Effects of exercise modalities on central hemodynamics, arterial stiffness and cardiac function in cardiovascular disease: Systematic review and meta-analysis of randomized controlled trials

Background Exercise is accepted as an important contribution to the rehabilitation of patients with cardiovascular disease (CVD). This study aims to better understand the possible causes for lack of consensus and reviews the effects of three exercise modalities (aerobic, resistance and combined exercise) on central hemodynamics, arterial stiffness and cardiac function for better rehabilitation strategies in CVD. Methods The electronic data sources, Cochrane Library, MEDLINE, Web of Science, EBSCO (CINAHL), and ScienceDirect from inception to July 2017 were searched for randomized controlled trials (RCTs) investigating the effect of exercise modalities in adult patients with CVD. The effect size was estimated as mean differences (MD) with 95% confidence intervals (CI). Subgroup analysis and meta-regression were used to study potential moderating factors. Results Thirty-eight articles describing RCTs with a total of 2089 patients with CVD were included. The pooling revealed that aerobic exercise [MD(95%CI) = -5.87 (-8.85, -2.88), P = 0.0001] and resistance exercise [MD(95%CI) = -7.62 (-10.69, -4.54), P<0.00001] significantly decreased aortic systolic pressure (ASP). Resistance exercise significantly decreased aortic diastolic pressure [MD(95%CI) = -4(-5.63, -2.37), P<0.00001]. Aerobic exercise significantly decreased augmentation index (AIx) based on 24-week exercise duration and patients aged 50–60 years. Meanwhile, aerobic exercise significantly improved carotid-femoral pulse wave velocity (cf-PWV) [MD(95%CI) = -0.42 (-0.83, -0.01), P = 0.04], cardiac output (CO) [MD(95% CI) = 0.36(0.08, 0.64), P = 0.01] and left ventricular ejection fraction (LVEF) [MD(95%CI) = 3.02 (2.11, 3.93), P<0.00001]. Combined exercise significantly improved cf-PWV [MD(95%CI) = -1.15 (-1.95, -0.36), P = 0.004] and CO [MD(95% CI) = 0.9 (0.39, 1.41), P = 0.0006]. Conclusions Aerobic and resistance exercise significantly decreased ASP, and long-term aerobic exercise reduced AIx. Meanwhile, aerobic and combined exercise significantly improved central arterial stiffness and cardiac function in patients with CVD. These findings suggest that a well-planned regime could optimize the beneficial effects of exercise and can provide some evidence-based guidance for those involved in cardiovascular rehabilitation of patients with CVD.

Introduction Cardiovascular disease (CVD) is the leading cause of death and the main risk factor for worldwide morbidity [1,2]. According to the World Health Organization global status report on Non-communicable Diseases (NCD) in 2014, 38 million of the world's 56 million deaths are from NCDs. Figures from the same report suggest that in 2012 an estimated 17.5 million (46%) of these deaths are due to CVD. Of these deaths from CVD, heart attacks are responsible for 7.4 million, and stroke, for 6.7 million [3]. Therefore, low-cost and effective prevention and treatment are urgently needed.
Insufficient physical activity is considered the fourth leading risk factor for global deaths, and in 2010, was responsible for 69.3 million Disability Adjusted Life Years (DALYs) [3,4]. Regular physical activity is accepted as an important contribution to the prevention and rehabilitation of CVD [5,6]. Exercise-based cardiac rehabilitation (aerobic endurance training, dynamic resistance training and both in combination) have been used to manage cardiovascular health in individuals with CVD [6].
Central hemodynamics and arterial stiffness have been recognized as strong independent predictors of all-cause mortality of cardiovascular (CV) events [7][8][9]. These parameters, such as central blood pressure, augmentation index (AIx) and carotid-femoral pulse wave velocity (cf-PWV)) were used to evaluate the exercise-based rehabilitation of patients with CVD [10][11][12]. Previous meta-analyses have reported the effects of exercise training on arterial stiffness. However, these analyses mainly focused on a range of adult subjects, including patients with CVD, diabetes, obesity and healthy people [13,14]; or the effect of one type of exercise training (such as aerobic or resistance exercise) on arterial stiffness [10,15,16].
In addition, some studies have investigated the effect of two or three types of exercise training on central hemodynamics and arterial stiffness. Croymans et al. found that aortic systolic blood pressure was decreased, while AIx and cf-PWV were not altered in response to highintensity resistance exercise [17]. Ashor et al. concluded that AIx and PWV were significantly reduced with aerobic exercise, while resistance or combined exercise had no significant effect on these variables [13]. Figueroa found that resistance exercise had no clear cut effects on central blood pressure and wave reflection in obese adults with prehypertension [18]. There was no consensus about the effects of different exercise modalities on the central hemodynamic and central arterial stiffness variables.
Moreover, central hemodynamics and central arterial stiffness are closely related to cardiac function [19,20]. Arterial stiffness, central hemodynamics and cardiac function contribute to the complex pathophysiological mechanism of CVD. Increased arterial stiffness (as expressed by PWV) leads to early arrival at the heart of reflected waves from peripheral sites, resulting in augmentation of central systolic pressure. This augmentation of central systolic pressure can lead to adverse changes in cardiac function such as elevation of left ventricular afterload and decreased coronary perfusion [21][22][23], which may in turn, lead to left ventricular hypertrophy and myocardial ischemia [23,24]. Previous studies have investigated these changes and reported the effects of exercise training on arterial stiffness/central hemodynamics, cardiac output (CO) and left ventricular ejection fraction (LVEF) in patients with CVD. However, it was controversial for the effects of exercise training on these parameters in CVD. Kitzman et al. found that exercise training did not increase the ejection duration (EF) or improve arterial stiffness [25], and Chrysohoou et al. showed that PWV and LVEF was not improved in response to combined exercise [26]. On the other hand, Molmen-Hansen et al. found that aerobic exercise increased CO, LVEF, and decreased blood pressure and total peripheral resistance (TPR) in patients with hypertension [27]. Understanding of the effects of different exercise training on the central hemodynamics, central arterial stiffness and cardiac function merits comprehensive analysis and evaluation.
Therefore, this systematic review and meta-analysis aimed to investigate the effect of different exercise modalities on central hemodynamics, central arterial stiffness and cardiac function in patients with CVD. Additionally, this meta-analysis provided an overall assessment of effect of different exercise modalities on cardiovascular system to evaluate the possible causes for aforementioned lack of consensus.

Protocol and registration
This meta-analysis was performed according to the guidelines of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). A completed PRISMA checklist was shown in S1 Text. The protocol of this study has been recorded in http://www.crd.york.ac.uk/ PROSPERO. PROSPERO registration number: CRD42016052379.
Search strategy. The search for relevant studies was performed via electronic searches of five databases (Cochrane Library, MEDLINE, Web of Science, EBSCO (CINAHL), and Scien-ceDirect from their inception to July 2017). This meta-analysis was only limited to RCTs. The electronic search strategies for all databases are provided in S2 Text. We also searched for eligible articles in reference citation of reviews and research articles.
Inclusion criteria.
stiffness and cardiac function, respectively. Subgroup analyses were also performed to explore the source of heterogeneity according to age, disease, exercise duration and gender. Metaregression was used to explore the relationships between study characteristics (such as age, disease, exercise duration and gender) and cardiovascular variables by using Stata software (version 12.0). In addition, funnel plot asymmetry estimation was conducted to evaluate possible publication bias using Egger's regression test [31].
Effect of different exercise modalities on the central hemodynamics-Aortic systolic pressure (ASP). Based on a fixed effects model, ASP was significantly improved by aerobic exercise [MD (95% CI) = -5.87 (-8.85, -2.88), P = 0.0001] and resistance exercise [MD (95% CI) = -7.62 (-10.69, -4.54), P<0.00001] for the exercise group when compared to the control group (Table 3 and Fig 2). There was no reduction in ASP in subjects undertaking combined exercise in patients with CVD [MD (95% CI) = -3.82 (-13.07, 5.43), P = 0.42] in this subgroup analysis due to one included study. Subgroup analyses of aerobic exercise according to age, disease, exercise duration and gender were listed in Table 4.
Aortic diastolic pressure (ADP). Using a fixed effects model, ADP was not significantly improved by aerobic exercise [MD (95% CI) = 0.06 (-2.19, 2.31), P = 0.96] or combined exercise [MD (95% CI) = -4.2 (-11.49, 3.09), P = 0.26] (Table 3 and Fig 3). However, Resistance exercise was found to reduce ADP by 4 mmHg [MD (95% CI) = -4 (-5.63, -2.37), P<0.001] for the exercise group when compared to the control group (Table 3 and Fig 3). Subgroup analyses of aerobic exercise according to age, disease, exercise duration and gender were reported in Table 4.  Augmentation index (AIx). A random effects model was selected as there were larger fluctuation of the variance between each included study in AIx. There was no significant difference in AIx after aerobic and resistance exercise in CVD, while there was a significant difference in AIx after combined exercise in this subgroup analysis due to one included study (Fig 4  and Table 3). Subgroup and regression analyses of effect of aerobic exercise on AIx were conducted according to age, disease, exercise duration and gender (Table 4). AIx was significantly decreased by 24-week aerobic exercise [MD (95% CI) = -12.00 (-19.92, -4.08), P = 0.003] or in patients aged 50-60 years [MD (95% CI) = -8.13 (-14.31, -1.95), P = 0.01]. AIx was significantly increased with combined exercise in patients with chronic heart failure (CHF) when compared to a control group with one included study. According to the meta-regression analysis, AIx was not affected by age (P = 0.981), gender (P = 0.703), disease (P = 0.206) and exercise duration (P = 0.133). In this meta-analysis, of 10 trials (aerobic exercise) that measured AIx, only 4 studies (152) reported heart rate-adjusted AIx. This result found that there was no significant difference [MD (95% CI) = 2.95 (-0.77, 6.67), P = 0.12] in heart rate-adjusted AIx after aerobic exercise (S1 Fig). Exercise rehabilitation and cardiovascular disease   (Table 3 and Fig 5). The result (mean difference or 95%CI or test for overall effect) was affected by one study [26] for cf-PWV after combined exercise in this sensitivity analysis. Therefore, this meta-analysis may provide weak evidence of the effect of combined exercise on cf-PWV. Subgroup analyses of aerobic exercise according to age, disease, exercise duration and gender were listed in Table 4.
Effect of different exercise modalities on cardiac function-Cardiac output (CO). Using a fixed effects model, seven studies with 372 patients were included to evaluate the effect of exercise on CO. Aerobic exercise and combined exercise were found to increase CO by 0.36 L/min [MD (95% CI) = 0.36 (0.08, 0.64), P = 0.01] and 0.9 L/min [MD (95% CI) = 0.9 (0.39, 1.41), P = 0.0006] in the exercise group when compared to the control group (Table 3 and Fig  6), respectively. However, CO was not significantly improved by resistance exercise in patients with CVD [MD (95% CI) = -0.02 (-0.6, 0.56), P = 0.95]. Subgroup analyses of aerobic exercise according to age, disease, exercise duration and gender were reported in Table 4. Age and disease affected the change of CO after aerobic exercise.
Left ventricular ejection fraction (LVEF). This meta-analysis of nineteen studies with 843 patients, using a fixed effects model, showed that aerobic exercise increased LVEF  7). A sensitivity analysis was conducted for LVEF after aerobic exercise, and the significance of the difference between the exercise and control groups was not changed when studies were removed 1 by 1. Meta-regression was used to explore the higher heterogeneity (I 2 = 48%) and showed evidence of a relationship between mean difference of LVEF and age, for patients with CVD (Beta = -3.47, P = 0.000, Fig 8A and Table 4), while mean difference of LVEF was not related to changes of exercise duration (Beta = 0.59, P = 0.389, Fig 8B and Table 4), disease (Beta = -0.55, P = 0.954, Table 4) and gender (Beta = 1.96, P = 0.276, Table 4).

Discussion
This meta-analysis, which gathered 2089 patients with CVD from 38 articles, provides evidence of the effects of exercise and differences between three types of exercise on central hemodynamics, central arterial stiffness and cardiac function. The results suggest that in patients with CVD, aerobic exercise significantly improved ASP, cf-PWV, CO and LVEF. Resistance exercise significantly reduced ASP and ADP. Combined exercise significantly improved cf-PWV and CO in patients with CVD. We have not found any meta-analysis or systematic review that had evaluated the effects of different exercise modalities on central hemodynamics, arterial stiffness and cardiac function of patients with CVD, and therefore there was no systematic description of evidence for these differences. Previous studies have focused primarily on either aerobic or resistance exercise rather than combined, or have considered only a few measures of arterial stiffness but not those associated with central blood pressure, AIx and cardiac function. In this study, only RCTs of patients with CVD were included, and the impact of these two types of exercise, both singly and combined, on a wider range of cardiovascular variables were investigated.   Exercise rehabilitation and cardiovascular disease Therefore, this study provides a broader evaluation of the effects of different exercise modalities on central hemodynamics, central arterial stiffness and cardiac function in patients with CVD as well as their application for the rehabilitation of CVD.
In this meta-analysis, we observed significant changes, in response to aerobic and resistance exercise, in central BP, (this being more predictive for target organ damage), cardiovascular morbidity and mortality in comparison with brachial BP [66,67] in patients with CVD. However, there was no significant difference in ASP or ADP after combined exercise in this subgroup analysis due to one included study [56]. Previous studies have supported the notion that aerobic or resistance exercise reduces ASP due to improvements in vasoactive substances and endothelial function [17,18,37,68], and agreed that there was no significant difference in ADP with aerobic exercise [59]. However, a study showed unfavorable effect of resistance exercise on central blood pressure and arterial compliance [69]. The high blood pressure in response to resistance exercise may be mainly related to arterial stiffness [70]. At the same time, increased arterial stiffness was associated with the promotion of vascular smooth muscle cell growth and inflammatory cytokines, increasing the central blood pressure [71]. However, several other mechanisms for the favorable effect of resistance exercise on central blood pressure have been proposed. Croymans et al. suggested that the beneficial effect of resistance exercise on central blood pressure is due to improved endothelial function and microvascular perfusion [17]. Whereas, Heffernan et al. found that resistance exercise significantly improved central blood pressure because of its impact on the reservoir pressure, which is proportional to the volume of blood stored in the aorta, and which in turn depends on the interactions of systemic arterial compliance and impedance to outflow [11,72]. However, Figueroa and Taaffe reported that it may be attributed to improved peripheral muscular artery dilation and peripheral vascular resistance (PVR) [18,68]. The reductions of PVR and arteriolar tone with Exercise rehabilitation and cardiovascular disease resistance exercise may change terminal impedance enabling greater runoff into peripheral microvascular beds during diastole (inflow< outflow) resulting in sustained reductions in reservoir pressure [72]. In the subgroup analysis of different exercise modalities, only one included study investigated effect of combined exercise on central blood pressure, with a weak evidence for this effect. Additionally, although there was no significant difference in effect of combined exercise on ASP due to one limited study (P = 0.42), different exercise modalities significantly decreased ASP according to the overall effect of total subgroup (P<0.00001).
In this study, subgroup analysis and meta-regression analysis of studies (effect of aerobic exercise on AIx) were conducted according to age, disease, exercise duration and gender. AIx was significantly decreased in response to 24-weeks aerobic exercise or in patients aged 50-60 years. However, the change of AIx was not affected by gender, disease and HR after aerobic exercise intervention. The previous work found that different durations and different age were associated with the effect of exercise training on the cardiovascular health [73]. In addition, some studies also reported that AIx may be influenced by age or disease, as arterial reservoir pressure increases rapidly with age and disease [74,75]. Furthermore, AIx may be affected by other factors including LV afterload, exercise intensity, frequency and duration, central wave reflection, forward wave genesis and impedance [11,13,75,76]. AIx was not only affected by wave reflection from pulse wave, but also affected by compliant properties of elastic arteries [77]. Combined exercise significantly increased AIx in patients with CHF in only one included study [26]. We found that aerobic exercise with resistance exercise just improved the aerobic capacity of patients with CHF [78], while a study found that increased AIx was related to improved ventricular-aortic coupling in response to combined exercise, and increased AIx and PWV can reveal the improved arterial function [26]. In this report, the result may be influenced by the fact that only one study [26] was included. Therefore, future experimental studies should investigate the AIx of patients with CHF in response to combined exercise.
This meta-analysis also reveals that aerobic and combined exercise significantly improved cf-PWV in patients with CVD. Our findings agree with those of some studies that aerobic exercise decreased cf-PWV in CVD patients or in all adults [13][14][15], and found that improved cf-PWV with aerobic exercise may be related to increased nitric oxide (NO) availability, higher conduit artery elastin content, decreased concentration of vasoconstrictor agents, increased oxygen uptake (VO 2) peak and reduction in ASP [13,43]. However, our findings did not agree with those of previous studies which reported no significant effect of aerobic training on PWV in patients with acute myocardial infarction (MI) due to the limited exercise duration [10,59]. In addition, some studies have reported that cf-PWV was not improved in hypertensive subjects or in patients with chronic kidney disease in response to aerobic exercise because of the high level of blood pressure regulated by sympathetic nervous system [10,79]. Combined exercise also significantly improved cf-PWV in patients with CVD in this study. Our findings support those of Li et al.
[80] that combined exercise may have favorable effect on arterial stiffness when aerobic and resistance exercise take place in the same exercise session. However, this contrasts with previous meta-analyses which found no significant difference in PWV with Exercise rehabilitation and cardiovascular disease combined exercise [13,14]. These meta-analyses pooled the results from subjects with disparate conditions (some were healthy, others were obese and others were CVD patients). Furthermore, peripheral and central arterial stiffness data were pooled. Montero et al. [10,14] included non-randomized controlled trials in their analysis. We emphasize that, in this study the response of central hemodynamic variables to different exercise modalities were comprehensively evaluated, and we restricted our analysis to RCTs of patients with CVD. The different results from these meta-analyses may be related to small number of studies in subgroup analysis. In addition, Montero et al. found that cf-PWV did not decrease with combined exercise due to the limitation of resistance exercise component [14]. It was controversial and complex for effects of resistance exercise on cf-PWV [16,17], and the effects of resistance were not only associated with exercise intensity and healthy status, but also with the changes of arterial compliance and central blood pressure regulated by sympathetic nervous system activity, and protein synthesis regulating muscle mass [14,16,80,81].
Aerobic exercise also significantly improved LVEF, and both aerobic exercise and combined exercise significantly improved CO in patients with CVD. However, there was no significant difference in LVEF in response to resistance exercise. A previous meta-analysis has shown that aerobic exercise improved EF, and reversed LV remodeling in patients with HF, while the benefit was not evident with combined exercise and resistance exercise [82]. In this study, meta-regression showed that the effects of aerobic exercise on LVEF were associated with age. LVEF is significantly increased with aerobic exercise due to its impact on the release of vasoconstrictive neurohormones, hemodynamic loading, peak exercise stroke volume (SV), myocardial contractility and diastolic filling [12,[82][83][84]. The increased SV with exercise was associated with a reduction in TPR driven by sympathetic activity and vagal tone [85]. At the same time, myocardial contractility contributed to the increased SV in response to exercise training [84]. It was controversial whether resistance training had any beneficial effect on LVEF. Some studies found that it may be associated with increased systolic and diastolic pressure loading, while others found that it may increase LV wall stress with resistance exercise leading to decreased LV wall stress and contractile and preload reserve [12]. This study supported the findings of the previous ones that CO was significantly increased with aerobic exercise because of improved VO 2 peak related to oxygen delivery and decreased TPR, HR and SV contributing to cardiac performance, and ventricular filling [48,79,86].
Although this meta-analysis has reported some novel findings concerning the disparate effects of different modalities on central hemodynamics, arterial stiffness and cardiac function in CVD, we note several limitations. First, some studies had small sample sizes, and according to the subgroup analysis of exercise modalities, only one study investigated the effects of resistance exercise on LVEF or combined exercise on ASP, ADP and AIx in this subgroup analysis. These reports had no heterogeneity results for these outcomes in the numerical results. Larger- Exercise rehabilitation and cardiovascular disease scale, good quality RCTs are needed for further investigating the effect of exercise on CVD. Second, a total of 10 articles (26.32%) had a long-term exercise intervention period (over 6 months), therefore, the long-term effects of different types of exercise on patients with CVD were not performed in this meta-analysis. Further studies are needed to explore the pathophysiological mechanisms of the relationship among central hemodynamics, arterial stiffness and cardiac function in CVD in response to different exercise training on the basis of controlling of external variables, such as age or disease.

Conclusions
Different exercise modalities have different effects on central hemodynamics, arterial stiffness and cardiac function in patients with CVD. Aerobic or resistance exercise significantly decreased ASP. Meanwhile, long-term aerobic exercise reduced AIx in patients with CVD. Aerobic exercise and combined exercise can effectively improve central arterial stiffness and cardiac function. The decreased central blood pressure in response to resistance exercise was mainly due to reservoir pressure reduction or improved microvascular perfusion. In contrast, improvement of central arterial stiffness and cardiac function in response to aerobic exercise and combined exercise was mainly due to their effects on cardiopulmonary fitness (VO 2 peak related to oxygen delivery or SV) and endothelial function. However, some heterogeneity in the results of the papers considered in this review remains unexplained, partly due to a paucity of high-quality studies, especially ones concerned with the effects of combined exercise on central hemodynamics and arterial stiffness for the rehabilitation of CVD. Finally, we note that these findings have important implications in the rehabilitation of these patients, not only for the patients themselves but also for medical professionals, allowing individual treatments to be tailored to specific cardiovascular pathologies.