Assessment of Arterial Stiffness by Cardio-Ankle Vascular Index for Prediction of Five-Year Cardiovascular Events After Coronary Artery Bypass Surgery

The study aim was to investigate the possibility of cardiovascular complications development predicting during a five-year follow-up of patients after coronary artery bypass grafting (CABG) using the cardio-ankle vascular index (CAVI) assessment. Methods: Three hundred and fifty–six patients after elective CABG were enrolled in the study. Prior to surgery, arterial stiffness was assessed in all patients using CAVI. The follow-up was performed five years after the surgery, information was obtained on 238 patients, who were divided into two groups: patients with pathological (≥9.0, n = 88), and normal (<9.0, n = 150) CAVI. Results: Pathological CAVI (≥9.0) was detected in 33% patients before CABG, in stepwise analyses only age and left atrium dimensions statistically significantly predicted CAVI. In patients with pathological CAVI the combined endpoint (major adverse cardiovascular events and hospitalization) and cardiovascular death developed more often in a five-year follow-up after CABG compared with normal CAVI (48.86% versus 34.9%, p = 0.034 and 4.55% versus 0.67%, p = 0.049, respectively). Pathological CAVI (p = 0.021) and the number of coronary bypass grafts (p = 0.023) were independent factors associated with the combined endpoint. Conclusions: Patients with pathological CAVI before CABG surgery are more likely to develop cardiovascular complications and cardiovascular death within a subsequent five-year follow-up. Evaluation of CAVI after CABG in dynamics deserves further study, it is important for monitoring the effects of secondary prevention and the possibility of influencing the prognosis.


Introduction
Arterial stiffness is an integral indicator reflecting the risk factors influence on the vascular wall state. This parameter is used in a clinical setting, as well as for additional risk assessment in population samples [1]. Nevertheless, the traditional indicator of arterial stiffness (carotid-femoral pulse wave velocity -cfPWV) has a number of disadvantages that complicate its practical use -dependence on blood pressure level, lack of measurement standardization, inconvenience for the patient and staff [1]. To overcome these limitations, a new indicator has been proposed and has been used for more than 10 years -the cardio-ankle vascular index (CAVI), which is based on the stiffness parameter β assessment [2]. This measure reflects the arterial stiffness of the local arterial segment and is not influenced by blood pressure during measurement. Accord-

Data collection
Demographic and perioperative patient data were obtained from the electronic database of the institute's CABG registry (certificate No. 2012020868 on registration of CABG database). For each patient who met the inclusion criteria, the following data were collected: age, sex, body mass index, present or absent history of myocardial infarction, hypertension, stroke, diabetes mellitus, heart failure, PCI, CABG, peripheral artery stenosis, dyslipidemia, smoking, creatinine and glucose levels. In addition, intraoperative parameters were recorded, including use of cardiopulmonary bypass; duration of surgery; aortic clamping time; number of grafts; and combined surgeries (ventriculoplasty, thrombectomy, radiofrequency ablation, carotid endarterectomy).

Echocardiography
It was performed on an expert class apparatus 'Vivid-7 Dimension' (General Electric, USA) by two-dimensional echocardiography, Doppler echocardiography in pulsed and continuous wave mode, color Doppler scanning (CDS) in accordance with the current [9]. Diameter and thickness of the LV walls were measured in the two-dimensional M-mode. Left heart analysis included assessment of the size and volumetric parameters of the left ventricle, left ventricular mass, the maximum transverse diameter of the left atrium in the diastole (LA). Ejection fraction of the left ventricle (LVEF) was calculated according to the Simpson method.
Left ventricular diastolic parameters were studied using pulsed-wave Doppler imaging, including peak velocity of early (E) and late (A) transmitral filling of the left ventricle and their ratio (E/A) and the isovolumetric relaxation time of the left ventricle. The study was carried out within the first two days from the moment the patient was admitted to the hospital.

Duplex ultrasonography assessment
The study was carried out on an expert class ultrasound device 'Vivid 7 Dimension' (General Electric, USA). Measurement of the thickness of the intima-media complex was performed in the common carotid artery. The patency of the vessel, the state of the vessel lumen, and the structure of the plaque were assessed. Doppler and B-mode according to the degree of stenosis measured the assessment of the narrowing of the arteries. The criteria for categorizing carotid artery stenosis were defined as: mild stenosis (30-49%); moderate and severe stenosis (50% and more).
Duplex ultrasonography of the lower extremity arteries, from common femoral to pedal arteries was performed for each patient using an expert class ultrasound device 'Vivid 7 Dimension' (General Electric, USA) and a 7.5 MHz linear-array transducer. The test was conducted by an experienced echocardiographer and followed standard imaging protocols. Examination was done with the patient lying supine after resting for at least 15 min. The common femoral and anterior tibial arteries were imaged with the patient in the supine position while those of the popliteal, peroneal and posterior tibial arteries were done in lateral position. The severity of stenosis was graded as follows: 30-49% -mild stenosis; 50% and more -moderate stenosis.
Multifocal atherosclerosis was established in the presence of subclinical lesions of peripheral arteries (stenosis 30% or more, or 50% or more).

Measurement of CAVI
All patients additionally assessed the stiffness of the peripheral arteries -cardio-ankle vascular index (CAVI) using a VaSera VS-1000 device (Fukuda Denshi, Japan) according to the previously described method [10]. This device is a portable sphygmomanometer, four blood pressure cuffs were placed on the limbs; ECG electrodes were attached to the wrists. A phonocardiographic microphone was installed in the second intercostal space on the left. The CAVI index reflects the rigidity of the entire arterial segment. This index is calculated by the device automatically on the right and left (R-CAVI and L-CAVI) and originates from the so-called stiffness parameter ß in combination with a modified the Bramwell-Hill equation CAVI = a {(2p / ∆P) • ln (Ps / Pd) PWV2} + ß, which estimates the relationship between the speed of propagation of the pulse wave and the elasticity of the vascular wall. Considering CAVI, the measurement uses the mean BP of the brachial artery. In addition, when assessing CAVI, the stiffness parameter ß is taken into account, which is defined as the ratio of the natural logarithm of pressure (In [Ps / Pd]) to the degree of change in the inner diameter (D / ∆D) of the vessel. It is known that this parameter does not depend on internal pressure, and the higher ß, the lower the extensibility, and the greater the vascular stiffness.
At the same time, the ankle-brachial index was assessed. With ABI values less than 0.9, the patient was excluded from the study. With a CAVI value ≥9.0 on at least one side, the index was considered pathological.

Follow-up
Patients included in the study for five years were observed by a cardiologist in their residence place. The follow-up was performed five years after the surgery by telephone surveys. We managed to find out information about the health status of 238 (65.2%) patients. To assess the prognosis, we analyzed coronary and non-coronary death, myocardial infarction, acute cerebrovascular accident/transient ischemic attack, percutaneous coronary intervention, carotid endarterectomy, pulmonary embolism, hospitalizations for cardiovascular diseases. The primary endpoint included all of these outcomes (combined endpoint). Secondary endpoints included cardiovascular and total mortality, nonfatal myocardial infarction, and nonfatal stroke.

Statistical analyses
Standard Statistica 8.0 (Tulsa, OK) and SPSS, Version 17.0 (SPSS Inc, Chicago, IL) software were used for statistical analyses. Baseline data and follow-up CABG outcomes were assessed in patients with normal and pathological CAVI values. Qualitative values were presented in absolute numbers (n) and percentage (%), and comparisons between the groups were performed using chi-square tests. The normality of the distribution was verified using the Kolmogorov-Smirnov test. For a distribution other than normal, all quantitative variables were presented as the median and quartiles (lower quartile, upper quartile). When comparing the quantitative data between the two groups, the Mann-Whitney test was used.
Multiple linear regression analysis (enter method and stepwise selection) was applied to evaluate significance correlation of CAVI with preoperative parameters in patients. Performance of preoperative indices for diagnosing the pathological CAVI presence was evaluated through receiver operating characteristic curve analysis. Logistic regression analysis was performed to estimate the odds ratio (OR) and 95% confidence interval (CI) for the effect of possible factors on the primary outcome (the combined endpoint development -total mortality, nonfatal MI, and nonfatal stroke, re-revascularization and hospitalization due to relapse or progression of angina pectoris). Binary logistic regression analysis (Forward LR method) was performed to estimate the effect of independent factors on the primary outcome. P < 0.05 was considered statistically significant.

Patient baseline characteristics
A flow diagram of study participants is presented in Figure 1. During the period from March 2012 to March 2013, 545 patients were admitted to the research institute's clinic for elective CABG. Exclusion criteria included recent acute coronary syndrome (n = 3); installed pacemaker (n = 4), valvular disease (n = 34), LVEF <30% (n = 7); ABI < 0.9 (n = 129), persistent atrial fibrillation/flutter (n = 6); and CABG rejection (n = 6). A total of 356 patients were included in the study. When evaluating the five-year results, it was not possible to  obtain information on 105 patients, 13 patients refused to participate in the study, information was obtained on 238 patients (115 patients -contact by phone, 123 -visit to the clinic). Further analysis was carried out in 238 patients, divided into groups with normal (n = 150) and pathological (n = 88) CAVI. Among patients without information about the five-year follow-up stage, the number of patients with pathological CAVI was comparable to the analyzed group (33% and 37%). When comparing the initial data among those who dropped out at the five-year stage in the groups with normal and pathological CAVI (Supplement Table 1), they were similar among those observed in the analyzed group ( Table 1).   Comparison of demographic data in patients with pathological and normal CAVI showed no gender differences between the groups (p = 0.243); men predominated in both groups. Patients in the group with pathological CAVI were older (p < 0.001), they more often had arterial hypertension (p = 0.011) and diabetes mellitus (p = 0.0043), but less often the prevalence of smoking (p = 0.0107) compared to the group with normal CAVI. Also, in this group, patients without angina pectoris (p = 0.038) and with the presence of multifocal atherosclerosis (p = 0.048) were more often detected (Table 1, Figure 2). For the remaining preoperative parameters (blood biochemical parameters, data on systolic and diastolic left ventricular function, the number of coronary arteries lesions, the severe coronary and heart failure presence) no significant differences were found between groups. We also did not reveal any intergroup differences according to the perioperative data (duration of operations, number of shunts, and frequency of operations with cardiopulmonary bypass). In the group with pathological CAVI, the atherogenic index was higher (p = 0.048).

Relationship between CAVI and preoperative parameters
Multiple regression was run to predict CAVI based on preoperative parameters: age, echocardiographic parameters, the presence of comorbidities, subclinical peripheral arterial lesions, BMI, blood biochemical parameters, the severity of coronary and heart failure, and the number of coronary arteries affected. In the model with the inclusion of all variables (Supplement Table 2 Table 2). The association of preoperative parameters (age, LA dimensions, two-and three-vessel CAD) with the pathological CAVI presence is presented in Figure 3. As shown in Table 3, the areas under the curves were only > 0.7 for age, indicating acceptable discrimination ability. The areas under the curves for the remaining indicators were <0.7, which indicated an unacceptable ability to distinguish. Three-vessel CAD was associated with pathological CAVI, and two-vessel disease -with normal CAVI.

Five-year follow-up CABG adverse events in patients with pathological and normal CAVI
The development of the primary endpoint (combined endpoint -death, stroke/TIA, PCI, CEE, PE, hospitalization for cardiovascular diseases) was more often observed in patients with pathological CAVI compared with patients with normal CAVI (48.86% and 34.9%, respectively, p = 0.034) (Figure 4). Among the secondary endpoints, death from all causes was equally common in both groups (p = 0.389), death from cardiac causes was more common in the group with pathological CAVI (4.55%) than in the group with normal CAVI (0.67%, p = 0.049). There were no statistically significant differences for other adverse cardiovascular events.

Logistic regression analyses
In the univariate analyses (Table 4), the OR for the development of combined endpoint in patients with pathological CAVI compared with patients with normal CAVI was 1.78 (95% CI 1.04-3.06; p = 0.031). The using three bypass grafts as a single predictor predicted a lower risk of CEP, with an OR of 0.8 (95% CI 0.67-0.96; p = 0.015). According to the results of multivariate analysis, the same factors remained independent predictors of the development of CEP: the pathological CAVI presence worsened the prognosis (OR 1.88; 95% CI 1.04-3.4, p = 0.033), and the using three bypass grafts improved it (OR 0.79; 95% CI 0.65-0.95, p = 0.013).

Discussion
In the present study, for the first time it was shown that in patients with pathological CAVI the combined endpoint and cardiovascular death developed more often in a five-year follow-up after CABG. Pathological CAVI and the number of coronary bypass grafts were independent factors associated with the combined endpoint.
In our previous work, the effect of CAVI on the annual prognosis after CABG was assessed. The presence of pathological CAVI was associated with a higher incidence of the combined endpoint (death, myocardial infarction, stroke, hospitalization, angina pectoris), which persisted in multivariate analysis (OR 2.5; 95% CI 1.26-5.08, p = 0.008) [8]. In other studies, the predictive value of pathological CAVI was studied in CAD patients without surgical treatment [5][6][7]. In patients with stable CAD, the incidence of cardiovascular events after 2.9 years in patients with initially elevated CAVI was significantly higher in the group without improvement in CAVI than in the group with its improvement [7]. When observing patients with ACS for 1.25 years, the presence of CAVI > 8.325 in patients was an independent predictor of cardiovascular events and nonfatal ischemic strokes [5]. Kirigaya J et al. [6], when observing patients with acute coronary  syndrome for five years, a significantly higher probability of MACE (cardiovascular death, recurrence of ACS, heart failure requiring hospitalization, or stroke) was demonstrated in the group with high CAVI than in the group with low CAVI (p < 0.001). In multivariate analysis, CAVI was an independent predictor of MACE and death from cardiovascular disease [6]. These results are similar to our study in terms of duration of observation and significant endpoints. A recent study showed that CAVI is one of the independent determinants of the hyperemic microvascular resistance index in patients with non-obstructive coronary artery disease, that is, it is associated with coronary microvascular dysfunction [11]. It was also shown a relationship of CAVI with the presence of obstructive coronary artery disease during CT angiography [12]. In this study revealed that, CAC grade and severe coronary stenosis showed a significant correlation with CAVI. In our study, we failed to identify associations of CAVI with the number of affected coronary arteries, apparently due to more pronounced lesion to the coronary and peripheral arteries.
On the other hand, an increase in the arterial wall stiffness is associated with an increase in afterload on the left heart, which can lead to corresponding morphological changes [3]: an increase in myocardial mass [13][14], the left ventricle diastolic dysfunction [15], the development of heart failure with preserved ejection fraction [16]. When examining individuals without overt cardiovascular disease, high CAVI values were associated with abnormal values of total left ventricular global longitudinal strain in speckle tracking analysis [17]. In the same group of participants, it was noted that the CAVI index was independently associated with the left atrium phase function [18]. The data of the present study on the independent association of CAVI with the left atrium dimention are quite consistent with the above results.
In addition to prognostic value, the assessment of CAVI in CAD patients has another important aspect. The dynamic assessment of this index provides additional opportunities. A recent study suggested that the rapid rise in CAVI in people with high baseline CAVI might be a prodrome of serious cardiovascular events. That is, this factor can mediate the influence of stressful situations on the risk of developing cardiovascular complications [19]. On the other hand, in the absence of a decrease after six months of the initially elevated CAVI in CAD patients, the incidence of cardiovascular events after 2.9 years was significantly higher than in the group with its decrease [7]. Thus, the possibility of improving CAVI in CAD patients has been shown, as well as the possibility of monitoring CAVI in secondary prevention programs in this patients.
Also, subgroup analysis of the TOXO-LIP study showed that a decrease in CAVI during the first year was positively correlated with the development of MACE during five years of follow-up and tended to be an independent predictor of MACE (P = 0.079), this effect was independent of the effect of lowering LDL cholesterol [21]. It remains unclear to what extent changes in CAVI over time may affect the prognosis. A Cardiovascular Prognostic Coupling study is currently underway in Japan to determine the effect of baseline CAVI and changes in CAVI on cardiovascular events [20]. Follow-up is planned for seven years, with an estimate of the time before the onset of a serious cardiovascular event. Data from this registry should provide information on the significance of baseline CAVI and changes in CAVI as indicators of cardiovascular prognosis in patients with at least one risk factor [20]. The significance of such changes in CAVI in patients after CABG remains unexplored, this, apparently, is the task of further research.

Study limitations
Information about the five-year follow-up was obtained in only 2/3 of patients, which could potentially lead to bias in the study results. However, firstly, we additionally analyzed the patients data who dropped out of the study, among them the percentage of pathological CAVI was comparable to patients with fiveyear follow-up data, that is, the very fact of the pathological CAVI presence could not affect the outcomes. Secondly, when comparing patients with pathological and normal CAVI among the dropped out patients, it can be noted that for a number of indicators in the pathological CAVI group, patients had more pronounced deviations (bypass duration, heart failure stage, and left ventricular end diastolic dimension) that could adversely affect the prognosis. That is, the inclusion of these patients in the analysis would rather strengthen the association of pathological CAVI with an unfavorable prognosis, than counterbalance it.
In our study, there is no information on the patients adherence to the therapy recommended upon discharge from the hospital after CABG. Nevertheless, all patients were followed up at the place of residence by a cardiologist and, in this respect, were in equal conditions, regardless of preoperative factors and CAVI values.
Also we did not use a number of additional markers of diastolic function, such as tissue Doppler imaging scores, which could strengthen the identified association of CAVI with left ventricular diastolic function based on analysis of LA dimensions. Since the study design did not provide for an extended protocol for assessing left ventricular diastolic function.