Factors Associated with Cognitive Outcomes After First-Ever Ischemic Stroke: The Impact of Small Vessel Disease Burden and Neurodegeneration

Abstract

Background:

Differences exist regarding post-stroke cognitive outcomes.

Objective:

The aim of this study investigates the potential factors associated with post-stroke cognitive performance and trajectories.

Methods:

We performed a prospective cohort study using serial monitoring of cognitive function over a 1-year period after a first-ever ischemic stroke. Small vessel disease (SVD) burden and hippocampal atrophy (HA) were evaluated using the modified cerebral small vessel disease scores (mCSVD) and medial temporal atrophy score (MTA) scores. A generalized estimating equation (GEE) model and a group-based trajectory model (GBTM) was used to analyze the potential factors associated with post-stroke cognitive outcomes.

Results:

A total of 112 patients were enrolled. The GEE model showed that all patients, regardless of initial cognitive performance, had a tendency to show an increase in the Montreal Cognitive Assessment over time. The cognitive performance was better in male patients with higher education levels (p = 0.046 and p < 0.001, respectively), but tended to be worse in patients with higher SVD burden and HA. The GBTM model grouped patients into low, intermediate, and high performance (LP, IP, and HP) after stroke. A higher SVD burden, rather than HA and initial stroke severity and location, independently predicted a higher odds of poor post-stroke cognitive trajectory (being in the LP group) after stroke (adjusted odds ratio 2.74, 95%CI 1.09–6.86).

Conclusion:

In patients with first-ever mild stroke, cognitive improvement over time was evident. The detrimental impact of the SVD burden may outweigh the effect of HA or acute stroke insult on the post-stroke cognitive trajectory during the 1-year follow-up.

Keywords

Cerebral small vessel diseaseburden hippocampal atrophy post-stroke cognitivetrajectory post-stroke dementia

INTRODUCTION

Stroke is the most common disease in the field of neurology, but its prevalence and prognosis differ between races, ages, and sexes [1]. Stroke causes numerous morbidities that result in a high socioeconomic burden for the patient, family, and society. Among the commonly recognized post-stroke disabilities, post-stroke cognitive impairment (PSCI) or dementia (PSD) is often being neglected [2]. PSD may result in a 7-fold risk of mortality, independent of age and other underlying diseases [3]. The prevalence of PSD varies widely across different study designs [4 –6], and some data suggest that cognitive impairment progressively worsens [7] after an acute stroke, even when physical disability improves.

Possible risk factors for PSCI/PSD include individual unmodifiable demographical features (age, sex, educational level, and genetic background), the characteristics of the index stroke (stroke localization, size, and severity), and some modifiable or controllable systemic comorbidities (diabetes mellitus, dyslipidemia, atrial fibrillation, hypertension, and chronic kidney disease) [7, 8]. The baseline brain reserve [9] at stroke may also influence the onset of PSCI/PSD. Chronic brain pathological changes, such as Alzheimer’s disease pathology or small vessel disease (SVD) changes, contribute to individual differences in baseline brain reserve, which diminishes the brain resilience to the detrimental effect of stroke on cognitive function [9]. Past studies have shown that leukoaraiosis and hippocampal atrophy (HA) contribute 2.5-fold and 2.7-fold to the risk of developing PSD, respectively [10]. However, an interesting issue may be the interplay between baseline brain reserve and acute stroke insult on PSCI and post-stroke cognitive trajectory. The present study evaluated the potential factors associated with post-stroke cognitive outcomes, including the cognitive performance over time and general cognitive trajectory after stroke during a 1-year follow-up period, especially the impact of the baseline status of SVD burden and HA on these outcomes.

METHODS

Study participants

This prospectively registered study was performed at the National Cheng Kung University Hospital (NCKUH). The study started in February 2015. All of the study participants were recruited from patients who were admitted consecutively to the Department of Neurology with a first-ever ischemic stroke. The diagnosis of ischemic stroke was confirmed by hyperintensity lesions on acute diffusion weighted magnetic resonance imaging (MRI). The following exclusion criteria were used: patients with stroke with other specific etiologies, such as vascular dissection, autoimmune disease, hypercoagulable state or hematologic disorders [11], for which stroke recurrence may be different; unstable clinical conditions, clinical symptoms or complications leading to poor cooperation or failure to receive a cognitive assessment during acute stroke; conscious disturbance or delirium, hemineglect, inattention, or mutism; moderate to severe vision or hearing loss leading to poor cooperation; documented premorbid cognitive dysfunction or psychiatric illness; premorbid brain insults or neurodegenerative disorders; and refusal to join the study. The study was approved by the Institutional Ethics Review Board of NCKUH (B-ER-104-134).

A total of 112 patients were enrolled according to the inclusion and exclusion criteria (Supplementary Figure 1). According to the exclusion criteria, most patients with moderate to severe stroke were excluded. The median NIHSS of our study cohort was 3.77 (IQR: 1.75–5).

Clinical assessment at admission and during follow-up

During admission for acute stroke, detailed social and demographic data were collected, including past functional status. Surveillance for stroke etiology and brain MRI were performed within 7 days after stroke onset. Blood samples from the study participants were also analyzed for the apolipoprotein E (APOE) genotype. The clinical follow-up schedule was arranged within an interval of at least 3 months after discharge. Prior to enrollment, a neurologist (P.S.S.) reviewed the medical history of all patients regarding past mental status, cognitive status, and functional status to exclude premorbid dementia, psychiatric illness, and the presence of obvious clinical symptoms of neurodegenerative disorder before index stroke. All patients were assessed for premorbid dementia, post-stroke dementia or depression during follow-up based on the criteria of the Diagnostic and Statistical Manuals of Mental Disorders 4th Edition (DSM-IV).

Neuropsychological assessment was arranged at baseline during the acute stage of stroke within 7 days of onset. To assess the cognitive outcomes after stroke, the second and last follow-up neuropsychological interviews were arranged for 3 months and 1 year after the index stroke. The validated Taiwanese version of the Montreal Cognitive Assessment (MoCA), which is a 10-minute cognitive screening tool, was used to assess general cognition. All patients underwent a battery of neuropsychological assessments based on the recommendation of NINDS-Canadian Stroke network vascular cognitive impairment harmonization standards [12]. These assessments included the Weschler Adult Intelligence Scale III (Similarities, Arithmetic, Matrix Reasoning, Digit Span, Working memory), the Wechsler Memory Scale III (Logic_memory I & Logic_memory II), the Semantic Association Of Verbal Fluency Test for semantic verbal fluency [13], and the Wisconsin Card Sorting Test to evaluate different cognitive domains, including executive function, memory, language, attention, visuospatial function, and mental speed. A single trained psychologist performed all of the neuropsychological tests. All patients completed the second neuropsychological assessment, but 14 patients (12.5%) refused the third follow-up assessment for several reasons, such as economic problems, family denial, personal factors or physical discomfort.

Assessment of the baseline SVD burden and HA

All patients underwent MRI examination to determine stroke location, vascular occlusion, and the baseline state of SVD burden and HA. Stroke location was assessed using diffuse weighted imaging (DWI). Trained neurologists assessed stroke severity using the National Institute of Health Stroke Scale (NIHSS) [14].

The baseline state of SVD burden was determined using the modified cerebral small vessel disease score (mCSVD) and the degree of HA was evaluated using the medial temporal lobe atrophy score (MTA). The mCSVD score is an extension of the CSVD score, which was proposed by Julia Staals et al. [15, 16]. A score from 0–4 was assigned to determine the SVD burden by counting the presence of each of the 4 MRI features of SVD. A point was given when there was any presentation of a previous lacunar infarction, microbleeds, moderate to severe perivascular space at the ganglionic level or a deep white matter Fazekas score≥2 [15]. However, later studies assigned different weights to some of these features [16]. Perivascular spaces with scores between 11–20 and > 20 were assigned scores of 1 and 2, respectively. Microbleeds were assigned a score of 1 only when they demonstrated a definite count > 5. Using these modifications, a high SVD burden was defined as a score≥3. The severity of HA was assessed by the MTA score designed by Scheltens [17], which focuses on the width of the choroid fissure, the width of the temporal horn and the height of hippocampal formation. Scores ranged from 0 to 4, with higher scores indicating a higher degree of HA. When correlated with normal aging, abnormal atrophy was defined as a score≥3 for patients younger than 75 years old and≥2 for patients over 75 years. When there was disproportionate atrophy from one side to the other, the atrophy with the highest score was recorded. To obtain objective data on the mCSVD score and MTA score and avoid bias from a single rater, in our study, a committee of two independent readers (K.P.L, P.Y.L) who were blinded to the cognitive results evaluated both scales, and the final scores of the mCSVD and MTA were given to each patient after discussion of the decision-making committee.

Post-stroke cognitive outcomes and statistical analysis

For demographics and clinical information, categorical variables are presented as numbers (percentages), and continuous variables are presented as the means (standard deviations) or medians (first quartile-third quartile). Due to the nonparametric distribution of these data, we used the chi-squared test or Fisher’s exact test for categorical variables and Kruskal-Wallis test for continuous variables to compare the differences between groups. The present study used two statistical methods to assess the impact of potential clinical and imaging parameters representing brain resilience on two post-stroke cognitive outcomes. We used a generalized estimating equation (GEE) model to examine the cognitive performance over time assessed by the serial MoCA tests during the 1-year follow-up period. The dependent variable was the serial MoCA scores, and the independent variables were the clinical and imaging parameters, including age, sex, education level, APOE genotype, the presence of comorbidities and depression, the severity of stroke, lesion involvement of the cortical region, the mCSVD score and the MTA score. These factors were reported to be associated with post-stroke cognitive function as mentioned above. We also assessed the serial MoCA changes with different time points. The calculated coefficient (slope) of time from the GEE model reflected the cognitive change during follow-up. We further analyzed the slopes for the participants with initial MoCA scores greater than or less than 24 points and assessed the impact of the abovementioned clinical factors on the difference in cognitive performance in patients with different initial MoCA scores. All of these factors were conceived as covariates and entered into the multivariable models.

Then, we used the group-based trajectory model (GBTM) and classified patients’ general post-stroke cognitive performances with similar statistical trajectories over 1 year after stroke according to the follow-up MoCA score data. SAS 9.4 was used to analyze the trajectory system. The “Proc Traj” command, which is compatible with GBTMs, was used, and the option “Normal” was selected to identify changes in the pattern of the trajectory [18, 19]. The Bayesian information criterion (BIC) was used as the criterion for model selection, and a smaller BIC value indicated a better fitting model. Multivariable logistic regression analysis was performed to examine the independent effects of age, sex, education, APOE genotype, the presence of depression and comorbidities, initial stroke severity, lesion involvement of the cortical region, mCSVD score and MTA score on the general cognitive trajectories after stroke. Factors with a p-value less than 0.05 in univariate analysis were entered into multivariable analysis to assess the association after adjusting for the abovementioned potential confounding factors. Similar methods were performed to analyze the impact of abovementioned factors on the trajectories of different domains of neuropsychological assessments.

All analyses were performed using the statistical software package SAS version 9.4 (SAS Institute Inc., Cary, North Carolina). A p-value less than 0.05 was defined as statistically significant.

RESULTS

Of the 112 patients, 72 were men (64.3%). The median age was 64.5 (IQR 57.0–73.5) years, and the educational level was 9.0 (IQR 6.0–12.0) years (Supplementary Table 1).

According to the GEE model (Table 1), all patients had an increase in MoCA scores over time during follow-up (p < 0.001). MoCA scores tended to increase in male patients of a younger age, with higher education, less stroke severity, and lack of stroke lesions involving the cortical region, APOE genotype, comorbidities, and the presence of depression. However, the cognitive performance were significantly better only in male patients and patients with higher education levels (p = 0.046 and p < 0.001, respectively). Although the MoCA scores tended to decrease in patients with a higher CSVD burden and abnormal HA, there was no significant difference between groups (Table 1). All patients, regardless of initial MoCA performance at the acute stage of stroke, had a tendency to show an increase in MoCA scores over time, but the trend was only significant in patients with initial MoCA < 24 after adjustment (p < 0.001) (Table 2). Age, sex, the presence of comorbidities and depression, the severity of stroke, and stroke lesions involving the cortical region did not change the course of cognitive performance in patients with different initial MoCA scores. The presence of APOE ɛ4 was associated with a decrease in cognitive performance in patients with an initial higher MoCA, and the association between high education levels and better cognitive performance over time remained similar between groups. A higher dementia occurrence rate was noted in patients with an initial MoCA less than 24 points (2.5%versus 6.9%in patients with initial MoCA ≥24 and < 24 points, respectively).

Table 1

The cognitive performance over time assessed by the generalized estimating equation (GEE) model among patients with different clinical factors, stroke etiologies, severity and location, and different severity of CSVD burden and HA

Clinical Variable	Multivariable model		Variable regarding index stroke and baseline mCSVD and HA burden	Multivariable model
	β±SE	p		β ±SE	p
Time^a	0.84±0.19	< 0.001	NIHSS scores at admission ^a	–0.09±0.13	0.496
Sex^a			Stroke location ^a
Female	Ref.		Lacunar infarction	Ref.
Male	2.61±1.31	0.046	Cortex involvement	–1.40±0.94	0.139
Age^a			Stroke etiologies ^a
< 65	Ref.		Large artery atherosclerosis	Ref.
≥65	–1.63±0.94	0.081	Cardioembolism	1.91±1.75	0.274
Education, years^a	0.63±0.13	< 0.001	Small vessel occlusion	1.49±2.24	0.506
APOE ɛ4 status^a			Undetermined etiology	0.86±2.02	0.672
0 ɛ4 carrier	Ref.		Modified CSVD burden ^c
1 or 2 ɛ4 carrier	–1.03±1.12	0.355	< 3	Ref.
Comorbidities^a			≥3	–0.99±1.22	0.419
No	Ref.		HA adjusted by age ^d
Yes	–1.63±1.32	0.217	No	Ref.
Depression^a			Yes	–1.89±1.41	0.180
No	Ref.		StatusofmCSVDandHA ^e
Yes	–5.10±3.62	0.159	mCSVD < 3 and no HA	Ref.
			mCSVD ≥ 3 and HA	–0.50±1.28	0.698
			mCSVD < 3 and HA	–1.29±1.42	0.365
			mCSVD ≥ 3 and HA	–2.85±2.05	0.164

^aWorking structure of model is adopted exchangeable structure,, QIC value is 246.2. ^bWorking structure of model is adopted unstructured structure, QIC value is 246.2. ^cWorking structure of model is adopted exchangeable structure, QIC value is 251.6. ^dWorking structure of model is adopted unstructured structure, QIC value is 251.4. ^eWorking structure of model is adopted unstructured structure, QIC value is 257.5. CSVD, cerebral small vessel disease; HA, hippocampal atrophy.

Table 2

Cognitive performance over time assessed by the generalized estimating equation (GEE) model among patients with different initial cognitive performance

	Multivariable model^a MOCA < 24 (N = 72)		Multivariable model^b MOCA ≥ 24 (N = 40)
MoCA scores at acute stage of stroke Worse MoCA during follow-up^c, N (%)	14 (19.4)		13 (32.5)
	β±SE	p	β±SE	p
Time	1.23±0.25	< 0.001	0.29±0.24	0.231
Sex
Female	Ref.		Ref.
Male	1.70±1.72	0.322	–0.09±0.82	0.916
Age
< 65	Ref.		Ref.
≥65	–0.33±1.32	0.805	–0.18±0.79	0.817
Education, years	0.53±0.15	< 0.001	0.22±0.10	0.036
APOE ɛ4 status
0 ɛ4 carrier	Ref.		Ref.
1 or 2 ɛ4 carrier	0.64±1.23	0.604	–1.82±0.85	0.034
Comorbidities
No	Ref.		Ref.
Yes	–0.59±1.86	0.751	0.47±0.73	0.514
Depression
No	Ref.		Ref.
Yes	–3.47±3.31	0.295	–	–
NIHSS scores at admission	0.07±0.16	0.665	–0.12±0.14	0.394
Stroke location
Lacunar infarction	Ref.	Ref.
Cortex involvement	–0.85±1.46	0.558	–1.43±0.75	0.058

^aWorking structure of model is adopted exchangeable structure, QIC value is 94.5. ^bWorking structure of model is adopted AR(1) structure, QIC value is 116.9. ^cWorse MoCA was defined as that follow-up final MoCA score worse than the initial MoCA score. MoCA, Montreal Cognitive Assessment.

The GBTM classified general cognitive performance post-stroke into three subgroups: high performance (HP), intermediate performance (IP), and low performance (LP), which included 49, 48, and 15 participants, respectively (Fig. 1). A significant difference was discovered in the distributions of sex (p = 0.004), age (p = 0.001), and educational level (p < 0.001), but no significant difference was found in the underlying comorbidities between groups (Table 3). Stroke severity and lesion involvement of the cortical region were similar between groups. No difference was found regarding the proportion of patients with higher mCSVD score and abnormal HA between groups. A higher incidence of dementia was also noted in the LP group (HP:IP:LP = 0 versus 8.33 versus 13.33%; p = 0.029). Multivariable logistic regression revealed that a higher mCSVD score (adjusted odds ratio (aOR) 2.74, 95%CI 1.09–6.86, p = 0.032) but not an abnormal MTA score (aOR 1.53, 95%CI 0.56–4.21, p = 0.405) independently predicted classification into the LP group. A combination of a higher mCSVD score and an abnormal MTA score resulted in the highest probability of classification in the LP group (aOR 4.18, 95%CI 1.05–16.66, p = 0.043). Higher educational level was a protector (aOR 0.79, 95%CI 0.70–0.88, p < 0.001), but sex, age, APOE status, the presence of depression during follow-up, initial stroke severity, and stroke location had a nonsignificant impact on the post-stroke general cognitive trajectory (Table 4).

Fig. 1

The distribution of cognitive trajectory over 1-year period after index stroke. The figure showed the poststroke cognitive trajectory in HP (solid line 3), IP (solid line 2), and LP (solid line 1) group. The serial MoCAs was shown in each time point (mean±SD). HP, high performance; IP, intermediate performance; LP, low performance; MoCA, the Montreal Cognitive assessment; Solid line, cognitive trajectory; Dash line, 95%CI of each trajectory in serial time points.

Table 3

Baseline characteristics of patients with different cognitive trajectory group assessed by the group-based trajectory model after acute stroke

	Cognitive trajectory during 1-year follow-up			p^e
	LP (N = 15)	IP (N = 48)	HP (N = 49)
Variables	Number (%)	Number (%)	Number (%)
Sex
Male	4 (26.67)	32 (66.67)	36 (73.47)	0.004*
Age
< 65	4 (26.67)	19 (39.58)	33 (67.35)	0.004*
≥65	11 (73.33)	29 (60.42)	16 (32.65)
Median (IQR)	72.0 (64.0, 78.0)	66.0 (60.0, 77.0)	61.0 (55.0, 68.0)	0.001^*
Education
Median (IQR)	1.0 (0.0, 6.0)	6.0 (6.0, 11.0)	12.0 (9.0, 12.0)	< 0.001^*
APOE ɛ4 status^a
0 ɛ4 carrier	11 (91.67)	25 (80.65)	38 (92.68)	0.269
1–2 ɛ4 carrier	1 (8.33)	6 (19.35)	3 (7.32)
Comorbidities
DM	8 (53.33)	20 (41.67)	12 (24.49)	0.065
HTN	13 (86.67)	37 (77.08)	40 (81.63)	0.686
HL	9 (60.00)	38 (79.17)	41 (83.67)	0.147
AF	2 (13.33)	5 (10.42)	2 (4.17)	0.274
ESRD	0 (0.00)	3 (6.25)	1 (2.04)	0.644
Dementia during follow-up	2 (13.3)	4 (8.3)	0 (0)	0.029^f
Depression during follow-up	1(6.67)	2(4.35)	4(8.51)	0.868
NIHSS at admission^b
≤5	11 (78.57)	35 (77.78)	36 (80.00)	1.000
6–13	3 (21.43)	9 (20.00)	9 (20.00)
≥14	0 (0.00)	1 (2.22)	0 (0.00)
Median (IQR)	4.0 (2.0, 5.0)	3.0 (2.0, 5.0)	3.0 (1.0, 5.0)	0.675
Stroke location^c
Lacunar infarction	9 (60.00)	26 (55.32)	34 (73.91)	0.175
Cortex involvement	6 (40.00)	21 (44.68)	12 (26.09)
Modified CSVD Score
< 3	7 (46.67)	29 (60.42)	36 (73.47)	0.126
≥3	8 (53.33)	19 (39.58)	13 (26.53)
HA adjusted by age
No	9 (60.00)	38 (79.17)	42 (85.71)	0.097
Yes	6 (40.00)	10 (20.83)	7 (14.29)
Group^d
C–H–	6 (40.00)	24 (50.00)	31 (63.27)	0.129
C + H–	3 (20.00)	14 (29.17)	11 (22.45)
C–H+	1 (6.67)	5 (10.42)	5 (10.20)
C + H+	5 (33.33)	5 (10.42)	2 (4.08)

^aAPOE E4 was not tested in 3, 17, and 8 patients in each group. ^bNIHSS in admission was missed in 1, 3, and 4 patients in each group. ^cStroke location was not shown on DWI in 0, 1, and 3 patients in each group, possibly due to rapid resolution of cerebral ischemia. ^dC: Modified cerebral small vessel disease (CSVD) scores (–: < 3;+: ≥3); H: Hippocampal atrophy adjusted by age. C-H-: patients do not have mCSVD ≥3 and hippocampal atrophy adjusted by age. C-H+: patients do not have mCSVD ≥3, but have hippocampal atrophy adjusted by age. C + H-: patients have mCSVD ≥3, but do not have hippocampal atrophy adjusted by age. C + H+: patients have mCSVD ≥3 and hippocampal atrophy adjusted by age. ^eChi-square test or Fisher’s exact test for categorical variables/ANOVA test or Kruskal-Wallis test for continuous variables. ^fFisher’s exact test. LP, low performance; IP, intermediate performance; HP, high performance; DM, diabetes mellitus; HTN, hypertension; HL, hyperlipidemia; AF, atrial fibrillation; ESRD, end-stage renal disease; CSVD, cerebral small vessel disease; HA, hippocampal atrophy.

Table 4

Potential predictors to predict patient being the low performance group in the group-based trajectory model after first-ever ischemic stroke

	Crude OR (95 %CI)	p	Adjusted OR^b (95 %CI)	p
Modified CSVD
< 3	Ref.		Ref.
≥3	2.93 (1.29–6.64)	0.01*	2.74 (1.09–6.86)	0.032^*
HA adjusted by age
No	Ref.		Ref.
Yes	2.41 (0.99–5.79)	0.05*	1.53 (0.56–4.21)	0.405
Group^a
C–H–	Ref.		Ref.
C + H–	3.13 (1.25–7.87)	0.015*	3.08 (1.09–8.66)	0.033^*
C–H+	3.54 (0.56–22.23)	0.178	2.38 (0.30–18.87)	0.411
C + H+	6.29 (1.90–20.83)	0.003*	4.18 (1.05–16.66)	0.043^*
Gender
Female	Ref.		Ref.
Male	0.35 (0.17–0.75)	0.007*	0.44 (0.19–1.06)	0.066
Age group
< 65	Ref.		Ref.
≥65	3.50 (1.66–7.35)	0.001*	1.83 (0.78–4.32)	0.165
Education
years	0.75 (0.67–0.83)	< 0.001*	0.79 (0.70–0.88)	< 0.001*
APOE ɛ4 status
0 ɛ4 carrier	Ref.
1–2 ɛ4 carrier	1.66 (0.48–5.72)	0.422
Depression
No	Ref.
Yes	0.62 (0.14–2.76)	0.531
Comorbidities
No	Ref.
Yes	1.25 (0.80–1.96)	0.325
NIHSS at admission
Per 1 points	1.01 (0.90–1.14)	0.818
NIHSS group at admission
≤5	Ref.
>5	1.10 (0.45–2.68)	0.831
Stroke location
Lacunar infarction	Ref.
Cortex involvement	1.85 (0.87–3.91)	0.109

^aC: Modified CVSD (–: < 3;+: ≥3); H: Hippocampal atrophy adjusted by age. C-H-: patients do not have mCSVD ≥3 and hippocampal atrophy adjusted by age. C-H+: patients do not have mCSVD ≥3, but have hippocampal atrophy adjusted by age. C + H-: patients have mCSVD ≥3, but do not have hippocampal atrophy adjusted by age. C + H+: patients have mCSVD ≥3 and hippocampal atrophy adjusted by age. ^bMultivariable ordinal logistic regression analysis of variables (p < 0.05 in univariate logistic regression analysis). CSVD, cerebral small vessel disease; HA, hippocampal atrophy.

Separate analyses were performed to investigate the trajectories of different cognitive subtests during follow-up, as shown in Supplementary Figure 2. The GBTM classified the data of various post-stroke cognitive subtests into two subgroups: HP (high performance) and LP (low performance) groups. Multivariable logistic regression revealed that education level played a protective role in predicting classification in the HP group for most cognitive subtests. A higher mCSVD independently predicted classification into the LP group in the subtests of immediate recall and executive function (immediate recall, aOR 3.46, 95%CI 1.40–8.54, p = 0.007; executive function, OR 3.55, 95%CI 1.32–9.55, p = 0.012) (Supplementary Tables 2–10). The abnormal MTA scores showed no significant impact on the cognitive trajectory of various cognitive subtests, including the memory subtests.

DISCUSSION

In patients with mild first-ever ischemic stroke, the present study demonstrated that post-stroke cognitive improvement was noted during the 1-year follow-up. Male patients and patients with higher educational levels were significantly associated with better cognitive performance over time. The GEE model showed that stroke severity, stroke lesions involving the cortical region, stroke etiologies, higher CSVD burden, which represent the baseline total brain damage associated with SVD, and age-disproportional HA did not contribute to a significant difference in post-stroke cognitive performance during follow-up. However, the cognitive trajectory model showed that baseline higher CSVD burden, rather than age-disproportional HA, significantly predicted a patient’s being poor general cognitive trajectory after stroke.

Our GEE model showed that male patients and patients with higher education levels may have better cognitive performance over time. Education may represent the baseline cognitive reserve of stroke patients [9]. Higher cognitive reserve may potentially lower the risk of PSD development [9, 10]. Regarding the sex difference on post-stroke cognitive performance, men were reported to have a reduced risk of PSD and a slower cognitive decline rate compared with women [10, 20]. A variety of mechanisms are proposed to explain the impact of sex on cognitive decline, including differences in genetics, lifestyle factors, sex hormones, and the brain structure, such as a higher burden of SVD and lower gray matter volume in women [21 –24].

The first cognitive assessments were arranged within 7 days of the index stroke in our study. Previous studies have shown that most patients experience an acute decline following a stroke, and post-stroke improvement of cognitive function is observed after several weeks [25, 26]. Our GEE model demonstrated that in all patients, regardless of initial MoCA performance, there was a tendency to show an increase in MoCA scores over time, indicating that regardless of initial cognitive performance, patients may generally have potential improvement in cognitive function. In addition, cognitive improvement was more significant in patients with an early MoCA score lower than 24 points, which may be contributed by the ceiling effect of patients with higher early MoCA scores during the acute stage of stroke. The dementia occurrence rate was still higher in patients with early lower MoCA scores, and the trajectory model also found that patients in the LP group had the highest risk of PSD development during follow-up than patients in the IP or HP groups. These findings were consistent with a previous study [27]. Although international guidelines suggest performing cognitive assessments for all stroke survivors, the arrangement of cognitive screening tests during follow-up is not routine for stroke survivors in real-world practice. Those with early lower MoCA scores and poor general cognitive performance after stroke should receive greater attention during follow-up.

A previous validation study of the Taiwanese version of the MoCA showed that the cutoff point for mild cognitive impairment was 23/24 [28]. In our study, patients in the IP and LP groups from the acute stage to the 1-year follow-up were all in the mild cognitive impairment stage according to this cutoff level. Although a higher percentage of patients were diagnosed with dementia in the IP and LP groups during follow-up, a proportion of patients in the LP group still remained independent for activities of daily living, which does not satisfy the criteria of dementia from the DSM-IV. This result may be explained by the lower education level and older age in the LP group in our study because performance on the MoCA is highly dependent on education level [29]. The general educational level of elderly people in southern Taiwan is between 0 and 6 years. The mean educational level in the LP group was only 1 year in our study, which explains their low MoCA scores. However, they continued to live independently without any difficulty in activities of daily living.

Our study found that a higher burden of CSVD, rather than abnormal HA, on baseline brain imaging predicted a higher odds of poor general cognitive trajectory after stroke. The combined effect of CSVD and HA had the highest odds of a poor general cognitive trajectory, even after adjustments for age, educational level, the involvement of cortical lesions, stroke severity, and other potential clinical factors. CSVD is a major vascular contributor to cognitive impairment and dementia. The higher burden of CSVD or severe HA represents poor baseline brain reserve and CSVD is also related to early onset PSD [9]. Our findings may support this point of view. However, our GEE model did not find significantly different cognitive performance over time between patients with different levels of CSVD severity, which may be related to the short follow-up interval since the rate of cognitive decline in patients with a higher CSVD burden is slow (a previous study showed little change in cognition over a 2- to 3-year period in patients with a severe SVD burden [30]). In addition to subcortical lesions, a higher CSVD burden may be negatively associated with reduced cortical thickness [31] and fronto-parieto-occipital gray matter atrophy [32], and may also be related to the severity of HA [33]. HA is a predictor of AD conversion, especially in patients of younger age, because HA in older populations is more often an age-related atrophy [17]. To exclude the age effect, the severity of HA was corrected for age in our study. The clinical meaning of HA may be more complicated in PSD because the vascular burden may also be related to HA severity as mentioned above. Contradictory findings on the impact of HA on PSD development were reported [33, 34]. These differences may be because HA is a chronic and complicated process and may have developed prior to clinical cognitive decline over a period of 3 to 5 years [35]. Therefore, the 1-year period in our study may not have been sufficient to observe the negative effects of HA on the post-stroke cognitive slope changes. However, following the application of amyloid PET, the in vivo detection of amyloid pathology after stroke was associated with an accelerated rate of post-stroke cognitive decline [36], which may be a more accurate predictor than HA. Notably, our study corroborates the synergistic effect of CSVD and HA on poor cognitive trajectory after stroke. Therefore, clinicians should pay closer attention to stroke patients with a high burden of CSVD either alone or in combination with HA.

There are some limitations in our study. First, our study was performed at a single center. The population was relatively small due to our strict exclusion criteria and inefficient patient recruitment due to the complex study design. The limited patient number may also potentially lead to type II error during statistical analysis. Then, we did not perform genetic tests for all patients, and 3, 17, and 8 patients dropped out of the LP, IP, and HP groups, respectively, due to refusal to continue with the study. The observation period may also not have been of sufficient length to observe complete cognitive transition, especially with regard to evaluating the correlation between HA, CSVD, and the cognitive slopes. We used the clinical diagnosis of depression as our adjusting factor rather than a screening test for depression at each time point. Finally, we primarily recruited patients with mild stroke, limiting the generalizability of the results. However, in patients without dementia-prone lesions (that is, large territorial infarcts, strategic infarcts, multiple or extensive infarctions), identifying the predictors or elucidating the underlying mechanism of poor cognitive trajectory or worse cognitive changes is more important since further preventive strategies against cognitive deterioration may be adopted or developed.

CONCLUSION

Cognitive recovery is evident after the acute stroke phase. Male patients and patients with higher education levels may have better cognitive performance over time after stroke. Among patients with mild stroke, the baseline CSVD burden, rather than HA, stroke lesions involving the cortical region, and stroke severity, is directly and significantly associated with a poor general cognitive trajectory. The combination of CSVD and HA had the largest negative impact on the general cognitive trajectory after stroke.

Footnotes

ACKNOWLEDGMENTS

We are grateful to the Biostatistics Consulting Centre, National Cheng Kung University Hospital for providing us with statistical consulting services.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Fitzmaurice

, Abate

, Abbasi

, Abbastabar

, Abd-Allah

, Abdel-Rahman

, Abdelalim

, Abdoli

, Abdollahpour

, Abdulle

ASM

, et al. (2019) Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 29 cancer groups, 1990 to 2017: A systematic analysis for the global burden of disease study. JAMA Oncol 5, 1749–1768.

Pollock

, St George

, Fenton

, Firkins

(2012) Top ten research priorities relating to life after stroke. Lancet Neurol 11, 209.

Gao

, Lian

, Zhang

, Fang

, Deng

, Li

, Xu

, Zhou

, Wang

(2016) Association of dementia with death after ischemic stroke: A two-year prospective study. Exp Ther Med 12, 1765–1769.

Madureira

, Guerreiro

, Ferro

(2001) Dementia and cognitive impairment three months after stroke. Eur J Neurol 8, 621–627.

Jacova

, Pearce

, Costello

, McClure

, Holliday

, Hart

, Benavente

(2012) Cognitive impairment in lacunar strokes: The SPS3 trial. Ann Neurol 72, 351–362.

Makin

SDJ

, Turpin

, Dennis

, Wardlaw

(2013) Cognitive impairment after lacunar stroke: Systematic review and meta-analysis of incidence, prevalence and comparison with other stroke subtypes. J Neurol Neurosurg Psychiatry 84, 893–900.

Mijajlovic

, Pavlovic

, Brainin

, Heiss

, Quinn

, Ihle-Hansen

, Hermann

, Assayag

, Richard

, Thiel

, Kliper

, Shin

, Kim

, Choi

, Jung

, Lee

, Sinanovic

, Levine

, Schlesinger

, Mead

, Milosevic

, Leys

, Hagberg

, Ursin

, Teuschl

, Prokopenko

, Mozheyko

, Bezdenezhnykh

, Matz

, Aleksic

, Muresanu

, Korczyn

, Bornstein

(2017) Post-stroke dementia - a comprehensive review. BMC Med 15, 11.

G-C

, Chen

Y-M

(2017) Post-stroke dementia: Epidemiology, mechanisms and management. Int J Gerontol 11, 210–214.

Mok

, Lam

, Wong

, Ko

, Markus

, Wong

(2017) Early-onset and delayed-onset poststroke dementia - revisiting the mechanisms. Nat Rev Neurol 13, 148–159.

10.

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.

11.

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.

12.

Hachinski

, Iadecola

, Petersen

, Breteler

, Nyenhuis

, Black

, Powers

, DeCarli

, Merino

, Kalaria

, Vinters

, Holtzman

, Rosenberg

, Wallin

, Dichgans

, Marler

, Leblanc

(2006) National Institute of Neurological Disorders and Stroke-Canadian Stroke Network vascular cognitive impairment harmonization standards. Stroke 37, 2220–2241.

13.

Hua

, Chang

, Chen

(1997) Factor structure and age effects with an aphasia test battery in normal Taiwanese adults. Neuropsychology 11, 156–162.

14.

Goldstein

, Bertels

, Davis

(1989) Interrater reliability of the NIH stroke scale. Arch Neurol 46, 660–662.

15.

Staals

, Makin

, Doubal

, Dennis

, Wardlaw

(2014) Stroke subtype, vascular risk factors, and total MRI brain small-vessel disease burden. Neurology 83, 1228–1234.

16.

Lau

, Li

, Schulz

, Simoni

, Chan

, Ho

, Cheung

RTF

, Küker

, Mak

HKF

, Rothwell

(2017) Total small vessel disease score and risk of recurrent stroke: Validation in 2 large cohorts. Neurology 88, 2260–2267.

17.

Scheltens

, Leys

, Barkhof

, Huglo

, Weinstein

, Vermersch

, Kuiper

, Steinling

, Wolters

, Valk

(1992) Atrophy of medial temporal lobes on MRI in “probable” Alzheimer’s disease and normal ageing: Diagnostic value and neuropsychological correlates. J Neurol Neurosurg Psychiatry 55, 967–972.

18.

Jones

, Nagin

, Roeder

(2001) A SAS procedure based on mixture models for estimating developmental trajectories. Sociol Methods Res 29, 374–393.

19.

Wang

, Hung

, Li

(2018) Development trajectories and predictors of the role commitment of nursing preceptors. J Nurs Res 26, 168–176.

20.

Levine

, Gross

, Briceño

, Tilton

, Giordani

, Sussman

, Hayward

, Burke

, Hingtgen

, Elkind

MSV

, Manly

, Gottesman

, Gaskin

, Sidney

, Sacco

, Tom

, Wright

, Yaffe

, Galecki

(2021) Sex differences in cognitive decline among US adults. JAMA Netw Open 4, e210169.

21.

Mielke

, Vemuri

, Rocca

(2014) Clinical epidemiology of Alzheimer’s disease: Assessing sex and gender differences. Clin Epidemiol 6, 37–48.

22.

Jack

Jr. , Wiste

, Weigand

, Knopman

, Vemuri

, Mielke

, Lowe

, Senjem

, Gunter

, Machulda

, Gregg

, Pankratz

, Rocca

, Petersen

(2015) Age, sex, and APOE ɛ4 effects on memory, brain structure, and β-amyloid across the adult life span. JAMA Neurol 72, 511–519.

23.

Prins

, van Dijk

, den Heijer

, Vermeer

, Jolles

, Koudstaal

, Hofman

, Breteler

(2005) Cerebral small-vessel disease and decline in information processing speed, executive function and memory. Brain 128, 2034–2041.

24.

Gennatas

, Avants

, Wolf

, Satterthwaite

, Ruparel

, Ciric

, Hakonarson

, Gur

(2017) Age-related effects and sex differences in gray matter density, volume, mass, and cortical thickness from childhood to young adulthood. J Neurosci 37, 5065–5073.

25.

Ballard

, Rowan

, Stephens

, Kalaria

, Kenny

(2003) Prospective follow-up study between 3 and 15 months after stroke: Improvements and decline in cognitive function among dementia-free stroke survivors >75 years of age. Stroke 34, 2440–2444.

26.

Wagle

, Farner

, Flekkoy

, Bruun Wyller

, Sandvik

, Fure

, Stensrod

, Engedal

(2011) Early post-stroke cognition in strokerehabilitation patients predicts functional outcome at 13 months. Dement Geriatr Cogn Disord 31, 379–387.

27.

Zietemann

, Georgakis

, Dondaine

, Müller

, Mendyk

, Kopczak

, Hénon

, Bombois

, Wollenweber

, Bordet

, Dichgans

(2018) Early MoCA predicts long-term cognitive andfunctional outcome and mortality after stroke. Neurology 91, e1838–e1850.

28.

Tsai

, Lee

, Wang

, Shia

, Nasreddine

, Fuh

(2012) Psychometrics of the Montreal Cognitive Assessment (MoCA) and its subscales: Validation of the Taiwanese version of the MoCA and an item response theory analysis. Int Psychogeriatr 24, 651–658.

29.

Zhou

, Ortiz

, Nuñez

, Elashoff

, Woo

, Apostolova

, Wolf

, Casado

, Caceres

, Panchal

, Ringman

(2015) Use of the MoCA in detecting early Alzheimer’s disease in a Spanish-speaking population with varied levels of education. Dement Geriatr Cogn Dis Extra 5, 85–95.

30.

Lawrence

, Brookes

, Zeestraten

, Barrick

, Morris

, Markus

(2015) Pattern and rate of cognitive decline in cerebral small vessel disease: A prospective study. PLoS One 10, e0135523.

31.

Tuladhar

, Reid

, Shumskaya

, de Laat

, van Norden

, van Dijk

, Norris

, de Leeuw

(2015) Relationship between white matter hyperintensities, cortical thickness, and cognition. Stroke 46, 425–432.

32.

Lambert

, Benjamin

, Zeestraten

, Lawrence

, Barrick

, Markus

(2016) Longitudinal patterns of leukoaraiosis and brain atrophy in symptomatic small vessel disease. Brain 139, 1136–1151.

33.

Arba

, Quinn

, Hankey

, Ali

, Lees

, Inzitari

(2017) Cerebral small vessel disease, medial temporal lobe atrophy and cognitive status in patients with ischaemic stroke and transient ischaemic attack. Eur J Neurol 24, 276–282.

34.

Cordoliani-Mackowiak

, Hénon

, Pruvo

, Pasquier

, Leys

(2003) Poststroke dementia: Influence of hippocampal atrophy. Arch Neurol 60, 585–590.

35.

Velickaite

, Ferreira

, Cavallin

, Lind

, Ahlström

, Kilander

, Westman

, Larsson

(2018) Medial temporal lobe atrophy ratings in a large 75-year-old population-based cohort: Gender-corrected and education-corrected normative data. Eur Radiol 28, 1739–1747.

36.

Liu

, Wong

, Au

, Yang

, Wang

, Leung

, Chen

, Ho

, Mok

(2015) Influence of amyloid-β on cognitive decline after stroke/transient ischemic attack: Three-year longitudinal study. Stroke 46, 3074–3080.