Arterial stiffness and its associations with left ventricular diastolic function according to heart failure types

Heart failure (HF) is a terminal state of almost all heart diseases, and its prevalence continues to increase with age. Mortality and medical costs due to HF are so enormous that they pose a huge burden to our human society [1, 2]. Therefore, it is important to understand the underlying pathophysiology, and to develop a treatment that can prevent the occurrence of HF based on this. For several decades, many effective drugs for HF have been developed, which have greatly improved the survival rate of patients with HF [3‐9]. However, since the mortality rate of HF is still very high, similar to that of some cancers, further efforts to treat HF are continuously required [1, 2].

HF is divided into HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF) according to left ventricular ejection fraction (LVEF) [10]. Although HFpEF is also a clinically serious disease due to its high prevalence and poor prognosis, as in HFrEF, underlying pathophysiology and effective long-term treatment has not been well elucidated [11]. Recently, emerging evidence has shown that increased arterial stiffness plays an important role in the development of HFpEF [12‐14]. However, most of the previous studies that conducted research into this issue analyzed subjects in the stage before clinically overt HF. Additionally, the role of arterial stiffness in HFrEF is still unknown. This study was performed to investigate in the degree of arterial stiffness and its association with left ventricular (LV) diastolic function in patients with HFrEF and HFpEF. We also compared results in patients with HF to control subjects without HF.

Comparisons of clinical characteristics among three groups are shown in Table 1. In overall study patients, mean age was 67.0 ± 12.9 years, and 94 (40.0%) were female. The patients in the HFpEF group were oldest, and the proportion of female patients was highest. The HFpEF group had the highest systolic blood pressure as well as the highest prevalence of cardiovascular risk factors including hypertension, diabetes mellitus, dyslipidemia, previous history of CAD, and stroke. In laboratory findings, patients with HFpEF showed better cholesterol profiles and worse renal function compared to those with HFrEF and control groups. The blood levels of glucose, glycated hemoglobin and C-reactive protein were higher in both HFrEF and HFpEF group compared to control group. HF patients were taking more cardiovascular drugs than the control group. Beta blockers, renin-angiotensin system blockers, and diuretics showed the highest frequency of use in the HFrEF group and calcium channel blockers in the HFpEF group, but there was no difference in the statin use rate among the three groups.

Results of transthoracic echocardiography are demonstrated in Table 2. Patients with HFrEF had the largest LV systolic and diastolic dimensions and LV mass index. The mean LVEF were 67.4% ± 4.6%, 29.7% ± 6.4%, and 63.6% ± 8.1%, in the control, HFrEF, and HFpEF groups, respectively. Compared to the control group, LV diastolic function was more severely impaired in both patients with HFrEF and HFpEF, which was shown by lower septal e′ velocity as well as by higher E/e′, TR Vmax, and LA volume index.

Comparison of baPWV values among three groups is demonstrated in Fig. 2. Mean baPWV values were 1,477 ± 296, 1,661 ± 390, and 1,909 ± 466 cm/sec, in the control, HFrEF, and HFpEF groups, respectively (ANOVA, P < 0.001). In post-hoc analysis, baPWV value was significantly higher in patients with HFrEF than in control subjects (P = 0.006). Also, baPWV value was significantly higher in the HFpEF group compared to the control and HFrEF groups (P < 0.05 for each). Differences in baPWV values between the control group and HFrEF or HFpEF groups were also significant when age-adjusted using ANCOVA analysis (age-adjusted, P < 0.005). However, after adjusting for age, no difference in baPWV values was observed between the HFrEF and HFpEF groups (age-adjusted, P = 0.948). Without stratification by heart failure type, baPWV was significantly higher in heart failure group than in control group (1,771 ± 439 vs. 1,477 ± 297 cm/sec) even after controlling for age (age-adjusted, P < 0.001) (Supplementary Figure S1).

Simple linear correlations between baPWV and diastolic parameters are shown in Fig. 3. The baPWV was significantly correlated with septal e′ velocity in all three groups (r = –0.572, P < 0.05 for the control group; r = –0.226, P = 0.040 for HFrEF group; r = –0.384, P = 0.001 for HFpEF group). baPWV was significantly correlated with E/e′ in control group (r = 0.551, P < 0.001), but not in the HFrEF (r = 0.049, P = 0.657) and HFpEF groups (r = 0.048, P = 0.702). Without stratification by heart failure type, baPWV was significantly correlated with septal e′ velocity (r = –0.213, P = 0.009) but not with septal E/e′ (r = –0.016, P = 0.842) (Supplementary Figure S2).

In the multiple linear regression analysis (Table 3), baPWV was significantly associated with both septal e′ velocity (β = –0.360, P = 0.001) and E/e′ (β = 0.344, P = 0.001) even after controlling for clinical confounders in the control group. However, baPWV was not associated with septal e′ velocity (β = –0.167, P = 0.177) and E/e′ (β = 0.063; P = 0.631) after controlling for confounders in the HFrEF group. The baPWV was associated with septal e′ velocity (β = –0.429, P = 0.002) but not with E/e′ (β = 0.117, P = 0.435) in the HFpEF group in the same multivariable analysis. Without stratification by heart failure type, baPWV was independently associated with septal e′ velocity (β = –0.281, P = 0.004) but not with E/e′ (β = 0.058, P = 0.551) (Supplementary Table S1).

Main findings of this study are as follows: (1) baPWV was significantly higher in patients with HFrEF and HFpEF compared to control subjects; (2) although univariable comparison showed that baPWV was significantly higher in patients with HFpEF than in those with HFrEF, it was similar between patients with HFrEF and HFpEF after adjusting for age; and (3) baPWV was significantly associated with septal e′ velocity and E/e′ in the control group, had no association with either of the LV diastolic indices in patients with HFrEF, and was associated only with septal e′ velocity, but not with E/e′ in patients with HFpEF.

Our results showed that baPWV was significantly higher in patients with HF than in control subjects who had no HF or other documented cardiovascular disease and stroke. The significance of this difference remains even after adjusting for age, a major determinant of arterial stiffness. It may be due to the fact that patients with HF had more various risk factors that could increase arterial stiffness compared to the control group, which was consistent with the characteristics of our study population. We also presented, for the first time, differences in the degree of arterial stiffness according to HF types. Although the baPWV value in the univariable comparison was higher in the HFpEF group than in the HFrEF group, there was no difference between the two groups after age adjustment. Large-scale data are required to verify our findings.

As arterial stiffness increases, LV diastolic function deteriorates. In a stiffened artery, the velocity of reflected wave is increased and merges with the forward wave early [22]. This raises systolic blood pressure and lowers diastolic blood pressure. Increased systolic blood pressure causes LV hypertrophy and decreased diastolic blood pressure reduces coronary perfusion. In addition, with the concept of a shared common risk factors, many cardiovascular risk factors related to arterial stiffening also exacerbate LV diastolic dysfunction [23]. Based on this hypothesis, many existing clinical studies have shown a significant association between increased arterial stiffness and LV diastolic dysfunction in the general population as well as in patients with certain diseases [18, 24‐31]. However, in studies demonstrating such ventricular-arterial (VA) coupling, the study subjects were mostly restricted to the general population or subjects without documented cardiovascular disease including HF [25‐27, 30]. To the best of our knowledge, there was only two studies showing the association between arterial stiffness and LV diastolic function in patients with established HF [28, 31]. Noguchi et al. [28] investigated 44 hypertensive patients with normal LVEF and 31 patients with reduced EF, and showed that cardio-ankle vascular index was correlated with septal e′ velocity in both groups. However, in that study, the definition of HFrEF and HFpEF depended only on LVEF and did not take into account clinical aspects such as hospitalization or symptoms. Additionally, the authors did not perform multivariable analysis [28]. More recently, another study of 107 patients with HFpEF revealed that ambulatory arterial stiffness index was correlated with E/e′ [31]. In our study, baPWV was correlated with e′ velocity in patients with HFpEF but not in those with HFrEF. It seems that the results of each individual study are inconsistent due to differences in the basic characteristics of the study subjects, including race and the method of measuring arterial stiffness. Our study is most meaningful in that it showed a relationship between arterial stiffness and LV diastolic function according to HF types and compared it with the control group.

In the comparisons between control and HF groups, our results showed that the degree of arterial stiffness is more severe in the HF groups, but the association between arterial stiffness and LV diastolic function was stronger in the control group. This implies that arterial stiffness has a greater impact on the diastolic function of LV in the stage before the HF onset, and that the effect is somewhat weakened when HF has already occurred. Therefore, this may suggest that strategy targeting arterial stiffness to improve LV diastolic function or prevent HF [32] should be implanted as early as possible. In addition, baPWV was associated with septal e′ velocity in patients with HFpEF but not in patients with HFrEF. This may be a consistent finding with the results of previous studies showing the important role of arterial stiffness in the development of HFpEF [32]. Our study also showed that baPWV was not associated with E/e′ in either type of HF. It has been suggested that e′ velocity is less affected by the LV loading condition than E/e′; thus, e′ velocity is a more reliable indicator of LV diastolic function [33]. Septal e′ may be a better indicator for response monitoring than E/e′ in treatment strategies targeting arterial stiffness especially in patients with HFpEF.

Our study has several limitations. First, the associations of baPWV with septal e′ velocity and E/e′ were determined with cross-sectional data; therefore, the causal relationship between arterial stiffness and diastolic function could not be confirmed. Second, although carotid-femoral pulse wave velocity (cfPWV) is the gold standard for the non-invasive measurement of arterial stiffness [34], baPWV was used in our study. However, baPWV is more simple to measure and has a good correlation with cfPWV [23]. Third, the unavoidable differences in clinical characteristics among the three groups might affect the study results. In order to overcome this, we enrolled consecutive subjects who visited the same institution during the same period and corrected for important confounding variables through multivariable analysis. Fourth, due to the small number of heart failure patients enrolled in the study, it was difficult to conduct a detailed analysis according to the etiology of heart failure, such as ischemic vs. non-ischemic. Finally, because our study target is limited to Korean adults, it may be difficult to directly apply our findings to other ethnic groups.

Compared to the control subjects, arterial stiffness was increased in HF patients, but the association of the arterial stiffness on LV diastolic function was weaker in HF patients compared to the control subjects. These results suggest early detection and effective intervention for reverse arterial stiffening may limit adverse cardiac remodeling and HF. The degree of arterial stiffness was similar between HFrEF and HFpEF, but the association between arterial stiffness and LV diastolic function was stronger in the HFpEF group. Given that baPWV correlated well with septal e′ velocity in HFpEF, septal e′ velocity could be useful for devising a therapeutic strategy targeting VA coupling. Further large-scale studies are needed to confirm our findings.