Impacts of Exercise Interventions on Inflammatory Markers and Vascular Adhesion Molecules in Patients With Heart Failure: A Meta-analysis of RCTs

Abstract

Background

The aim of this meta-analysis was to investigate the effects of concurrent, aerobic, and resistance exercise on markers of inflammation and vascular adhesion molecules (high-sensitivity C-reactive protein [hs-CRP], interleukin [IL]-6, tumour necrosis factor-alpha [TNF-α], soluble intercellular adhesion molecule-1 [sICAM-1], soluble vascular cell adhesion molecule-1 [sVCAM-1], fibrinogen, IL-1-β, IL-10, IL-18, and E-selectin) in patients with heart failure (HF).

Methods

The PubMed, Scopus, Web of Science, and Google Scholar databases were searched for dates up to August 31, 2022. Randomized controlled trial studies for exercise interventions on circulating inflammatory and vascular adhesion markers in patients with HF were included. Standardized mean difference (SMD) and 95% confidence interval (CI) were calculated.

Results

A total of 45 articles were included. Exercise training significantly reduced hs-CRP (SMD –0.441 [95% CI: –0.642 to –0.240], P = 0.001), IL-6 (SMD –0.158 (95% CI: –0.303 to –0.013], P = 0.032), and sICAM-1 (SMD –0.282 [95% CI: –0.477 to –0.086], P = 0.005) markers. Analysis of subgroups revealed that a significant reduction occurred in hs-CRP level for the following subgroups: middle-aged, elderly, overweight status, aerobic exercise, concurrent training, both high and moderate intensity, and short-term, long-term, and very long-term follow-up, compared to a control group (P < 0.05). A significant reduction occurred in IL-6 and sICAM-1 levels for those in the following subgroups, compared to a control group (P < 0.05): middle-aged, aerobic exercise, moderate-intensity exercise, and short-term follow-up. A reduction in TNF-α level occurred for middle-aged patients, compared to a control-group (P < 0.05).

Conclusions

These exercise-related changes (improved inflammation and vascular adhesion markers) as clinical benefits in general, and for exercise-based cardiac rehabilitation in a more-specific format, improve clinical evolution and survival in patients with HF of different etiologies (registration number = CRD42021271423).

Heart failure (HF) is a pathophysiological state of ventricular dysfunction with a high risk of mortality and morbidity globally.^1,2 HF is related to the following: an elevation in inflammation^1,3,4 and inflammatory cytokines (interleukin [IL]-6, tumor necrosis factor-alpha [TNF-α], IL-1-β, IL-18, high-sensitivity C reactive protein [hs-CRP], fibrinogen); vascular adhesion molecules (VAMs) (soluble intercellular adhesion molecule-1 [sICAM-1], soluble vascular cell adhesion molecule-1 [sVCAM-1], E-selectin); and a reduction in anti-inflammatory cytokines such as IL-10 and adiponectin.^3,4 Numerous studies have reported that cytokine and VAM levels increase significantly in response to clinical HF.^3,4 Lifestyle changes and exercise are important treatment approaches to HF.⁵ Regular exercise training can restore endothelial function and improve neo-angiogenesis; it can also reduce production of inflammatory cytokines, reactive oxygen species (ROS), and peripheral vascular resistance with improved cardiac output, maximal oxygen consumption (VO₂max), maximal heart rate (HRmax), and systolic blood pressure in HF.⁶ Data on HF are conflicting. A number of studies have reported that aerobic exercise,7, 8, 9, 10, resistance training,¹¹ and concurrent exercise12, 13, 14 can reduce the level of inflammatory cytokines and VAMs in patients with HF, whereas other studies have reported no change in these markers.^3,15, 16, 17 A prior meta-analysis demonstrated a favourable impact of exercise on cytokines in patients with type-2 diabetes mellitus (T2DM) ¹⁸ and metabolic syndrome,¹⁹ and in postmenopausal women,⁴ but these did not include patients with HF. Therefore, the purpose of this meta-analysis was to clarify the effects of concurrent, aerobic, and resistance exercise on inflammatory markers and VAMs in patients with HF.

Methods

Search strategy

Our meta-analysis protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO) at the University of York (registration number = CRD42021271423) and was designed based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.²⁰ The PubMed, Scopus, Web of Science, and Google Scholar databases were searched to identify original published full-text articles up to August 31, 2022. The search strategy for exercises, inflammatory markers, VAMs, and patients with HF included the following keywords (with hs-CRP for high-sensitivity C-reactive protein, and TNF for tumour necrosis factor): [concurrent AND resistance AND aerobic AND inflammation OR cytokines OR hs-CRP IL-6 OR TNF-α OR sVCAM-1 OR sICAM-1 OR fibrinogen OR IL-1 beta OR IL-10 OR IL-18 OR E-selectin AND heart failure AND randomized controlled trial (RCT)]. After duplicate publications were removed, the abstracts and titles of articles were screened, and then articles were reviewed for eligibility by 4 reviewers (A.M., A.K., M.N., and N.G.).

Study selection

Only RCT studies were considered for eligibility. Inclusion criteria were as follows: (i) English-language original research; (ii) only human patients aged ≥ 18 years with HF; (iii) HF with maintenance of routine medications, standard and usual care, home-based exercise, and optimal medical therapy; (iv) measurement of serum or plasma levels of hs-CRP, IL-6, TNF-α, sICAM-1, sVCAM-1, fibrinogen, IL-1-β, IL-10, IL-18, and E-selectin at baseline and after intervention; (v) duration of exercise ≥ 2 weeks; vi) having at least one exercise group (aerobic, resistance, concurrent), with HF vs control group with HF; and (vii) usual care or routine medications for control group with and without exercise prescription and/or home-based exercise intervention. In this meta-analysis, the type of exercise included the following: aerobic (aerobic training; endurance training; aerobic exercise-based cardiac rehabilitation; cardiac rehabilitation; low and moderate-intensity inspiratory; aerobic interval training; Tai chi); resistance (resistance training; functional electrical stimulation; inspiratory muscle training; peripheral resistance training; power training); and concurrent (concurrent training; combined; aerobic plus resistance). Exclusion criteria are as described previously.²

Data extraction

The data extraction process was performed, and any disagreement was resolved by discussion among all the reviewers (A.M., A.K., M.N., N.G., and M.G.). The parameters of each study were extracted as follows: (i) study design; (ii) participant characteristics, including age, sex, body mass index (BMI), and sample size; (iii) study characteristics, including exercise (type, frequency, duration, training protocol, and supervised/unsupervised) and control group; (iv) outcome markers, including hs-CRP, IL-6, TNF-α, sICAM-1, sVCAM-1, fibrinogen, IL-1-β, IL-10, IL-18, and E-selectin. Pretest and posttest values are presented as mean ± standard deviation (SD), and mean differences were considered in order to generate forest plots. The data for standard error, median, range, and interquartile range were converted to mean ± SD.^21,22 Data figures or graphs were extracted using Getdata Graph Digitizer software (Version 2.26, Krasnoyarsk, Russia). Exercise studies with multiple arms vs a control group were included; the control group was divided by the number of intervention arms to avoid multiple counts of sample size. In addition, for the studies with more than one evaluated posttest, only the last period of the posttest was considered. To obtain additional information about the articles, the corresponding author was contacted.

Quality assessment and sensitivity analysis

The Pedro scale was used to assess the methodological quality of included studies (Pedro scores ranged from 7 to 15, with a maximum of 15 scores) and the risk of bias (high risk of bias = Pedro score of < 5; these studies were removed;²² Table 1), as described previously.²

Table 1.

Risk of bias assessment

Study (reference)	Eligibility criteria specified	Random allocation of participants	Allocation concealed	Groups similar at baseline	Assessors blinded	Outcome measures assessed in 85% of participants∗	Intention-to-treat analysis	Reporting of between-group statistical comparison†	Point measures and measures of variability reported for main effects	Activity monitoring in control group	Relative exercise intensity reviewed	Supervised /nonsupervised	Total Pedro score
Abolahrari-Shirazi et al.7 (2018)	✓	✓	✓	✓	✓	✓✓✓	✓	✓✓	✓	✓	✓	✓	15
Adamopoulos et al.8 (2001)	✓	✓	✓	✓	✓	✓✓✓	-	✓✓	✓	✓	✓	-	13
Adamopoulos et al.38 (2002)	✓	✓	✓	✓	✓	✓✓✓	-	✓✓	✓	✓	✓	-	13
Adamopoulos et al.12 (2014)	✓	✓	✓	✓	-	✓✓	-	✓✓	✓	✓	✓	✓	12
Ahmad et al.31 (2014)	✓	✓	✓	✓	✓	✓✓	-	✓✓	✓	-	✓	✓	12
Aksoy et al.9 (2015)	✓	✓	-	✓	-	-	-	✓✓	✓	-	-	✓	7
Balen et al.39 (2008)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	✓	✓	✓	12
Butts et al.10 (2018)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	✓	✓	✓	12
Byrkjeland et al.40 (2011)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	✓	✓	✓	12
de Meirelles et al.13 (2014)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	-	✓	✓	11
Eleuteri et al.33 (2013)	✓	✓	✓	✓	-	✓✓✓	✓	✓✓	✓	-	✓	✓	13
Erbs et al.34 (2010)	✓	✓	-	✓	✓	✓✓✓	✓	✓✓	✓	-	✓	✓	13
Feiereisen et al.30 (2013)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	-	✓	✓	11
Fernandes-Silva et al.41 (2017)	✓	✓	✓	✓	✓	✓✓	✓	✓✓	✓	-	-	✓	12
Fu et al.25 (2013)	✓	✓	-	✓	-	✓✓	-	✓✓	✓	-	✓	✓	10
Giallauria et al.42 (2011)	✓	✓	✓	✓	✓	✓✓✓	-	✓✓	✓	✓	✓	✓	14
Gielen et al.26 (2012)	✓	✓	✓	✓	✓	✓✓✓	✓	✓✓	✓	-	✓	✓	14
Jalaly et al.43 (2015)	✓	✓	✓	✓	-	✓✓	✓	✓	✓	-	✓	✓	11
Karavidas et al.11 (2006)	✓	✓	✓	✓	✓	✓✓	✓	✓✓	✓	-	-	✓	12
Kim et al.44 (2008)	✓	-	-	✓	-	✓✓	✓	✓✓	✓	-	✓	✓	10
Kobayashi et al.45 (2003)	✓	✓	-	✓	-	✓✓✓	✓	✓✓	✓	-	-	✓	11
Lara Fernandes et al.46 (2011)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	✓	✓	✓	12
Larsen et al.35 (2001)	✓	-	-	✓	-	✓✓✓	✓	✓	✓	-	✓	✓	10
Linke et al.36 (2005)	✓	✓	-	✓	-	✓✓	-	✓✓	✓	-	✓	✓	10
Marco et al.16 (2013)	✓	✓	✓	✓	✓	✓✓	✓	✓✓	✓	-	-	✓	12
Masterson-Creber et al.47 (2015)	✓	✓	✓	✓	-	✓✓	-	✓✓	✓	-	✓	✓	11
McDermott et al.48 (2004)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	-	-	✓	10
Melo et al,29 (2019)	✓	✓	✓	✓	-	✓	✓	✓✓	✓	-	✓	✓	11
Munk et al.49 (2011)	✓	✓	✓	✓	✓	✓✓✓	✓	✓✓	✓	-	✓	✓	14
Myers et al.14 (2010)	✓	✓	-	✓	✓	✓✓✓	✓	✓✓	✓	-	✓	✓	13
Niebauer et al.3 (2005)	✓	✓	-	✓	-	✓✓	-	✓✓	✓	-	✓	✓	10
Parrinello et al.50 (2010)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	✓	-	✓	11
Pierce et al.32 (2008)	✓	✓	-	✓	-	✓✓	-	✓✓	✓	✓	-	✓	10
Prescott et al.56 (2009)	✓	✓	-	✓	-	✓✓	-	✓✓	✓	-	✓	✓	10
Pullen et al.51 (2008)	✓	✓	-	✓	✓	✓✓✓	✓	✓✓	✓	-	-	✓	12
Ranković et al.52 (2009)	✓	-	-	✓	-	✓✓	-	✓✓	✓	-	✓	✓	9
Redwine et al.17 (2020)	✓	✓	✓	✓	✓	✓✓✓	✓	✓✓	✓	-	-	✓	13
Sandri et al.27 (2016)	✓	✓	✓	✓	✓	✓✓	-	✓✓	✓	-	✓	✓	12
Tisi et al.53 (1997)	✓	✓	✓	✓	-	✓✓✓	✓	✓✓	✓	✓	-	✓	13
Trippel et al.57 (2017)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	-	✓	✓	11
Tsarouhas et al.54 (2011)	✓	-	-	✓	-	✓✓	✓	✓✓	✓	-	✓	✓	10
Walther et al.37 (2008)	✓	✓	✓	✓	✓	✓✓✓	✓	✓✓	✓	-	-	✓	13
Wosornu et al.28 (1992)	✓	✓	-	✓	-	✓✓✓	✓	✓✓	✓	✓	-	✓	12
Yeh et al.55 (2011)	✓	✓	-	✓	-	✓✓	✓	✓✓	✓	-	-	✓	10
Zaidi et al.58 (2019)	✓	✓	-	✓	-	✓✓	-	✓✓	✓	-	-	✓	9

Total Pedro score out of 15 points; (✓) = 1 point; (-) = not reported or unclear.

^∗Three points possible—1 point if adherence > 85%; 1 point if adverse events reported, 1 point if exercise attendance is reported.

^†Two points possible—1 point if primary outcome is reported; 1 point if all other outcomes are reported. Pedro score < 5 = high risk of bias.

Statistical analysis

Comprehensive meta-analysis (CMA) software (Version 2, Englewood, New Jersey) was used for data analysis and for calculating the standardized mean difference (SMD) and 95% confidence intervals (CIs) using fixed-effect and random-effect models. The significance level was considered to be at P < 0.05. The effect size was calculated to compare the effects of exercise vs the control group on circulating hs-CRP, IL-6, TNF-α, sICAM-1, sVCAM-1, fibrinogen, IL-1-β, IL-10, IL-18, and E-selectin markers. The Cochrane guidelines for interpreting effect sizes were used, as follows: large (> 0.8); medium (0.5-0.79); and small (0.2-0.49) effect sizes.²³ Heterogeneity was assessed by using the I-squared (I²) statistic. The following Cochrane guidelines were used in the interpretation of the I² statistic: high heterogeneity (75%); medium heterogeneity (50%); and low heterogeneity (25%). The visual interpretation of funnel plots was considered to identify publication bias. In addition, Egger’s test was used as a secondary determinant test; significant publication bias was considered to be apparent if P < 0.1.²⁴

Results

Included and excluded studies

The initial searches in the PubMed, Scopus, Web of Science, and Google Scholar databases identified 278,160; 1008; 4,434,264; and 174,000 articles, respectively. After removing duplicates and screening articles based on the title/abstract, 426 full-text articles were included for final screening based on the inclusion and exclusion criteria. Forty-five full-text articles (of those 426 articles) met the inclusion criteria, and 381 articles were excluded (Fig. 1). In this meta-analysis, 8 of the included articles had 2 arms of exercise interventions. ^7,9,17,25, 26, 27, 28, 29 One article had 3 arms of exercise interventions.³⁰ Forty-five full-text articles (total intervention arms = 55) were included, with a total of 41 aerobic intervention arms, 10 concurrent intervention arms, and 4 resistance intervention arms. A total of 3403 participants (exercise = 1868 and control = 1535) were included. The study flowchart is shown in Figure 1.

Figure 1.

Flowchart of study selection. HF, heart failure; hs-CRP, high-sensitivity C-reactive protein; IL, interleukin; RCT, randomized controlled trial; sVCAM-1, soluble vascular cell adhesion molecule-1; SICAM-1, soluble intercellular adhesion molecule-1; TNF, tumour necrosis factor; VAM, vascular adhesion molecule.

Participant characteristics

The participant characteristics of included articles are presented in Table 2. The sample size, mean age, and BMI for each article, respectively, ranged between 7 and 477 participants,^11,31 49 ± 19.36 and 73.1 ± 6 years,^14,26 and 23.3 ± 3.9 and 33 ± 8 kg/m².^17,32 In the total of 45 included articles, 36 included both male and female participants; 6 included only male participants;^28,33, 34, 35, 36, 37 and for 3 articles, the gender was unknown.^8,28,38

Table 2.

Characteristics of patients with HF at baseline.

Source, yrs	Country	Exercise + control = total sample size (baseline)	Sex	Participant characteristics	Groups	Age, y (baseline) Mean ± SD	BMI, kg/m2 (baseline) Mean ± SD	Inflammatory markers and VAMs
Abolahrari-Shirazi et al.7 (2018)	Iran	25 (Concurrent) + 25 (Aerobic) + 25 (Control) = 75	Male/female	Patients with HF (NYHA class I-III)	Concurrent Aerobic Control	Concurrent: 56.76 ± 8.71 Aerobic: 57.64 ± 7.85 Control: 57.32 ± 9.41	Concurrent: 25.69 ± 3.65 Aerobic: 26.71 ± 2.91 Control: 26.10 ± 3.86	hs-CRP
Adamopoulos et al.8 (2001)	Greece	12 + 10 = 22	Unknown	Patients with chronic HF (NYHA class II-III)	Aerobic Control	Aerobic: 59.6 ± 6.92 Control: 00.0 ± 00.0	Unknown	sICAM-1, sVCAM-1
Adamopoulos et al.38 (2002)	Greece	24 + 20 = 44	Unknown	Patients with chronic HF (NYHA class II-III)	Aerobic Control	Aerobic: 55.0 ± 9.79 Control: 00.0 ± 00.0	Unknown	TNF-α, IL-6
Adamopoulos et al.12 (2014)	Belgium	21+22 = 43	Male/female	Patients with chronic HF (NYHA class I-III)	Concurrent Control	Concurrent: 57.8 ± 11.7 Control: 58.3 ± 13.2	Concurrent: 28.6 ± 6.7 Control: 27.2 ± 2.9	CRP
Ahmad et al.31 (2014)	England	477 + 451 = 928	Male/female	Patients with chronic HF (NYHA class II-IV)	Aerobic Control	Aerobic: 59.36 ± 12.41 Control: 59.23 ± 12.86	Unknown	hs-CRP
Aksoy et al.9 (2015)	Turkey	15 (Continuous) + 15 (Intermittent) + 15 (Control) = 45	Male/female	Patients with chronic HF (NYHA class II-III)	Aerobic Aerobic Control	Aerobic: Continuous: 63.7 ± 8.8 Intermittent: 59.6 ± 6.9 Control: 57.5 ± 11.2	Aerobic: Continuous:28.4 ± 4.9 Intermittent:30.1 ± 5.1 Control: 29.1 ± 4.2	sVCAM-1, sICAM-1, fibrinogen, CRP
Balen et al.39 (2008)	Croatia	30 + 30 = 60	Male/female	Patients with MI	Aerobic Control	Aerobic: 59 ± 9 Control: 61 ± 10	Aerobic: 28.8 ± 3.8 Control: 28 ± 3.8	Fibrinogen, hs-CRP, IL-10, TNF-α
Butts et al.10 (2018)	USA	38 + 16 = 54	Male/female	Patients with HF (NYHA class II-III)	Aerobic Control	Aerobic: 60 ± 8.7 Control: 58.19 ± 12.8	Aerobic: 31.51 ± 7.1 Control: 31.03 ± 6.1	IL-1β, IL-18
Byrkjeland et al.40 (2011)	Norway	40 + 40 = 80	Male/female	Patients with chronic HF (NYHA class I-IIIB)	Aerobic Control	Aerobic: 68.8 ± 7.9 Control: 71.5 ± 7.8	Unknown	CRP, TNF-α, IL-6, IL-18, E-selectin, ICAM-1, VCAM-1
de Meirelles et al.13 (2014)	Brazil	15 + 15+30	Male/female	Patients with HF (NYHA class II and III)	Concurrent Control	Concurrent: 54 ± 3 Control: 55 ± 2	Concurrent: 28.6 ± 0.9 Control: 27.9 ± 0.7	Fibrinogen, CRP, IL-6, TNF-α
Eleuteri et al.33 (2013)	Italy	11 + 10 = 21	Male	Patients with chronic HF (NYHA class II)	Aerobic Control	Aerobic: 66 ± 2 Control: 63 ± 2	Unknown	IL-6, CRP
Erbs et al.34 (2010)	Germany	18 + 19 = 37	Male	Patients with advanced chronic HF (NYHA class IIIB)	Aerobic Control	Aerobic: 60 ± 11 Control: 62 ± 10	Unknown	TNF-α
Feiereisen et al.30 (2013)	Luxembourg	15 (Concurrent) + 15 (Resistance) + 15 (Aerobic) + 15 (Control) = 60	Unknown	Patients with chronic HF (NYHA class II-III)	Concurrent Resistance Aerobic Control	Concurrent: 60.6 ± 5.6 Resistance: 57.9 ± 5.8 Aerobic: 59.4 ± 6.5 Control: 55.5 ± 7.5	Unknown	TNF-α, IL-6
Fernandes-Silva et al.41 (2017)	Brazil	28 +16 = 44	Male/female	Patients with HF (NYHA class IV)	Aerobic Control	Aerobic: 51 ± 7 Control: 48 ± 7	Aerobic: 29 ± 4 Control: 28 ± 4	IL-6, TNF-α
Fu et al.25 (2013)	Taiwan	15 (Aerobic interval) + 15 (Aerobic continuous) + 15 (Control) = 45	Male/female	Patients with HF (NYHA class II-III)	Aerobic Aerobic Control	Aerobic: Interval: 67.5 ± 1.8 Continuous: 66.3 ± 2.1 Control: 67.8 ± 2.5	Unknown	IL-6
Giallauria et al.42 (2011)	Italy	37 + 38 = 75	Male/female	Patients with acute MI (AHA Class IIB or III)	Aerobic Control	Aerobic: 61 ± 7 Control: 60 ± 8	Aerobic: 27.3 ± 2.2 Control: 28.2 ± 2.8	hs-CRP
Gielen et al.26 (2012)	Germany	15 + 15 = 30 15 + 15 = 30	Male/female	Patients with chronic HF ≤ 55 yrs and ≥ 65 yrs (NYHA class II-III)	Aerobic Control Aerobic Control	Aerobic: 50 ± 19.36 Control: 49 ± 19.36 Aerobic: 72 ± 15.49 Control: 72 ± 11.61	Aerobic: 29 ± 7.74 Control: 30 ± 11.61 Aerobic: 28 ± 11.61 Control: 28 ± 7.74	TNF-α
Jalaly et al.43 (2015)	Iran	20 + 20 = 40	Male/female	Patients with stable angina pectoris	Aerobic Control	45-65	Unknown	s-ICAM-1, E-selectin
Karavidas et al. 11(2006)	Greece	16 + 8 = 24	Male/female	Patients with chronic HF (NYHA class II-III)	Resistance/functional electrical stimulation Control	Resistance: 57.4 ± 15.3 Control: 63.8 ± 8.1	Resistance: 26.57 ± 4.80 Control: 28.07 ± 3.68	TNF-α, IL-6, s-ICAM-1, s-VCAM-1, IL-10
Kim et al.44 (2008)	Korea	29 + 10 = 39	Male/female	Patients with CAD	Aerobic Control	Aerobic: 59.9 ± 8.61 Control: 52.8 ± 11.70	Aerobic: 25.6 ± 3.23 Control: 26.6 ± 2.21	hs-CRP, fibrinogen, TNF-α, IL-1β, IL-6
Kobayashi et al.45 (2003)	Japan	14 + 14 = 28	Male/female	Patients with chronic HF (NYHA class II-III)	Aerobic Control	Aerobic: 55 ± 7.48 Control: 62 ± 7.48	Unknown	IL-6
Lara Fernandes et al.46 (2011)	Brazil	15 + 19 = 34	Male/female	Patients with CAD	Aerobic Control	Aerobic: 60.7 ± 6.7 Control: 59.5 ± 7.3	Aerobic: 28.6 ± 5.9 Control: 27.6 ± 3.6	CRP, VCAM-1
Larsen et al.35 (2001)	Norway	28 + 16 = 44	Male	Patients with HF (NYHA class II-III)	Aerobic Control	Aerobic: 67 ± 8 Control: 62 ± 5	Unknown	TNF-α, IL-6
Linke et al.36 (2005)	Germany	12 + 11 = 23	Male	Patients with chronic HF (NYHA class II-III)	Aerobic Control	Aerobic: 55 ± 6.92 Control: 52 ± 9.94	Unknown	TNF-α, IL-1-β
Marco et al.16 (2013)	Spain	11 + 11 = 22	Male/female	Patients with chronic HF (NYHA class II-III)	Concurrent Control	Concurrent: 68.5 ± 8.88 Control: 70.1 ± 10.75	Concurrent: 28.4 ± 3.64 Control: 26.3 ± 2.4	hs-CRP
Masterson-Creber et al.47 (2015)	USA	163 + 157 = 320	Male/female	Patients with chronic HF (NYHA class II-IV)	Aerobic Control	58.66 ± 11.91	Unknown	hs-CRP
McDermott et al.48 (2004)	USA	24 + 8 = 34	Male/female	Peripheral arterial patients	Aerobic Control	Aerobic: 69.4 ± 9.6 Control: 65.9 ± 6.2	Aerobic: 28.6 ± 5.1 Control: 28.8 ± 6.1	hs-CRP, fibrinogen, IL-6
Melo et al.29 (2019)-a	Portugal	7 + 9 = 16	Male/female	Patients with chronic HF (atrial fibrillation) (NYHA class II-IV)	Aerobic Control	69.4 ± 7.2	28.2 ± 4.8	TNF-α, IL-6
Melo et al.29 (2019)-b	Portugal	11 + 10 = 21	Male/female	Patients with chronic HF (sinus rhythm) (NYHA class II-IV)	Aerobic Control	66.2 ± 14.57	26.7 ± 4.58	TNF-α, IL-6
Munk et al.49 (2011)	Norway	18 + 18 = 36	Male/female	Patients with angina pectoris	Aerobic Control	Aerobic: 59.5 ± 10 Control: 60.7 ± 9	Aerobic: 26.1 ± 4 Control: 28.4 ± 3.3	IL-6, TNF-α, IL-10, VCAM, E-selectin
Myers et al.14 (2010)	USA	26 + 31 = 57	Male/female	Patients with abdominal aortic aneurysm	Concurrent Control	Concurrent: 73.1 ± 6 Control: 70.4 ± 9	Concurrent: 28.2 ± 4.4 Control: 26.9 ± 3.4	CRP
Niebauer et al.3 (2005)	UK	18 + 9 = 27	Male/female	Patients with chronic HF	Concurrent Control	Concurrent: 53.6 ± 9.2 Control: 51.3 ± 6.9	Unknown	TNF-α, IL-6, soluble E-selectin, sICAM-1
Parrinello et al.50 (2010)	Italy	11 + 11 = 22	Male/female	Patients with compensated congestive HF (NYHA class II-III)	Aerobic Control	Aerobic: 62.3 ± 4.9 Control: 63.2 ± 5	Unknown	CRP
Pierce et al.32 (2008)	USA	8 + 6 = 14	Male/female	Patients with heart transplant recipients	Aerobic Control	Aerobic: 53.5 ± 13.6 Control: 54.2 ± 6.4	Aerobic: 23.3 ± 3.9 Control: 25.8 ± 3.8	CRP, IL-6, TNF-α, sICAM-1
Prescott et al.56 (2009)	Denmark	20 + 23 = 43	Male/female	Patients with chronic systolic HF (NYHA class II–IV)	Concurrent Control	Concurrent: 68 ± 11 Control: 66.9 ± 12.5	Concurrent: 27.7 ± 4.12 Control: 27.7 ± 5.92	hs-CRP, IL-6, TNF-α
Pullen et al.51 (2008)	USA	9 + 10 = 19	Male/female	Patients with chronic HF (NYHA class II-III)	Aerobic Control	Aerobic: 52.1 ± 3.3 Control: 50.5 ± 12.8	Unknown	IL-6, hs-CRP
Ranković et al.52 (2009)	Serbia	22 + 30 = 52	Male/female	Patients with ischemic heart disease	Aerobic Control	Aerobic: 62.7 ± 7.1 Control: 58.4 ± 7.6	Aerobic: 29.3 ± 3.2 Control: 29.1 ± 2.7	hs-CRP, VCAM-1, ICAM-1
Redwine et al.17 (2020)	USA	24 (Aerobic) + 22 (Resistance) + 23 (Control) = 69	Male/female	Patients with HF (AHA Class III)	Aerobic Resistance Control	Aerobic: 63 ± 9 Resistance: 65 ± 9 Control: 67 ± 7	Aerobic: 32 ± 8 Resistance: 33 ± 8 Control: 31 ± 6	CRP, IL-6, TNF-α
Sandri et al.27 (2016)	Germany	30 + 30 = 60 30 + 30 = 60	Male/female	Patients with chronic HF ≤ 55 yrs (LVEF > 55%) Patients with chronic HF aged ≥ 65 y (LVEF > 55%)	Aerobic Control Aerobic Control	≤ 55 ≥ 65	Unknown	sVCAM-1, sICAM-1
Tisi et al.53 (1997)	UK	67 + 15 = 82	Male/female	Patients with intermittent claudication	Aerobic Control	69.3 66.2	Unknown	Fibrinogen, CRP
Trippel et al.57 (2017)	Germany	43 + 19 = 62	Male/female	Patients with HF	Concurrent Control	64.4 ± 7.2	Unknown	TNF-α, IL-1β, IL-6, IL-10
Tsarouhas et al.54 (2011)	Greece	27 + 12 = 39	Male/female	Patients with chronic HF (NYHA class II-III)	Aerobic Control	Aerobic: 66.8 ± 13.1 Control: 67 ± 5.6	Aerobic: 24.1 ± 7.1 Control: 25.5 ± 4.5	TNF-α
Walther et al.37 (2008)	Germany	51 + 50 = 101	Male	Patients with CAD	Aerobic Control	Aerobic: 62 ± 7.14 Control: 60 ± 7.07	Aerobic: 27.2 ± 2.85 Control: 28 ± 3.53	hs-CRP, IL-6
Wosornu et al.28 (1992)	UK	15 (Aerobic) + 20 (Resistance) + 20 (Control) = 55	Male	Patients with coronary artery surgery (CCS I–III)	Aerobic Resistance Control	Aerobic: 57 ± 9 Resistance: 60 ± 6 Control: 56 ± 7	Unknown	Fibrinogen
Yeh et al.55 (2011)	USA	50 + 50 = 100	Male/female	Patients with chronic HF (NYHA class I-III)	Aerobic Control	Aerobic: 68.1 ± 11.9 Control: 66.6 ± 12.1	Unknown	CRP, TNF-α
Zaidi et al.58 (2019)	Norway	69 + 68 = 137	Male/female	Patients with combined CAD and type 2 diabetes mellitus	Concurrent Control	Concurrent: 64 ± 8 Control: 63 ± 8	Concurrent: 28.83 ± 4.61 Control: 28.56 ± 4.62	IL-18

AHA, American Heart Association; CAD, coronary artery disease; CCS, Canadian Cardiovascular Society; CRP, C-reactive protein; HF, heart failure; hs-CRP, high-sensitivity CRP; IL, interleukin; LVEF, left ventricular ejection fraction; MI, myocardial infarction; NYHA, New York Heart Association; SD, standard deviation; sICAM-1, soluble intercellular adhesion molecule-1; sVCAM-1, soluble vascular cell adhesion molecule-1; TNF-α, tumor necrosis factor alpha; UK, United Kingdom; VAM, vascular adhesion molecule.

Characteristics of interventions in the exercise and control groups

The characteristics of the exercise intervention and control groups are provided in Table 3. In our meta-analysis, we included studies with the following types of exercise: aerobic;7, 8, 9, 10^,17,25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55 concurrent;^3,7,12, 13, 14, 15, 16^,30,56, 57, 58 and resistance.^11,17,28,29 The intensity ranges for aerobic intervention performance were 40% VO₂peak/ HRmax/ HRmax reserve^7,43,54 to 95% HRmax,²⁹ with the most common exercise intensity being 60%-75% or 80% VO₂peak/ HRmax/ HRmax reserve.7, 8, 9, 10^,25, 26, 27^{,30,31,35,36,38,39,42,43,47,52,54} The intensity ranges for resistance intervention performance were 60% of one-repetition maximum (1RM)³⁰ and an 11 rating on the Borg perceived exertion (RPE) scale¹⁷ to 75% of 1RM³⁰ and a 13 RPE,¹⁷ with the most common exercise intensity being 60%-75% of 1RM³⁰ and an 11-13 RPE.¹⁷ The intensity ranges for concurrent interventions were performed from 40% VO₂peak plus 40% of 1RM⁷ to 100% of 10RM.¹⁶ The duration of exercise varied from 3 weeks³⁹ to 24 months,³⁷ with the most common period being 3 months (12 weeks).^8,10,12,25,31, 32, 33, 34, 35^{,38,41,43,45,48,54,55,57} Durations of 4 weeks,^16,26,27 6 weeks,^11,52 7 weeks,⁷ 8 weeks,^3,51,56 10 weeks,^9,50 14 weeks,^30,44 16 weeks,¹⁷ 4 months,^15,40,46 6 months,^{13,28,29,36,42,49} 12 months,^14,47,53,58 and 24 months³⁷ were used in the other studies. The exercise frequency was performed from 2 days per week^{17,29,40,43,44,51,55,56,58} to 7 days per week,^32,53 with the most common exercise frequency being 3 days per week.^7,9,10,12, 13, 14, 15^{,25,26,30,31,35,36,39,41,42,}45, 46, 47, 48, 49^,52 In addition, the exercise frequency in the other studies was 4 days per week,^26,27 5 days per weeks,^{3,8,11,33,38,50,54} 6 days per weeks,³⁴ and 7 days per week.^16,32,53 The duration of each session for aerobic intervention ranged from 20 minutes^26,27,34 to 70 minutes,⁵¹ with the most common session duration being 30-45 minutes.7, 8, 9, 10^,25,30, 31, 32, 33^,35,38,39,41, 42, 43, 44, 45^,47,48,50,52, 53, 54 The duration of resistance exercise intervention sessions ranged from 30 minutes¹¹ to 60 minutes,^17,28 with the most common session duration being 30-60 minutes.^11,17,28,30 The duration of concurrent exercise sessions ranged from 20 minutes¹⁶ to 120 minutes,¹² with the most common duration being 60 minutes.^7,14,15 The exercise sessions and participants were supervised in 40 included studies, were unsupervised in 2 studies,^37,54 and were both supervised and unsupervised in 1 study.⁵⁸

Table 3.

Study characteristics in patients with heart failure.

Study	Exercise intervention					Control group
	Type	Frequency, d/wk	Follow-up (duration)	Training protocol	Supervised or Unsupervised
Abolahrari-Shirazi et al.7 (2018)-a	Combined (cycle + weight training)	3	7 wk	Endurance = 45 min, 40%–70% peak VO2; 20 min cycle ergometer, 10 min arm ergometer, 15 min treadmill; Resistance = knee extension, knee flexion, elbow flexion, and shoulder abduction. 40% 1RM-60% 1RM and duration of 15 min	Supervised	Pamphlet for daily exercising at home
Abolahrari-Shirazi et al.7 (2018)-b	Endurance (cycle)	3	7 wk	Endurance = 45 min, 40%–70% peak VO2; 20 min cycle ergometer, 10 min arm ergometer, 15 min treadmill	Supervised	Pamphlet for daily exercising at home
Adamopoulos et al.8 (2001)	Aerobic (bicycle)	5	12 wk	30 min bicycle, 5 d/wk at 50 rpm with 70%–80% of HRmax	Unknown	Home-based bicycle exercise similar to the aerobic group
Adamopoulos et al.38 (2002)	Aerobic (bicycle)	5	12 wk	30 min home-based bicycle, 5 d/wk at 50 rpm with 60%–80% of HRmax	Unknown	Home-based bicycle similar to the aerobic group
Adamopoulos et al.12 (2014)	Concurrent (cycle + inspiratory muscle training)	3	12 wk	Aerobic = 45 min ergometer at 70%–80% HRmax with 5 min warm-up and cool down periods. Resistance = 30 min inspiratory-incremental resistive loading device at 60% of individual sustained maximal inspiratory pressure with 6 inspiratory efforts at each level. Initially, 60s rest intervals over its 6 inspiratory efforts, but at the second level through the sixth level, this rest period was reduced to 45, 30, 15, 10, and 5 s. After the sixth level, the rest period was kept at 5 s	Supervised	Control group was exercised at only 10% of their sustained maximal inspiratory pressure
Ahmad et al.31 (2014)	Aerobic (walking, treadmill or cycling)	3	12 wk	15-30 min per session at 60% of HRmax reserve	Supervised	Usual care
Aksoy et al.9 (2015)	Aerobic (cycle)	3	10 wk	35 min of aerobic exercise (bicycle ergometer at a constant pedal rate of 50 rpm) including 10 min of warm-up and cool down at 50%–75% peak VO2. Intermittent: 60-s bouts of cycling, 30-s intervals of low-intensity cycling at 30 W, 17 cycles of low- and high-intensity bouts. Continuous: worked without any change in the intensity.	Supervised	Optimal medical therapy without any particular regular physical activity before
Balen et al.39 (2008)	Aerobic (cycle)	3	3 wk	45 min cycle-ergometer with 50%–60% VO2peak plus 30-min organized program of supervised walking on a standardized track	Supervised	Standard care
Butts et al.10 (2018)	Aerobic (walking)	3	3 mo	30 min at 60% HRmax for the first 2 weeks, 45 minutes 3 times per wk at 60% Hrmax for weeks 3 and 4, and 45 min at 70% HRmax for the remaining 8 wk	Supervised	Education and flexibility and stretching exercises
Byrkjeland et al.40 (2011)	Aerobic (walking)	2	4 mo	Three intervals of high intensity (15–18 on the Borg scale) and 2 periods of moderate intensity (11–13 on the Borg scale), in addition to warm-up and cool-down periods, 50 min walking per session	Supervised	Standard follow-up care by their primary physician
de Meirelles et al.13 (2014)	Concurrent (walking + weight training)	3	6 mo	90 min aerobic, resistance, and stretching exercises; 30 min of treadmill with 5%–15% above the ventilator threshold, whole body skeletal muscle strength with 2–3 sets of 10–15 RM of 8–10 exercises for the major muscle groups, and stretching or cool-down period for the major muscle groups.	Supervised	Optimal medical therapy and usual care
Eleuteri et al.33 (2013)	Aerobic (cycle)	5	3 mo	30-min cycle ergometer (60 rpm) at a power and heart rate corresponding to ventilatory anaerobic threshold (VAT), preceded and followed by a 5-min warm-up and cool-down unloaded period, respectively	Supervised	Normal activities without exercise intervention
Erbs et al.34 (2010)	Aerobic (bicycle)	3–6	12 wk	During the first 3 weeks, patients exercised 3 to 6 times daily for 5 to 20 min on a bicycle ergometer at 50% of VO2max in-hospital and 60% of VO2max on discharge	Supervised	Without exercise intervention
Feiereisen et al.30 (2013)-a	Concurrent (bicycle + weight training)	3	14 wk (40 sessions)	20 min of bicycle and 20 min of strength training on 5 different weight machines by a warm-up period of 5 min of bicycle training at 30% of VO2peak	Supervised	Without exercise intervention
Feiereisen et al.30 (2013)-b	Resistance (weight training)	3	14 wk (40 sessions)	10 different strength exercises on weight machines during 40 min, starting at 60% of 1RM, and progressively increasing to 75% of 1RM; by a warm-up period of 5 min of bicycle training at 30% of VO2peak	Supervised	Lifestyle activities without exercise intervention
Feiereisen et al.30 (2013)-c	Aerobic (bicycle)	3	14 wk (40 sessions)	40 min of bicycle and treadmill, starting at 60% of VO2peak, which they progressively adapted to reach 75% of the VO2 peak; by a warm-up period of 5 min of bicycle training at 30% of VO2peak	Supervised	Without exercise activities
Fernandes-Silva et al.41 (2017)	Aerobic (cycle)	3	12 wk	30 min cycle ergometer, 11–14 Borg scale, and 5-min warmup and cool-down	Supervised	Medical therapy without exercise activities
Fu et al.25 (2013)	Aerobic (cycle)	3	12 wk	Aerobic interval training (AIT) = 3 min at 30% of VO2peak and five 3-min intervals at 80% of VO2peak; Each interval was separated by 3-min exercise at 40% of VO2peak and 3-min cool-down at 30% of VO2peak. Moderate continuous training (MCI) = warmup at 30% of VO2peak for 3 min, 60% of VO2peak for 30 min, then a cool-down at 30% of VO2peak for 3 min	Supervised	General home-based health care
Giallauria et al.42 (2011)	Aerobic (bicycle)	3	6 mo	30 min bicycle ergometer, 60%–70% of peak VO2, 5-min warmup and 5-min cool-down	Supervised	Generic instructions for maintaining physical activity and a correct lifestyle
Gielen et al.26 (2012)	Aerobic (bicycle)	4	4 wk	20 min of bicycle ergometer (excluding 5 min of warmup and cool-down) at 70% of VO2max	Supervised	Usual clinical care
Jalaly et al.43 (2015)	Aerobic (walking or jogging)	2	12 wk	20–30 min walking or jogging on treadmill at 40%–60% of heart rate reserve (HRR) and RPE of 11–13 (on 6–20 Borg scale)	Supervised	Without exercise intervention
Karavidas et al.11 (2006)	Resistance (electrical stimulation)	5	6 wk	30 min/d at 25 Hz for 5 s followed by 5 s of rest. When the muscles of the right leg were contracted, the muscles of the left leg were relaxing and vice versa.	Supervised	Control group was exposed to the same regimen as the functional electrical stimulation group
Kim et al.44 (2008)	Aerobic (treadmill or bicycle)	2	14 wk	A warm-up, 30–40 min of treadmill or bicycle ergometer at 50%–85% of VO2max, and a cool-down	Supervised	Standard care
Kobayashi et al.45 (2003)	Aerobic (cycle)	2-3	12 wk	15 min of cycle, adjusted to maintain heart rate equivalent to the ventilatory threshold level	Supervised	Normal lifestyle without exercise intervention
Lara Fernandes, et al.46 (2011)	Aerobic (cycling)	3	4 mo	60 min including 5 min stretching, 40 min of cycling with a target heart rate between anaerobic threshold and respiratory compensation point, 10 min of local strengthening and 5 min of cool down	Supervised	Recommendations for lifestyle modification
Larsen et al.35 (2001)	Aerobic (walking and jogging)	3	12 wk	10 min of warmup, 25 min of walking and jogging at 80% of maximum capacity, and 10 min of cooling down and stretching	Supervised	Without exercise intervention
Linke et al.36 (2005)	Aerobic (walking and bicycle)	3	6 mo	20 min of walking, noncompetitive ball games, and calisthenics at 70% of VO2max; during the first 2 weeks, aerobic group exercised in hospital 4 to 6 times daily for 10 minutes each on a bicycle ergometer at 70% of VO2max	Supervised	Without exercise intervention
Marco et al.16 (2013)	Concurrent (inspiratory muscle endurance and strength)	7	4 wk	20 min of high-intensity inspiratory muscle training with 10 consecutive RM, 5 sets of 10 repetitions followed by 1–2 min of unloaded recovery breathing off the device and with 100% of their 10RM twice a day	Supervised	Sham- inspiratory muscle training at an initial workload of 10 cmH2O which was increased 2.5 cmH2O every wk
Masterson-Creber et al.47 (2015)	Aerobic (walking or cycling)	3	12 mo	30–35 min of walking or cycling at 60%–70% HRmax reserve	Supervised	Usual care
McDermott et al.48 (2004)	Aerobic (walking)	3	12 wk	50 min of step back on the treadmill and walk continuously, speed 0.5 miles/h and grade 2%, and 11–12 RPE	Supervised	Usual clinical care
Melo et al.29 (2019)	Aerobic (walking)	2	6 mo	60 min, 4 interval training periods, 90%–95% HRmax with 3 lower-intensity active periods (60%–70% of HRmax) between interval training periods as well as a 10-min warmup and a 5–7 min cool-down.	Supervised	Usual care
Munk et al.49 (2011)	Aerobic (bicycle or running)	3	6 mo	60 min, warmup period, followed by four 4-minute intervals at 80%–90% of HRmax, intervals were interrupted by 3 min of active recovery at 60%–70% of HRmax, 5 min cool-down, 10 min abdominal- and spine-resistance exercises, and 5 min of stretching and relaxing	Supervised	Usual care
Myers et al.14 (2010)	Concurrent (treadmill, cycling, stair climbing, elliptical training and rowing)	3	12 mo	45 min of treadmill, cycle ergometry, stair climbing, elliptical training, and rowing by 10 min of resistance exercise at 60%–80% of HRmax reserve	Supervised	Usual care
Niebauer et al.3 (2005)	Concurrent (calisthenics and bicycle)	5	8 wk	Exercise training consisted of at least 5 d/wk of a 20-min/d of calisthenics and bicycle ergometer at home, first 9 exercises in the Canadian airforce XBX (10 basic exercises) program with 25W at 50 rpm, resistance 70%–80% HRmax, and 2-3 min cool down	Supervised	Without exercise intervention
Parrinello et al.50 (2010)	Aerobic (walking)	5	10 wk	30 min of mild–moderate walking exercise over the usual physical activity	Supervised	Medical therapy and dietary recommendations with routine activities
Pierce et al.32 (2008)	Aerobic (walking)	7	12 wk	30 min of continuous treadmill walking and progressed to 35–40 min as tolerated after the initial 4 weeks, and exercise intensity 12–14 Borg scale	Supervised	Standard medical care
Prescott et al.56 (2009)	Concurrent (walking, cycling, step machine, and step board + weight training)	2	8 wk	1.5-h including 20 min warmup, four 6-min series of aerobic training (walking, cycling, step machine, and step board) and 2 posts of resistance endurance exercises (leg press and exercises with rubber bands for quadriceps, gluteus/ hamstring region, and arms; 3 sets of 20 repetitions with each arm/leg), 70%–80% of peak VO2 (4–5 modified Borg scale	Supervised	Usual care
Pullen et al.51 (2008)	Aerobic (yoga)	2	8 wk	10-min warmup, 40-min standing or seated yoga postures (Asanas), and finally 20-min relaxation including breathing exercises (pranayama) and meditation	Supervised	Standard medical therapy
Ranković et al.52 (2009)	Aerobic (treadmill, bicycle or walking)	3	6 wk	45 min treadmill, room bicycle or walking at 70%–80% of HRmax	Supervised	Without exercise intervention
Redwine et al.17 (2020)-a	Aerobic (Tai chi)	2	16 wk	60 min of Tai chi chuan movements (Yang–style short form–first third) with 11–13 Borg scale, and warmup; 10–20 min per d on nonclass days at home	Supervised	Usual care
Redwine et al.17 (2020)-b	Resistance (resistance band)	2	16 wk	Resistance band (upper back, tricep extension, bicep curl, chest press, internal obliques, standing hip abduction, standing hip extension, seated leg extension, bent over rows, lateral rows) with 8- 10 repetitions on each side), with 11–13 Borg scale, and warmup; 10–20 min per d on nonclass days at home	Supervised	Usual care
Sandri et al.27 (2016)	Aerobic (bicycle)	4	4 wk	15-20 min cycle ergometer at 70% VO2max, 5 min of warmup and cool down	Supervised	Usual clinical care
Tisi et al.53 (1997)	Aerobic (active and passive leg exercises+walking)	7	12 mo	45 min of active and passive leg exercises performed to the limit of claudication pain with daily walks of at least 1 mile	Supervised	Normal volunteers and/or at least 5 mo following elective minor or intermediate non-arterial elective general surgery
Trippel et al.57 (2017)	Concurrent (bicycle+weight training)	3	12 wk	30–60 min, the first 4 weeks of bicycle ergometer at 50% peakVO2 in the first 2 wk and 70% peak VO2 after 1 mo, 10-min warmup and cool-down. After the initial 4 wk, 7 resistance training for major muscle groups; 12–15 repetitions at 60% of 1-RM with one repetition lasting 3s; After 3 mo, increase to 2 sets, allowing 90s of rest between sets.	Supervised	Usual care
Tsarouhas et al.54 (2011)	Aerobic (walking)	5	12 wk	40 min of walking at 40% of HRmax for 10 min progressing to reach 60% of HRmax	Unsupervised	Control group received usual care
Walther et al.37 (2008)	Aerobic (bicycling)	Unknown	24 mo	Daily bicycling.	Unsupervised	Control group received usual care and percutaneous coronary intervention
Wosornu et al.28 (1992)-a	Aerobic (running and bicycle)	3	6 mo	12–60 min of modified Canadian Airforce PBX (plan for 5 basic exercises) training; running on the spot, step ups, arm circling, star jumps, standing trunk curls, bridging, trunk rotation, side lying with hip abduction, arm raising, trunk side flexion, and crook lying with trunk rotation, and ended the session with a ride on a stationary bicycle	Supervised	Control group had no formal exercise training but continued with their leisure-time activities
Wosornu et al.28 (1992)-b	Resistance (weight training)	3	6 mo	12–60 min of leg extensions, hamstrings curl, biceps curl, push down, pull down, press behind neck, bench press, pulley row, military press, and sit-ups (rest periods of 45s between each station)	Supervised	Without formal exercise with their leisure-time activities
Yeh et al.55 (2011)	Aerobic (Tai chi)	2	12 wk	1-h Tai chi exercises by standard protocol of a pilot trial	Supervised	Time-matched education without exercise
Zaidi et al.58 (2019)	Concurrent (walking and bicycling + weight training)	2	12 mo	10–15 min of warmup and 5–10 min of cool down and concurrent program: 1) circuit training containing 10 aerobic and resistance exercises of large muscle groups (40 s work, 20 s break); 2) interval training (RPE ≥ 15) uphill walking (running) outdoors (20 s on/off for 3–4 min, 5–6 sets); 3) interval step training indoors (3-min series with basic steps, side steps and crossover steps, 4–5 sets) and resistance training; and 4) spinning on a bike (including pyramid intervals 6 × 20, 4 × 40, and 2 × 60 s) and resistance training of chest, biceps, shoulder, triceps, back and front, and 10–15 repetitions. Unsupervised home-based exercise session (eg, walking, swimming, bicycling, cross-country skiing and resistance training in health studios).	Supervised and unsupervised	Normal follow-up by their general practitioner or without exercise intervention

HRmax, maximal heart rate; peak VO₂, maximal oxygen consumption; RM, repetition maximum; RPE, rate of perceived exertion; rpm, revolutions per minute; VO₂max, maximal oxygen consumption.

The characteristics of control group-related interventions were as follows: use of non-exercise interventions;^3,30,33, 34, 35, 36^{,42,43,45,46,52,58} optimal medical therapy; usual care or standard therapy;^{9,13,14,17,26,27,29,31,32,37,}39, 40, 41^,44,47, 48, 49, 50, 51^,54,56,57 general home-based healthcare;²⁵ a home-based exercise group;^8,38 time-matched education;⁵⁵ leisure-time activities;²⁸ a pamphlet for daily exercising at home;⁷ received education and performed flexibility and stretching exercises;¹⁰ aerobic plus resistance exercises at only 10% of their sustained maximal inspiratory pressure;¹² functional electrical stimulation;¹¹ normal volunteers or at least 5 months following elective minor;⁵³ and sham-inspiratory muscle training.¹⁶

Inflammatory markers

Inflammatory markers in patients with HF including serum/plasma hs-CRP, IL-6, TNF-α, sVCAM-1, sICAM-1, fibrinogen, IL-1-β, IL-10, IL-18, and E-selectin levels were measured in 24 articles,^7,9,12, 13, 14^,16,17,31, 32, 33^{,37,39,40,42,44,}46, 47, 48^,50, 51, 52, 53^,55,56 19 articles,^{3,11,13,17,25,30,32,33,35,38,40,41,44,45,48,49,51,56,57} 20 articles,^{3,11,13,17,26,30,32,}34, 35, 36^,38, 39, 40, 41^,44,49,54, 55, 56, 57 8 articles,^{8,9,11,27,40,46,49,52} 9 articles,^{3,8,9,11,27,32,40,43,52} 7 articles,^{9,13,28,39,44,48,53} 4 articles,^10,36,44,57 4 articles,^11,39,49,57 3 articles,^10,40,58 and 4 articles,^3,40,43,49 respectively.

Meta-analysis

hs-CRP

A significant reduction occurred in hs-CRP level (SMD –0.441 [95% confidence interval (CI): –0.642 to –0.240], P = 0.001) with exercise training (Fig. 2). Significant heterogeneity was present, as shown by an I² > 50% (I² = 73.638%, P = 0.001). Subgroups by age and BMI showed that a significant reduction occurred in hs-CRP level in middle aged participants (SMD –0.490 [95% CI: –0.841 to –0.138], P = 0.006), elderly participants (SMD –0.421 [95% CI: –0.674 to –0.167], P = 0.001), those with overweight status (SMD –0.550 [95% CI: –0.847 to –0.254], P = 0.001) and those with unknown status (SMD –0.337 [95% CI: –0.670 to –0.004], P = 0.048), in patients with HF compared to those in the control group. In addition, subgroup by type of exercise revealed that a significant reduction occurred in hs-CRP level with aerobic exercise and with concurrent training, compared to the control group (SMD –0.415 [95% CI: –0.636 to –0.194], P = 0.001) and (SMD –0.659 [95% CI: –1.259 to –0.059], P = 0.031), respectively. In the subgroups of exercise intensity, a significant reduction was seen in hs-CRP for both high-intensity exercise (SMD –0.462 [95% CI: –0.648 to –0.277], P = 0.001) and moderate-intensity exercise (SMD –0.189 [95% CI: –0.286 to –0.091], P = 0.001), compared to the control group. Analysis of subgroups by duration of follow-up showed that a significant reduction in hs-CRP level occurred for 3 durations, as follows: short-term (SMD –0.177 [95% CI: –0.219 to –0.014], P = 0.025); long-term (SMD –0.753 [95% CI: –1.478 to –0.028], P = 0.042); and very long-term (SMD –0.555 [95% CI: –1.005 to –0.104], P = 0.016).

Figure 2.

Forest plot of high-sensitivity C-reactive protein. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

IL-6

A significant reduction occurred in IL-6 (SMD –0.158 [95% CI: –0.303 to –0.013], P = 0.032) with exercise training (Fig. 3). No significant heterogeneity was present, as shown by an I² < 50% (I² = 0.0%, P = 0.815). Analysis of subgroup by age showed that a significant reduction occurred in IL-6 in middle-aged participants (SMD –0.304 [95% CI: –0.519 to –0.089], P = 0.006), whereas no significant change occurred in elderly participants, by BMI status, or by type, intensity, or follow-up duration of exercise in patients with HF (P > 0.05).

Figure 3.

Forest plot of interleukin-6. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

TNF-α

Exercise intervention did not significantly change TNF-α values (SMD –0.067 [95% CI: –0.370 to 0.235], P = 0.663; Fig. 4). Significant heterogeneity was present, as shown by an I² > 50% (I² = 79.24%, P = 0.001). Analysis of subgroups by age and BMI showed that a significant reduction occurred in TNF-α values in middle-aged participants (SMD –0.430 [95% CI: –0.838 to –0.022], P = 0.039) and those with unknown status (SMD –0.318 [95% CI: –0.519 to –0.117], P = 0.002], whereas no significant change occurred in elderly participants, those with overweight status, or by type, intensity, or follow-up duration of exercise, in patients with HF (P > 0.05).

Figure 4.

Forest plot of tumour necrosis factor-α. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

sICAM-1

Exercise intervention significantly reduced sICAM-1 values (-0.282 [SMD and 95% CI: –0.477 to –0.086], P = 0.005; Fig. 5). No significant heterogeneity was present, as shown by an I² < 50% (I² = 18.632%, P = 0.266). Analysis of subgroups by age, BMI, and type, intensity, and duration of exercise showed that a significant reduction occurred in sICAM-1 level in middle aged participants (SMD –0.395 [95% CI: –0.693 to –0.098], P = 0.009), those with overweight status (SMD –0.378 [95% CI: –0.752 to –0.003], P = 0.048), and with aerobic exercise (SMD –0.277 [95% CI: –0.484 to 0.069], P = 0.009), moderate-intensity exercise (SMD –0.645 [95% CI: –1.194 to –0.096], P = 0.021), and with short-term follow-up (SMD –0.318 [95% CI: –0.536 to –0.100], P = 0.004). By contrast, no significant change occurred in elderly participants, those with normal or unknown status, with resistance and concurrent exercises, at low and high intensities, and with long-term duration, in patients with HF (P > 0.05).

Figure 5.

Forest plot of soluble intercellular adhesion molecule-1. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

sVCAM-1

No significant difference with exercise intervention occurred in sVCAM-1 values (-0.156 [SMD and 95% CI: –0.476 to 0.164], P = 0.339; Fig. 6). Heterogeneity analysis showed significant heterogeneity, as indicated by an I² > 50% (I² = 58.196%, P = 0.010). Analysis of subgroups by age, BMI, and type, intensity, and duration of exercise revealed that no significant differences occurred in sVCAM-1 values in patients with HF (P > 0.05).

Figure 6.

Forest plot of soluble vascular cell adhesion molecule-1. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

Fibrinogen

No significant difference with exercise intervention occurred in fibrinogen level (-0.186 [SMD and 95% CI: –0.422 to 0.050], P = 0.122; Fig. 7). No significant heterogeneity was present, as shown by an I² < 50% (I² = 0.0%, P = 0.671). Analysis of subgroups by age, BMI, type, and intensity and duration of exercise demonstrated that no significant differences occurred in fibrinogen level (P > 0.05).

Figure 7.

Forest plot of fibrinogen. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

IL-1-β

No significant difference with exercise intervention occurred in IL-1-β level (SMD –0.265 [95% CI: –0.586 to 0.106], P = 0.780; Fig. 8). No significant heterogeneity was present, as shown by an I² < 50% (I² = 0.0%, P = 0.691). Analysis of subgroups by age, BMI, and type, intensity, and duration of exercise showed that no significant differences occurred in IL-1-β level (P > 0.05).

Figure 8.

Forest plot of interleukin-1-β. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

IL-10

No significant difference with exercise intervention occurred in IL-10 level (SMD 0.267 [95% CI: –0.038 to 0.566], P = 0.086; Fig. 9). No significant heterogeneity was present, as shown by an I² < 50% (I² = 0.0%, P = 0.964). Analysis of subgroups by age, BMI, and type, intensity, and duration of exercise demonstrated that no significant differences occurred in IL-10 level (P > 0.05).

Figure 9.

Forest plot of interleukin-10. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

IL-18

No significant difference with exercise intervention occurred in IL-18 level (SMD –0.015 (95% CI: –0.257 to 0.228], P = 0.906; Fig. 10). No significant heterogeneity was present, as shown by an I² < 50% (I² = 0.0%, P = 0.851). Analysis of subgroups by age, BMI, and type, intensity, and duration of exercise showed that no significant differences occurred in IL-18 level (P > 0.05).

Figure 10.

Forest plot of interleukin-18. CI, confidence interval; Std diff, standard difference.

E-selectin

No significant difference with exercise intervention occurred in E-selectin level (SMD –0.293 [95% CI: –0.587 to 0.002], P = 0.051; Fig. 11). No significant heterogeneity was present, as shown by an I² < 50% (I² = 0.0%, P = 0.638). Analysis of subgroups by age, BMI, and type, intensity, and duration of exercise showed that no significant differences occurred in E-selectin level (P > 0.05).

Figure 11.

Forest plot of E-selectin. BMI, body mass index; CI, confidence interval; Std diff, standard difference.

Quality assessment and publication bias

The quality assessment of included studies (Pedro scores ranged from 7 to 15, with a maximum of 15 scores; high risk of bias = score < 5, and low risk of bias = score > 10) is illustrated in Table 1. Visual interpretation of a funnel plot and Egger’s test were used to assess publication bias (in each case that follows, the asymmetry or symmetry of the funnel plot was confirmed by Egger’s test as significant or nonsignificant). An asymmetric distribution suggested the presence of a publication bias for hs-CRP (Fig. 12), confirmed as significant (P = 0.027). An asymmetric distribution (ie, publication bias was present) was seen for IL-6 (Fig. 13), confirmed as significant (P = 0.013). A symmetric distribution (ie, publication bias was not present) was seen for TNF-α (Fig. 14), confirmed as nonsignificant (P = 0.258). An asymmetric distribution (significant publication bias) was seen for sICAM-1 (Fig. 15), confirmed as significant (P = 0.015). A symmetric distribution (ie, publication bias was not present) was seen for sVCAM-1 (Fig. 16) and for fibrinogen (Fig. 17), confirmed in each case as nonsignificant (sVCAM-1 P = 0.249; fibrinogen P = 0.646). An asymmetric distribution (ie, publication bias was present) was seen for IL-1-β (Fig. 18), confirmed as significant (P = 0.029). A symmetric distribution (ie, publication bias was not present) was seen for IL-10 (Fig. 19), IL-18 (Fig. 20), and E-selectin (Fig. 21), each confirmed as nonsignificant (P = 0.464 for IL-10, P = 0.188 for IL-18, and P = 0.374 for E-selectin).

Figure 12.

Funnel plot for high-sensitivity C-reactive protein. Std diff, standard difference.

Figure 13.

Funnel plot for interleukin-6. Std diff, standard difference.

Figure 14.

Funnel plot for tumour necrosis factor-α. Std diff, standard difference.

Figure 15.

Funnel plot for soluble intercellular adhesion molecule-1. Std diff, standard difference.

Figure 16.

Funnel plot for soluble vascular cell adhesion molecule-1. Std diff, standard difference.

Figure 17.

Funnel plot for fibrinogen. Std diff, standard difference.

Figure 18.

Funnel plot for interleukin-1-β. Std diff, standard difference.

Figure 19.

Funnel plot for interleukin-10. Std diff, standard difference.

Figure 20.

Funnel plot for interleukin-18. Std diff, standard difference.

Figure 21.

Funnel plot for E-selectin. Std diff, standard difference.

Discussion

Our results indicate that exercise interventions with durations of 3weeks to 24 months, with an exercise frequency of 2-7 days per week, intensity ranges at 60%-80% HRmax, and a duration per session of 30-60 minutes significantly reduced hs-CRP, IL-6, TNF-α, and sICAM-1 values in patients with HF who had overweight status and were middle-aged. In other words, our results showed small and medium-sized effects of reduced hs-CRP level in patients with HF in the following subgroups: middle-aged (d = 0.490); elderly (d = 0.421); with overweight status (d = 0.550); with aerobic training (d = 0.415); with concurrent exercise (d = 0.659); high-intensity (d = 0.462); long-term (d = 0.753); or very long-term (d = 0.555). In addition, small and medium-sized effects of reduced IL-6 occurred in middle-aged patients (d = 0.304), of reduced TNF-α in middle-aged patients (d = 0.430), and reduced sICAM-1 in middle-aged patients (d = 0.395), those with overweight status (d = 0.378), those who performed aerobic exercise (d = 0.277), those who performed moderate-intensity exercise (d = 0.645), and with short-term follow-up (d = 0.318), in patients with HF. However, age, BMI, and type, intensity, and duration of exercise were not associated with any significant changes in IL-1-β, IL-10, IL-18, fibrinogen, sVCAM-1, or E-selectin values in patients with HF.

Given that cardiomyocytes are an important source of inflammatory markers in heart damage, an increase in cardiac cytokines can be associated with reduced cardiac function, especially in patients with HF.⁵⁹ HF is associated with an increase in systemic inflammation, pro-inflammatory cytokines, and endothelial activation.^59,60 In contrast, exercise is an effective nonpharmacologic intervention to reduce inflammatory markers and VAMs, and it is certainly important in terms of clinical exercise physiology, given that the inflammation process plays a major role in the development of many metabolic disorders.⁴ These findings suggest that cardiomyocyte-related inflammatory markers and endothelial activation may play an important role in the pathologic and physiological responses of the heart, especially clinical HF. In addition, novel biomarkers such as IL-1 receptor-like 1 (IL1RL1), soluble ST2, B-type natriuretic peptide (BNP), and N-terminal-pro hormone BNP (NT-proBNP) were considered for the prevention, management, and assessment of HF,^2,61 as cardiovascular peptides were described in our previous meta-analysis.² On the other hand, a decrease in inflammation, as demonstrated by a noted reduction in hs-CRP, IL-6, TNF-α, and sICAM-1 following exercise, is associated with improved cardiovascular function in patients with HF, as observed after aerobic and concurrent interventions in this meta-analysis.

Our findings were consistent with the results of Pearson et al.,⁶² Monteiro-Junior et al.,⁶³ and Hayashino et al.¹⁸, which indicated that exercise training decreased TNF-α, CRP, and IL-6 values in HF patients, those with T2DM, and older persons. Alizaei-Yousefabadi et al.⁶⁴ reported that exercise was effective at lowering CRP and TNF-α values in patients with metabolic syndrome. In addition, Xing et al.⁶⁵ demonstrated that combined exercise led to a significant reduction in CRP, IL-6, and TNF-α values in middle-aged and older adults with T2DM, compared to the control group. Huang et al.⁶⁶ also demonstrated that aerobic exercise led to a significant reduction in IL-6 and TNF-α markers in patients with dementia. In contrast, other meta-analysis studies^19,67 reported no significant differences in inflammatory markers (IL-6 and CRP) between patients with metabolic disorders undergoing high-intensity interval training/exercise, compared with controls. Kim and Yeun⁶⁸ reported that resistance training was effective in reducing levels of CRP, IL-6, and TNF-α in elderly adults, a result inconsistent with our findings.

Chronic inflammation plays an important role in the pathogenesis of HF.⁶¹ After cardiac dysfunction, the adaptive immune system-induced inflammatory response upregulates cytoprotective responses, and it provides the heart with a short-term adaptation to elevated stress and physiological inflammation. However, this acute inflammatory response can turn into a dysregulated inflammatory response and chronic inflammation, resulting in left ventricular systolic and diastolic dysfunction, as well as HF.^61,69,70 The inflammatory process is associated with the activation of the renin-angiotensin-aldosterone and sympathetic nervous systems in HF.² On the other hand, cytokines and inflammatory markers have been found to regulate VAMs, such as ICAM and VCAM mRNA and protein, in both cardiomyocytes and fibroblasts.⁷¹ In addition, higher concentrations of systemic inflammation and cardiomyocyte-induced cytokines are associated with a higher risk of morbidity and mortality from any cause, pump failure and sudden cardiac death, as well as potential pathogenic effects of endothelial activation in HF.^60,72 In addition, circulating inflammatory cytokines and VAMs are associated with serum insulin,⁷³ visceral fat,⁴ BMI,^4,74 waist circumference,⁷⁴ overweight/obesity,^4,73,74 and fat-free mass or lean mass.^4,75

Thus, the effects of pathologic and physiological conditions on circulating inflammatory levels appear to be contradictory. In other words, acute exercise results show that circulating levels of inflammatory markers significantly increase immediately after exercise, with a reduction on the following day, and return to baseline values within 72 hours.⁷⁰ It is possible that an increase in circulating inflammation levels is related to the exercise-induced physiological endocrine responses to an increase in myocardial stress,^61,70 suggesting that inflammatory response releases during and after exercise are related to the cytoprotective and growth-regulating responses as well as the physiological reaction of cardiomyocytes,⁶¹ but not myocardial damage.⁷⁰ However, the physiological mechanism of exercise-induced inflammatory responses in HF remains unclear.

The results of these meta-analyses demonstrate that reduction of cardiovascular peptide levels by aerobic training is associated with improved cardiorespiratory function, increased VO₂max, improved workload, and increased left ventricular ejection fraction in HF.^2,76 consistent with the results of our meta-analysis. In addition, these positive physiological adaptations and decreased levels of inflammatory cytokines were reported to occur with concurrent and resistance exercises in HF.11, 12, 13, 14 Given that the exercise, especially aerobic exercise, plays an important role in improving myocardial stretch markers, regulated cardiac function, optimal reduced inflammatory cytokines, and improved endothelial function⁷³, considering aerobic exercise as part of the therapeutic approach to patients with HF, initiated through structured cardiac rehabilitation programs, may be appropriate. Additionally, concurrent exercise can be effective in reducing inflammation in patients with HF. Inflammatory cytokines have been reported to be secreted by cardiomyocytes, adipose tissue, lung epithelial cells, skeletal muscle mass, and obesity.^4,55,77 In contrast, a number of studies reported that exercise interventions lead to a reduction in adipose tissue by decreased expression of toll-like receptors (TLRs) on macrophages and monocytes,^4,78 an increase in blood flow to the respiratory and skeletal muscles, improved pulmonary function, VO₂max, and muscle mass,⁷⁶ as well as improved left ventricular ejection fraction in HF.^2,76 These positive physiological adaptations from aerobic and concurrent exercises may be associated with a reduction in production of pro-inflammatory cytokines, such as IL-6, TNF-α, and liver-induced CRP in patients with HF^2,79 as confirmed by the results of our meta-analysis. In addition, the reduction of hs-CRP, IL-6, and TNF-α levels by moderate- and high-intensity aerobic and concurrent exercise interventions with short-term, long-term, and very long-term follow-ups may regulate the VAMs of ICAM and VCAM in both cardiomyocytes and fibroblasts in patients with HF who are middle-aged and have overweight status,^62,71 consistent with the results of our meta-analysis. The downregulation of inflammatory markers of hs-CRP, IL-6, TNF-α, and regulation of sICAM-1 can be confirmed to be anti-inflammatory effects of both aerobic exercise and concurrent training in patients with HF. The mechanism of increased anti-inflammatory properties induced by aerobic and concurrent interventions may counteract the adverse adaptations of the activation of the renin-angiotensin-aldosterone system and sympathetic nervous activity in HF.² Based on the science of training in clinical exercise physiology, optimal protocol and intensity, frequency, and duration of exercise interventions are related to the release of cardiac inflammatory markers and/ or VAMs,^2,74 although the differential effects of exercise on inflammatory responses in patients with HF who are in different age groups and have different BMIs are not fully understood.

Limitations

The subgroup analysis based on age, BMI, and type, intensity, and duration of exercise, as well as using the most common period of 3 months (12 weeks) in patients with HF, were strengths of our meta-analysis. However, limitations of this meta-analysis are as follows: the lack of studies examining the association of resistance training with IL-1-β, IL-8, IL-18, and E-selectin markers; the lack of data on concurrent training and its association with IL-8 and sVCAM-1 markers; the low number of included original articles based on subgroup by sex, and type of exercise; and heterogeneity and publication bias in the data.

Conclusions

Aerobic and concurrent exercises (based on the type, intensity, and duration of exercise) in patients with HF of middle-aged and overweight status were effective in improving inflammatory and vascular adhesion markers, including levels of hs-CRP, IL-6, TNF-α, and sICAM. These results can be considered an exercise prescription in exercise and cardiac rehabilitation programs to improve the inflammation process in patients with HF, although IL-1-β, IL-10, IL-18, fibrinogen, sVCAM-1, and E-selectin markers remained unchanged. This meta-analysis established that exercise interventions had the anti-inflammatory effects of improved inflammatory and vascular inflammation profiles, and these clinical benefits in general and exercise-based cardiac rehabilitation improve clinical evolution and survival in patients with HF (Figure 22).

Figure 22.

Graphical results of our meta-analysis. HF, heart failure; HRmax, maximal heart rate; HRmaxR, maximal heart rate reserve; hs-CRP, high-sensitivity C-reactive protein; IL, interleukin; RCT, randomized controlled trial; RM, repetition maximum; sICAM-1, soluble intercellular adhesion molecule-1; sVCAM-1, soluble vascular cell adhesion molecule-1; TNF-α, tumor necrosis factor alpha; VO₂max, maximal oxygen consumption.

Clinical perspectives

Clinical implications of our meta-analysis are as follows: a reduction in circulating hs-CRP, IL-6, TNF-α, and sICAM values occur after aerobic and concurrent exercise, based on the type, intensity, and duration of exercise in those patients with HF who are middle aged, elderly, or of overweight status. In other words, our meta-analysis established that reduction in inflammation and improved vascular inflammatory profiles were both anti-inflammatory effects of exercise interventions. These results indicate that these exercise-related changes have general clinical benefits, and exercise-based cardiac rehabilitation, in a more specific format, will improve clinical evolution and survival in patients with HF of different etiologies. Therefore, in future research, exercise prescriptions, such as aerobic and concurrent exercises, can be considered nonpharmacological interventions in exercise and cardiac rehabilitation programs for improving inflammation and vascular inflammatory profiles in patients with HF.