Table 1.

Association of dysfunction of the neurovascular coupling unit, as quantified by cerebral imaging, with cognitive performance.

N	Study population	Brain imaging technique	Cognitive performance assessment	Confounders taken into account	Main results	Reference
Cerebral perfusion at rest or cerebrovascular vasoreactivity
Prospective studies
N=103 (2 years follow-up)	N= 35 patient with AD; n=33 patients with MCI; and n=35 elderly controls.	SPECT	Cognitive performance was assessed with the 15-objects test, which consists of the following: Temporal, Spatial and Personal Orientation; Digit span forwards and backwards, Block Design and Similarities subtests of Wechsler Adult Intelligence Scal (Third Edition); The Word List Learning test from the Wechsler Memory Scale (Third Edition); the RBANS visual memory subtest; Verbal comprehension (2 simple, 2 semi-complex and 2 complex commands); an abbreviated 15 item confrontation naming test from the Boston Naming Test; the Poppelreuter test; Luria’s Clock test; Ideomotor and Imitation praxis; the Automatic Inhibition subtest of the Syndrom Kurtz Test (SKT); Phonetic Verbal Fluency (words beginning with ‘P’ during one minute); Semantic Verbal Fluency (‘animals’ during one minute), and the Spanish version of the Clock Test.	None	The decline in performance on the 15-objects test between baseline and 2-year follow-up was significantly correlated (P= 0.001) with a lower cerebral perfusion at rest in the left Broadman Area 48.	Alegret, 2012 (20);
N=116 (4-year follow-up)	116 cognitively-normal adults ranging from 20 to 88 years old.	BOLD- fMRI	Four domains of cognitive function were assessed, i.e. processing speed, working memory, reasoning, and episodic memory. Processing speed was measured using the Digit Comparison Task, adapted from the Letter Comparison Task of Salthouse & Babcock, and Wechsler Adult Intelligence Scale-III Digit Symbol; working memory was evaluated from the Cambridge Neuropsychological Test Automated Battery Spatial Working Memory, and Wechsler Adult Intelligence Scale-III Letter Number Sequencing); reasoning was estimated using the Raven’s progressive matrices, ETS letter sets and Cambridge Neuropsychological Test Automated Battery Stockings of Cambridge; and episodic memory was assessed using the modified Hopkins Verbal Learning Task, and the Cambridge Neuropsychological Test Automated Battery Verbal Recognition Memory Task.	Baseline cerebrovascular vasoreactivity and baseline cognitive performance	Lower whole brain cerebrovascular vasoreactivity was significantly associated with a greater decline in processing speed (P<0.05) and episodic memory (P<0.05), but not with other domains.	Peng, 2018 (21);
N=309; 4-years follow-up	Individuals aged 20 to 89 years old.	ASL-MRI	Change in general cognitive performance over time was calculated, where general cognitive ability was estimated from four domains of cognitive function: processing speed, working memory, reasoning, and episodic memory. Processing speed was evaluated using the Digit Comparison Task, adapted from the Letter Comparison Task of Salthouse & Babcock, and Wechsler Adult Intelligence Scale-III Digit Symbol test. Working memory was measured using the Cambridge Neuropsychological Test Automated Battery Spatial Working Memory test and the Wechsler Adult Intelligence Scale-III Letter Number Sequencing test. Reasoning was estimated using Raven’s progressive matrices, ETS letter sets and the Cambridge Neuropsychological Test Automated Battery Stockings. Episodic memory was assessed using the modified Hopkins Verbal Learning Task and the Cambridge Neuropsychological Test Automated Battery Verbal Recognition Memory Task.	Age, systolic blood pressure, physical activity, and baseline global cognitive performance (or domain-specific performance in analyses with individual domains as outcome).	Greater whole-brain cerebral blood flow at baseline was significantly associated with greater general cognitive performance over time (P=0.031). Greater frontal cerebral blood flow at baseline was significantly associated with greater general cognitive performance over time in the older group (>53 years old; P=0.021). Cerebral blood flow in other brain lobes (occipital lobe, parietal lobe, and temporal lobe) was not associated with general cognitive performance. Additionally, in the older group (>53 years old) frontal cerebral blood flow was associated with reasoning ability (P=0.006) and episodic memory (P=0.009).	De Vis, 2018 (22);
Cross-sectional studies
N=104	Men; 81y old; without dementia or stroke.	SPECT	Low versus high achievers, estimated from the MMSE, verbal ability, verbal memory, Digit Symbol Substitution Test, and the Benton Visual Retention Test.	None	Frontal, temporal, parietal, occipital, basal nucleus, thalamus, and subcortical regional cerebral blood flow were significantly lower in low achievers than high achievers.	André-Petersson, 2017 (23);
N=12	Patients >45 years old from a stroke center.	SPECT	Repeatable Battery for Assessment of Neuropsychological Status, which consisted of the following tests: the Digit Span subtest of the Wechsler Adult Intelligence Scale (third edition), Trail Making Test parts A and B, Clock Drawing Tests CLOX1 and CLOX2, and the Controlled Oral Word Association Test	None	Cerebral perfusion at rest in the left limbic lobe, right sub lobar lobe, right temporal lobe, left sub lobar lobe, right cerebellum (anterior lobe), and left frontal lobe were mostly all significantly correlated with the repeatable Battery for Assessment of Neuropsychological Status score.	Baker, 2013 (24);
N=78	n=24 individuals with MCI; and n=54 normal controls.	BOLD fMRI and ASL-MRI	MMSE	Age, sex, race, body-mass index, years of education, hypertension status, diabetic status, hypertension medication status, diabetic medication status, and either gray matter (for analyses in white matter regions) or white matter volume (for analyses in grey matter regions)	Cerebral blood flow was not associated with MMSE score (neither in the white matter nor in the gray matter). A higher MMSE score was significantly associated with a higher cerebrovascular vasoreactivity (in gray matter: beta per point higher MMSE score, 0.689 per unit change in end-tidal CO2, P = 0.005; and in white matter: beta per point higher MMSE score, 0.578 per unit change in end-tidal CO2, P= 0.016).	Kim, 2021 (25);
N =218	N=68 individuals with cognitive impairment and moderate to severe white matter hyperintensities; N=63 individuals without cognitive impairment and with no white matter hyperintensities; N=87 normal controls.	BOLD- fMRI	General cognitive performance was assessed with the MMSE, and a Chinese version of the MoCa. Executive function was assessed with the Trail Making Test-B and Stroop Color and Word Tests C (Stroop C); information processing speed was assessed with the Trail Making Test-A and Stroop Color and Word Tests A and B (Stroop A and B); memory was assessed with the Wechsler Memory Scale-visual reproduction-delayed recall and Auditory Verbal Learning Test-long delayed recall, representing visual memory and verbal memory, respectively; visuospatial performance was assessed with the clock drawing test and via a visual reproduction-copy test; and language abilities were assessed with the category verbal fluency test and Boston naming test.	Age, sex, years of education, and vascular risk factors (history of chronic diseases such as diabetes mellitus, hypertension, and dyslipidemia, and smoking status).	The group with cognitive impairment and white matter hyperintensities showed further decreased cerebrovascular vasoreactivity in the left frontal area, in comparison with the group with white matter hyperintensities without cerebrovascular vasoreactivity (P < 0.05). In the group with cognitive impairment and with white matter hyperintensities lower cerebrovascular vasoreactivity in the left frontal area was significantly associated with lower performance on tests for general cognition (r = 0.311), executive function (r = 0.362), and information processing speed (r = 0.399; all P < 0.05).	Ni,2020 (26);
N=59	N=30 younger healthy individuals (men and women; mean age 30 years); and n= 29 older healthy individuals (mean and women; mean age 65 years).	BOLD- fMRI	Cognitive performance (i.e. memory and attention) was assessed with the Swinburne University Computerized Cognitive Aging Battery.	Age, sex, and years of education	Lower cerebrovascular vasoreactivity in the temporal lobes was significantly associated with a lower memory score (P<0.05). In the older, but not the younger, group lower cerebrovascular vasoreactivity in the hippocampus was significantly associated with a lower memory score (P<0.05).	Catchlove, 2018 (27);
N=162	N=51 individuals with MCI; N=31 individuals with objectively-defined subtle cognitive decline; and N=80 individuals without cognitive impairment.	ASL-MRI	With MCI was defined as follows: >1 SD below the age-/education-/sex-adjusted mean on: (1) two neuropsychological measures within the same cognitive domain, or (2) at least one measure across all three sampled cognitive domains, or (3) a score of 6 or higher on the Functional Activities Questionnaire score. With objectively-defined subtle cognitive decline was defined as follows: 1 SD below the age-/education-/sex-adjusted mean on (1) one impaired total test score in two different cognitive domains (memory, language, attention/executive), or (2) two impaired neuropsychological process scores from the Auditory Verbal Learning Test, or (3) one impaired total test score and one impaired process score. Tests: Memory measures were the Rey Auditory Verbal Learning Test, delayed free recall correct responses and the Auditory Verbal Learning Test recognition test); language measures were: 30-item Boston Naming Test total correct and the Animal Fluency total score); and attention/executive functioning measures were: Trail Making Test Parts A and B.	None	Hippocampal cerebral blood flow was higher in individuals with objectively-defined subtle cognitive decline than in individuals without cognitive impairment (P = 0.007) or with MCI (P = 0.016). However, hippocampal cerebral blood flow did not differ between individuals with objectively-defined subtle cognitive decline and individuals with MCI (P = 0.974). A similar pattern was found for the inferior parietal lobe and the inferior temporal gyrus.	Thomas, 2020 (28);
N=232	N=33 individuals with AD; N= 87 individuals without dementia and who are amyloid positive; and N = 112 without dementia and who are amyloid negative.	ASL-MRI	Global cognitive performance was measured with the MMSE. Verbal memory encoding was assessed using the Rey Auditory Verbal Learning Test recognition subtest. Executive function was evaluated using Trails B of the Trail Making Test.	Age, APOe4 carrier status, BMI, sex, and fludeoxyglucose uptake	Cerebral blood flow was significantly lower in individuals with AD versus amyloid-negative individuals without dementia in the left hippocampus (P = 0.03) and left inferior temporal cortex (P = 0.01). In addition, cerebral blood flow was significantly lower in the inferior parietal cortex in individuals with AD than in individuals without AD who were amyloid positive (P < 0.001) or amyloid negative (P < 0.001). Associations of cerebral blood flow with MMSE score, verbal memory score and executive function score were not shown.	Yew, 2017 (29);
N=182	N=24 individuals with AD; N=66 individuals with early MCI; N=41 individuals with late MCI; and N=51 healthy controls.	ASL-MRI	MMSE	Age, sex and reference cerebral blood flow (precentral cortex cerebral blood flow).	Mini-Mental State Examination score was positively correlated with a higher cerebral blood flow in the entorhinal cortex (P = 0.034), the hippocampus (P = 0.028), and the inferior temporal (P = 0.0072) cortex. Cerebral blood flow was reduced in individuals with AD as compared with control subjects in most regions where effects were expected (entorhinal cortex, P < 0.01; hippocampus, P = 0.01; inferior temporal, P < 0.01; inferior parietal cortex, P < 0.01; posterior cingulate cortex, P = 0.07; precuneus, P < 0.01; medial-orbital frontal cortex, P = 0.03). There was no significant difference in cerebral blood flow between controls and individuals with early or late MCI.	Mattsson, 2014 (30);
N=179	N= 71 individuals with AD; N=35 patients with MCI; and N=73 subjects with subjective memory complaints who visited a memory clinic.	ASL-MRI	MMSE	Age, and sex	Both greater total and regional cerebral blood flow were significantly associated with a higher MMSE score (this association was mainly driven by the strong association between cerebral blood flow and cognitive performance within the AD group, most markedly in the parietal and precuneus and posterior cingulate regions). There were no significant differences in cerebral blood flow between individuals with MCI and individuals with subjective memory complaints (though values of cerebral blood flow were numerically in between those of individuals with subjective memory complaints and AD).	Binnewijzend, 2013 (31);
Blood-brain barrier
Prospective studies
N=51; 2-years follow up.	Patients with lacunar stroke or MCI.	DCE-MRI	Global cognitive performance was estimated from memory, executive function, and information processing speed.	Age, sex, educational level, baseline white matter hyperintensity volume, and baseline brain volume	Greater blood-brain barrier leakage rate was significantly associated with a greater cognitive decline (in normal white matter, standardized beta [95% CI]: 0.65 [0.065–1.23]; and in grey matter: 1.34 [0.64–2.00]).	Kerkhofs, 2021 (32);
N=57 12-years follow-up.	Community-dwelling participants	DCE-MRI	Cognitive performance was estimated from memory function, processing speed, executive functioning, and executive functioning. Memory function was measured using the verbal learning test in which the immediate recall score gives an indication of short-term episodic memory and learning, and the delayed recall score is considered to be a measure of long-term episodic memory. Processing speed was measured with the letter-digit substitution test. Executive functioning was measured with the Stroop color-word test, in which the interference score is considered to be a measure of inhibition.Cognitive decline was calculated by subtracting the participants’ current score from their previous score in the last measure approximately 12 years ago, so that a larger difference score would correspond to more cognitive decline.	Age, sex, and education level	There was a significant association between blood-brain barrier leakage (ki) in the white and grey matter and decline in delayed recall (white matter, beta per 10−6 · min−1 greater leakage 0.389 SD, P = 0.006; grey matter, beta per 10−6 · min−1 greater leakage = 0.287, P = 0.044).	Verheggen,2020 (33);
N=118; 34-months follow up	Community-dwelling individuals	CSF/blood albumin	Cognitive decline was assessed with the Clinical Dementia Rating sum of boxes.	Age, sex, APOe4 status, and baseline clinical dementia rating	Each unit increase in the CSF/blood albumin index was significantly associated with an increase in the Clinical Dementia Rating–Sum of Boxes by 0.09 units (P = .015).	Bowman, 2018 (34);
Cross-sectional
N=80	N=14 patients with AD; N=34 patients with MCI; and N=32 normal controls.	DCE-MRI	Memory and information processing speed.	Age, sex, educational level, lag time between MRI and neuropsychological assessment, diagnostic group	Greater blood-brain barrier leakage rate was significantly associated with lower information processing speed (P<0.05), and in the same direction, but not statistically significantly, associated with lower memory function.	Freeze, 2020 (35);
N=47	N=26 patients with MCI; and N=21 normal controls.	DCE-MRI	Overall cognitive performance assessed with the MoCA.	Age, sex, vascular risk factors, education	Greater leakage rate (per 10−4 min−1) was significantly associated with lower MoCA score (in points) in white matter hyperintensities (beta, −4.363 [95% CI NR], but not significantly associated with lower MoCa score (in points) in normal white matter volume (-4.718 [95%CI NR]); cortical grey matter (-0.571 [95%CI NR]); and in deep grey matter (-1.346 [95%CI NR]).	Li, 2021 (36);
N=75	N=39 patients with MCI; and n=36 normal controls.	DCE-MRI	MMSE, Clinical Dementia Rating Scale global and Sum of Boxes Score.	Education level	In the female MCI group greater blood-brain barrier leakage (Ktrans) of the occipital cortex was significantly associated with lower MMSE score (beta, per 10−3 min−1,−0.397 points, P = 0.025). However, there was not association with the Clinical Dementia Rating Scale global and Sum of Boxes Score. The difference in blood-brain barrier leakage between individuals with MCI and healthy controls was not reported.	Moon, 2021 (37);
N=102	N=36 with low white matter hyperintensity burden; N=35 with medium white matter hyperintensity burden; and N=31 with a high white matter hyperintensity burden. Individuals were enrolled from a neurology clinic.	DCE-MRI	MMSE, MoCA.	Age, sex, education years, and white matter hyperintensity burden	Scores on MMSE and MoCA decreased with increasing leakage rate in white matter hyperintensities (beta per unit increase in Ki −0.857 points, P = 0.029; and −1.492, P = 0.006; respectively) and in deep grey matter (beta, −1.216, P = 0.024; and −1.875, P = 0.012).	Li, 2017 (38);
N=45	N=21 patients with MCI; and N=24 normal controls.	DCE-MRI	Diagnostic criteria were not reported	None	In individuals with MCI, as compared to age-matched older normal controls, there was a significant increase in the blood-brain barrier permeability (Ktrans) in the hippocampus and its CA1 and dentate gyrus regions, but not CA3. In individuals with MCI, as compared to age-matched older normal controls, there were no significant differences in the blood-brain barrier permeability in cortical, subcortical and white matter regions.	Montagne, 2015 (39);
N=70	N=53 patients with vascular cognitive impairment; and N=17 normal controls.	DCE-MRI	Classified as with or without vascular cognitive impairment.	None	In individuals with vascular cognitive impairment DCE-MRI assessed permeability was significantly greater than in healthy controls (P<0.05).	Taheri, 2011 (40);
N=45	N=21 patients with MCI; and N=24 normal controls.	CSF/blood albumin	Diagnostic criteria not reported.	None	In individuals with MCI, as compared to age-matched older normal controls, there was a significant 30% increase in the CSF/blood albumin ratio.	Montagne, 2015 (39);
N=70	N=53 patients with vascular cognitive impairment; and N=17 normal controls.	CSF/blood albumin	Classified as with or without vascular cognitive impairment.	None	The CSF/blood albumin index was significantly higher in the individuals with vascular cognitive impairment than in the healthy controls (P<0.05)	Taheri, 2011 (40);
N=118	Community-dwelling individuals	CSF/blood albumin	Cognitive state was classified using the Clinical Dementia Rating	None	The CSF/blood albumin index was higher in individuals with MCI versus individuals without cognitive impairment (P = 0.005).	Bowman, 2018 (34);

Table 1 shows the association of dysfunction of the neurovascular coupling unit, as quantified by cerebral imaging, with cognitive performance.

Fully adjusted results are shown (where appropriate, i.e. when associations where adjusted for confounders).

AD, Alzheimer’s disease; MCI, mild cognitive impairment; MRI, magnetic resonance imaging; MMSE, mini-mental state examination; MoCA, Montreal Cognitive Assessment; BOLD, blood oxygen level dependent; ASL, arterial spin labelling; DCE, dynamic contrast enhanced; CSF, cerebrospinal fluid; BMI, body-mass index; Apoe4, apolipoprotein E4; NR, not reported; SD, standard deviation; CI, confidence interval.