Introduction

Cerebrovascular diseases and cognitive dysfunction are both global health concerns, and these burdens are expected to increase with population aging [1]. The World Alzheimer’s Disease 2015 report shows that the number of people with dementia in Asia is as high as 22.9 million, and this number continues to increase over time [2].

It was reported that the global prevalence of increased carotid intima-media thickness (CIMT) in people aged 30–79 years in 2020 is estimated to be 27.6%, indicating that ~1667 million people are affected (an increase of 57.46% compared with 2000) [3]. Carotid intima-media thickening reflects different severities of the atherosclerotic process [4]. Previous studies have shown inconsistent results regarding whether increased CIMT was associated with cognitive impairment [5,6,7]. Most studies on CIMT and cognitive impairment have been conducted in Western countries, but research in Asian populations is sparse.

Previous studies have also shown a positive relationship between hypertension and cognitive impairment [8], Alzheimer’s disease (AD) [9], and vascular dementia. Of note, the possible effect modifiers for the CIMT-cognitive impairment relationship have not been fully investigated. The current study aimed to explore the association between CIMT and cognitive impairment and examine possible effect modifiers among Chinese hypertensive patients.

Methods

Our article adheres to the American Heart Association Journals’ implementation of the Transparency and Openness Promotion Guidelines. The China Stroke Primary Prevention Trial (CSPPT) was approved by the Ethics Committee of the Institute of Biomedicine, Anhui Medical University, Hefei, China (FWA assurance number: FWA00001263) and registered with Clinical Trials.gov (NCT00794885). All participants provided written informed consent. The current study was approved by the Ethics Committee of the First People’s Hospital of Lianyungang, and the requirement for informed consent was waived due to the retrospective nature of the study and the lack of participant interaction.

Participants

All data were obtained from the CSPPT study. Details regarding the inclusion/exclusion criteria, treatment assignments and outcome measures of the trial have been thoroughly described elsewhere [10,11,12]. Briefly, the CSPPT was a multicenter, randomized, double-blind trial conducted from May 19, 2008, to August 24, 2013, in 32 communities in Jiangsu (20 communities) and Anhui (12 communities) provinces with 20,702 hypertensive adults who were randomly assigned to 2 double-blind treatment groups: 10-mg enalapril+0.8-mg folic acid or 10-mg enalapril, daily.

In this cross-sectional analysis of the CSPPT, we included 15,924 hypertensive individuals, with data on exit carotid artery ultrasonography in 2013, as well as MMSE score. After excluding subjects who did not finish the 30 questions for MMSE (n = 779) and with new-onset stroke or dementia (n = 383), and 3% of the participants with extreme value scores (MMSE scores ≤ 11) (n = 440), a total of 14,322 hypertensive subjects were included in the final analysis (Fig. 1).

Fig. 1
figure 1

Smoothing curves of the association between carotid intima-media thickness (CIMT) and MMSE; (a): male (b): female; adjusted for age, education, body mass index, systolic and diastolic blood pressure, methylenetetrahydrofolate reductase C677T genotype, smoking status, alcohol consumption, total cholesterol, triglycerides, high-density lipoprotein cholesterol, fasting blood glucose, total homocysteine, estimated glomerular filtration rate, total homocysteine, treatment group and research center

Full size image

Assessment of MMSE

Cognitive assessment was conducted using the Chinese version of the Mini-Mental State Examination (MMSE) at the final follow-up visit. The Chinese version of the MMSE has been widely used as a reliable and standardized tool to screen cognitive impairment and dementia [13, 14]. The MMSE is a 30-point questionnaire that estimates the severity of cognitive impairment and can allow for the documentation of serial cognitive changes. The MMSE is often used in community epidemiological studies to assess participants’ current mental function. It examines functions including registration (repeating named prompts), attention and calculation, recall, language, ability to follow simple commands, and orientation [15].

Assessment of CIMT

Carotid artery ultrasonography was performed in 2013 by using a Terason 3000 with an ultrasound scanner equipped with a 12L5A linear-array transducer. Participants were asked to lie on the scanner bed in a supine position with their head resting comfortably and rotate their neck in the direction opposite to the probe at a 45-degree angle. CIMT was measured by certified sonographers with an MIA-Carotid Analyzer 6.0, which allows for semiautomatic edge detection of the echogenic lines of the intima-media complex, from the far walls of the right and left mean CCA when lumen diameter was minimal on multiple cycles of images. The measured segment of interest was 10 mm in length in CCA near the bulb and free of plaques.

The interobserver and intraobserver reliability tests of the CIMT value were reported by another post hoc analysis of CSPPT [16]. In brief, carotid artery ultrasound was performed by 3 observers (1 physician and 2 technicians) who received training prior to the start of the study. Fifty individuals from the entire study population were randomly sampled. Each observer performed the measurements independently at least 4 weeks later, and all measurements were repeated by the 3 observers. According to the Bland–Altman analysis, the average difference in interobserver agreement (95% limits of agreement) was 0.003 (95% CI − 0.113 to 0.122) mm and in intraobserver agreement was 0.049 (95% CI − 0.196 to 0.286) mm [16].

Laboratory tests and clinical data collection

Fasting blood samples were collected from all follow-up patients. Serum total homocysteine (tHcy), total cholesterol, triglyceride, high-density lipoprotein cholesterol (HDL-C), fasting blood glucose (FBG), and serum creatinine levels were measured by automatic clinical analyzers (Beckman Coulter, Brea, CA, USA) at the National Clinical Research Center for Kidney Disease, Nanfang Hospital, Guangzhou. The estimated glomerular filtration rate (eGFR) was calculated with serum creatinine using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation. MTHFR C677T (rs1801133) polymorphisms were detected on an ABI Prism 7900HT sequence detection system (Life Technologies) using the TaqMan assay. Serum folate and vitamin B12 were measured by a commercial laboratory (New Industrial).

A standardized physical examination and accurate medical history were available for all participants. Height and weight were obtained using standard operating procedures. Body mass index (BMI) was calculated as the weight (in kilograms) divided by the squared height. Self-reported smoking and alcohol consumption statuses were recorded, which were coded into three categories: current, former, or never.

Blood pressure was recorded by trained medical staff after the subjects had been seated for 10 minutes. Triplicate measurements on the same arm were taken with at least a 2-min break between readings.

Statistical analysis

The population characteristics are described according to sex. N (%) is used to describe categorical variables, and the mean ± SD is used to describe continuous variables. Characteristics of all participants were compared using analysis of variance (ANOVA) for continuous variables and Kruskal–Wallis rank-sum test for categorical variables. Variables known as traditional or suspected risk factors for carotid atherosclerosis or those showing significant differences across cognitive impairment levels were selected as covariates. Two models were constructed. Model 1 was adjusted for age, sex, and education. The following additional adjustments were performed in Model 2: BMI, systolic blood pressure (SBP), diastolic blood pressure (DBP), lipids [total cholesterol, triglycerides, HDL-C], MTHFR C677T genotype, tHcy, FBG, smoking status, alcohol consumption, folic acid treatment, and research center. Generalized additive regression models and smoothing curves (penalized spline method) were used in the fully adjusted model to estimate the relationship between CIMT and MMSE scores. We performed a multivariate linear regression analysis of CIMT and MMSE in the adjusted model for male and female patients with hypertension, respectively. Next, stratified analyses of potential covariates, including age (<60 vs. ≥60 years), total cholesterol (<5.2 mmol/L vs. ≥5.2 mmol/L), HDL-C [abnormal versus normal: male HDL-C ≥ 1.04 mmol/L, female HDL-C ≥ 1.3 mmol/L] [17], tHcy (median, <12 μmol/L vs. ≥12 μmol/L), and education (illiterate vs. primary vs. secondary and above), were performed in male and female populations.

All tests were two sided, and a p value < 0.05 was considered statistically significant. All analyses were performed using EmpowerStats (www.empowerstats.com; X&Y Solutions, Inc., Boston, MA) and the statistical software package R (http://www.r-project.org).

Results

Baseline characteristics of participants

As shown in the flow chart (Supplementary Fig. 1), 14,322 hypertensive patients with a mean age of 64.2 ± 7.4 years and among whom 40.9% (5854) were male were included in this analysis. The mean BMI of those participants was 25.0 ± 3.8 kg/m2, and the mean CIMT and MMSE scores were 0.74 ± 0.11 mm and 23.5 ± 4.8, respectively. Participant characteristics for all subjects and by sex are listed in Table 1. Compared with females, males were older; had lower BMI, SBP, total cholesterol, triglycerides and eGFR; were more likely to have received longer-term education; and had higher tHcy, current smoking and alcohol consumption rates. No significant difference was found between males and females in terms of FBG, MTHFR C677T genotype or treatment group.

Table 1 Participant characteristics stratified by sex
Full size table

Association between CIMT and MMSE

Overall, in the multivariate regression models, there was a significant negative association between the mean CIMT and MMSE in males (β: −1.26, 95% CI: −2.05, −0.47, P = 0.002) but not in females (Fig. 1 and Table 2). Consistently, when CIMT was assessed in quartiles, the adjusted β values for participants in the third and fourth quartiles were −0.31 (95% CI: −0.56, −0.05) and −0.46 (95% CI: −0.72, −0.19), respectively. However, a significant association was no longer observed in females.

Table 2 Association of CIMT and MMSE for males and females
Full size table

Subgroup analysis

Stratified analysis of potential covariates was conducted in male and female populations. There was a significant interaction (P for interaction < 0.05) of subgroups stratified by age <60 vs. ≥60 years. A significant association of CIMT with MMSE was found among participants, both male and female, among those aged 60 years and above [male: β = −2.29, 95% CI −3.23 to −1.36, P < 0.001; female: β = −1.96, 95% CI −2.97 to −0.95, P < 0.001] (Table 3), while no significant association was found among those aged <60 years. In addition, males with abnormal HDL-C showed a stronger negative association (β = −3.16, 95% CI −4.85 to −1.47, P < 0.001) than those with normal HDL-C (normal vs. abnormal, P for interaction = 0.004) (Fig. 2). Further subgroup analysis defined by potential covariates was performed in males and females stratified by age of 60 years. There were no significant interactions in most subgroups (P for interaction > 0.05), except that HDL-C did appear to interact with the association of CIMT with MMSE in males aged 60 years and above (Supplementary Fig. 2).

Table 3 The association of CIMT and MMSE in males and females stratified by age of <60 years or ≥60 years
Full size table
Fig. 2
figure 2

Subgroup analyses of the association between MMSE score and carotid intima-media thickness (CIMT) in males (a) and females (b). The multivariate model was adjusted for age, education, body mass index, systolic and diastolic blood pressure, methylenetetrahydrofolate reductase C677T genotype, smoking status, alcohol consumption, total cholesterol, triglycerides, high-density lipoprotein cholesterol, fasting blood glucose, total homocysteine, estimated glomerular filtration rate, total homocysteine, treatment group and research center, with the exception of the variable that was stratified

Full size image

Discussion

Our study explored the modification effect of age on the association between CIMT and MMSE scores in both male and female hypertensive populations. CIMT was significantly and negatively associated with MMSE in people aged ≥60 years but not in those aged <60 years. Notably, abnormal HDL-C could improve the negative correlation between CIMT and MMSE in males aged ≥60 years.

Previous studies frequently suggested that thickened CIMT was inversely associated with cognitive impairment and dementia [5, 6, 18,19,20], but exceptions have been seen as well. A cross-sectional study based on the Heinz Nixdorf Recall (HNR) cohort revealed that the ankle-brachial index, but not CIMT, was significantly associated with mild cognitive impairment [21]. The Northern Manhattan Study found that an inverse association of CIMT with cognition existed only among people at higher risk for Alzheimer’s disease [22]. Our study, which focused on Chinese hypertensive adults, demonstrated a significant negative correlation between CIMT and cognitive function only in people aged 60 and older. The mechanism underlying the correlation between CIMT and cognitive function is still unclear. In the Framingham Offspring study, Romero et al. found an inverse association of carotid atherosclerosis with cognitive impairment performance in subjects without stroke, which was independent of structural brain changes [23]. It was suggested that the correlation between IMT thickening and cognitive impairment provides valuable information about vascular risk factors rather than directly reflecting cerebrovascular lesions [23]. A similar negative association of CIMT with cognitive function was reported in a dialysis population, but magnetic resonance imaging of the brain suggested that the impact of CIMT on global cognitive impairment was induced by brain atrophy, not by cardiovascular disease [24]. In animal research, the induction of increased carotid artery intima-media thickness and calcification caused arterial stiffening with a decrease in arterial compliance and dilatation, an increase in the production of cerebral superoxide anion, and neurodegeneration in the hippocampus, which is a key region involved in cognitive function [25].

Notably, the role of age in the relationship between CIMT and cognitive function has not been well recognized in previous studies. Most studies have focused on the association of CIMT with cognition in the elderly, but the results from studies involving the mid-life population are limited and inconsistent. Baseline data from the ELSA-Brasil study found that CIMT was associated with worse memory function in middle-aged adults without stroke [26]. A similar result was also found in The Beaver Dam Offspring Study (BOSS), which showed that CIMT was associated with the MMSE in the population aged 21–84 years after adjusting for multiple potential confounders [27], whereas another longitudinal study reported that CIMT was not a risk factor for cognitive decline among middle-aged adults in the Atherosclerosis Risk in Communities (ARIC) Brain MRI cohort that included middle-aged participants (45–64 years) [28, 29]. In the Coronary Artery Risk Development in Young Adults (CARDIA) study of stroke-free middle-aged adults, higher IMT was not associated with performance on the verbal memory test, which is an important component of cognitive functioning [18]. A recent quantitative study conducted in 182 patients with AD revealed that the most involved cognitive components based on age were recalling language and space orientation [30], while CIMT in our analysis was significantly associated with worse performance on cognitive tests only in participants aged ≥60 years rather than in those aged <60 years. In addition to the direct impact of age factors on MMSE, the inconsistent conclusions might be attributed to the “threshold effect” in CIMT. Previous studies have demonstrated that the significant association of CIMT with cognitive function and dementia was only established in the top quartile [31], even quintile [32], of CIMT. In our study, the CIMT thickness in each of the quartile groups was more severe in participants aged ≥60 years, both for males and females, than in those aged <60 years. This result indicated that CIMT might need to reach a certain severity to obtain statistically predictive ability for cognitive impairment.

HDL-C levels were found to have a modifying effect on the relationship between CIMT and MMSE among males aged ≥60 years in the current study. The significant association of CIMT with MMSE was present in those with abnormal HDL-C. No significant interaction was found in females, but abnormal HDL-C showed a similar trend of a negative association. Wang et al. showed that plasma HDL-C levels were lower in patients with Alzheimer’s disease and vascular dementia than in healthy subjects [33]. HDL-C reduction had a high diagnostic value for Alzheimer’s disease (AUC = 0.731, P < 0.001) and vascular dementia (AUC = 0.800, P < 0.001) in the elderly population. Our study shows that abnormal HDL-C enhances the association of increased CIMT with cognitive impairment. This was demonstrated by epidemiological data showing that HDL-C levels were negatively associated with the risk of cardiovascular disease [34, 35]. It is widely accepted that HDL-C deficiency contributes to cognitive impairment by increasing the risk of atherosclerosis [36]. Moreover, as the predominant lipoprotein in the human brain, HDL-C can maintain cognitive function and delay the onset of dementia by preventing amyloid-β protein aggregation and polymerization [37, 38]. However, no evidence has supported that supplementation with HDL-C improves cognitive function. Thus, additional prospective trials and basic research are needed to confirm the mechanism by which HDL-C affects cognitive function and atherosclerosis.

There are several limitations that are worth mentioning in the current study. First, the sole use of the MMSE test to measure global cognitive function, as the main limitation of this study, may not be accurate enough to detect mild cognitive impairment (MCI). A meta-analysis including 149 studies with 49,000 participants comparing 11 screening tests for dementia indicated that the MMSE has limited performance in identifying MCI, whereas MoCA might be more preferable [39]. Nevertheless, the MoCA may have evident floor effects for participants with low education [40]. Thus, MoCA combined with MMSE may contribute to the persuasive power of our study. Second, given the limitations of our cross-sectional data obtained from the last follow-up of the CSPPT, which means that our participants were treated with strict drug interventions, the blood pressure control rate in CSPPT was well above the levels among general hypertensive patients in China [10, 41, 42]. Furthermore, no data of an unaffected control group could be analyzed. Thus, caution is required in generalizing to general hypertension populations. Third, the population that we analyzed was composed of hypertensive patients, and the above results may not be generalizable to other populations. However, as an essential risk factor for cognitive function, our analysis of hypertensive patients may have a more value than a study of the general population. Finally, participants in our study had a lower education level with a high illiteracy rate, especially among females, which may affect the sensitivity of cognitive assessment. Compared with other cognitive assessment scales, the MMSE has lower educational requirements for subjects [43]. In terms of statistical analysis, we analyzed males and females independently, which can reduce bias in the results caused by differences in education levels between males and females.

Conclusions

We observed that increased CIMT was associated with cognitive decline in the hypertensive population aged 60 years and over. HDL-C deficiency improved the negative association between CIMT and cognitive function in males aged ≥60 years. Although the current study had several limitations, our study undoubtedly suggested that attention should be given to carotid atherosclerosis in high-risk groups with older age and abnormal HDL-C. Overall, upon diagnosis of hypertension, treatment should start at the earliest opportunity to prevent end-organ damage and cognitive decline.