A single-center, cross-sectional study design was used. Healthy volunteers from the medical examination center of Jiading Branch of Shanghai General Hospital between January 2022 and January 2023 were selected as the study subjects. Demographic information including age, gender, height and weight was recorded for each participant using electronic data capture technology. BMI for each participant was calculated as weight (kg)/square height (m${}^2$), and BSA was defined as 0.0061 $\times$ height (cm) + 0.0128 $\times$ weight (kg) – 0.1529. Healthy volunteers aged 18–80 years with complete echocardiographic data were recruited to the study. Subjects with the following conditions were excluded: (1) age 18 years or $>$80 years; (2) endocrine system diseases such as diabetes mellitus or hyperthyroidism; (3) hypertension with systolic arterial pressure $>$140 mmHg and/or diastolic arterial pressure $>$90 mmHg and a value of 125/80 mmHg on 24-hour ambulatory blood pressure monitoring; (4) hematological disease, malignant tumor, or serious liver, kidney and lung disease; (5) a previous history of cardiovascular disease; (6) patients with atrial fibrillation; (7) congenital heart disease. All subjects were divided into different groups according to age, gender and BMI to explore the effect of these variables on the novel index of ventricular-vascular coupling. Specifically, patients were stratified into three age groups according to the World Health Organization age criteria. These were defined as the young group (44 years), the middle-age group (45 years $\le$ age 60 years) and the old group ($\ge$60 years) [20]. Gender groups were defined as male or female, and BMI groups as overweight/obese (BMI $\ge$24) or controls (BMI 24) according to The Chinese National Health Commission weight criteria [21]. The study protocol was approved by the Ethics Committee of Shanghai General Hospital (2021KY057) and registered on the official website of China Clinical Trial Registration Center (ChiCTR2000035937). All participants provided written informed consent.

A commercially available system (EPIQ7, Philips, S5 probe, Amsterdam, Netherlands) with a frequency of 1–5 MHz, a frame frequency of $\ge$60 frames/sec, and an inspection depth of 13–16 cm was used to acquire the original images. The electrocardiogram was connected and recorded synchronously. Prior to the measurements, subjects were rested for more than 5 minutes to ensure stable hemodynamic conditions. The left ventricular end-diastolic diameter (LVEDD) and left ventricular end-systolic diameter (LVESD) were obtained using M-mode technique. Pulsed Doppler was used in an apical 4-chamber view to measure early mitral orifice diastolic flow velocity (E) and late mitral orifice diastolic flow velocity (A). Tissue Doppler imaging was employed to obtain the motion velocity of mitral annulus (TDI e${}^{\mathrm{\prime }}$) of left ventricular lateral wall in early diastole. The index of left ventricular diastolic function E/A (normal cutoff values: late e${}^{\mathrm{\prime }}$ 10 cm/sec, E/A 0.8) was calculated [22]. Left ventricular end-diastolic volume (LVEDV), left ventricular end-systolic volume (LVESV), stroke volume (SV) and LVEF were measured using the biplane Simpson method.

Mean arterial pressure (MAP) was calculated as [(2 $\times$ diastolic blood pressure (DBP)) + SBP/3] [23]. The left ventricular mass (LVM) was calculated using the formula 0.8 $\times$ 1.04 $\times$ [(LVEDD + IVS + LVPW)${}^3$ – LVEDD${}^3$] + 0.6. The left ventricular end-systolic pressure (LVESP) was calculated as 0.9 $\times$ SBP, the Ees was calculated as LVESP/(LVESV – V${}_0$) = LVESP/LVESV, V${}_0$ is the left ventricular volume when the end-systolic pressure is 0. Ea was calculated as LVESP/SV and the VVI was calculated as Ea/Ees [24].

Transthoracic acquisitions of three consecutive long-axis images (apical four-chamber, three-chamber, and two-chamber) were performed during image capture, with careful attention paid to the absence of arrhythmia during image capture. Quantitative analysis software was used for off-line analysis. Specifically, the endocardial border of the apical four-chamber, two-chamber, and three-chamber views was manually traced and adjusted at end-systole. Longitudinal strain curves were automatically processed and the LVGLS value from all three views was calculated. Negative strain values indicate left ventricular myofiber shortening. To ensure the accuracy and reliability of the measurements, LVGLS was measured independently by two experienced doctors (WL with 10 years of experience, and XL with 12 years of experience) in a blinded fashion and without knowledge of the subjects’ information.

The AVI was obtained using PASESA AVE-2000Pro (Shisei Datum, Tokyo, Japan). Participants were seated and allowed to rest for 5 minutes before placing their left upper arm on the cuff. Their name, age, gender, height, and weight information were inputted. The instrument automatically obtained the AVI after pressing the “measure” button. Each subject’s measurements were repeated at 5-minutes intervals, and the average of three measurements was calculated (Fig. 1).

Fig. 1.

Analysis of AVI and LVGLS. (A,B) Male, 39 years, AVI = 7, LVGLS = –21.5%, AVI/LVGLS = –32.56. (C,D) Female, 48 years, AVI = 9, LVGLS = –22.4%, AVI/LVGLS = –40.18. (E,F) Female, 65 years, AVI = 13, LVGLS = –24.9%, AVI/LVGLS = –34–52.21. AVI, arterial velocity pulse index; LVGLS, left ventricular global longitudinal strain; API, arterial pressure volume index; CSBP, central systolic blood pressure; CAPP, central artery pulse pressure; SYS, systolic; DIA, diastolic.

The proposed VVI index was calculated as the ratio of the arterial stiffness measured by AVI, and the myocardial performance estimated with LVGLS (AVI/LVGLS ratio). This was compared with the widely used VVI calculated by echocardiography as described above.

Twenty subjects were randomly selected and AVI and LVGLS were measured by two physicians for inter-group repeatability. One week later, AVI and LVGLS measurements were repeated by one of the physicians on the same 20 subjects to assess intra-group repeatability.

SPSS 23.0 (IBM, Armonk, NY, USA) statistical software was used for statistical analyses. Normality and homogeneity of variance were tested for quantitative data, expressed as the mean and standard deviation. Comparison of the means between two groups was performed using t-test, while comparison of the means among three groups was performed using one-way analysis of variance. Pairwise comparison of means between groups was conducted using LSD-q test. Qualitative data were represented as examples, and the chi-square test was used for comparison. Pearson correlation analysis was used for univariate correlation analysis, and multiple stepwise linear regression analysis was used for multivariate correlation analysis. Bland-Altman and linear regression analysis were used to assess the repeatability of AVI and LVGLS measurements. The diagnostic efficacy of AVI and Ea, LVGLS and Ees, AVI/LVGLS and VVI for differentiating changes in cardiovascular function in subjects from different age groups was evaluated by ROC curve. A two-sided p-value of 0.05 was considered statistically significant in all analyses.

A total of 239 subjects were included in this study, consisting of 109 males (45.6%) and 130 females (54.4%) with a mean age of 46.16 $\pm$ 13.51 years.

The general factors, cardiac structure and functional indicators of subjects from the different age groups are summarized in Table 1. The old group had 52 subjects, of which 48 (96.2%) were aged 60–69 years and 4 (3.8%) were aged 70–80 years. Compared to the young group, the middle-age and old groups had significantly higher Ea values, indicating increased arterial stiffness. No significant differences in Ees and VVI were observed between the different age groups. AVI increased progressively with age and the highest value was observed in the elderly group, suggesting that arterial stiffness was greatest in this group. LVGLS was lower in both the middle-age and old groups compared with the young group, indicating impaired myocardial function. The AVI/LVGLS ratio was smallest in the young group, followed by the middle-aged and then the old group. The difference in AVI/LVGLS ratio between the three groups was statistically significant, suggesting it can identify and reflect the ventricular-vascular coupling relationship in subjects with different ages (Table 1).

We next investigated possible gender differences in AVI/LVGLS. Males showed significantly greater values for the general factors (height, weight, BMI, SBP, DBP, MAP) and left ventricular structural indices (LVEDD, LVESD, LVEDV, LVESV, SV and LVM) compared to the female subjects (Table 2). Both Ea and Ees were greater in female subjects than in male subjects. No significant gender difference was observed for VVI. Females showed higher LVGLS than males, suggesting their myocardial function was stronger than in men. No significant differences were observed between males and females for AVI and AVI/LVGLS (Table 2).

To investigate the impact of overweight/obesity on ventricular-vascular coupling, we divided subjects into two groups according to a BMI threshold of 24, namely an overweight/obese group (BMI $\ge$24) and a control group (BMI 24). The overweight/obese group had significantly higher values for general factors (height, weight, BSA, DBP and MAP) and left ventricular structural indexes (LVEDD, LVESD, LVED, LVESV, SV and LVM) compared to the control group (Table 3). However, no significant differences were observed between the two BMI groups for traditional VVIs (Ea, Ees and VVI) and for novel VVIs (AVI, LVGLS and AVI/LVGLS) (Table 3).

As shown in Fig. 2A, a significant inverse relationship was observed between AVI/LVGLS and SBP in young subjects (r = –0.399, p 0.05), but not in middle-age and old subjects (r = –0.034 and –0.011, respectively; both p $>$ 0.05). Furthermore, AVI/LVGLS was positively correlated with LVEF in all three age groups (r = 0.428, 0.266 and 0.455, respectively; all p 0.05). A positive association was also observed between AVI/LVGLS and TDI e${}^{\mathrm{\prime }}$ in the young and the middle-age groups (r = 0.308 and 0.331, respectively; p 0.05), but not in the old group (r = 0.297; p $>$ 0.05) (Fig. 2A–C).

Fig. 2.

Correlation between the AVI/LVGLS ratio and SBP, LVEF, TDI e${}^{\mathrm{\prime }}$ in different age, gender and BMI subgroups. AVI/LVGLS was negatively correlated with SBP (A), and positively correlated with LVEF (B) and TDI e${}^{\mathrm{\prime }}$ (C) in different age groups. AVI/LVGLS did not correlate with SBP (D), but was positively correlated with LVEF (E) and TDI e${}^{\mathrm{\prime }}$ (F) in different gender groups. AVI/LVGLS was positively correlated with SBP (G), LVEF (H) and TDI e${}^{\mathrm{\prime }}$ (I) in different BMI groups. AVI, arterial velocity pulse index; LVGLS, left ventricular global longitudinal strain; SBP, systolic blood pressure; LVEF, left ventricular ejection fraction; TDI, Tissue Doppler Imaging; e${}^{\mathrm{\prime }}$, early diastolic velocity of the mitral annulus by TDI; BMI, body mass index.

No significant correlation was found between AVI/LVGLS and SBP in either males or females (r = –0.086 and –0.193, respectively; p $\mathrm{>}$ 0.05). However, AVI/LVGLS showed a positive correlation with LVEF in both males and females (r = 0.424 and 0.286, respectively; p 0.05), and also with TDI e${}^{\mathrm{\prime }}$ (r = 0.437 and 0.596, respectively; p 0.05) (Fig. 2D–F).

AVI/LVGLS was not significantly correlated with SBP in either the control or overweight/obese groups (r = –0.171 and –0.103, respectively; p $>$ 0.05), but was positively correlated with LVEF (r = 0.290 and 0.503, respectively; p0.05) and with TDI e${}^{\mathrm{\prime }}$ (r = 0.579 and 0.363, respectively; p 0.05) in both BMI groups (Fig. 2G–I).

In all subjects, AVI/LVGLS was negatively correlated with age, SBP, LVEDV, LVESV, Ea and VVI, (r = –0.454, –0.153, –0.149, –0.323, –0.132 and –0.371, respectively; all p 0.05), and positively correlated with LVEF, E/A, TDI e${}^{\mathrm{\prime }}$ and Ees (r = 0.367, 0.361, 0.504 and 0.146, respectively; p 0.05). Age, LVEDV, LVESV and TDI e${}^{\mathrm{\prime }}$ were identified as factors that influence AVI/LVGLS ($\beta$ = –0.482, –1.492, 2.953 and –8.365, respectively; p 0.05) (Table 4).

AVI/LVGLS had significantly superior diagnostic performance compared to VVI in young and middle-age subjects, with areas under the ROC curve of 0.709 and 0.519, respectively (p 0.05) (Fig. 3G). However, no significant difference in diagnostic performance was found between AVI and Ea in young and middle-age subjects (0.721 and 0.754, respectively; p $>$ 0.05; Fig. 3A), nor between LVGLS and Ees (0.587 and 0.585, respectively; p $>$ 0.05; Fig. 3D).

Fig. 3.

ROC curves for the parameters of arterial and ventricular coupling. ROC curves for Ea and AVI in young and middle-age groups (A), young and old groups (B), and middle-age and old groups (C). ROC curves for Ees and LVGLS in young and middle-age groups (D), young and old groups (E), and middle-age and old groups (F). ROC curves for VVI and AVI/LVGLS in young and middle-age groups (G), young and old groups (H), and middle-age and old groups (I). Ea, effective arterial elastance; Ees, left ventricular end-systolic elastance; AVI, arterial velocity pulse index; LVGLS, left ventricular global longitudinal strain; VVI, ventricular-vascular coupling index; ROC, receiver operating characteristic curve; AUC, area under curve.

In the young and elderly subjects, the diagnostic efficacy of AVI/LVGLS was higher than that of VVI (Fig. 3H), while that of AVI was higher than Ea (Fig. 3B). There was no significant difference in diagnostic efficacy between LVGLS and Ees (Fig. 3E).

In middle-age and old subjects, no significant differences in diagnostic efficacy were found between AVI/LVGLS and VVI, AVI and Ea, and LVGLS and Ees (all p $>$ 0.05) (Fig. 3C,F,I, respectively).

The repeatability of AVI was evaluated by inter- and intra-group comparisons. These revealed a high consistency in the measurements [between: R${}^2$ = 0.206, p 0.05, mean difference (0.15 $\pm$ 3.06); within groups: R${}^2$= 0.569, p 0.05, mean difference (–0.05 $\pm$ 2.55)]. The Bland-Altman plots indicated the 95% limits of agreement were –6.19–6.49 for inter-observer, and –5.04–4.94 for intra-observer comparisons (Fig. 4).

Fig. 4.

Repeatability test of AVI and LVGLS by Bland-Altman plot and linear correlation analysis. Bland-Altman analysis (A,B) showed a consistent trend and linear regression analysis (E,F) showed good agreement for AVI inter-observer and intra-observer respectively. Bland-Altman analysis (C,D) showed a consistent trend and linear regression analysis (G,H) showed good agreement for LVGLS inter-observer and intra-observer respectively. AVI, arterial velocity pulse index; LVGLS, left ventricular global longitudinal strain.

Similarly, the repeatability of LVGLS was assessed by between- and within-group comparisons. These revealed a high consistency in the measurements [between: R${}^2$ = 0.221, p 0.05, mean difference (–2.98 $\pm$ 1.87) %; within groups: R${}^2$ = 0.262, p 0.05, mean difference (0.16 $\pm$ 1.47) %]. Bland-Altman analysis further confirmed the high consistency of LVGLS repeat measurements (Fig. 4).

The main findings of this study were: (1) AVI, LVGLS, and AVI/LVGLS showed age-related changes, with AVI and AVI/LVGLS increasing with age and LVGLS decreasing. Our study suggests that arterial stiffness increases with age, while left ventricular systolic function decreases and the ratio of the two increases, thereby reflecting age-related changes in the cardiovascular system. AVI/LVGLS was found to increase with age and correlated inversely and independently with age. LVESV and Ea were positively correlated with TDI e${}^{\mathrm{\prime }}$, which is an indicator of left ventricular diastolic function. (2) Males had lower Ea, Ees and LVGLS than females, while AVI and AVI/LVGLS did not differ between the genders. (3) AVI, LVGLS and AVI/LVGLS are novel VVI systems that are more effective than traditional indexes at detecting differences in cardiovascular function in different age groups.

AVI provides a more direct assessment than Ea of the total elastance of the arterial tree and the resistance of peripheral arteries, both of which are closely related to cardiac function. The present study revealed that the novel arterial stiffness index AVI was more sensitive than Ea in identifying the degree of arterial aging in subjects with different ages. Furthermore, it was more stable in its assessment of differences in arterial elastance between genders. Arterial stiffness increases with age [9] and is also promoted by various factors associated with aging, such as increased arterial wall stress, oxidative stress, and inflammatory stimuli. These factors decrease the elastic fibers in arteries that maintain vascular compliance, whereas the collagen fibers increase. This leads to changes in the structure and function of the arterial wall, resulting in increased vascular stiffness and decreased compliance [25, 26]. AVI, a novel index of arterial stiffness, was found in this study not only to reflect the stiffness of the central artery, but also to correlate significantly with total peripheral vascular impedance and left ventricular contractility [15]. Previous studies have shown that AVI is a sensitive predictor of the risk of subclinical coronary atherosclerosis without significant obstruction in the lumen [22]. Furthermore, AVI is independently associated with the concentration of plasma atrial natriuretic peptide and with a history of congestive heart failure [27], suggesting that enhanced AVI may reflect an increase in cardiac workload. The present study also demonstrated that Ea could differentiate vascular elastance between young and middle-age individuals, but not between middle-age and old individuals. This result suggests that Ea has limitations for evaluating the elastance of arteries in middle-age and elderly people. Ea reflects the elastance of the arterial tree and the resistance of peripheral vessels, as well as reflecting the overall afterload of the heart [4]. It is determined by the formula Ea = LVESP/SV, which reflects only the reserve function of the heart SV [24]. Ea may have reduced diagnostic efficacy in middle-age and elderly individuals due to a decrease in cardiac stroke reserve [9]. Additionally, the inadequate assessment of systemic common artery elastance by Ea limits its clinical application [24].

LVGLS has greater sensitivity than Ees for the evaluation of impaired cardiac systolic function [5]. Of note, LVGLS has the ability to differentiate cardiac function across different age groups, as well as between genders. Various studies have confirmed that LVGLS enables earlier detection of impaired systolic function and can thus function as an early warning for cardiovascular and cerebrovascular events [28, 29]. LVGLS, as a contrasting predictor of cardiovascular events, has been incorporated into guidelines for the early assessment of cardiac ventricular disorders [30]. Reduced LVGLS serves as a partial manifestation of impaired systolic function in patients with HFpEF [31]. Moreover, a follow-up study of 4172 patients with acute heart failure revealed that LVGLS alone was an independent predictor of all-cause mortality at 5 years, regardless of whether or not LVEF was normal [32]. Myocardial compliance decreases as cardiomyocytes undergo apoptosis or necrosis with aging and as the extracellular matrix undergoes fibrotic remodeling, thus leading to functional changes [33]. The present study found that Ees was higher in middle-age and old subjects compared to young subjects. Ees reflects ventricular contractility and systolic stiffness, and is an essential parameter for assessing cardiac systolic function and hemodynamic status. However, Ees does not take into account myocardial geometry and tissue structure (such as cardiomyocytes and elastic fibers), both of which have a significant impact on ventricular systolic and diastolic function [6]. The present study found that ventricular end-diastolic volume decreased with increasing age, while the ejection fraction remained unchanged. This finding explains why Ees behaves differently across different age groups and in obese individuals.

An increased AVI is indicative of increased arterial stiffness, while decreased absolute LVGLS values suggest impaired ventricular myocardial systolic function and are also associated with diastolic dysfunction [31]. The present study examined the interaction between AVI and LVGLS, which represents ventricular-vascular interactions. Possible reciprocal mechanisms include increased arterial stiffness, decreased peripheral vascular compliance, increased ventricular afterload, decreased coronary perfusion and myocardial oxygen delivery. These result in cardiomyocyte hypertrophy, fibroblast growth and interstitial fibrosis, leading to ventricular hypertrophy, and systolic and diastolic dysfunction [7]. Based on this, the AVI/LVGLS ratio is closely associated with left ventricular diastolic function indicators such as E/A and TDI e${}^{\mathrm{\prime }}$. We found no significant differences in VVI between different age groups, whereas AVI/LVGLS gradually increased with age. Higher AVI was associated with higher vascular stiffness in the elderly, which may be accompanied by subclinical left ventricular dysfunction. This leads to lower LVGLS and subsequently to an increased AVI/LVGLS ratio. In addition, we compared the diagnostic performance of AVI/LVGLS and traditional VVI in different age groups. AVI/LVGLS was found to have high area under curve (AUC), sensitivity and specificity, and a superior diagnostic performance to that of VVI. Therefore, AVI/LVGLS is a valuable tool for early detection of subclinical cardiac-vascular coupling mismatch.

It is worth noting that the AVI/LVGLS ratio, a novel index system for ventricular-vascular coupling, is consistent with previous VVI obtained through echocardiography. Moreover, in healthy populations it does not differ significantly according to gender or BMI. There are several potential explanations for this observation. Firstly, overweight/obese individuals and those with metabolic diseases may have impaired subclinical myocardial contractility which can affect LVGLS [30]. Studies have shown that bariatric surgery can improve LVGLS in severely obese individuals (BMI $\ge$35), and that subclinical myocardial dysfunction is associated with visceral adiposity and inflammatory markers rather than with BMI or fat mass [34, 35]. Secondly, gender differences may play a role, as LVGLS is typically higher in females than males. This is due to more deformation of the smaller LVM in females, resulting in increased intraventricular blood flow efficiency and decreased energy expenditure [36, 37]. In contrast, males may compensate for relatively low blood flow efficiency by increasing kinetic energy. Thus, the interaction between arteries and ventricles as reflected by the AVI/LVGLS ratio appears to conform to physiological characteristics and is not solely determined by numerical values.