Association of enlarged perivascular spaces with cognitive function in dementia‐free older adults: A population‐based study

Abstract

Introduction

We sought to characterize cognitive profiles associated with enlarged perivascular spaces (EPVS) among Chinese older adults.

Methods

This population‐based study included 1191 dementia‐free participants (age ≥60 years) in the MIND‐China MRI Substudy (2018–2020). We visually evaluated EPVS in basal ganglia (BG) and centrum semiovale (CSO), white matter hyperintensities (WMHs), lacunes, cerebral microbleeds (CMBs), and cortical superficial siderosis. We used a neuropsychological test battery to assess cognitive function. Data were analyzed using general linear models.

Results

Greater BG‐EPVS load was associated with lower z‐scores in memory, verbal fluency, and global cognition (p < 0.05); these associations became non‐significant when controlling for other cerebral small vessel disease (CSVD) markers (e.g., WMHs, lacunes, and mixed CMBs). Overall, CSO‐EPVS load was not associated with cognitive z‐scores (p > 0.05); among apolipoprotein E (APOE) ‐ε4 carriers, greater CSO‐EPVS load was associated with lower verbal fluency z‐score, even when controlling for other CSVD markers (p < 0.05).

Discussion

The associations of BG‐EPVS with poor cognitive function in older adults are largely attributable to other CSVD markers.

HIGHLIGHTS

The association of enlarged perivascular spaces (EPVS) with cognitive function in older people is poorly defined. The association of basal ganglia (BG)‐EPVS with poor cognition is attributed to other cerebral small vessel disease (CSVD) markers. In apolipoprotein E (APOE) ε4 carriers, a higher centrum semiovale (CSO)‐EPVS load is associated with poorer verbal fluency.

BACKGROUND

Perivascular spaces (PVS) are fluid-filled cavities that surround small arterioles, capillaries, and venules in the brain traveling from the brain surface into the parenchyma.¹ Enlarged perivascular spaces (EPVS) visible on magnetic resonance imaging (MRI) scans, as an emerging marker of cerebral small vessel disease (CSVD),² are indicative of impaired clearance of cerebrospinal fluid (CSF) and metabolic products (i.e., glymphatic dysfunction) in the brain parenchyma.^3,4 The population-based Vanderbilt Memory and Aging Project Study of dementia-free older adults suggested that a higher EPVS burden was associated with poorer information-processing speed and executive function.⁵ However, a meta-analysis of five population-based cohort studies and an additional community-based cohort study of older adults found no clear evidence supporting the associations of EPVS with cognitive impairment among dementia-free people.^6,7 Thus, the potential associations between EPVS load and the function of specific cognitive domains are still unclear in the general population settings. Moreover, EPVS are distributed predominantly in the basal ganglia (BG) and centrum semiovale (CSO), which are considered to reflect primarily hypertensive microangiopathy and cerebral amyloid angiopathy (CAA), respectively.^8,9 Therefore, investigating the potential differential associations of EPVS by brain regions with specific cognitive domains might shed light on the specific brain pathologies that underlie poor performance in different cognitive domains.

The burden of EPVS has been associated with older age, male sex, and apolipoprotein E (APOE) ε4 allele.^10–12 In addition, EPVS is correlated with white matter hyperintensities (WMHs), cerebral microbleeds (CMBs), and lacunes in older adults.^10,13,14 Given that other CSVD markers and demographic and genetic factors are closely related to cognitive function,^5,14–16 it is worth exploring whether the associations of EPVS load with domain-specific cognitive function are present independent of other CSVD markers or vary by demographic factors and APOE genotype.

Therefore, in this population-based study, we aimed to investigate the associations of EPVS with the function of multiple cognitive domains in dementia-free rural older adults in China while considering other CSVD markers and further exploring the potential effect modification of age, sex, and APOE genotype.

METHODS

Study design and participants

This population-based cross-sectional study used data from the Multimodal Intervention to delay Dementia and disability in rural China (MIND-China, registration no.: ChiCTR1800017758),¹⁷ which targeted people who were ≥60 years of age and living in the 52 villages of Yanlou Town, Yanggu County, western Shandong Province. In March–September 2018, a total of 5765 (74.9% of all eligible) participants were examined for MIND-China. Of these, using the cluster (village)–based sampling approach, 1844 participants from 26 villages that were randomly selected from all the 52 villages were invited for the MRI substudy. Of these, 1304 (70.7%) signed informed consent and underwent the structural brain MRI scans at either Southwestern Lu Hospital (n = 1178) or Liaocheng People's Hospital (n = 126) from September 2018 to November 2020. Of these, we excluded 113 persons due to incomplete MRI scans (n = 39), suboptimal image quality (n = 28), missing APOE genotype data (n = 27), or missing data on all cognitive domains (n = 19), leaving 1191 participants who were free of dementia for the current analysis. Figure 1 shows the flowchart of the study participants.

[IMAGE OMITTED. SEE PDF]

Data collection and assessments

Following administration of a structured questionnaire, data were collected by trained staff through face-to-face interviews, clinical examinations, neuropsychological testing, or laboratory tests. These data included demographic characteristics (e.g., age, sex, and education), lifestyle factors (e.g., smoking and alcohol drinking), health conditions (e.g., hypertension, diabetes, heart disease, and stroke), use of medication, history of head injury/trauma, and genetic factors (e.g., APOE genotype).¹⁷ Education was categorized as no formal school education, primary school, and middle school or above. Smoking and alcohol consumption were categorized as never versus ever smoking or alcohol consumption, respectively. Body mass index (BMI) was calculated as weight (kg) divided by height (m) squared, and obesity was defined as a BMI ≥ 28 kg/m² according to the national guidelines for Chinese adults.¹⁸ All medications were classified and coded following the Anatomical Therapeutic Chemical (ATC) Classification System, as described previously.¹⁹ Hypertension was defined as self-reported physician diagnosis of hypertension, blood pressure ≥140/90 mmHg, or current use of antihypertensive drugs (ATC codes C02, C03, and C07–C09); diabetes as self-reported physician diagnosis of diabetes, fasting blood glucose ≥7 mmol/L, or current use of oral hypoglycemic medication or insulin injection (ATC code A10); and hyperlipidemia as total cholesterol ≥6.2 mmol/L, triglycerides ≥2.3 mmol/L, low density lipoprotein cholesterol (LDL-C) ≥4.1 mmol/L, high density lipoprotein cholesterol (HDL-C) ≤1.0 mmol/L, or current use of hypolipidemic agents (ATC code C10).²⁰ Coronary heart disease and clinical stroke were ascertained according to self-reported history, clinical and neurological examinations, and electronic annual health check-up records.¹⁹ APOE genotype was determined using multiple-polymerase chain reaction amplification, as described previously,²¹ and was dichotomized into carriers versus non-carriers of the APOE ε4 allele.

Neuropsychological assessments

The Chinese version of a neuropsychological test battery was used to assess the function of four cognitive domains, as described previously.^17,21 Briefly, memory function was assessed using the Auditory Verbal Learning Test, immediate recall, long-delayed free recall, and long-delayed recognition; verbal fluency was assessed using the Verbal Fluency Test categories of animals, fruits, and vegetables; attention function was assessed using the Digit Span Test and the Trail Making Test (Part A); and executive function was assessed using the Digit Span Test (backward test) and Trail Making Test (Part B). For the Trail Making Test, we used the score divided by the time taken to complete the test. We carefully chose those tests (e.g., carton and picture) that were suitable for persons with no or limited education. The raw test scores were standardized according to the means and SDs among all participants who were free from dementia. The composite z-score for each cognitive domain was calculated by averaging the standard z-scores of all tests for that domain. A composite z-score for global cognitive function was computed as the mean of all cognitive z-scores for individuals with data available in at least two of the four cognitive domains.

RESEARCH IN CONTEXT

• Systematic review: Several clinical-based studies have linked enlarged perivascular spaces (EPVS) with poor cognition. The potential associations between EPVS load and the function of specific cognitive domains in the general population setting remain unclear.

• Interpretation: In this population-based cross-sectional 3T magnetic resonance imaging (MRI) study (n = 1191) of rural-dwelling Chinese older adults who were free of dementia, we revealed that overall, a greater basal ganglia (BG)-EPVS load was associated with poorer performance in global cognition, memory, and verbal fluency. We further revealed that the observed associations could be explained largely by other cerebral small vessel disease (CSVD) markers. In addition, a greater centrum semiovale (CSO)-EPVS load was associated with lower verbal fluency z-score in apolipoprotein E (APOE) ε4 carriers independent of other CSVD markers (i.e., deep white matter hyperintensity [DWMH], strict lobar lacunes, and strict lobar cerebral microbleeds [CMBs]).

• Future directions: Future longitudinal studies are warranted to further elucidate the cognitive consequences of EPVS, which will contribute to the development of clinical guidelines for the optimal management of visible perivascular spaces in older adults.

MRI acquisition and assessment protocols

All eligible participants were scanned on either the Philips Ingenia 3.0T MR System (Philips Healthcare, Best, The Netherlands) in Southwestern Lu Hospital or the Philips Archiva 3.0T MR System (Philips Healthcare, Best, The Netherlands) in Liaocheng People's Hospital, as described previously.¹⁷ A 16-channel head-neck coil was used at both hospitals. The core MRI sequences included the sagittal three-dimensional (3D) T1-weighted, axial T2-weighted, and sagittal 3D fluid-attenuated inversion recovery (FLAIR) images. The parameters of these core MRI sequences were provided elsewhere.¹⁷ Briefly, in the Southwestern Lu Hospital, the parameters for the T1-weighted sequence were: repetition time (TR) = 8.30 ms, echo time (TE) = 3.80 ms, field of view (FOV) = 240 × 219 mm², matrix = 240 × 219, flip angle (FA) = 8°, and slice thickness = 1.00 mm; the parameters for T2-weighted sequence were: TR = 4000.00 ms, TE = 106.00 ms, FOV = 230 ×  210 mm², matrix = 384 × 330, FA = 90°, and thickness = 5.00 mm; the parameters for T2-FLAIR sequence were: TR = 4800.00 ms, TE = 296.00 ms, FOV = 250 × 250 mm², matrix = 224 × 224, and thickness = 1.12 mm; and the parameters for susceptibility-weighted image (SWI) sequence were: TR = 18.00 ms, TE = 25.00 ms, FOV = 221 × 182 mm², matrix = 220 × 183, FA = 10°, and thickness = 1.2 mm. In the Liaocheng People's Hospital, the parameters for T1-weighted sequence were: TR = 7.00 ms, TE = 3.30 ms, FOV = 240 × 240 mm², matrix = 220 × 218, FA = 8°, and thickness = 1.10 mm; the parameters for T2-weighted sequence were: TR = 7583.00 ms, TE = 107.00 ms, FOV = 230 × 230, matrix = 232 × 228, FA = 90°, and thickness = 5.00 mm; the parameters for T2-FLAIR sequence were: TR = 4800.00 ms, TE = 259.00 ms, FOV = 250 × 250 mm², matrix = 224 × 224, and thickness = 1.12 mm; and the parameters for SWI sequence were: TR = 16.00 ms, TE = 22.00 ms, FOV = 221 × 182 mm², matrix = 220 × 182, FA = 15°, and thickness = 1.0 mm.

EPVS were manually assessed on the axial T2-weighted sequence following a validated protocol.²² Briefly, EPVS appeared linear when imaged parallel to the course of the vessel, and round or ovoid with a diameter <3 mm when imaged perpendicular to the course of the vessel.¹ The trained rater (M.Z., a junior neurologist), blinded to the clinical information, visually counted EPVS in BG and CSO bilaterally under the supervision of an experienced clinical neurologist (L.S.). The rater first reviewed all MRI slices for EPVS in the areas of BG and CSO and then counted bilateral BG-EPVS and CSO-EPVS on all slices. We categorized EPVS in BG and CSO according to the highest number of EPVS on the slice and hemisphere with the most EPVS, by following the scale as described previously²³: 0 = no EPVS, 1 = 1–10 EPVS, 2 = 11–20 EPVS, 3 = 21–40 EPVS, and 4 = >40 EPVS. Figure 2 shows EPVS in example images. For EPVS in CSO and BG, the scale collapsed categories 0 and 1 into “none/mild” severity, scale category 2 was called “moderate” severity, and scale categories 3 and 4 were collapsed to “severe” EPVS in line with a previous study.²⁴ Three months after the initial assessment, EPVS were re-assessed on MR images of 30 randomly selected subjects, which yielded the Cohen's kappa value of 0.75 for BG-EPVS and 0.74 for CSO-EPVS.

[IMAGE OMITTED. SEE PDF]

We used FLAIR images to evaluate periventricular white matter hyperintensity (PWMH) and deep white matter hyperintensity (DWMH) following the Fazekas scale,²⁵ with a grade ≥2 indicating severe degree. Lacunes were defined as focal lesions of 3–15 mm with the same signal intensities as CSF and a hyperintense rim on the FLAIR images, according to the STandards for ReportIng Vascular changes on nEuroimaging (STRIVE) criteria.¹ CMBs were defined as small (<10 mm in diameter), homogeneous, and round foci of low signal on SWIs.²⁶ Lacunes and CMBs were categorized as strict lobar and mixed areas, respectively. Strict lobar lacunes or CMBs refer to those only in the frontal, parietal, temporal, and occipital lobes. Lacunes or CMBs in mixed areas referred to those in the deep brain regions (BG, thalamus, internal capsule, and external capsule) or infratentorial regions (brainstem and cerebellum) with or without concomitant lobar lacunes or lobar CMBs.^27,28 We defined cortical superficial siderosis (cSS) as a linear gyriform pattern of hypointense signals on SWIs.²⁹ The trained rater (J.W.) who was blinded to clinical data assessed lacunes, PWMH, and DWMH under the supervision of a senior neurologist (L.S.) and an experienced neuroradiologist (T.G.). Six months later, MR images of 200 randomly selected subjects were re-evaluated for lacunes, PWMH, and DWMH, which yielded Cohen's kappa value of 0.84 for lacunes, 0.89 for PWMH, and 0.86 for DWMH. The trained rater (M.Z.) who was blinded to clinical data assessed CMBs and cSS under the supervision of an experienced neuroradiologist (T.G.). Two weeks after the initial assessment, CMBs and cSS were re-assessed on MR images of all subjects, which yielded Cohen's kappa value of 0.84 for CMBs and 0.91 for cSS.

Statistical analysis

EPVS in both BG and CSO regions were considered as categorical variables, categorized by severity of EPVS, as described above. Clinical and neuroimaging characteristics of the study participants by the severity of EPVS were compared using one-way analysis of variance (ANOVA) for continuous variables and the chi-square test or Fisher's exact test for categorical variables. We used the general linear regression models to examine the associations of EPVS load with cognitive z-scores. The assumptions of the models were verified to be satisfied. We tested the statistical interaction of EPVS burden with age (<75 vs ≥75 years), sex, and APOE genotypes on cognitive z-scores. When a statistical interaction was detected, further stratifying analysis was performed to verify the direction and magnitude of the interaction. We reported the main results from two models: Model 1 was adjusted for sociodemographic variables (age, sex, and education), APOE genotype, and vascular risk factors that showed a relevant association with EPVS at p < 0.20 in Table 1; and Model 2 was additionally adjusted for other relevant CSVD markers that were associated with EPVS at p < 0.20 in Table 1. Stata Statistical Software: Release 15 for Windows (Stata Corp LLC., College Station, TX, USA) was used for all analyses.

TABLE 1 Characteristics of study participants by severity of enlarged perivascular spaces.

Characteristics	Total sample, n = 1191	BG-EPVS	CSO-EPVS
None/mild (grade 0–1) (n = 476)	Moderate (grade 2) (n = 437)	Severe (grade 3–4) (n = 278)	p	None/mild (grade 0–1) (n = 280)	Moderate (grade 2) (n = 474)	Severe (grade 3–4) (n = 437)	p
Age, years	69.39 (4.27)	68.30 (3.98)	69.58 (4.36)	70.97 (4.08)	<0.001	69.46 (4.47)	69.12 (4.33)	69.64 (4.07)	0.176
Female, n (%)	694 (58.27)	291 (61.13)	245 (56.06)	158 (56.83)	0.257	180 (64.29)	275 (58.02)	239 (54.69)	0.039
Education, n (%)					0.022				0.005
Illiterate	410 (34.42)	181 (38.03)	143 (32.72)	86 (30.94)		116 (41.43)	161 (33.97)	133 (30.43)
Primary school	541 (45.42)	196 (41.18)	198 (45.31)	147 (52.88)		120 (42.86)	225 (47.47)	196 (44.85)
Middle school or above	240 (20.15)	99 (20.80)	96 (21.97)	45 (16.19)		44 (15.71)	88 (18.57)	108 (24.71)
APOE ε4 carrier, n (%)	177 (14.86)	71 (14.92)	67 (15.33)	39 (14.03)	0.891	47 (16.79)	66 (13.92)	64 (14.65)	0.559
Ever smoking, n (%)	421 (35.35)	155 (32.56)	172 (39.36)	94 (33.81)	0.083	90 (32.14)	161 (33.97)	170 (38.90)	0.131
Ever alcohol intake, n (%)	430 (36.10)	159 (33.40)	175 (40.05)	96 (34.53)	0.093	93 (33.21)	167 (35.23)	170 (38.90)	0.265
Obesity, n (%)	230 (19.31)	86 (18.07)	89 (20.37)	55 (19.78)	0.662	52 (18.57)	85 (17.93)	93 (21.28)	0.414
Diabetes, n (%)	178 (14.95)	77 (16.18)	65 (14.87)	36 (12.95)	0.487	44 (15.71)	73 (15.40)	61 (13.96)	0.763
Hyperlipoidemia, n (%)	284 (23.85)	112 (24.49)	107 (24.49)	65 (23.38)	0.924	72 (25.71)	110 (23.21)	102 (23.34)	0.702
Hypertension, n (%)	805 (67.59)	277 (58.19)	310 (70.94)	218 (78.42)	<0.001	177(63.21)	325(68.57)	303 (69.34)	0.196
Coronary heart disease, n (%)	215 (18.05)	79 (16.60)	76 (17.39)	60 (21.58)	0.207	48 (17.14)	78 (16.46)	89 (20.37)	0.279
Stroke, n (%)	143 (12.01)	39 (8.19)	59 (13.50)	45 (16.19)	0.002	35 (12.50)	59 (12.45)	49 (11.21)	0.814
Severe PWMH, n (%)	654 (54.91)	139 (29.20)	261 (59.73)	254 (91.37)	<0.001	149 (53.21)	253 (53.38)	252 (57.67)	0.347
Severe DWMH, n (%)	584 (48.67)	156 (32.43)	219 (49.89)	209 (74.64)	<0.001	114 (40.71)	208 (43.88)	259 (59.27)	<0.001
Strict lobar lacunes, n (%)	97 (8.14)	23 (4.83)	39 (8.92)	35 (12.59)	0.001	13 (4.64)	36 (7.59)	48 (10.98)	0.009
Mixed lacunes, n (%)	247 (20.74)	46 (9.66)	92 (21.05)	109 (39.21)	<0.001	54 (19.29)	107 (22.57)	86 (19.68)	0.443
Strict lobar CMBs, n (%)	160 (13.43)	55 (11.55)	61 (13.96)	44 (15.83)	0.232	25 (8.93)	59 (12.45)	76 (17.39)	0.004
Mixed CMBs, n (%)	209 (17.55)	40 (8.04)	76 (17.39)	93 (33.45)	<0.001	51 (18.21)	81 (17.09)	77 (17.62)	0.925
Presence of cSS, n (%)	19 (1.60)	6 (1.26)	10 (2.29)	3 (1.08)	0.342	3 (1.07)	6 (1.27)	10 (2.29)	0.341

RESULTS

Characteristics of the study participants

Of the 1191 participants, the mean age was 69.4 years (SD, 4.3; age range, 60–82 years), 694 (58.3%) were female, 410 (34.4%) were illiterate, and 177 (14.9%) carried the APOE ε4 allele. Participants with a higher BG-EPVS burden were older and less educated, and more likely to have hypertension, a history of stroke, severe PWMH and DWMH, strict lobar and mixed lacunes, and mixed CMBs, whereas those with a higher CSO-EPVS burden were more likely to be male and to have high education, severe DWMH, strict lobar lacunes, and strict lobar CMBs (p < 0.05) (Table 1).

Associations of EPVS burden with cognitive z-scores

With controlling for sociodemographic variables, APOE ε4 allele, and relevant vascular risk factors, severe BG-EPVS, compared with none or mild BG-EPVS, was significantly associated with lower z-scores of memory, verbal fluency, and global cognition (p < 0.05), but not with z-scores of attention and executive function (p > 0.05) (Table 2, Model 1). However, the associations were substantially diluted and became non-significant after further adjusting for other relevant CSVD markers (PWMH, DWMH, strict lobar and mixed lacunes, and mixed CMBs) (Table 2, Model 2). The burden of CSO-EPVS was not significantly associated with any of the examined cognitive z-scores (p > 0.05, Table 2).

TABLE 2 Associations of enlarged perivascular spaces with cognitive function (n = 1191).

EPVS severity by regions	β-coefficient (95% confidence interval), cognitive z-scores
Memory (n = 1184)	Verbal fluency (n = 1185)	Attention (n = 1190)	Executive function (n = 1182)	Global cognition (n = 1190)
Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
BG-EPVS
None/mild	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)
Moderate	–0.01 (–0.11−0.10)	0.02 (–0.09−0.13)	0.04 (–0.06−0.13)	0.05 (–0.05−0.15)	–0.02 (–0.11−0.08)	–0.01 (–0.11−0.09)	–0.03 (–0.13−0.07)	–0.01 (–0.11−0.09)	–0.01 (–0.08−0.07)	0.01 (–0.06−0.08)
Severe	–0.17 (–0.30−–0.05)**	–0.10 (–0.24−0.04)	–0.16 (–0.27−–0.04)**	–0.12 (–0.25−0.01)	–0.04 (–0.16−0.07)	0.02 (–0.15−0.11)	–0.07 (–0.19−0.05)	–0.03 (–0.17−0.10)	–0.12 (–0.20−–0.03)**	–0.08 (–0.17−0.02)
CSO-EPVS
None/mild	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)	0 (reference)
Moderate	0.01 (–0.11−0.13)	0.01 (–0.11−0.14)	0.03 (–0.07−0.14)	0.04 (–0.07−0.15)	0.06 (–0.05−0.16)	0.06 (–0.05−0.17)	0.07 (–0.04−0.18)	0.07 (–0.04−0.18)	0.04 (–0.04−0.12)	0.05 (–0.03−0.13)
Severe	–0.01 (–0.13−0.12)	0.01 (–0.12−0.13)	0.02 (–0.09−0.13)	0.03 (–0.08−0.15)	0.08 (–0.04−0.18)	0.08 (–0.03−0.19)	0.05 (–0.07−0.16)	0.05 (–0.07−0.16)	0.04 (–0.05−0.12)	0.04 (–0.04−0.13)

Interactions of EPVS burden with age, sex, and APOE genotype on cognitive z-scores

We detected statistical interactions of greater burdens of BG-EPVS with sex on the z-scores of attention (multivariable-adjusted p for interaction < 0.05) (Figure 3). Stratified analyses by sex revealed that the linear association between a higher burden of BG-EPVS and lower attention z-scores was statistically significant in male but not in female participants; however, the linear associations in male participants were substantially attenuated and became non-significant when entering PWMH, DWMH, strict lobar and mixed lacunes, and mixed CMBs variables into the model (Figure 3A). In addition, we detected statistical interactions of CSO-EPVS with APOE genotype on verbal fluency z-score (p for interaction < 0.05). Further analysis stratified by APOE ε4 allele suggested that severe CSO-EPVS were significantly associated with a lower z-score of verbal fluency among APOE ε4 carriers, even in Model 2 when DWMH, strict lobar lacunes, and strict lobar CMBs were controlled for, whereas there was no significant association between EPVS load and verbal fluency z-score among APOE ε4 non-carriers in either model (Figure 3B). We detected no statistical interaction of EPVS load with age on cognitive z-scores.

[IMAGE OMITTED. SEE PDF]

DISCUSSION

The main findings from this population-based study of rural-dwelling dementia-free older adults in China can be summarized as follows: (1) the higher burden of BG-EPVS was associated with poorer performance in global cognition, memory, and verbal fluency, but the observed associations were largely attributable to other CSVD markers (e.g., WMHs, lacunes, and mixed CMBs); and (2) the associations of a greater EPVS load with worse cognitive function varied with sex and APOE genotype, such that the associations were evident mainly in male participants and the APOE ε4 allele carriers. Notably, a greater CSO-EPVS load was associated with poorer verbal fluency in APOE ε4 allele carriers independent of a range of potential confounders as well as other CSVD markers.

We found that the associations between a greater burden of BG-EPVS with poorer cognitive performance in dementia-free older people were largely attributable to other CSVD markers (e.g., WMHs, lacunes, and mixed CMBs), which was in line with the results from the meta-analysis of five population-based studies from the community-based clinical-neuropathological studies in United States, and the memory clinic-based study in Korea.^6,7,14 In contrast, the community-based Vanderbilt Memory and Aging Project suggested that a higher BG-EPVS burden was associated with worse perceptual speed and executive function among older adults independent of other CSVD markers.⁵ The differences in the findings across studies could be due in part to the wide variety of definitions and rating scales used to quantify EPVS, and MRI sequences for assessing EPVS (e.g., T2-weighted vs T1-weighted and FLAIR images). BG-EPVS are closely correlated to WMHs, lacunes, and mixed CMBs, and these CSVD markers may share similar pathophysiological mechanisms of arteriolosclerosis, which predominantly affect deep perforating arterioles.^8,30 Moreover, previous studies have suggested that other CSVD markers were associated with worse performance in cognitive domains such as memory and language.^5,31,32 Taken together, these studies support the notion that the association between a higher BG-EPVS burden and poorer performances in memory, verbal fluency, and global function could be explained, at least in part, by other CSVD markers.

Overall, we found no association between CSO-EPVS load and cognitive performance, which was again in accordance with results from the meta-analysis of five population-based studies and the Lothian Birth Cohort 1936 study.^6,33 However, the community-based Sydney Memory and Ageing Study did show that having more CSO-EPVS, but not BG-EPVS, was associated with a faster decline in global cognition, whereas having more EPVS in either region was not associated with any of the examined cognitive domains (i.e., memory, language, attention, visuospatial function, and executive function).³⁴ Differences in the study design (cross-sectional vs longitudinal study), demographic features of the study sample (rural residents with no or very limited education vs highly educated urban residents), and different quantitative approaches of EPVS might partly contribute to the discrepant findings across studies. The potential cognitive consequences of EPVS in older people should be further characterized in large-scale population-based prospective cohort studies.

Furthermore, we detected the independent association of the greater BG-EPVS load with poorer cognition only in men, although the linear trend of the association was attenuated substantially after further adjustment for other CSVD markers. Indeed, previous studies showed that compared to women, men were more vulnerable to cardiovascular risk factors (e.g., hypertension and smoking),^35, and had more cerebral microvascular lesions.^10,36 These cerebral microvascular lesions might disrupt the fronto-subcortical circuits and then lead to impaired attention.³⁷ This may partly account for the sex-varying associations of a higher BG-EPVS burden with worse cognitive function. We also found that the association of a higher CSO-EPVS burden with poorer verbal fluency was evident only among APOE ε4 carriers. This was consistent with the reports from a population-based study from Sweden and the clinic-based Sunnybrook Dementia Study from Canada.^38,39 The load of CSO-EPVS was correlated with strict lobar lacunes and strict lobar CMBs, which are indicative of CAA.^26,27,40 CAA, which results from amyloid beta (Aβ) deposition within small cortical and leptomeningeal arteries, is associated with increased blood–brain barrier permeability, lobar CMBs, cortical microinfarcts, and alterations of structural connectivity. All these neuropathological features have been linked with cognitive impairment.^41,42 Meanwhile, the APOE ε4 allele is associated with CAA and disruption of soluble Aβ clearance in the brain,^43,44 which might partly explain the finding that carrying the APOE ε4 allele could strengthen the association of severe CSO-EPVS with poor cognitive performance.

The major strengths of our study include the population-based design with a relatively large sample, and the integration of high-quality brain 3.0T MRI data with comprehensive clinical and neurocognitive data. In addition, our study engaged older adults who were living in rural communities in China and who had received no or very limited formal education, a sociodemographic group that has been substantially underrepresented in research on brain aging, cognition, and dementia.⁴⁵ However, our study has some limitations. First, the cross-sectional nature of the study cannot determine a temporal relationship for the observed associations. Furthermore, we assessed the EPVS burden by counting the total number of typical EPVS (<3 mm), which might be less precise than EPVS volume, computational count, or other computational parameters. Indeed, automatic segmentation techniques for identifying and quantifying EPVS have been developed in recent years, but the visual rating scores were correlated well with the automatically segmented EPVS count and volume.⁴⁶ Finally, participants in the brain MRI substudy were relatively younger, healthier, and more likely to be male than in the MIND-China total sample,²¹ which should be kept in mind when generalizing our findings to other rural populations in China.

In summary, our population-based study of rural-dwelling older adults in China supports the association of a greater BG-EPVS burden with poorer performance in global cognition and multiple cognitive domains (memory and verbal fluency), but the observed associations are largely attributable to other CSVD markers (e.g., WMHs, lacunes, and mixed CMBs). Notably, a greater CSO-EPVS load is associated with worse cognition among APOE ε4 allele carriers independent of other CSVD markers. This result suggests that patients with severe CSO-EPVS who are APOE ε4 carriers may have cognitive decline due to cerebral amyloid angiopathy or Alzheimer's disease pathology. Therefore, in the clinical setting, if severe CSO-EPVS is incidentally observed, it may be worthwhile to consider cognitive function testing. Future longitudinal studies are warranted to further characterize cognitive consequences associated with EPVS, which will contribute to the development of clinical guidelines for the optimal management of visible PVSs in older adults.

ACKNOWLEDGMENTS

We would like to thank all the participants in the MIND-China project as well as the MIND-China Research Group for their collaboration in data collection and management. MIND-China was supported in part by the grant from the National Key R&D Program of the Ministry of Science and Technology of China (grant no.: 2017YFC1310100), and this study was supported by additional grants from the National Nature Science Foundation of China (grant no.: 81861138008 and 82001120), the Academic Promotion Program of Shandong First Medical University (grant no.: 2019QL020 and 2020RC009), the Alzheimer's Association Grant (AACSFD-22-922844), the Taishan Scholar Program of Shandong Province (grant no.: tsqn201909182 and ts201712094), the Nature Science Foundation of Shandong Province (grant no.: ZR2020QH098), and the Integrated Traditional Chinese and Western Medicine Program in Shandong Province (YXH2019ZXY008), STI2030-Major Projects (grant no.: 2021ZD0201808 and 2022ZD0211600). J.M. Wardlaw receives funding from the UK Dementia Research Institute, which is funded by the UK Medical Research Council, Alzheimer's Society, and Alzheimer's Research UK, UK. L.J. Launer is supported by the Intramural Research Program, National Institute on Aging, National Institutes of Health, Maryland, USA. C. Qiu received grants from the Swedish Research Council (grant no.: 2017-05819 and 2020-01574), the Swedish Research Council for Health, Working Life and Welfare (program grant no.: 2023-01125. M Kivipelto as program PI; C Qiu as work-package leader), the Swedish Foundation for International Cooperation in Research and Higher Education (STINT) (grant no.: CH2019-8320) for the Joint China-Sweden Mobility program, and Karolinska Institutet (grant no.: 2020-01456), Stockholm, Sweden.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest. Author disclosures are available in the Supporting information.

CONSENT STATEMENT

The MIND-CHINA protocol was approved by the ethics committee at Shandong Provincial Hospital affiliated with Shandong University in Jinan, Shandong, China. Written informed consent was obtained from all participants, or in the case of cognitively impaired persons, from a proxy (usually a guardian or a family member).

References

Wardlaw JM, Smith EE, Biessels GJ, et al. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol. 2013;12(8):822‐838.

Doubal FN, MacLullich AM, Ferguson KJ, Dennis MS, Wardlaw JM. Enlarged perivascular spaces on MRI are a feature of cerebral small vessel disease. Stroke. 2010;41:450‐454.

Brown R, Benveniste H, Black SE, et al. Understanding the role of the perivascular space in cerebral small vessel disease. Cardiovasc Res. 2018;114:1462‐1473.

Wardlaw JM, Benveniste H, Nedergaard M, et al. Perivascular spaces in the brain: anatomy, physiology and pathology. Nat Rev Neurol. 2020;16:137‐153.

Passiak BS, Liu D, Kresge HA, et al. Perivascular spaces contribute to cognition beyond other small vessel disease markers. Neurology. 2019;92:e1309‐e1321.

Hilal S, Tan CS, Adams HHH, et al. Enlarged perivascular spaces and cognition: a meta‐analysis of 5 population‐based studies. Neurology. 2018;91:e832‐e842.

Javierre‐Petit C, Schneider JA, Kapasi A, et al. Neuropathologic and cognitive correlates of enlarged perivascular spaces in a community‐based cohort of older adults. Stroke. 2020;51:2825‐2833.

Shams S, Martola J, Charidimou A, et al. Topography and determinants of magnetic resonance imaging (MRI)‐visible perivascular spaces in a large memory clinic cohort. J Am Heart Assoc. 2017;6(9): [eLocator: e006279].

Charidimou A, Boulouis G, Pasi M, et al. MRI‐visible perivascular spaces in cerebral amyloid angiopathy and hypertensive arteriopathy. Neurology. 2017;88:1157‐1164.

Zhu YC, Tzourio C, Soumaré A, Mazoyer B, Dufouil C, Chabriat H. Severity of dilated Virchow–Robin spaces is associated with age, blood pressure, and MRI markers of small vessel disease: a population‐based study. Stroke. 2010;41:2483‐2490.

Pinheiro A, Demissie S, Scruton A, et al. Association of Apolipoprotein E varepsilon4 allele with enlarged perivascular spaces. Ann Neurol. 2022;92(1):23‐31.

Evans TE, Knol MJ, Schwingenschuh P, et al. Determinants of perivascular spaces in the general population: a pooled cohort analysis of individual participant data. Neurology. 2023;100:e107‐e122.

Li Y, Kalpouzos G, Laukka EJ, et al. Progression of neuroimaging markers of cerebral small vessel disease in older adults: a 6‐year follow‐up study. Neurobiol Aging. 2022;112:204‐211.]

Choe YM, Baek H, Choi HJ, et al. Association between enlarged perivascular spaces and cognition in a memory clinic population. Neurology. 2022;99(13):e1414‐e1421.

Jack CR Jr, Wiste HJ, Weigand SD, et al. Age, sex, and APOE epsilon4 effects on memory, brain structure, and beta‐amyloid across the adult life span. JAMA Neurol. 2015;72:511‐519.

Hu HY, Ou YN, Shen XN, et al. White matter hyperintensities and risks of cognitive impairment and dementia: a systematic review and meta‐analysis of 36 prospective studies. Neurosci Biobehav Rev. 2021;120:16‐27.

Wang Y, Han X, Zhang X, et al. Health status and risk profiles for brain aging of rural‐dwelling older adults: data from the interdisciplinary baseline assessments in MIND‐China. Alzheimers Dement. 2022;8: [eLocator: e12254].

Chen C, Lu FC, Department of Disease Control Ministry of Health, PR China. The guidelines for prevention and control of overweight and obesity in Chinese adults. Biomed Environ Sci. 2004; 17 Suppl:1‐36.

Cong L, Ren Y, Hou T, et al. Use of cardiovascular drugs for primary and secondary prevention of cardiovascular disease among rural‐dwelling older Chinese adults. Front Pharmacol. 2020;11: [eLocator: 608136].

Joint committee for guideline r. 2016 Chinese guidelines for the management of dyslipidemia in adults. J Geriatr Cardiol. 2018;15:1‐29.

Cong L, Ren Y, Wang Y, et al. Mild cognitive impairment among rural‐dwelling older adults in China: a community‐based study. Alzheimers Dement. 2022;19(1):56‐66.

Potter GM, Chappell FM, Morris Z, Wardlaw JM. Cerebral perivascular spaces visible on magnetic resonance imaging: development of a qualitative rating scale and its observer reliability. Cerebrovasc Dis. 2015;39:224‐231.

MacLullich AMJ, Wardlaw JM, Ferguson KJ, Starr JM, Seckl JR, Deary IJ. Enlarged perivascular spaces are associated with cognitive function in healthy elderly men. J Neurol Neurosurg Psychiatry. 2004;75(11):1519‐1523.

Banerjee G, Kim HJ, Fox Z, et al. MRI‐visible perivascular space location is associated with Alzheimer's disease independently of amyloid burden. Brain. 2017;140:1107‐1116.

Fazekas F, Chawluk JB, Alavi A, Hurtig HI, Zimmerman RA. MR signal abnormalities at 1.5 T in Alzheimer's dementia and normal aging. AJR Am J Roentgenol. 1987;149(2):351‐356.

Greenberg SM, Vernooij MW, Cordonnier C, et al. Cerebral microbleeds: a guide to detection and interpretation. Lancet Neurol. 2009;8(2):165‐174.

Pasi M, Boulouis G, Fotiadis P, et al. Distribution of lacunes in cerebral amyloid angiopathy and hypertensive small vessel disease. Neurology. 2017;88(23):2162‐2168.

Ding J, Sigurdsson S, Garcia M, et al. Risk factors associated with incident cerebral microbleeds according to location in older people. JAMA Neurology. 2015;72(6):682‐688.

Vernooij MW, Ikram MA, Hofman A, Krestin GP, Breteler MM, van der Lugt A. Superficial siderosis in the general population. Neurology. 2009;73(3):202‐205.

Fazekas F, Kleinert R, Offenbacher H, et al. The morphologic correlate of incidental punctate white matter hyperintensities on MR images. AJNR Am J Neuroradiol. 1991;12:915‐921.

Zeng W, Chen Y, Zhu Z, et al. Severity of white matter hyperintensities: lesion patterns, cognition, and microstructural changes. J Cereb Blood Flow Metab. 2020;40:2454‐2463.

Thong JY, Hilal S, Wang Y, et al. Association of silent lacunar infarct with brain atrophy and cognitive impairment. J Neurol Neurosurg Psychiatry. 2013;84:1219‐1225.

Valdés Hernández MDC, Ballerini L, Glatz A, et al. Perivascular spaces in the centrum semiovale at the beginning of the 8th decade of life: effect on cognition and associations with mineral deposition. Brain Imaging Behav. 2020;14:1865‐1875.

Paradise M, Crawford JD, Lam BCP, et al. Association of dilated perivascular spaces with cognitive decline and incident dementia. Neurology. 2021;96:e1501‐e1511.

Altermatt A, Gaetano L, Magon S, et al. Clinical associations of T2‐weighted lesion load and lesion location in small vessel disease: insights from a large prospective cohort study. Neuroimage. 2019;189:727‐733.

Qiu C, Zhang Y, Bronge L, et al. Medial temporal lobe is vulnerable to vascular risk factors in men: a population‐based study. Eur J Neurol. 2012;19:876‐883.

Dey AK, Stamenova V, Turner G, Black SE, Levine B. Pathoconnectomics of cognitive impairment in small vessel disease: a systematic review. Alzheimers Dement. 2016;12:831‐845.

Wang R, Laveskog A, Laukka EJ, et al. MRI load of cerebral microvascular lesions and neurodegeneration, cognitive decline, and dementia. Neurology. 2018;91: [eLocator: e1487].

Mirza SS, Saeed U, Knight J, et al. APOE ε4, white matter hyperintensities, and cognition in Alzheimer and Lewy body dementia. Neurology. 2019;93:e1807‐e1819.

Charidimou A, Hong YT, Jäger HR, et al. White matter perivascular spaces on magnetic resonance imaging: marker of cerebrovascular amyloid burden? Stroke. 2015;46:1707‐1709.

Charidimou A, Boulouis G, Gurol ME, et al. Emerging concepts in sporadic cerebral amyloid angiopathy. Brain. 2017;140:1829‐1850.

Freeze WM, Bacskai BJ, Frosch MP, et al. Blood‐brain barrier leakage and microvascular lesions in cerebral amyloid angiopathy. Stroke. 2019;50:328‐335.

Alonzo NC, Hyman BT, Rebeck GW, Greenberg SM. Progression of cerebral amyloid angiopathy: accumulation of amyloid‐beta40 in affected vessels. J Neuropathol Exp Neurol. 1998;57:353‐359.

Bales KR, Dodart JC, DeMattos RB, Holtzman DM, Paul SM. Apolipoprotein E, amyloid, and Alzheimer disease. Mol Interv. 2002;2(6):363‐375.

Wiese LAK, Gibson A, Guest MA, et al. Global rural health disparities in Alzheimer's disease and related dementias: state of the science. Alzheimers Dement. 2023;19(9):4204‐4225.

Ballerini L, Lovreglio R, Valdés Hernández MDC, et al. Perivascular spaces segmentation in brain MRI using optimal 3D filtering. Sci Rep. 2018;8:2132.