Intima Media Thickness and Cognitive Function Among Adults: Meta-Analysis of Observational and Longitudinal Studies.

Background Carotid structural changes measured by intima media thickness (IMT) have been related to cognitive complaints during aging. Therefore, the aims of this meta-analysis were (1) to elucidate the relationship between vascular status, measured as IMT, and cognitive domains distinguishing between global cognition, executive functions, memory and attention; and (2) to explore whether demographic (ie, age and sex), clinical (ie, body mass index and IMT baseline values), and procedure characteristics influence this association. Methods and Results We performed a systematic review of MEDLINE (via PubMed), Scopus, and Web of Science databases from their inception to June 2021. Studies meeting the following inclusion criteria were included: (1) the participants were adults; (2) the exposure was carotid IMT; (3) the outcome was cognitive function, including global cognition, executive function, memory, and attention measured using standardized tests; and (4) the study design was cross-sectional or longitudinal including unadjusted and adjusted analyses. A total of 19 cross-sectional and 15 longitudinal studies were included and demographic (age and sex), clinical (body mass index and baseline IMT values), and procedure characteristics were analyzed as potential mediator or moderators of the association. Conclusions Our data support negative associations between IMT and cognitive function in cross-sectional studies. The association between IMT and cognition lost significance in longitudinal studies and when controlling for covariates in cross-sectional studies. Finally, the strength of these associations seems not to be modified by age, sex, body mass index, and baseline IMT values. This systematic review and meta-analysis adds to the evidence supporting the use of IMT as a measure for identifying patients at risk of cognitive decline.

effect size

intima media thickness

Clinical Perspective

What Is New?

A cross‐sectional negative association between intima media thickness (IMT) and global cognition, executive functions, and memory has been identified.

This association is not supported for longitudinal studies and when controlling for covariates in cross‐sectional studies.

In addition, variables including age, sex, body mass index, and baseline IMT values do not seem to modify the strength of this association.

What Are the Clinical Implications?

The results of this systematic review and meta‐analysis do not support a strong association between IMT and longitudinal change in cognitive function.

Although IMT could be used as a measure for identifying patients at risk of cognitive decline, future work is needed to address the association of IMT with the onset of cognitive impairment.

According to World Health Organization estimates, by the end of 2020, the number of people aged >60 years might outnumber children younger than 5 years and reach 22% of the global population by 2050. As population life expectancy continues growing worldwide, health systems, social systems, and governments have to face the aging‐related burden of chronic diseases. Among the needs of elderly people are changes in several physical and mental health domains (eg, somatic diseases, physical function aging, and psychological and cognitive changes). In particular, cognitive complaints are among the most common reasons for consultation with older patients and their primary caregivers and have been described as a predictor of cognitive decline.

To date, cognitive decline lacks effective treatment, and the search for approaches to prevent or delay its progression and the onset of cognitive impairment has become an important clinical target. , The concept of cognitive decline as a vascular disease is being increasingly accepted, together with the evidence that the early detection and treatment of classical cardiovascular risk factors could reduce the impact of cognitive decline. , , In this framework, the monitoring of subclinical cardiovascular risk markers may have an important role in detecting individuals at risk to develop cognitive impairment and in tracking changes induced by treatments. A number of structural alterations that are an expression of vascular aging, reflecting the integrated burden of known and unknown cardiovascular risk factors on the vasculature, such as large artery stiffness , , and carotid structural changes, , have been associated with a steeper cognitive decline. Among subclinical cardiovascular risk markers, carotid structural changes have been specifically associated with chronic cerebral hypoperfusion, silent micro‐ and macrocerebrovascular disease, and cortical atrophy, silent cerebral small vessel lesions, , , which in turn are associated with reduced cognitive function.

The utility of the measurement of carotid structural changes by intima media thickness (IMT) has been debated; these subclinical biomarkers of vascular aging are currently not recommended in international guidelines for risk stratification, though recent meta‐analyses demonstrate that the extent of intervention effects on carotid IMT predicted the degree of cardiovascular diseases reduction, thus supporting the usefulness of IMT as surrogate biomarkers in interventional trials. Furthermore, a relationship between IMT and cognitive performance has been demonstrated in many cross‐sectional and longitudinal studies. , , Discrepancies among studies, including differences in the design and population characteristics and the measurement of a wide variety of cognitive domains, make the evidence of this relationship inconsistent. Furthermore, not every cognitive domain is equally affected during aging, because the brain does not age uniformly, and several factors could protect or damage specific cognitive functions. ,

Therefore, the aims of this meta‐analysis were the following: (1) to elucidate the relationship between vascular status, measured as IMT, and cognitive domains distinguishing between global cognition, executive functions, memory, and attention; and (2) to explore whether demographic (ie, age and sex), clinical (ie, body mass index (BMI) and baseline IMT values), and procedure characteristics influence this association.

METHODS

This systematic review and meta‐analysis was conducted following the Cochrane Collaboration Handbook and was reported following the Preferred Reporting Items for Systematic Reviews and Meta‐Analysis of Observational Studies in Epidemiology statement. The protocol for this systematic review and meta‐analysis was previously registered on PROSPERO. The methods used in the analysis, and materials used to conduct the research are available to other researchers upon request.

Data Sources and Searches

Studies on the association between carotid IMT and cognitive function in adults were searched on Medline (via PubMed), Web of Science, and Scopus from database inception to June 2021. The search strategy included the following terms: "endothelial function," "atherosclerosis," "IMT," ‘"intima thickness," "intima media thickness," "carotid plaque," "cognition," "executive," "executive function," "cognitive control," "memory," "attention," "metacognition," "life skills," "goal setting," "problem solving," "self‐regulation," "brain development," and "brain health". We completed the search by screening previous systematic reviews and meta‐analyses in the field and checking the reference lists of the included studies.

Study Selection

This systematic review includes studies on the relationship between vascular status, measured by IMT, and cognitive function in adults. The inclusion criteria were as follows: (1) the participants were adults, (2) the exposure was carotid IMT, (3) the outcome was cognitive function, including global cognition, executive function, memory and attention, measured using standardized tests, and (4) the study design was cross‐sectional or longitudinal including unadjusted and adjusted analyses.

Studies were excluded when they were (1) focused on children or adolescents, (2) focused on patients with dementia, or (3) written in languages other than English, French, Portuguese, or Spanish.

Data Extraction and Risk of Bias Assessment

The main characteristics of the included studies are summarized in Table 1 and Table S1, including information on (1) subject characteristics (sample size, percentage of women, mean age, BMI, systolic blood pressure, and diastolic blood pressure; (2) exposure (technique used to measure IMT); and (3) outcome information (test used to measure cognitive function and cognitive domain measured).

Table 1

Characteristics of the Studies Included in the Systematic Review and Meta‐Analysis on the Association Between Cognition Parameters and IMT

	Subjects characteristics						Exposure		Outcome
References	n, female (%)	Age	BMI	SBP	DBP	Depressive s (%)	IMT device	IMT average	Cognitive measurement	Cognitive construct
Al Hazzouri et al, 201528	2618 (57.1)	45.3 (3.6)	NR	NR	NR	16.2%	GE‐Logiq‐700	0.8 (0.1)	Rey AVLT Digit Symbol Substation Stroop Interference	Delayed verbal memory Processing speed Executive function
Arntzen et al, 201223	4371 (51.5)	M: 58.6 (9.3) W: 59.5 (9.9)	M: 26.1 (3.1) W: 25.8 (4.2)	M: 143.1 (19.2) W: 142.4 (22.9)	NR	M: 2.1% W: 5.5%	High‐ resolution B‐mode ultrasonography	M: 0.88 (0.18) W: 0.81 (0.16)	12‐word memory Test Digit Symbol‐Coding Test Finger tapping Test	Immediate free recall Psychomotor speed, attention, and mental flexibility. Psychomotor tempo
Casado‐Naranjo et al, 201629	181 (45.9)	MCI: 77.5 (4.6) C: 75.3 (5.2)	NR	NR	NR	MCI: 18.1% C: 4.0%	Philips HD 11	MCI: 1.03 (0.27) C: 0.85 (0.32)	MMSE	Global cognitive
Cohen‐Manheim et al, 201630	507 (32.3)	49.9 (0.8)	24.7 (3.6)	112.0 (10.0)	68.0 (9.0)	Depressive symptoms score (HADS 0–21): 3.6 (2.9)	Philips IU22	0.63 (0.11)	Digit arithmetic problems Go‐NoGo, Stroop and Catch Game Abstract spatial ability test Immediate and delayed memory tests	Global cognitive Attention Information processing speed Executive Visual spatial processing Memory
Cohen et al, 200931	88 (NR)	72.2 (7.7)	NR	129.3 (19.5)	68.5 (10.1)	Exam Score Beck Depression Inventory: 4.2 (2.5)	Agilent 5500 machine	0.88 (0.13)	MMSE, Dementia Rating Scale BNT, Animals Block Design, Hooper Visual Organization, Rey Complex Figure Test‐ Copy California Verbal Learning delayed, Rey Complex Figure Test delayed, Brief Visual Memory Test–Revised Stroop, TMT A‐B, controlled oral Word association, Letter search, Digit Symbol Coding, Digit Span and Pegs‐D	Global cognitive Language Visual‐spatial Learning and memory functions Attention‐executive‐psychomotor functions
Del Brutto et al, 202024	561 (58)	57.8 (11.9)	NR	NR	NR	10%	Terason Smart 3300 NexGen	NR	MoCA	Global cognition
Feinkohl et al, 201332	831 (48.3)	67.7 (4.2)	31.3 (5.6)	132.5 (15.9)	69.0 (8.9)	HADS Depression: 3 (1–5)	Sonoline Elegra Ultrasound Im‐ aging System	1.0 (0.2)	Borkowski Verbal Fluency Test Logical Memory Faces Subtest TMT‐B Digit Symbol Coding Letter‐Number Sequencing Matrix Reasoning Mill Hill Vocabulary Scale MMSE	Executive function Immediate and delayed memory Nonverbal memory Mental flexibility and Executive function Speed of information processing Working memory Nonverbal reasoning Vocabulary abilities Global cognition
Frazier et al, 201433	251 (46.0)	78.0 (6.4)	NR	NR	NR	NR	High‐resolution B‐mode ultrasound	0.9 (0.1)	Dementia Rating Scale Initiation‐perseveration subscale, Wechsler Memory Scale Memory Scale‐Revised Digit and Visual Span backwards, COW‐Fluency Task Word List Learning Test, short and long delayed recall	Executive function Verbal memory
Gardener et al, 201834	1166 (60.0)	70. 0 (9.0)	28.0 (5.0)	NR	NR	NR	GE LogIQ 700	0.9 (0.1)	List‐learning, Delayed Recall, Delayed Recognition Color Trail Test, Odd‐man‐out subtest Grooved Pegboard Task, Color Trial test, Visual–motor Integration BNT, Animal Naming, Phonetic fluency test	Episodic memory Executive function Processing speed Semantic memory
Gatto et al, 200935	504 (38.9)	60.8 (9.9)	28.1 (5.0)	129.6 (16.9)	80.8 (10.4)	7.3%	High‐resolution B‐mode ultrasound	0.75 (0.15)	Symbol Digit Modalities, Trial‐B, Letter‐Number Sequencing, Judgment of Line Orientation, Block design, Shipley Institute of Living Scale California Verbal Learning Test, immediate recall and delayed recall Paragraph recall‐ immediate recall and delayed recall Faces I and II Category fluency and BNT	Executive function Verbal learning Logical memory Visual memory Semantic memory Global cognitive
Geijselaers et al, 201636	722 (44.9)	60.0 (8.0)	27.2 (4.4)	137.0 (19.0)	77.0 (11.0)	3.9%	MyLab 70	0.85 (0.15)	Verbal Learning Test Stroop Colour/Word, Concept Shifting Test Part A and B, Letter‐Digit Substitution Test Stroop Colour‐Word, Concept Shifting	Immediate and delayed recall Processing speed Executive function and attention
Haley et al, 200737	109 (43.0)	69.2 (7.43)	NR	NR	NR	NR	High‐ resolution B‐mode ultrasonography	0.88 (0.13)	MMSE, Dementia Rating Scale BNT, Category Fluency for Animals HVOT, WAIS‐III Block Design Subtest, CFT copy CVLT, CFT immediate recall, delayed recall, and recognition discrimination, BVMT‐R, recognition discrimination TMT (A and B), Stroop Word/Color, COW‐Association Test, Letter Search, WAIS‐III Block Coding Subtest, WAIS‐III Digit Span Subtest, Grooved Pegboard Dominant Hand	Global cognitive functioning Language Visual–spatial Memory Attention‐executive‐psychomotor
Imran et al, 202038	79 (NR)	55.5 (12.7)	NR	NR	NR	NR	Mindray Z6	0.81 (0.22)	RAVLT ROCFT	Memory function
Jiang et al, 201739	357 (65.0)	57.2 (9.3)	25.3 (3.3)	132.2 (16.6)	83.4 (9.3)	NR	Sequoia scanner	0.8 (0.2)	MoCA	Global cognition
Kemp et al, 201640	8114 (56.3)	51.2 (8.8)	NR	NR	NR	Depression severity: 8.0 (7.8)	Toshiba (Aplio XG)	0.8 (0.2)	TMT‐B	Executive function
Komulainen et al, 200741	91 (100)	63.8 (3.2)	27.6 (4.4)	NR	NR	Zung self‐report 20‐item scale: 36.4 (5.5)	Carotid ultrasonography	1.02 (0.26)	MMSE Word Recall Test Stroop test and Letter‐Digit Substitution Test	Global cognition Memory Cognitive Speed
Lim et al, 201642	463 (43.2)	MP: 63.0 (6.1) NP: 64.2 (6.4)	MP: 25.0 (4.1) NP: 24.6 (3.5)	NR	NR	NR	High‐resolution B‐mode ultrasound	MP: 0.8 (0.5–1.8) NP: 0.7 (0.5–1.8)	MMSE Digit Span‐Forward, Colour Trails Test Rey Auditory Verbal Learning Test, Story Memory and Recall BNT Brief Visuospatial Memory Test‐Revised Digit Span‐ Backward, Block Design, Colour Trails Test 2 and Categorical Verbal Fluency (Animal Naming)	Global cognition Attention Verbal memory Language Visuospatial ability Executive function
Masley et al, 201443	536 (27.4)	48.0 (7.5)	27.4 (4.7)	117.7 (15.3)	75.7 (10.4)	NR	High‐resolution B‐mode ultrasound	0.7 (0.1)	Index score Verbal memory and visual memory components Symbol Digit Coding Stroop Test, Continuous Performance Test Finger Tapping Test, Stroop Test	Global cognition Memory Executive function Attention Motor speed
Matsumuto et al, 201825	176 (55.7)	CI: 67.7 (12.3) Non‐CI: 64.6 (9.6)	CI: 23.8 (2.8) Non‐CI: 23.4 (3.4)	CI: 125.4 (20.9) Non‐CI: 119.9 (16.1)	CI: 76.5 (12.0) Non‐CI: 76.7 (10.2)	NR	Xario SSA‐660A	CI: 2.0 (1.0) Non‐CI: 1.7 (0.6)	MMSE, Hasegawa Dementia Scale‐revised Logical Memory IA and IIA of the WMS‐R Clock‐drawing Test	Global cognition Immediate and delayed recall Visuo‐constructional function, executive function and planning ability
Muela et al, 201844	211 (54.8)	NoHTA: 52.2 (13.9) HTA 1: 52.1 (13.0) HTA 2: 51.3 (10.1)	NoHTA:26.7 (4.2) HTA 1: 28.5 (4.6) HTA 2: 30.1 (4.6)	NoHTA: 121.9 (8.3) HTA 1: 135.0 (13.5) HTA 2: 147.5 (26.1)	NoHTA: 76.5 (6.9) HTA 1: 83.1 (9.9) HTA 2: 90.3 (14.5)	NR	Wall Track System, Medical Systems Arnhem)	NoHTA: 0.7 (0.1) HTA 1: 0.8 (0.1) HTA 2:0.8 (0.1)	MMSE, MoCA BNT RAVLT REY, Clock Drawing Test Semantic Verbal Fluency animal category, Phonological Verbal Fluency Forward and Backward Digit Span Tess, Trail Making Test, and Digit Symbols Substitution Test	Global cognition Language Memory Visuospatial abilities Attention and executive functions
Muller et al, 200745	396 (0.0)	N‐CVD: 54.5 (10.3) S‐CVD: 66.8 (8.1) P‐CVD: 67.7 (8.8)	N‐CVD: 25.9 (0.3) S‐CVD: 26.5 (0.3) P‐CVD: 27.3 (0.5)	N‐CVD: 134.2 (1.3) S‐CVD: 145.5 (1.7) P‐CVD: 140.2 (2.5)	NR	NR	Acuson Aspen	N‐CVD: 0.77 (0.01) S‐CVD: 0.89 (0.01) P‐CVD: 0.89 (0.02)	MMSE, Rey auditory verbal learning test, and door test Digit span, TMT‐A TMT‐B, Word fluency test Dutch adult reading test	Memory Processing/speed Executive function
Roberts et al, 201346	278 (54.3)	49.0–51.0	NR	NR	NR	NR	High‐resolution B‐mode ultrasound	NR	IQ test Moray House Tests 57 & 58, the English and Arithmetic tests and the Mill Hill and Raven’s Progressive Matrices	IQ Language Arithmetic
Rogne et al, 201347	1577 (47.3)	57 (52–61)	25.6 (3.2)	138.0 (18.0)	NR	NR	High‐ resolution B‐mode ultrasonography	0.78 (0.69–0.89)	Digit Symbol Test Finger tapping test 12‐word test parts 1 and 2 test (modification of the CVL test) MMSE	Executive function Motor speed Verbal episodic memory Global cognition
Romero et al, 200948	1971 (53.0)	58.0 (10.0)	NR	126.0 (18.0)	NR	NR	Doppler spectral analyzer (Model SSH‐140A)	NR	Wechsler Memory Scale Logical Memory, Paragraph A subtest, Immediate and Delayed Recall Halstead Reitan TMT (A and B) BNT, WAIS Similarities subtest, HVOT	Verbal memory Executive function Non‐verbal memory
Schwerdtfeger et al, 201549	124 (49.0)	37.5 (7.9)	23.8 (4.0)	NR	NR	NR	High‐resolution B‐mode ultrasound	0.5 (0.1)	Mainz Coping Inventory	Cognitive avoidance
Singh‐Manoux et al, 200850	3896 (27.9)	H‐SES‐M: 62.3 (5.6) H‐SES‐W: 59.8 (5.5) I‐SES‐M: 60.0 (5.8) I‐SES‐W: 60.1 (5.9) L‐SES‐M: 60.8 (6.5) L‐SES‐W: 62.0 (5.7)	H‐SES‐M: 26.2 (3.7) H‐SES‐W: 25.6 (5.1) I‐SES‐M: 26.5 (3.8) I‐SES‐W: 26.4 (5.3) L‐SES‐M: 26.5 (4.2) L‐SES‐W: 27.6 (4.9)	H‐SES‐M: 127.9 (15.5) H‐SES‐W: 122.8 (17.7) I‐SES‐M: 127.2 (15.2) I‐SES‐W: 124.7 (16.9) L‐SES‐M: 124.5 (14.7) L‐SES‐W: 126.5 (17.3)	H‐SES‐M: 74.2 (10.3) H‐SES‐W: 71.9 (10.3) I‐SES‐M: 74.3 (10.1) I‐SES‐W: 72.6 (10.4) L‐SES‐M: 72.2 (10.1) L‐SES‐W: 72.6 (10.6)	NR	Aloka 5500	H‐SES‐M: 0.8 (0.2) H‐SES‐W: 0.8 (0.1) I‐SES‐M: 0.8 (0.2) I‐SES‐W: 0.8 (0.1) L‐SES‐M: 0.8 (0.2) L‐SES‐W: 0.8 (0.1)	20‐word Free Recall Test Alice Heim 4‐I Mill Hill Vocabulary Test “S” words and “animal” words MMSE	Short term verbal memory Inductive reasoning Verbal meaning and Encompasses Phonetic and Semantic fluency Global cognition
Smith et al, 201151	124 (63.7)	52.3 (9.6)	32.8 (3.8)	138.3 (8.4)	86.1 (6.5)	8%	High‐ resolution B‐mode ultrasonography	0.70 (0.14)	TMT (A and B), Stroop Test, Verbal Paired Associates test, COW‐Association Test, Digit Span Test, Animal Naming Ruff 2 & 7 Test, Digit Symbol Substitution Test	Executive Function Psychomotor Speed
Suemoto et al, 201552	8208 (55.9)	49.6 (7.3)	26.6 (4.5)	NR	NR	4.0%	Toshiba ultrasound machine	0.7 (0.2)	CERAD Delayed Word Recall test Category Fluency Test TMT‐B	Verbal learning and recent memory Language and executive function Executive function, speed of processing, and attention
Wang et al, 201653	3227 (43.4)	57.9 (10.9)	24.9 (3.3)	130.8 (19.9)	82.6 (11.0)	NR	Philips iU‐22 ultra‐ sound system	NR	MMSE	Global cognition
Wendell et al, 200915	538 (60.2)	54.9 (14.0)	26.3 (4.5)	NR	NR	CES‐D: 6.6 (9.6)	Ultramark 9 HDI	0.5 (0.1)	Blessed Information‐Memory‐ Concentration (I‐M‐C) Test, MMSE Digits Forward, Digits Backward CVL learning slope CVL immediate, short‐delay and long‐delay free recall BVRT, RCFT immediate and long‐delay recall TMT (A and B) Letter Fluency Category Fluency BNT RCFT copy Card Rotations Test	Attention and concentration Verbal learning Memory Nonverbal memory Attention, perceptuo‐motor speed, visuomotor scanning, and mental flexibility Phonemic fluency Semantic fluency Language Visuospatial constructional ability Visuospatial function
Wendell et al, 201654	1712 (55.1)	46.9 (9.3)	29.4 (7.4)	119.5 (17.2)	NR	CES‐D: 14.0 (10.8)	Acuson CV 70	0.7 (0.1)	MMSE Benton Visual Retention Test CVL Animal fluency Brief Test of Attention Digit Span subtest of the Wechsler Adult Intelligence Scale‐Revised TMT (A and B)	Global cognition Visuospatial memory Verbal learning and memory Language and semantic association fluency Auditory divided attention Attention and working memory Attention, visual scanning, psychomotor speed, and mental flexibility
Yue et al, 201655	1826 (35.2)	63.2 (11.9)	NR	CInt: 147.2 (22.4) CImp: 149.0 (22.7)	CInt: 85.5 (12.5) CImp: 84.0 (13.6)	NR	High‐resolution B‐mode ultrasonography	1.4 (0.7)	MMSE	Global cognition
Zhong et al, 201156	2794 (54.0)	49.0 (9.8)	NR	NR	NR	NR	Biosound AU4	0.65 (0.15)	TMT (A and B) Grooved Pegboard Test MMSE	Executive, attention and psychomotor function Executive, and psychomotor function General cognitive function
Zhong et al, 201222	1651 (59.2)	66.8 (NR)	30.1 (5.9)	NR	NR	NR	Biosound AU4	0.86 (0.21)	MMSE TMT (A and B) Digit Symbol Substitution Test AVLT Verbal Fluency Test (VFT)	Executive function, attention and speed Psychomotor speed and sustained attention Memory Spontaneous production of words

ADAS‐cog indicates Alzheimer Disease Assessment Scale‐cognitive subscale; AVLT, Rey Auditory Verbal Learning Test; BNT, Boston naming test; BVMT‐R, Brief Visual Memory Test–Revised; COWAT, Controlled Oral Word Association Test; CVLT, California Verbal Learning Test; HVOT, Hooper Visual Organization Test; M, men; MMSE, Mini‐Mental State Examination; MoCA, Montreal cognitive assessment; NR, not reported; RAVLT, Osterrieth Auditory Verbal Learning Test; ROCFT, Rey Osterrieth Complex Figure Test; TMT, Trail Making Test; W, women; WAIS, Wechsler Adult Intelligence Scale; and WMS‐R, Wechsler Memory Scale Revised.

The Quality Assessment Tool for Observational Cohort and Cross‐Sectional Studies was used to evaluate the risk of bias. This tool evaluates 14 criteria for longitudinal studies. For cross‐sectional designs, only 11 were applied. Each criterion could be scored as “yes” when the study achieves the criterion, “no” when the study does not achieve the criterion, and “not reported” when the studies do not clearly report the required information. Following this risk of bias tool, studies could be rated as good (ie, at least 11 criteria for longitudinal studies and 8 criteria for cross‐sectional studies were met), fair (ie, from 6–10 criteria for longitudinal studies and 4–7 criteria for cross‐sectional studies were met), or poor (ie, from 1–5 criteria for longitudinal studies and 1–3 criteria for cross‐sectional studies were met).

The literature search (inter‐rater agreement Kappa , index 0.93 [95% CI, 0.99–0.95]), data extraction (inter‐rater agreement Kappa index 0.89 [95% CI, 0.87–0.91]), and risk of bias assessment (inter‐rater agreement Kappa index 0.92 [95% CI, 0.90–0.94]) were independently performed by 2 researchers (C.A.‐B. and I.C.‐R.), and disagreements were solved by consensus or involving a third researcher (V.M.V.).

Data Synthesis and Statistical Analysis

To perform the meta‐analysis, 4 cognitive domains were considered: (1) global cognition, (2) executive functions, (3) memory, and (4) attention. Separate analyses were conducted for unadjusted and adjusted estimations of cross‐sectional and longitudinal associations. For cross‐sectional associations, effect sizes (ESs) and 95% CIs were calculated for each observed correlation and regression coefficient using Cohen’s d index. A pooled ES was estimated for each cognitive domain using a random‐effects model based on the Der Simonian and Laird method. Random effects models were used as they provide more conservative results than fixed effects models and assume that each sample comes from a different population and that the effects in these populations may also differ. , The ES was interpreted following Cohen’s suggestions; d=0.2 was considered a “small” ES, 0.5 represented a “medium” ES, and 0.8 a “large” ES. The Cochran’s Q statistics were used to estimate the between‐study heterogeneity in ES. The proportion of the total variation across studies because of heterogeneity was assessed using the I2 statistic, whose values were not important (0% to <30%), moderate (≥30% to <50%), substantial (≥50% to <75%), or considerable (≥75% to 100%). Moreover, the corresponding P values were also considered. Following similar procedures, we estimated the pooled ES for the longitudinal associations between the baseline IMT and the pre–post change in the cognitive domains. The Z scores and corresponding P values against the hypothesis that IMT has no effect on cognitive function were also reported.

The following methodological considerations for data collection and analysis should be noted. When longitudinal studies reported baseline associations between IMT and cognitive function, these reports were included in the cross‐sectional pooled ES estimates. When studies provided ≥2 measurements for the same cognitive domain, these measurements were combined to calculate a single pooled ES for the corresponding domain. For unadjusted analyses, those associations including the shortest number of covariates were considered, and for the adjusted analyses, those associations including the largest number of covariates were considered. Finally, when studies reported mean value trends by groups or associations using regression models or correlation coefficients, ES values were calculated.

Sensitivity analyses were performed excluding all the studies 1 at a time from the pooled estimates to evaluate whether any particular study modified the original summary estimate. Subgroup analyses were performed based on characteristics of the procedures to measure IMT, including (1) manual, automated, or not specified measurements; (2) right, left, or bilateral carotid artery measurement; and (3) frequency of the ultrasound (defined as a range, >7 or not specified). Additional subgroup analyses were performed based on the method used to measure cognitive function (global test of cognition or domain‐specific assessments). Random effect meta‐regressions were calculated based on sample characteristics: percentage of women, mean age, BMI, and baseline IMT values. Finally, publication bias was estimated using Egger’s test. All the statistical analyses were performed using STATA 15 (StataCorp) software.

RESULTS

Systematic Review

The literature search retrieved 6879 studies, of which 19 cross‐sectional and 15 longitudinal studies were included in this systematic review and meta‐analysis , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , (Figure S1). The studies included a total of 50 779 participants aged 45.3 to 78.0 years. Table 1 and Table S1 summarize the characteristics of included studies. Table S2 summarizes the covariates used in the analyses of the included studies.

Risk of Bias Assessment

Cross‐sectional studies scored between 5 and 11 points, and longitudinal studies scored between 8 and 12 points. The 4 criteria in which most articles lacked information were sample size justification, power description, variance, and outcome blinding of the assessors to the participants’ exposure status (Table S3).

Meta‐Analysis

The pooled ES for the unadjusted cross‐sectional association between carotid IMT and cognitive function was small for global cognition (−0.25 [95% CI, −0.36 to −0.14]; Q: 134.40; I2: 89.6%), for executive function (−0.18 [95% CI, −0.29 to −0.07]; Q: 37.15; 81.2%), for memory (−0.14 [95% CI, −0.25 to −0.04]; Q: 151.33; 90.1%), and for attention (−0.12 [95% CI, −0.29 to 0.04]; Q: 54.16; 87.1%). Considering the adjusted cross‐sectional data, the pooled ES was small for global cognition (−0.15 [95% CI, −0.24 to −0.07]; Q: 78.08; I2: 82.1%), for executive function (−0.12 [95% CI, −0.21 to −0.03]; Q: 27.04; 74.1%), for memory (−0.09 [95% CI, −0.15 to −0.03]; Q: 33.88; I2: 55.7%), and for attention (−0.13 [95% CI, −0.26 to 0.01]; Q: 32.46; 78.4%) (Figures 1 and 2).

Figure 1

Forest plot for the unadjusted cross‐sectional association between intima media thickness and cognitive function domains.

SES indicates socioeconomic status.

Figure 2

Forest plot for the adjusted cross‐sectional association between intima media thickness and cognitive function domains.

SES indicates socioeconomic status.

Additionally, for the longitudinal associations, the pooled ES for the unadjusted longitudinal association between IMT and cognitive function was small for global cognition (−0.21 [95% CI, −0.38 to −0.04]; Q: 16.37; I2: 81.7%), for executive function (−0.14 [95% CI, −0.29 to 0.01]; Q: 50.25; I2: 90.0%), for memory (−0.15 [95% CI, −0.30 to 0.00]; Q: 255.80; I2: 96.5%), and for attention (−0.23 [95% CI, −0.62 to 0.17]; Q: 215.95; 98.6%). Considering the adjusted longitudinal data, the pooled ES was small for global cognition (−0.09 [95% CI, −0.20 to 0.02]; Q: 7.30; I2: 58.9%), for executive function (−0.04 [95% CI, −0.12 to 0.04]; Q: 12.78; I2: 60.9%), for memory (−0.00 [95% CI, −0.05 to 0.04]; Q: 23.32; I2: 61.4%); and for attention (−0.04 [95% CI, −0.09 to 0.01]; Q: 3.81; 21.3%) (Figures 3 and 4).

Figure 3

Forest plot for the unadjusted longitudinal association between intima media thickness and cognitive function domains.

SES indicates socioeconomic status.

Figure 4

Forest plot for the adjusted longitudinal association between intima media thickness and cognitive function domains.

 

The Z scores and corresponding P values against the hypothesis that IMT has no effect on cognitive function are reported in Table S4.

Sensitivity Analysis

Sensitivity analyses were performed excluding all the studies 1 at a time from the pooled estimates to evaluate whether any particular study modified the original summary estimate. The sensitivity analyses for the cross‐sectional estimates showed that the adjusted association between IMT and attention became significant after excluding the studies performed by Geijselaers et al and Zhong et al.

The sensitivity analyses for the longitudinal estimates showed that the unadjusted association between IMT and cognitive function became significant after excluding the studies performed by (1) Cohen‐Manheim et al, Del Brutto et al, and Feinkohl et al, for global cognition; (2) Gardener et al, and Romero et al, for executive functions; and (3) Gardener et al, Romero et al, Wendell et al, and Zhong et al, for memory (Tables S5–S8).

Meta‐Regressions and Subgroup Analyses

Meta‐regressions showed that none of the considered variables (ie, % females, age, BMI, and baseline IMT values) influenced the relationship between IMT and cognitive function for the cross‐sectional or longitudinal models (Table S9). Additionally, when considering the procedure characteristics, the ESs for the subgroup analyses were similar to the pooled ESs when (1) automated and not specified measurements of IMT were used for cross‐sectional studies reporting on global cognition and memory; (2) bilateral carotid artery measurement was used for cross‐sectional and longitudinal studies reporting on global cognition, executive functions, and memory; (3) >7 frequency of the ultrasound was used for cross‐sectional studies reporting on global cognition, memory, and attention; and (4) domain‐specific assessments to measure cognitive function were used for cross‐sectional studies reporting on global cognition and memory (Table S10).

Publication Bias

As evaluated by Egger’s test and funnel plot asymmetry, publication bias was found for (1) global cognition in the unadjusted longitudinal analysis; (2) memory in the unadjusted and adjusted cross‐sectional analyses; and (3) attention in the unadjusted cross‐sectional analysis (Table S11 and Figures S2–S9).

DISCUSSION

The purpose of this meta‐analysis was to identify associations between IMT and cognitive function in cross‐sectional and longitudinal studies. Our data support a small negative association between IMT and cognitive function in cross‐sectional studies, especially for global cognition, executive functions, and memory. The association between IMT and cognition lost significance in the longitudinal studies and after controlling for covariates in cross‐sectional studies. Finally, our data suggest that demographic (age and sex), clinical (BMI and baseline IMT values), and procedure characteristics do not influence the strength of this association.

Data from cross‐sectional studies suggest an association between IMT and cognition, indicating that cognition is reduced in people with increased IMT. The association was not significant when ESs of longitudinal studies were pooled; therefore, the pooled results did not support that IMT level predicted cognitive performance over time. This asymptomatic carotid atherosclerosis could be the image of the arterial remodeling that occurs during the natural process of vascular aging. In addition, several covariates could negatively impact vascular aging, including age, sex, diabetes, hypertension, and patient education, resulting in atherosclerosis as a maladaptive process of vascular remodeling.

Significant heterogeneity was observed in the pooled analyses and explored by subgroup analyses and meta‐regressions based on demographic, clinical, and procedure characteristics. Although sex, age, and BMI have been proposed to be factors affecting IMT, the results of the meta‐regressions do not suggest that the relationship between IMT and cognitive functions could be influenced by these factors. In addition, the different effects of hemodynamic and biochemical changes on the left and right IMT could be sources of heterogeneity; data from the subgroup analyses suggest that the bilateral IMT measurement is the most reported and reproducible method when assessing the relationship between IMT and cognitive functions. The influence of other procedure characteristics including the method used to measure IMT (automated or manual methods) and cognitive functions (domain‐specific assessment or global test of cognition), and the mHz applied could not be confirmed because of the lack of studies to draw conclusions.

Different mechanisms have been proposed to explain the effect of IMT on cognitive function. Blood support to neural cells could be compromised as a result of 2 interrelated events: (1) the thickening of the arterial wall, which could reduce the vessel lumen and produce chronic cerebral hypoperfusion; and (2) the promotion of endothelial dysfunction by the increased wall stress. Additionally, the increased wall stress has been related to an increased risk of plaque fissuring, increasing the subsequent risk of neural ischemia. Atherosclerosis and embolization of the vascular microcirculation and subsequent chronic inflammation , have been suggested to precipitate cerebral small vessel disease. Furthermore, the local thickening of the arterial wall could produce a microturbulent flow, reducing the supply of blood and nutrients and leading to neuronal dysfunction and cell death.

The results of this systematic review and meta‐analysis should be cautiously considered, as some limitations should be mentioned. In addition to the specific limitations of the meta‐analysis design, other sources of bias could be that (1) we did not use the original data but the data as reported by the studies to estimate the pooled ES; (2) substantial heterogeneity was found when pooled ESs were estimated, and diverse methods and tools were used to measure cognitive function across the included studies; (3) publication bias was found in some of the analyses; (4) language restrictions may have limited the number of included studies; and (5) our data might be limited by the use of 3 databases, and additional studies may have been found by checking other databases.

In conclusion, the pooled analysis of cross‐sectional studies suggests a negative association between IMT and global cognition, executive functions, and memory. The association between IMT and cognition lost significance in longitudinal studies and when controlling for covariates in cross‐sectional studies. Finally, age, sex, BMI, and baseline IMT values do not seem to modify the strength of this association. The results of this systematic review and meta‐analysis do not support a strong association between IMT and longitudinal change in cognitive function, and future work is needed to address this association with the onset of cognitive impairment.

Sources of Funding

This study was funded by European Regional Development Fund.

Disclosures

None.

Tables S1–S11

Figures S1–S9

Click here for additional data file.