Ischemic heart disease (IHD) and stroke remain the leading causes of the global burden of disease in the population older than 49 years.1 Arterial stiffness has been proposed as a noninvasive imaging method to assess arterial wall damage and vascular aging, as well as a predictive biomarker of vascular risk. Although current guidelines do not recommend its use due to measurement difficulties,2,3 a meta-analysis including 10 studies (n=22 472 individuals) showed that carotid arterial stiffness predicts cardiovascular and overall mortality and future cerebrovascular events and improves the predictive capacity of classic risk functions.4

Arterial stiffness can be estimated by measuring regional or local stiffness and several indices are available.3 The most popular index for calculating regional arterial stiffness is pulse wave velocity (PWV), generally the carotid-femoral PWV to estimate aortic stiffness. Local arterial stiffness can be quantified by indices based on changes in arterial diameters and their relationship to changes in arterial blood pressure.5 The linear and nonlinear pattern of the association of these indices with vascular outcomes and mortality has not been properly analyzed to date.

The aim of this analysis was to determine the dose-response association of several measurements of local carotid arterial stiffness with vascular outcomes and overall mortality in a population cohort in a region with a low incidence of cardiovascular disease (figure 1). We also assessed the added predictive capacity value of carotid arterial stiffness indices for those outcomes.

Figure 1.

Central illustration. Summary of the design, aim and main results of the study. D, carotid artery diameter in diastole; D’, carotid artery diameter in systole; DBP, diastolic blood pressure; PP, pulse pressure; SBP, systolic blood pressure.

(0.24MB).

The flow chart of the individuals finally included in the analysis is shown in figure 2 (n=6468; 73.67% of the originally eligible participants included). No clinically significant differences between those included (n=6468) and not included (n=2312) were observed (table 1 of the supplementary data).

Figure 2.

Flow chart of the participants included in this study.

(0.38MB).

The median follow-up was 6.5 years, and 97, 85, 181 and 311 IHD, cerebrovascular, overall vascular, and fatal events, respectively, were observed. These events represent a cumulative incidence rate of 1.50%, 1.31%, 2.80%, and 4.81%, respectively, during the follow-up. The characteristics of the participants included in the study, stratified by the incidence of overall vascular disease, are shown in table 1. Participants with overall vascular disease were older, with a higher proportion of men and higher prevalence of vascular risk factors. They also presented with higher levels of stiffness, Peterson's elasticity coefficient, and PWV and lower strain, compliance, and distensibility coefficients. Participant characteristics stratified by the presence of IHD, cerebrovascular disease, and overall mortality are shown in tables 2 to 4 of the supplementary data. The correlation between carotid stiffness indices and vascular risk factors is shown in table 2.

The dose-response association between carotid artery stiffness indices and the outcomes of interest showed 3 patterns (figure 3; figure 1 of the supplementary data): a) compliance and distensibility coefficients were not associated with any of the outcomes of interest; b) PWV, stiffness, and Peterson elasticity coefficient were linearly and directly related to cerebrovascular and overall vascular disease (figure 3A,B,D); and c) strain showed a nonlinear U-shaped relationship to overall vascular disease and IHD (figure 2C; figure 1C of the supplementary data), with an initial risk decrease up to values of 0.09 and a subsequent risk increase from strain values> 0.09.

Figure 3.

Predicted relative risk (with 95% confidence limits – dotted lines) of cerebrovascular disease, and overall vascular disease related to several stiffness indices (A-D). The graph was produced using a Cox regression model with spline smoothing functions and 2 degrees of freedom. PWV, pulse wave velocity.

(0.5MB).

In the multivariate Cox regression model, the risk of cerebrovascular disease increased by 9% (95%CI, 3% to 16%) per stiffness unit increase, by 8% (95%CI, 2-14) per 10-unit increase in Peterson elasticity coefficient, and by 29% (95%CI, 10-51) per PWV unit increase (table 3). When overall vascular disease was analyzed, the risk increase per corresponding stiffness and Peterson elasticity was 6% (95%CI, 0-12), and 5% (95%CI, 1-10%), respectively (table 3).

As the association between strain and overall vascular disease was not linear, the analysis was stratified by the point that best differentiated the 2 patterns of the association: ≤ 0.09 and> 0.09 (figure 2C, table 3). When strain was ≤ 0.09 units, each 0.01 increase was associated with a lower risk of vascular events (HR, 0.80; 95%CI, 0.69-0.93), whereas when strain was> 0.09, each 0.01 increase in strain was associated with a higher risk of vascular events (HR, 1.09; 95%CI, 1.00-1.19). This nonlinear pattern was mainly dependent on IHD risk: when strain was ≤ 0.09 units, each 0.01 increase was associated with a lower risk of coronary events (HR, 0.85; 95%CI, 0.66-1.08), whereas above this value, each 0.01 increase in strain was associated with a higher risk of coronary events (HR, 1.16; 95%CI, 1.05-1.28). Carotid artery stiffness indices were not associated with all-cause mortality.

The addition of strain, stiffness, Peterson elasticity coefficient, or PWV on validated IHD or cerebrovascular predictive models did not improve the discrimination or the reclassification of the clinical risk models (table 4).

The results stratified by sex (tables 5 and 6 of the supplementary data) and the sensitivity analyses were consistent with those observed in the main analysis.

In this study, we analyzed the linear and nonlinear dose-response association of several indices of local carotid artery stiffness with vascular clinical outcomes and overall mortality. We report an independent, linear, and direct association between cerebrovascular disease and carotid stiffness, Peterson elasticity coefficient, and PWV. We also report an independent and nonlinear association between carotid strain and overall vascular disease and IHD, with a clear change in the direction of the association at a strain value of 0.09: an inverse association below this value and direct association above this cutoff. The addition of these indices to specific coronary or cerebrovascular risk functions did not improve predictive capacity.

The association between local carotid stiffness indices and vascular outcomes has been analyzed in previous studies. Three systematic reviews published from 2015 to 2016 summarize the evidence,4,13,14 supporting the association between stiffer carotid arteries and cerebrovascular disease risk in most of the studies15-19 but not in all of them.20-21 These studies also report a lack of association between most of the carotid stiffness indices and IHD risk.4,13,22

In our study, we incorporated one of the largest sample sizes and one of the longest follow-ups to evaluate the correlation between carotid stiffness and vascular disease. Furthermore, we examined both linear and nonlinear dose-response patterns in this relationship. This analysis is crucial in identifying nonlinear patterns that may define specific stiffness values where the effect size of the association shows exponential increase or decline, reaches a plateau, or even changes the direction of the association. We replicated the linear association between some carotid stiffness indices and cerebrovascular disease. The mechanisms explaining this association have been related to an increased pulsatile pressure and flow load, characterized by increased pulse pressure.4

It is difficult to disentangle whether carotid stiffness is a cause or consequence of vascular dysfunction. On the one hand, antihypertensive drugs reduce pulse pressure, improving vascular function, and on the other hand, increased pulse pressure induces vascular and endothelial dysfunction,23,24 and artery hypertrophic remodeling and rarefaction of small cerebral arteries, decreasing vascular relaxation capacity.25 Other proinflammatory, proliferative, pro-oxidant and hemostatic factors26 and carotid intraplaque hemorrhage through rupture-prone atherosclerotic carotid plaques27 have also been proposed as potential actors in this complex relation between arterial stiffness, vascular dysfunction, and clinical events.

Interestingly, we also report a nonlinear association between carotid strain and overall vascular disease, mainly dependent on IHD risk. Carotid strain has not been previously related to IHD risk,4,14 although usually the nonlinear dose-response relationship has not properly been explored, and a linear relation has been assumed. This finding was unexpected (mainly the direct association observed above values of 0.09 units of strain and IHD risk) and we cannot exclude a spurious association as the local carotid stiffness indices could not reflect the local coronary vascular dysfunction. Nevertheless, this nonlinear pattern of association deserves assessment and potential replication in other studies.

Finally, the inclusion of carotid stiffness indices in specific coronary or cerebrovascular risk functions did not improve the predictive capacity of those functions. A previous meta-analysis with individual data reported an improvement in reclassification (continuous NRI, integrated discrimination index) but not in the discrimination capacity of the original model.4 Therefore, the limited added predictive value along with the lack of standardized techniques and reference values could hamper the usefulness of these indices in clinical practice.

The main strength of our study is the population-based sampling, the standardized methodology and the complete follow-up of the participants. However, there are some limitations. First, despite the recommendations of expert consensus to measure pulse pressure at the site of distension,3 we determined pulse pressure from brachial artery measurements, which may not reflect the local pressure conditions at the carotid arteries. Carotid stiffness indices were measured in 2 laboratories, which could introduce some misclassification of the main exposure variables, even though the same methodology was used. However, no differences between laboratories were observed and the results were consistent in sensitivity analyses. Inter- and intrasonographer variability were evaluated using repeated intima-media thickness (IMT) measurements but not stiffness indices. Cerebrovascular events included fatal and nonfatal stroke, but we did not have information related to the type of stroke.

Carotid stiffness, Peterson elasticity coefficient, and PWV are linearly and directly associated with cerebrovascular disease risk. Carotid strain presents a nonlinear U-shaped relation with IHD risk; an initial inverse association up to values of 0.09 and a direct association above this value were observed. However, the inclusion of these stiffness indexes did not improve the predictive capacity of risk functions.

This work was supported by the Carlos III Health Institute–European Regional Development Fund [CIBERCV]; PERIS from Agència de Gestió d’Ajuts Universitaris i de Recerca [SLT002/16/00088]; and the Government of Catalonia through the Agency for Management of University and Research Grants [2017SGR946].

All the participants were duly informed and signed a written consent form. The study was approved by the Parc de Salut Mar ethics committee (2016/7075/I). Gender dimensions have been considered in the recruitment, data collection and analyses.

During the preparation of this work the authors did not use any artificial intelligence tool.

Substantial contribution to the conception and design of the work: R. Elosua, R. Arnold, E. de Groot, I. R. Degano, J. Marrugat, R. Ramos. Substantial contribution to the acquisition of the data of the work: R. Elosua, R. Arnold, E. de Groot, R. Martí-Lluch, I. R. Degano, J. Marrugat, R. Ramos. Substantial contribution to the analysis and interpretation of the work: R. Elosua, A. Toloba, R. Ramos. Drafting the work: R. Elosua, R. Ramos. Reviewing the work critically for important intellectual content: A. Toloba, R. Arnold, E. de Groot, R. Martí-Lluch, I. R. Degano, J. Marrugat. Final approval of the version to be published: R. Elosua, A. Toloba, R. Arnold, E. de Groot, R. Martí-Lluch, I. R. Degano, J. Marrugat, R. Ramos. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: R. Elosua, A. Toloba, R. Arnold, E. de Groot, R. Martí-Lluch, I. R. Degano, J. Marrugat, R. Ramos.

None.