGR sequence using the FreeSurfer V5.1 image analysis suite, well documented and freely available software (http://surfer.nmr.mgh.harvard.edu/) [6 –8] with careful manual correction. Based on the automated segmentation, we extracted the volumes of interest: hippocampus, thalamus, global gray matter (GM), global white matter (WM), white matter hyperintensity volume (WMHV), ventricular cerebrospinal fluid (CSF) volume, and total intracranial volume. Quantification of total brain volume (TBV) was performed using the freely available Functional MRI of the Brain Software Library (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/) [9].

Cortical thickness measurement

Reconstruction of the cortical surfaces and measurement of cortical thickness for each lobe were performed using the FreeSurfer toolkit (version 5.1). This method of estimating the thickness of the cortex has been validated histologically [10] and by manual measurement on MRI sections [11, 12]. As previously described, the pial surface of each hemisphere was computed by deforming the tessellated WM surface outward toward the GM-CSF boundary. The distance between the WM surface and pial surface yields an estimate of cortical thickness at each vertex [7 , 13]. Cortical thickness estimates are obtained as follows. For each point on the gray/white surface, the shortest distance to the pial surface is first computed. Next, for each point on the pial surface, the shortest distance to the gray/white is identified, and the cortical thickness at that location is set to the average of these two values.

Ischemic infarct identification

The identification of the ischemic lesion was previously described in Kliper et al. [14]. Briefly, ischemic lesions were defined as cortical, sub-cortical, or subtentorial infarcts according to their anatomical locations based on structural MRI images. Cortical infarcts were defined as any infarct that includes the cortex; subtentorial infarcts were defined as cerebellar or brainstem infarctions.

DTI analysis and tissue segmentation

Calculation of the DTI maps and tissue segmentation into ischemic lesions, WMHV, and normal-appearing white matter (NAWM), were performed as previously described by Kliper et al. [14]. The analyses were based on multi-modal view with an in-house semi-automatic method [15] based on the FMRIB Software Library (FSL, http://www.fmrib.ox.ac.uk/fsl/) and SPM5 software (www.fil.ion.ucl.ac.uk/spm/software/spm5). Four different DTI maps were calculated: fractional anisotropy (FA), mean diffusivity (MD), axial and radial diffusivities (Da and Dr). Next, volumes (in mm³) of both ischemic lesions and WMHV were calculated across the whole brain. Mean FA, MD, Dr, and Da values were calculated for WML and NAWM tissue clusters. All segmentation results were visually inspected and manually corrected.

Small vessel disease (SVD) markers

All MRI scans were fully evaluated for SVD radiological markers according to the STRIVE protocol [16]. Images were rated by three raters (EA, JM, and EK) who were blinded to the clinical outcome [17]. SVD markers included: 1) White matter hyperintensities (WMH) were graded using Fazekas score [18, 19]; 2) Old lacunar infarcts were defined as sharply demarcated hypointense lesions sized between 3 mm and 15 mm in diameter on T1-weighted images with corresponding hypointense lesions with hyperintense rim on T2 dark fluid [16]; 3) Cerebral microbleeds (CMB) were defined as round hypointense lesions on SWI with a diameter <10 mm. In order to rule out CMB mimics, other sequences including DWI and T1 were evaluated. CMBs were divided to lobar and deep [16, 20]; 4) Enlarged perivascular spaces were defined as smooth margin, round, oval or linear-shaped lesions, sized up to 3 mm, with signal intensity equal to CSF on T2-weighted images. Enlarged perivascular spaces at the level of the basal ganglia as well as at the level of centrum semiovale in the most involved hemisphere were counted [16, 21].

Baseline and follow-up cognitive assessments

Patients underwent a baseline neuropsychological assessment including the Montreal Cognitive Assessment (MoCA) [22] and the NeuroTrax computerized cognitive testing (NeuroTrax Corp., Bellaire, TX) [23] during their first days of hospitalization post stroke. These comprehensive neuropsychological evaluations were again conducted 24 months after the stroke. A Global Cognitive Score was computed as the average of the six index scores (memory, executive functions, visuospatial perception, verbal function, attention, and motor skills). Data for each outcome parameter of the NeuroTrax battery were normalized according to stratifications of age (50 to 70 years, >70 years) and education (≤12 years, >12 years) to give a distribution with a mean of 100 and a standard deviation of 15 (i.e., an IQ-style scale).

Criteria for cognitive impairment

Patients with cognitive decline were diagnosed as having either MCI or dementia. In order to diagnose MCI, the modified Petersen criteria [24] was applied: the subject had to be impaired (≥1.5 SD) on at least 1 cognitive domain compared with age and education matched published norms on the NeuroTrax computerized cognitive testing, as previously described [25, 26], to have no impairment of basic functional activities, and to not fulfill the DSM IV-TR criteria for dementia.

Participants with suspected cognitive impairment were referred to an experienced cognitive neurologist (ADK) for assessment. Included were patients who could not complete the MoCA or whose NeuroTrax or MoCA tests had fallen by >1.5 SD in follow-up examinations, those with subjective cognitive complaints or who were suspected by a senior neurologist of having cognitive impairment.

Assessments were further reviewed by a consensus forum to determine whether the participant had dementia or MCI. The forum included the assessor, three senior neurologists specializing in memory disorders and a neuropsychologist.

Vascular risk factors

Vascular risk factors were assessed according to the Framingham Stroke Risk Profile (FSRP) score [27]. FSRP is based on the following risk factors: age, systolic blood pressure, antihypertensive medication, diabetes, cigarette smoking, history of cardiovascular disease, atrial fibrillation, and left ventricular hypertrophy as determined by ECG. As ECG data were not available, and since FSRP is purely additive in nature, the left ventricular hypertrophy was removed from the present analysis. Vascular risk factor data was recorded as described in Ben Assayag et al. [5] and for each patient the FSRP score was calculated.

Statistical analysis

Baseline characteristic and risk factors are presented as means and standard deviations for continuous variables and as frequencies and percentages for categorical variables. Chi-square tests and independent t-tests were performed to compare demographic data and risk factors, for categorical and continuous variables, respectively. Multivariate analysis for prediction of cognitive outcome two years post-stroke was performed by logistic regression model, and odds ratios with 95% confidence intervals were calculated. The analyses included independent variables/covariates that were statistically significant in the univariate analyses. Model suitability to the events rates observed was evaluated by the Hosmer-Lemeshow statistics at baseline. We used multivariable Cox regression to examine the associations of MRI and cognitive measures with incidence of cognitive impairment two years post-stroke. Independent variables were defined as continuous or dichotomous. Brain MRI measures were divided into cohort-specific quartiles, and the most abnormal quartiles (for GM, hippocampal, and cortical volume, this is the bottom quartile and for WMHV and CSF the top quartile) were compared with the rest. The models were adjusted for independent variables/covariates that were statistically significant in the univariate analyses: age, years of education, and stroke severity (NIHSS at admission). We ran Cox-proportional hazards models relating the number of identified predictor variables for stroke to risk of PSCI and illustrate these using cumulative hazard function. We used a multistage multiple regression approach to evaluate the contributions of vascular pathology measures or pathologic changes/measures known as associated with AD to cognitive performance. The independent variables were the Framingham risk score and neuroimaging measures of vascular pathology (WMHV, lacunes, and CMB) and pathologic changes/measures known as associated with AD (apolipoprotein E4 (APOE4) status and hippocampal volume). The baseline stage of the model included demographic variables: sex, age, and educational level. In the second stage, we evaluated whether each measure considered separately, was associated with cognition. In the third stage, we evaluated the joint and independent effects of pathologic markers by adding all to the baseline demographic model. This approach was adopted to describe the effects of each type of pathologic change when considered alone, then to investigate their independent contributions when considered together. Finally, receiver operating characteristic (ROC) curves with corresponding areas under the curves (AUCs; neutral value 0.50 = risk prediction by pure chance) and 95% CIs were constructed for the vascular versus AD-related measures. Significant difference between AUC for the vascular versus AD-related measures was tested using χ² test. A 2-tailed α level of 0.05 was used to determine statistical significance. Analyses were performed using SPSS software version 25.0. (SPSS Inc. Headquarters, 233 S. Wacker Drive, 11th floor Chicago, Illinois 60606, USA).

RESULTS

Study population

A total of 575 consecutive eligible cognitive intact patients who were admitted to the Department of Emergency Medicine at Tel-Aviv Medical Center between April 1, 2008 to December 1, 2014, within 72 h from onset of symptoms of TIA or stroke were initially evaluated. Of these, cognitive assessments at baseline and 2 years later were available for 507 subjects. Brain MRI scans at baseline were available for 397 of the subjects who had cognitive data, and these were included in the final analysis. The mean age was 66.9±9.7 years; 56.9% were male.

Stroke etiologies according to the TOAST criteria [28] were as follows: 207 lacunar stroke (52.2%), 49 cardioembolic stroke (12.3%), 40 large-artery atherosclerotic stroke (10%), 101 stroke of other or undetermined etiology (25.5%). Table 1 lists the demographic information, neuroimaging characteristics, and average cognitive scores for the cohort.

Table 1

Demographics of intact and impaired patients’ groups

	Remain cognitively intact	Developed PSCI	p
N	317	80
Age, y (SD)	65.5 (9.4)	72.4 (8.8)	<0.001
Male gender, n (%)	185 (58.4)	41 (51.3)	0.251
Education, y (SD)	13.6 (3.7)	11.3 (3.9)	<0.001
Body mass index, kg/m² (SD)	26.9 (4.4)	27.3 (3.7)	0.586
Current smokers, n (%)	76 (24)	13 (16.3)	0.049
Diabetes mellitus, n (%)	75 (23.7)	34 (42.5)	0.001
Dyslipidemia, n (%)	157 (49.5)	54 (67.5)	0.004
Hypertension, n (%)	175 (55.2)	57 (71.3)	0.009
Framingham Risk Score for Stroke (SD)	10.4 (5.6)	14 (5.3)	<0.001
APOE ɛ4 allele, n (%)	55 (17.4)	15 (18.8)	0.875
Admission NIHSS, median (IQR)	2 (0–3)	3 (1–6)	0.001
Cognitive scores
Admission computerized global cognitive score (SD)	94 (10.8)	83 (15.4)	<0.001
Admission MoCA score (SD)	24.2 (3.2)	21.4 (3.8)	<0.001
Computerized total cognitive score at 24 m (SD)	98.8 (10.8)	86.6 (11.5)	<0.001
MoCA score at 24 m (SD)	26 (31)	21.2 (4.8)	<0.001
Neuroimaging measures
Infarct type
No infarct, n (%)	112 (35.3)	15 (11.8)	0.012
Cortical infarct, n (%)	71 (22.4)	26 (32.5)	0.060
Sub-cortical infarct, n (%)	93 (29.3)	29 (36.3)	0.231
Subtentorial infarct, n (%)	39 (12.3)	9 (11.3)	0.796
Total brain volume, mm³	1351386.3 (142735)	1289373.7 (21162.7)	0.016
Intracranial volume, mm³	1450648.1 (169509.3)	1418950 (174364.7)	0.172
Total WM volume, mm³	447905.3 (6433.1)	415643.3 (59139.6)	<0.001
Total GM volume, mm³	582525.1 (61156.5)	546638.2 (54983.9)	<0.001
Total CSF volume, mm³	32449.4 (17542.7)	38075.1 (17794.8)	0.019
Total hippocampus volume, mm³	7534.8 (1106.3)	7075.9 (855.4)	<0.001
Total thalamus volume, mm³	11828.4 (1509.2)	11442.7 (1338.9)	0.055
Frontal cortex thickness, mm²	2.43 (0.13)	2.37 (0.14)	0.001
Temporal cortex thickness, mm²	2.70 (0.18)	2.61 (0.17)	<0.001
Parietal cortex thickness, mm²	2.26 (0.12)	2.21 (0.13)	0.003
Occipital cortex thickness, mm²	2 (0.13)	1.95 (0.13)	0.010
Total WMHV	16912.8 (18215.9)	31632.2 (27992.7)	0.003
Ischemic lesion volume, mm³	3217.2 (8656.7)	5404.9 (12146.4)	0.199
NAWM FA values, mm²/s	0.4050 (0.0131)	0.3983 (0.0088)	<0.001
NAWM MD values, mm²/s	0.00085 (0.00004)	0.00089 (0.00008)	0.006
NAWM Dr values, mm²/s	0.00066 (0.00004)	0.00069 (0.00006)	<0.001
NAWM Da values, mm²/s	0.00123 (0.00005)	0.00128 (0.00011)	0.014
WMH score	0.4 (0.5)	0.6 (0.5)	0.012
Lobar Microbleeds yes/no (%)	36 (11.4)	16 (20)	0.041
Deep Microbleeds yes/no (%)	14 (4.4)	5 (6.3)	0.492
Lacunes yes/no (%)	118 (37.2)	33 (41.3)	0.507
Perivascular spaces yes/no (%)	294 (92.7)	71 (88.8)	0.241

Entries are mean (SD) or n and %, as indicated. IQR, interquartile range; NIHSS, National Institutes of Health Stroke Scale; GDS, Geriatric Depression Score; MoCA, Montreal Cognitive Assessment; MRI, magnetic resonance imaging; CSF, cerebrospinal fluid; WMHV, white matter hyperintensity volume; NAWM, normal-appearing white matter; FA, fractional anisotropy; MD, mean diffusivity; Dr, radial diffusivity; Da, axial diffusivity.

Association of brain MR imaging measurements with specific cognitive domains

Table 2 summarizes the association of MR imaging brain volume measurements with specific cognitive domains as measured by the NeuroTrax computerized cognitive testing battery. All associations adjusted for age, sex, education, and NIHSS at admission. The domains most strongly correlated with individual brain structural measures were executive function and attention. Memory had a significant interaction with frontal cortex thickness only. The WM and GM volumes and WM microstructural integrity measures (reflected as FA and MD) were associated with executive function, attention and the global cognitive score, while hippocampal volume was only associated with the global cognitive score.

Table 2

Magnetic resonance imaging brain volumes and association with domain-specific cognitive scores 24 months post-stroke, adjusted for age, sex, years of education and NIHSS at admission

	Global score		Memory		Executive function		Visual-spatial		Verbal function		Attention		Motor skills		MoCA score
	β	p	β	p	β	p	β	p	β	p	β	p	β	p	β	p
ICV, mm³	0.118	0.138	0.012	0.882	0.151	0.062	0.129	0.120	0.076	0.381	0.127	0.128	–0.035	0.749	0.092	0.193
WM volume, mm³	0.142	0.072	0.003	0.969	0.245	0.002	0.079	0.332	0.034	0.697	0.198	0.017	0.073	0.508	0.123	0.079
GM volume, mm³	0.218	0.010	0.092	0.296	0.278	0.001	0.151	0.087	0.099	0.292	0.233	0.009	0.034	0.774	0.048	0.473
CSF volume, mm³	–0.030	0.700	–0.076	0.337	–0.072	0.369	0.012	0.883	–0.069	0.417	0.084	0.308	0.020	0.843	0.168	0.029
Hippocampal volume, mm³	0.164	0.049	0.041	0.622	0.147	0.084	0.089	0.302	0.121	0.185	0.100	0.238	0.06	0.557	0.033	0.643
WMH score	–0.049	0.452	–0.113	0.089	–0.095	0.154	–0.063	0.342	0.057	0.424	–0.013	0.851	–0.185	0.030	–0.045	0.426
Ischemic lesion volume, mm³	0.004	0.954	–0.055	0.422	–0.001	0.987	0.013	0.856	0.045	0.540	–0.056	0.424	0.037	0.706	–0.093	0.109
Frontal cortex thickness, mm²	0.166	0.019	0.161	0.017	0.092	0.201	0.055	0.450	0.144	0.064	0.140	0.057	0.116	0.236	0.015	0.802
NAWM volume, mm³	0.106	0.245	0.105	0.263	0.178	0.057	0.152	0.116	–0.152	0.137	0.142	0.139	0.020	0.879	0.043	0.594
NAWM FA values, mm²/s	0.170	0.037	0.145	0.084	0.205	0.015	0.064	0.464	0.016	0.865	0.182	0.028	0.064	0.555	0.132	0.061
NAWM MD values, mm²/s	–0.175	0.049	–0.137	0.134	–0.256	0.004	–0.028	0.765	–0.063	0.526	–0.146	0.120	–0.082	0.488	–0.159	0.040

NIHSS, National Institutes of Health Stroke Scale; ICV, intracranial volume; WM, white matter; GM, gray matter; CSF, cerebrospinal fluid; WMH, white matter hyperintensity; NAWM, normal-appearing white matter; FA, fractional anisotropy; MD, mean diffusivity.

Predictors of post-stroke cognitive impairment (PSCI)

During the two years follow-up period, most patients improved their global cognitive function from admission to 24 months (208, 52.4%), while 109 (27.5%) retained the same cognitive results and 80 participants developed clinically significant cognitive impairment (20.1%). Of these, 9 patients (2.3%) developed dementia and 71 patients (17.9%) developed MCI. Univariate and multivariate predictors for cognitive impairment are shown in Table 3. In this sample of stroke patients, MRI measures that predicted future PSCI were: TBV, GM, WMH, hippocampal, cortex, and CSF volume, as well as WM microstructural integrity and lobar microbleeds on MRI. However, only low GM and frontal cortex volume, high WMHV, and high CSF volumes remain strong predictors following adjustment for age, sex, education, and NIHSS at admission and doubled the risk to develop PSCI two years from the index event (HR, 2.55; 95% CI, 1.16–5.59, HR, 2.65; 95% CI, 1.15–6.11 and HR, 3.06; 95% CI, 1.45–4.9, respectively) (Table 3). There was a dose-dependent relationship between the number of MRI indicators and the risk of developing PSCI (p = 0.001) (Fig. 1). The new ischemic lesion volume did not predict development of PSCI, nor did the location (anatomical region or side) of the new lesion.

Fig.1

Cumulative incidence of post-stroke cognitive impairment based on age, and education-adjusted Cox models by number of neuroimaging indicators: bottom GM volume quartile, bottom frontal cortex volume quartile, top WMHV quartile and top CSFV quartile. GM, gray matter; WMHV, white matter hyperintensity volume; CSFV, cerebrospinal fluid volume; TIA, transient ischemic attack.

Table 3

Hazard Ratios (95% CI) for cognitive status within 24 months from stroke with quartiles of MRI measures (bottom versus others for Gray, Hippocampal and Frontal cortex volumes, as well as White matter FA, and top versus others for White matter hyperintensity and Cerebrospinal fluid volume)

	Unadjusted (Model A)		Model B^†
	Hazard ratio with 95% CI	p	Hazard ratio with 95% Cl	p
Total brain volume	2.15 (1.03–4.5)	0.042	1.14 (0.42–3.1)	0.804
Gray matter volume	2.92 (1.7–5.04)	<0.001	2.37 (1.07–5.26)	0.034
White matter hyperintensity volume	3.42 (1.64–7.13)	0.001	2.65 (1.15–6.11)	0.022
Total hippocampal volume	1.74 (1.02–2.98)	0.043	1.02 (0.52–2.02)	0.953
Cerebrospinal fluid volume	2.39 (1.39–4.11)	0.002	2.21 (1.12–4.34)	0.022
Frontal cortex thickness	2.38 (1.35–4.17)	0.003	2.22 (1.14–4.32)	0.019
White matter microstructural integrity (reflected as fractional anisotropy)	2.24 (1.08–4.61)	0.029	1.94 (0.82–4.59)	0.133
New lesions on MRI (Y/N)	2.22 (1.08–4.57)	0.006	1.42 (0.68–2.97)	0.350
Ischemic lesion volume	1.09 (0.62–1.94)	0.761	1.11 (0.54–2.28)	0.776
Lobar Microbleeds (Y/N)	1.98 (1.04–3.79)	0.039	1.83 (0.84–3.99)	0.127
Deep Microbleeds (Y/N)	1.46 (1.45–4.18)	0.483	—	—
Age	4.04 (2.24–7.3)	<0.001	3.83 (1.92–7.67)	<0.001
Sex	1.36 (0.83–2.23)	0.224	—	—
Education	2.89 (1.71–4.89)	<0.001	2.66 (1.45–4.89)	0.002
Admission NIHSS	1.14 (1.05–1.25)	0.002	1.12 (1.02–1.23)	0.021

CI, confidence interval; MRI, magnetic resonance imaging; NIHSS, National Institutes of Health Stroke Scale. Significant results are shown in bold (p < 0.05). ^†Model B: adjusted for age, education, and NIHSS at admission.

As for non-imaging parameters, vascular risk factor represented by the Framingham risk score were found to be associated with the development of PSCI (HR, 2.3; 95% CI, 1.06–5.02). APOE4 status was not associated with PSCI development (Table 4).

Table 4

Hazard Ratios (95% CI) for cognitive status within 24 months from stroke with vascular pathology measures and parameters known as associated with AD

	Unadjusted (Model A)		Model B^†
	Hazard ratio with 95% CI	p	Hazard ratio with 95% Cl	p
Vascular pathology
Framingham risk score for stroke	3.02 (1.62–5.64)	0.001	2.3 (1.06–5.02)	0.036
White matter hyperintensity volume	3.49 (1.67–7.27)	0.001	2.68 (1.16–6.19)	0.021
Presence of lacunes and/or microbleeds in MRI	1.58 (0.96–5.6)	0.069	1.5 (1.82–2.72)	0.194
Parameters known as associated with AD
Total hippocampal volume	1.75 (1.03–2.99)	0.04	1.02 (0.52–2.02)	0.955
APOE ɛ4 status	1.07 (0.56–2.04)	0.847	1.0 (0.45–2.26)	0.994

CI, confidence interval; AD, Alzheimer’s disease; APOE ɛ4, apolipoprotein E4. Significant results are shown in bold (p < 0.05). ^†Model B: adjusted for age, education, and NIHSS at admission.

In order to explore the contribution of vascular pathology-related markers versus separate from and in conjunction with AD-related markers, we employed multivariate logistic regression model including both vascular markers (Framingham risk score for stroke, WMHV, lacunes, and CMB) and AD associate markers (APOE4 status and hippocampal volume), along with known confounders related to cognitive performance (age, education).

Figure 2 shows the receiver operating characteristic (ROC) curve for vascular pathology markers and AD-associated markers in prediction of PSCI. Vascular related markers had discriminative power to predict PSCI (AUC, 0.67 (95% CI: 0.56–0.78)), while AD related markers did not (AUC, 0.58 (95% CI: 0.45–0.67)). The AUC for vascular pathology measures was significantly higher than the AUC for parameters known as associated with AD (p < 0.001). Combining both measures together (AD and vascular markers) had slightly less discriminative power to predict PSCI than vascular related markers alone (AUC, 0.66 (95% CI: 0.55–0.77)). In addition, to explore whether the association between vascular pathology and risk of PSCI was not only dependent on subjects with very high vascular morbidities, we used a sensitivity analysis, excluding subjects with Framingham risk score >18 (n = 30) and found similar findings.

Fig.2

Receiver operating characteristic curves showing performance of vascular pathology measures and parameters known as associated with AD for predicting 2-year risk of post- stroke cognitive impairment. CI, confidence interval; AUC, area under the curve; AD, Alzheimer’s disease.

DISCUSSION

PSCI is a common and significant cause of disability and impaired quality of life. This study examined a wide spectrum of MRI markers, in order to shed more light on PSCI etiology—whether PSCI is a result of the vascular disease or rather secondary to an ongoing neurodegenerative disease unrelated to the acute event. Of course, the acute ischemic event may trigger or accelerate neurodegenerative processes. Various MRI markers were found to be associated with PSCI; however, after adjustments, only low GM, frontal cortex volume, high WMHV, and high CSF volume remained strong predictors.

The association between low GM, cortex volume, and high CSF volume to PSCI is not surprising. These markers represent brain damage secondary to various processes, and therefore the association to cognitive state does not support specific mechanisms. These findings along with the association of other markers prior to adjustment (lobar microbleeds, hippocampal volume, temporal and occipital cortex thickness) may imply that PSCI is a result of mixed pathologies. The cumulative model (Fig. 1) demonstrates the increased risk for PSCI by the number of markers involved, strengthens this hypothesis and emphasizes the dramatic influence of combined pathologies on the risk for PSCI.

The strong association found between WMHV and PSCI is an important finding supporting vascular processes to have a key role in PSCI. Similar findings were previously described [29]. Lobar microbleeds were also associated with the cognitive state, possibly as a marker for cerebral amyloid angiopathy contribution. Nevertheless, this association was lost after adjustments. It should be noted that other cerebral small vessel disease markers were not found to be associated with PSCI in this study. Previous studies have demonstrated conflicting results examining the association between SVD markers and cognition. As SVD measures in neuroimaging constitute indirect markers, their sensitivity and specificity to estimate SVD contribution may not be sufficient, and they most probably do not represent the total burden of SVD. Future studies including pathological correlations or use of high definition 7T MRI may reflect the true burden of SVD and its contribution to PSCI. In order to further establish vascular processes contribution to PSCI, we evaluated the combined effect of vascular imaging markers and risk factor, which showed powerful prediction effect. In contrast, both ApoE4 and hippocampal volume were not significantly associated with PSCI in our cohort. Thus, it seems that although various pathologies contribute to PSCI, the vascular disease is the more significant underlying pathology.

Although our data show that vascular disease is a major contributor to PSCI, the acute lesion itself does not have a direct effect. The acute ischemic event is probably a marker representing the ongoing ischemic pathology, leading to high ischemic burden and PSCI. There is no doubt that acute ischemic stroke may facilitate cognitive impairment. However, as some patients in our cohort demonstrated progression of cognitive performances over a 2-year period, it is unlikely that the acute event is the cause for PSCI, and indeed no association was found. This is also supported by the lack of association between lesion location or volume and PSCI. Acute cognitive impairment may be a result of massive ischemic event or of stroke in specific (strategic) locations [30, 31]. However, our cohort included only mild to moderate stroke survivors. As for the possible thalamic-induced cognitive impairment, such effect was not demonstrated in our cohort, is probably a rare cause for PSCI, and has an immediate and not progressive effect.

This study strengths include the prospective cohort, repetitive extensive cognitive evaluations, and the vast range of MRI markers examined. Study limitations include the lack of follow-up imaging, the inclusion of only patients with mild clinical stroke manifestations and the missing of neuroimaging data for 110 patients (of note, these patients had similar incidence of PSCI, 19.2%, compared to those who had brain MRI scans at baseline, 20.2%).

Conclusion

Our findings show that PSCI is a result of mixed damage mechanisms, but point out that ongoing ischemic vascular processes are the main mechanism beyond PSCI. This emphasizes the importance of maximal treatment of cardiovascular risk factors and ischemic stroke prevention, also in order to prevent the high burden of morbidity caused by PSCI.

Footnotes

ACKNOWLEDGMENTS

This study is supported by grants RAG11482 from the American Federation for Aging Research, grant 2011344 From the U.S.- Israel Bi-national Science Foundation and grant AARG-16-442861 from the Alzheimer’s Association. These funding agencies had no role in the conduct and publication of this study.

Authors’ disclosures available online ().

References

Pendlebury

, Rothwell

, Oxford Vascular

(2019) Incidence and prevalence of dementia associated with transient ischaemic attack and stroke: Analysis of the population-based Oxford Vascular Study. Lancet Neurol 18, 248–258.

Pendlebury

, Rothwell

(2009) Prevalence, incidence, and factors associated with pre-stroke and post-stroke dementia: A systematic review and meta-analysis. Lancet Neurol 8, 1006–1018.

METACOHORTS Consortium (2016) METACOHORTS for the study of vascular disease and its contribution to cognitive decline and neurodegeneration: An initiative of the Joint Programme for Neurodegenerative Disease Research. Alzheimers Dement 12, 1235–1249.

Brainin

, Tuomilehto

, Heiss

, Bornstein

, Bath

, Teuschl

, Richard

, Guekht

, Quinn

, Post Stroke Cognition Study Group (2015) Post-stroke cognitive decline: An update and perspectives for clinical research. Eur J Neurol 22, 229–238, e213-226.

Ben Assayag

, Korczyn

, Giladi

, Goldbourt

, Berliner

, Shenhar-Tsarfaty

, Kliper

, Hallevi

, Shopin

, Hendler

, Baashat

, Aizenstein

, Soreq

, Katz

, Solomon

, Mike

, Usher

, Hausdorff

, Auriel

, Shapira

, Bornstein

(2012) Predictors for poststroke outcomes: The Tel Aviv Brain Acute Stroke Cohort (TABASCO) study protocol. Int J Stroke 7, 341–347.

Dale

, Fischl

, Sereno

(1999) Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage 9, 179–194.

Fischl

, Salat

, Busa

, Albert

, Dieterich

, Haselgrove

, van der Kouwe

, Killiany

, Kennedy

, Klaveness

, Montillo

, Makris

, Rosen

, Dale

(2002) Whole brain segmentation: Automated labeling of neuroanatomical structures in the human brain. Neuron 33, 341–355.

Fischl

, Sereno

, Dale

(1999) Cortical surface-based analysis. II: Inflation, flattening, and a surface-based coordinate system. Neuroimage 9, 195–207.

Woolrich

, Jbabdi

, Patenaude

, Chappell

, Makni

, Behrens

, Beckmann

, Jenkinson

, Smith

(2009) Bayesian analysis of neuroimaging data in FSL. Neuroimage 45, S173–186.

10.

Rosas

, Liu

, Hersch

, Glessner

, Ferrante

, Salat

, van der Kouwe

, Jenkins

, Dale

, Fischl

(2002) Regional and progressive thinning of the cortical ribbon in Huntington’s disease. Neurology 58, 695–701.

11.

Kuperberg

, Broome

, McGuire

, David

, Eddy

, Ozawa

, Goff

, West

, Williams

, van der Kouwe

, Salat

, Dale

, Fischl

(2003) Regionally localized thinning of the cerebral cortex in schizophrenia. Arch Gen Psychiatry 60, 878–888.

12.

, Motzkin

, Philippi

, Kirk

, Newman

, Kiehl

, Koenigs

(2012) Cortical thinning in psychopathy. Am J Psychiatry 169, 743–749.

13.

Fischl

, Dale

(2000) Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad Sci U S A 97, 11050–11055.

14.

Kliper

, Ben Assayag

, Tarrasch

, Artzi

, Korczyn

, Shenhar-Tsarfaty

, Aizenstein

, Hallevi

, Mike

, Shopin

, Bornstein

, Bashat

(2014) Cognitive state following stroke: The predominant role of preexisting white matter lesions. PLoS One 9, e105461.

15.

Artzi

, Aizenstein

, Jonas-Kimchi

, Myers

, Hallevi

, Ben Bashat

(2013) FLAIR lesion segmentation: Application in patients with brain tumors and acute ischemic stroke. Eur J Radiol 82, 1512–1518.

16.

Wardlaw

, Smith

, Biessels

, Cordonnier

, Fazekas

, Frayne

, Lindley

, O’Brien

, Barkhof

, Benavente

, Black

, Brayne

, Breteler

, Chabriat

, Decarli

, de Leeuw

, Doubal

, Duering

, Fox

, Greenberg

, Hachinski

, Kilimann

, Mok

, Oostenbrugge

, Pantoni

, Speck

, Stephan

, Teipel

, Viswanathan

, Werring

, Chen

, Smith

, van Buchem

, Norrving

, Gorelick

, Dichgans

, STandards for ReportIng Vascular changes on nEuroimaging (STRIVE v1) (2013) Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 12, 822–838.

17.

Molad

, Kliper

, Korczyn

, Ben Assayag

, Ben Bashat

, Shenhar-Tsarfaty

, Aizenstein

, Shopin

, Bornstein

, Auriel

(2017) Only white matter hyperintensities predicts post-stroke cognitive performances among cerebral small vessel disease markers: Results from the TABASCO Study. J Alzheimers Dis 56, 1293–1299.

18.

Fazekas

, Kleinert

, Offenbacher

, Schmidt

, Kleinert

, Payer

, Radner

, Lechner

(1993) Pathologic correlates of incidental MRI white matter signal hyperintensities. Neurology 43, 1683–1689.

19.

Wahlund

, Barkhof

, Fazekas

, Bronge

, Augustin

, Sjogren

, Wallin

, Ader

, Leys

, Pantoni

, Pasquier

, Erkinjuntti

, Scheltens

, European Task Force on Age-Related White Matter Changes (2001) A new rating scale for age-related white matter changes applicable to MRI and CT. Stroke 32, 1318–1322.

20.

Greenberg

, Vernooij

, Cordonnier

, Viswanathan

, Al-Shahi Salman

, Warach

, Launer

, Van Buchem

, Breteler

, Microbleed Study Group (2009) Cerebral microbleeds: A guide to detection and interpretation. Lancet Neurol 8, 165–174.

21.

Doubal

, MacLullich

, Ferguson

, Dennis

, Wardlaw

(2010) Enlarged perivascular spaces on MRI are a feature of cerebral small vessel disease. Stroke 41, 450–454.

22.

Nasreddine

, Phillips

, Bedirian

, Charbonneau

, Whitehead

, Collin

, Cummings

, Chertkow

(2005) The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. J Am Geriatr Soc 53, 695–699.

23.

Doniger

, Dwolatzky

, Zucker

, Chertkow

, Crystal

, Schweiger

, Simon

(2006) Computerized cognitive testing battery identifies mild cognitive impairment and mild dementia even in the presence of depressive symptoms. Am J Alzheimers Dis Other Demen 21, 28–36.

24.

Arnaiz

, Almkvist

, Ivnik

, Tangalos

, Wahlund

, Winblad

, Petersen

(2004) Mild cognitive impairment: A cross-national comparison. J Neurol Neurosurg Psychiatry 75, 1275–1280.

25.

Dwolatzky

, Whitehead

, Doniger

, Simon

, Schweiger

, Jaffe

, Chertkow

(2004) Validity of the Mindstreams computerized cognitive battery for mild cognitive impairment. J Mol Neurosci 24, 33–44.

26.

Stuart

, Turman

, Shaw

, Walsh

, Nguyen

(2003) Effects of aging on vibration detection thresholds at various body regions. BMC Geriatr 3, 1.

27.

Wolf

, D’Agostino

, Belanger

, Kannel

(1991) Probability of stroke: A risk profile from the Framingham Study. Stroke 22, 312–318.

28.

Adams

Jr. , Bendixen

, Kappelle

, Biller

, Love

, Gordon

, Marsh

3rd (1993) Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 24, 35–41.

29.

Molad

, Ben-Assayag

, Korczyn

, Kliper

, Bornstein

, Hallevi

, Auriel

(2018) Clinical and radiological determinants of transient symptoms associated with infarction (TSI). J Neurol Sci 390, 195–199.

30.

Gorelick

, Scuteri

, Black

, Decarli

, Greenberg

, Iadecola

, Launer

, Laurent

, Lopez

, Nyenhuis

, Petersen

, Schneider

, Tzourio

, Arnett

, Bennett

, Chui

, Higashida

, Lindquist

, Nilsson

, Roman

, Sellke

, Seshadri

, American Heart Association Stroke Council, Council on Epidemiology and Prevention, Council on Cardiovascular Nursing, Council on Cardiovascular Radiology and Intervention, and Council on Cardiovascular Surgery and Anesthesia (2011) Vascular contributions to cognitive impairment and dementia: A statement for healthcare professionals from the american heart association/american stroke association. Stroke 42, 2672–2713.

31.

Zhao

, Biesbroek

, Shi

, Liu

, Kuijf

, Chu

, Abrigo

, Lee

, Leung

, Lau

, Biessels

, Mok

, Wong

(2018) Strategic infarct location for post-stroke cognitive impairment: A multivariate lesion-symptom mapping study. J Cereb Blood Flow Metab 38, 1299–1311.