Acute Exercise before Dialysis Is as Cardioprotective as during Dialysis

Visual Abstract

Abstract

Key Points

• Physical exercise before hemodialysis is as cardioprotective as intradialytic exercise.

• Predialytic exercise potentially addresses several modality-specific barriers and challenges encountered by both health care providers and patients.

• These are likely mediated by mechanisms inherent to exercise itself, rather than by transient central and/or systemic hemodynamic alterations.

Background

Hemodialysis induces left ventricular regional wall motion abnormalities (RWMAs) due to myocardial hypoperfusion. Although acute intradialytic exercise (IDE) has shown cardioprotective effects, its routine implementation faces feasibility challenges, and the potential of predialysis exercise as an alternative remains unexplored. This study aimed to compare the effect of predialysis exercise and IDE on hemodialysis-induced myocardial stunning.

Methods

In this open-label, randomized cross-over trial, 25 patients with ESKD underwent to each of three hemodialysis conditions, administrated in random order: standard hemodialysis (HD-_CONT), hemodialysis with IDE (HD-_PER), and hemodialysis preceded by exercise (HD-_PRE). Two-dimensional echocardiography and whole blood viscosity (WBV) measurements were performed both immediately before hemodialysis onset (T₀) and at peak stress of hemodialysis (T_peak). Left ventricular longitudinal strain from an 18-segment model was used to assess the presence of RWMAs. Regular monitoring of cardiovascular hemodynamics was set up with measurements staggered every 30 minutes.

Results

Compared with HD-_CONT, there was a significant reduction in RWMAs during both HD-_PER (estimated difference, 1.60 segments; 95% confidence interval, 0.09 to 3.10; P = 0.04) and HD-_PRE (estimated difference, 1.72 segments; 95% confidence interval, 0.21 to 3.22; P = 0.02). The magnitude of the exercise-induced reduction in myocardial stunning did not differ between HD-_PER and HD-_PRE (P = 0.86). Apart from the exercise period itself, kinetics of all hemodynamic variables were similar between HD-_CONT and HD-_PER, whereas they were totally similar between HD-_CONT and HD-_PRE. No associations of changes in RWMAs and hemodynamics variables between HD-_CONT versus HD-_PRE or HD-_PER were found (P > 0.42). Comparing HD-_CONT versus HD-_PER, WBV was preserved in HD-_PER and changes in RWMAs were associated with changes in WBV at high shear rates (225 s⁻¹: P = 0.006; 90 s⁻¹: P = 0.04).

Conclusions

Physical exercise performed before hemodialysis provides cardioprotective effects comparable with those of IDE. The mechanisms behind these benefits do not seem to involve hemodynamic factors.

Clinical Trial registry name and registration number:

NCT04831021.

Introduction

Cardiovascular disease is the leading cause of death in people with ESKD.¹ Hemodialysis is recognized as an independent risk factor for major cardiac events and mortality.^2,3 Previous studies have shown that hemodialysis is associated with disturbances in left ventricular (LV) myocardial segmental perfusion, severe enough to cause transient LV regional wall motion abnormalities (RWMAs), well-known as myocardial stunning.^4–6 The prevalence of RWMAs is associated with a high risk of heart failure or cardiovascular mortality in patients with ESKD,⁷ highlighting the need for strategies to minimize RWMAs during the hemodialysis procedure. Intradialytic exercise (IDE) has emerged as a safe and effective nondrug therapeutic approach to improve cardiovascular health and is now recommended for people undergoing hemodialysis.^8,9 Using speckle tracking echocardiography, two exploratory studies have demonstrated that acute IDE can reduce RWMAs compared with usual care.^10,11 Our team has further shown that acute IDE can prevent the decline in LV myocardial mechanics (i.e., reduction in LV longitudinal and circumferential strains as well as torsional mechanics) typically observed during standard hemodialysis.¹² Despite these benefits, integrating IDE into routine hemodialysis care remains challenging because of feasibility issues. Key barriers include patient-related factors such as fatigue and safety concerns (e.g., hemodialysis-induced fatigue, muscle cramps, and hypotension) along with logistical constraints at dialysis centers (e.g., insufficient equipment, competing clinical priorities during hemodialysis).¹³ Predialysis exercise may offer a promising alternative to IDE, potentially overcoming some of these modality-specific barriers from both patient and health care provider perspectives. However, to the best of our knowledge, no studies have yet evaluated the effect of predialysis exercise on hemodialysis-induced myocardial stunning. We therefore conducted a randomized cross-over trial comparing three conditions: standard hemodialysis, IDE, and predialysis exercise. Our objective was to determine whether both exercise interventions could similarly reduce RWMAs and confer cardioprotective effects relative to standard hemodialysis.

Methods

Study Design and Participants

The EXp-DIAL (Predialytic or per-dialytic physical exercise: a cardioprotective role) study was a randomized, open-label, monocentric, and proof-of-concept clinical trial. Untrained ESKD patients, aged 20–79 years, undertaking maintenance hemodialysis for >3 months were eligible to participate. Exclusion criteria were orthopedic complication, ejection fraction <45%, severe heart or respiratory diseases, severe obesity, and poor echogenicity. All participants received detailed information about the procedures and signed written informed consent before enrollment. The study was approved by the ethics committee North West III and registered at ClinicalTrials.gov (NCT04831021). Recruitment began in August 2021 and ended in March 2023, which also marked the completion of the study. Figure 1 shows the flow chart of patient recruitment. This study used a cross-over design, in which all participants underwent each of the three hemodialysis conditions, administrated in a randomized order: standard hemodialysis (HD-_CONT), hemodialysis incorporating 30 minutes of per-dialytic exercise (HD-_PER), and hemodialysis preceded by 30 minutes of exercise before the beginning of hemodialysis (HD-_PRE; Figure 2). Randomization was conducted using the Capture system software Clinsight. During HD-_PRE and HD-_PER, subjects underwent a continuous cycling test (50–70 rpm) of moderate intensity (11–14 on the 20-point Borg scale¹⁴). Exercise was performed in a semirecumbent position on a calibrated cycle ergometer (oxycycle 3—Physiomed) attached to each patient's bed. It started 30 minutes after hemodialysis onset for HD-_PER and 60 minutes before hemodialysis for HD-_PRE. The effort lasted 30 minutes and included each time 5 minutes of warm-up and 5 minutes of cool-down at 50% of the exercise workload. Echocardiography was performed twice during each session: immediately before hemodialysis (T₀) and at peak stress of hemodialysis (i.e., 30 minutes before hemodialysis ending, T_Peak). Central hemodynamic data and BP were collected at T₀ and every 30 minutes throughout hemodialysis (Figure 2). HD-_CONT, HD-_PRE, and HD-_PER were conducted consistently 1 week apart, starting from the first hemodialysis-day of the week. Clinical routine examinations were conducted by separate medical staff. Echocardiographic scans were acquired by one experienced echographer (M. Josse) and analyzed offline without knowledge of patient identities or examination order, ensuring blinded assessment.

Figure 1.

Study consort diagram. Study screening, echo window, consent, participation, and dropout.

Figure 2.

Schematic for HD-_CONT, HD-_PER, and HD-_PRE sessions. Each participant underwent the three sessions in a randomized order. EX, exercise; HD, hemodialysis.

Outcome Measures

The primary outcome was the within-patient difference in the number of RWMAs during HD-_PRE compared with both HD-_PER and HD-_CONT. Secondary outcomes included within-patient difference in the number of intradialytic hypotension episodes per 100 hours of hemodialysis as well as changes during hemodialysis (from T₀ to T_peak) in global longitudinal strain (GLS; i.e., mean of four, three, and two chamber views), central hemodynamics (i.e., heart rate, stroke volume, and cardiac output), BP, as well as whole blood viscosity (WBV).

Echocardiography

Transthoracic echocardiography was performed using a Vivid Q system (GE healthcare, Horten, Norway) equipped with a 3.4-MHz Transducer.

Standard Echocardiography

Acquisitions were recorded at rest just before the hemodialysis session, in accordance with the guidelines from the American Society of Echocardiography.¹⁵ LV internal diameters and wall thicknesses were measured from the parasternal long axis view. LV mass was calculated by the Devereux formula and indexed to body surface area. LV end-diastolic and end-systolic volumes and LV ejection fraction were measured using the Simpson's biplane method. LV diastolic function was assessed using early (E) and atrial transmitral flow velocities, recorded in apical four-chamber view. Myocardial early diastolic velocities (e′) were measured with a color-coded Doppler Tissue Imaging at the mitral annular level in apical four-chamber view. Peak e′ (i.e., average recorded on the LV septal and lateral wall) and the E/e′ ratio were used as indexes of LV relaxation and filling pressure, respectively.¹⁶

Echocardiography during Hemodialysis

Ultrasound scans were obtained at T₀ and T_peak (Figure 2). Cine loops triggered by the QRS complex in parasternal long-axis and apical four, two, and three chamber views were recorded. For each view, the three cycles with the best image quality were saved and used for blinded offline analysis using dedicated software (EchoPAC 203TM—GE Healthcare, Chicago). Speckle tracking echocardiography was performed in accordance with recent guidelines.¹⁷ Segmental and GLS were calculated using a 18-segment model from apical four, two, and three chamber views, as previously described in our laboratory.^18,19 Data were processed with a specific toolbox (MATLAB 2021a) and normalized to the percentage of systole (i.e., aortic valve closure representing 100% of systole) using interpolations. RWMAs were evaluated using longitudinal strain derived from speckle tracking echocardiography, the most accurate and sensitive echocardiographic technique for detecting wall motion abnormalities.^20,21 RWMAs were identified from the 18-segment model. For each segment, a reduction >20% in peak longitudinal strain at T_peak relative to its peak value at T₀ was an indication of RWMAs.⁵ The presence of myocardial stunning was defined as the presence of two or more segments with RWMAs.¹⁰ LV volumes and internal dimensions and wall thicknesses were also measured at T₀ and T_peak. LV end-diastolic volume was obtained as preload index, while the systolic meridional wall stress (σes) was used as an index of afterload.²²

Hemodynamics and WBV during Hemodialysis

Regular monitoring of cardiovascular hemodynamics was set up with measurements staggered every 30 minutes from hemodialysis onset for aortic blood flow, heart rate, and BP. Cardiac output and stroke volume were measured, as previously described.¹² Brachial BP was measured on the nonaccess arm using an automated BP cuff integrated into the dialysis unit. Intradialytic hypotension was defined as fall of systolic BP of more than 20 and/or 10 mm Hg of mean arterial pressure from the initial BP (at T₀) associated with symptoms.²³ WBV was determined after complete blood oxygenation, at native hematocrit and different shear rates (2.25, 11.25, 22.5, 45, 90, and 225 s⁻¹) using a cone-plate viscometer (Brookfield DVII+ with CPE40 spindle, Brookfield Engineering Labs, Natick, MA). Hemorheologic analyses were performed within 30 minutes after blood sampling to avoid any blood alteration and international guidelines were followed.²⁴

Clinical Examination and Hemodialysis Settings

On the first hemodialysis session, biometric measurements (height, weight, body mass index) and comprehensive clinical examination was conducted. Routine biochemistries and hematologic data were obtained from patient charts (Table 1). Routine treatment parameters followed the medical prescription. Net ultrafiltration was determined clinically based on ideal dry weight for each session. Dialysis information was extracted from medical records at the conclusion of each hemodialysis assessment session.

Table 1.

Participant characteristics

Variable	ESKD (n=25)
Age (yr)	59±16
Sex. n (male/female)	17/8
Dry weight (kg)	74.5±14.9
Body mass index (kg/m2)	25.9±4.3
Dialysis vintage (yr)a	2.4 (1.6–10)
Hemoglobin (g/100 ml)	11.7±2.1
Comorbidities, n (%)
Diabetes	10 (40)
Hypertension	14 (56)
Hyperparathyroidism	5 (20)
Peripheral vascular disease	2 (8)
CKD etiology, n (%)
GN	4 (16)
IgA nephropathy	1 (4)
Hypertensive nephropathy	8 (32)
Diabetic nephropathy	8 (32)
Congenital	1 (4)
Indeterminate	1 (4)
Other	7 (28)
Medication, n (%)
Angiotensin-converting enzyme inhibitor	1 (4)
Anticoagulants	8 (32)
Nitrates	1 (4)
Statins	14 (56)
Diuretics	6 (24)
Antiarrhythmic	3 (12)
Calcium channel blockers	2 (8)
β-Blockers	12 (45)
Erythropoietin	1 (4)
Nonsteroidal anti-inflammatory	3 (12)
Aspirin	11 (44)

^aMedian (lower quartile–upper quartile).

Statistical Analysis

Statistical analysis was performed using IBM SPSS statistics version 26.0 (IBM, Armonk, NY). Continuous data were expressed as mean±SD or median and interquartile range, and categorical data were expressed with frequency count. Statistical significance was defined as a P value < 0.05. Sample size calculation was based on the results from Penny et al. comparing RWMAs at peak stress hemodialysis during a standard hemodialysis and a hemodialysis session incorporating IDE.¹⁰ A reduction from 5.8±2.7 to 4.0±1.8 (P = 0.01) in the condition with exercise was reported in 19 patients with ESKD. We estimated that a total number of 21 individuals would be required, considering an effect size of 0.64 with a statistical power of 0.80 and α (two-sided) of 0.05 (G-Power Version 3.1.9.2, University of Düsseldorf, Germany). Further inflating this for an estimated drop-out rate of 10%, we calculated an overall sample size of 23. All analyses regarding the effect of exercise were based on intention-to-treat principle. The primary outcome was analyzed using a generalized linear mixed model (with a Poisson regression model) with condition (i.e., HD-_CONT, HD-_PRE, and HD-_PER) and session sequence as fixed effects and a random effect for patients. There were no missing data in the three sessions for the primary outcome. Analyses of prespecified secondary outcomes were undertaken using a linear mixed-effects regression model, with condition and session sequence as fixed effects and a random effect for patients. For WBV, the same model was used with time (i.e., T₀ and T_peak) and shear rate as covariates. Differences between sessions for intradialytic hypotension were assessed using the same statistical model as the primary outcome. A generalized linear model (using a Poisson regression model) was used to assess possible relationships between change in RWMAs (i.e., HD-_CONT versus HD-_PRE and HD-_CONT versus HD-_PER) and changes in hemodynamics, WBV, hemodialysis and biologic parameters, intradialytic hypotension episodes, and secondary cardiovascular outcomes.

Results

Baseline Characteristics

A total of 100 individuals receiving hemodialysis were screened. Of the 48 eligible patients approached, 32 completed the inclusion visit. Five were excluded for poor echogenicity or one for aortic stenosis, one decided to withdraw after echocardiography, and 25 completed the three sessions (Figure 1). Baseline characteristics are presented in Table 1. The mean age was 59±16 years, and 68% of the subjects were male. Body weight at prehemodialysis and posthemodialysis as well as all hemodialysis parameters were strictly similar between HD-_CONT, HD-_PRE, and HD-_PER, and no condition by sequence interaction was noticed (Table 2). Standard echocardiographic data are presented in Table 3.

Table 2.

Hemodialysis parameters for the three sessions

Variable	HD-CONT	HD-PER	HD-PRE	Condition P Value	Condition by Sequence P Value
Weight (kg)
Pre	76.8±14.9	76.6±14.6	77.1±15.1	0.99	0.08
Post	74.2±14.7	74.0±14.5	74.2±14.8	0.99	0.18
Ultra-filtrated volume (L)	2.78±0.80	2.78±0.73	2.78±0.74	0.98	0.72
Duration (min)	241±10	241±10	240±12	0.99	0.13
Ultrafiltration rate (ml/h)	702±187	715±196	693±203	0.93	0.45
Kt/V	1.72±0.36	1.76±0.38	1.71±0.31	0.89	0.62
Urea (mmol/L)	24.49±5.83	24.06±5.93	23.50±5.83	0.40	0.30
Calcium (mmol/L)	2.20±0.14	2.18±0.14	2.19±0.15	0.74	0.88
Potassium (mmol/L)	4.98±0.91	4.93±0.97	5.19±0.94	0.29	0.71
Sodium (mmol/L)	137.7±2.5	137.8±2.6	137.5±2.5	0.81	0.25
Creatinine (μmol/L)	916±278	901±280	900±279	0.64	0.21
CRP us (mg/L)a	3.3 (1.4–8.0)	3.0 (1.2–6.0)	2.6 (1.1–7.5)	0.80	0.24

P values from analysis of log-transformed values. CRP us, C-reactive protein ultrasensitive.

^aMedian (lower quartile–upper quartile).

Table 3.

Echocardiographic characteristics

Variable	ESKD (n=25)
Two dimensions
LV end-diastolic volume (ml)	119±37
LV end-systolic volume (ml)	52±18
Relative wall thickness	0.34±0.06
LV mass index (g/m2)	100±28
LV ejection fraction (%)	55±8
Pulsed-wave Doppler
E (m/s)	0.85±0.26
E/A	1.12±0.41
TDI parameters
e′ (cm/s)	9.0±2.4
E/e′ ratio	10.6±5.9

A, peak late transmitral flow velocity; E, peak early transmitral flow velocity; e′, peak early diastolic mitral annular velocity; LV, left ventricle; TDI, tissue Doppler imaging.

Cardiac Function during Hemodialysis

Primary Outcome

Data are presented in Table 4. There was evidence for a reduction in RWMAs in HD-_PRE when compared with HD-_CONT (estimated difference, 1.60 segments; 95% confidence interval [CI], 0.09 to 3.10; P = 0.04) but not to HD-_PER (estimated difference, −0.12; 95% CI, −1.55 to 1.30; P = 0.86). There was also a reduction in RWMAs in HD-_PER when compared with HD-_CONT (estimated difference, 1.72 segments; 95% CI, 0.21 to 3.22; P = 0.02). There was no sequence effect (P = 0.66) nor condition by sequence interaction (P = 0.47), underscoring the absence of a carryover effect between HD-_PRE, HD-_PER and HD-_CONT conditions. During HD-_CONT, all ESKD individuals had evidence of myocardial stunning (25 out of 25), with a cumulative total of 199 segments presenting with RWMAs. During HD-_PRE and HD-_PER, 23 of 25 and 21 of 25 ESKD developed myocardial stunning, respectively, and there were fewer RWMAs when compared with HD-_CONT (156 and 159, respectively).

Table 4.

Myocardial mechanics at each time point during the three sessions

Variable	HD-CONT		HD-PER		HD-PRE		Condition	Between-Session Difference
							P Valuea		Effect (95% CI)	P Valueb
Primary outcome	Tpeak		Tpeak		Tpeak
RWMAs (n)	7.8±3.2		6.1±2.9		6.2±3.1		0.02c	HD-CONT versus HD-PER	1.72 (0.21 to 3.22)c	0.02c
								HD-CONT versus HD-PRE	1.60 (0.09 to 3.1)c	0.04c
								HD-PER versus HD-PRE	−0.12 (−1.55 to 1.30)	0.86
Secondary outcomes	T0	Tpeak	T0	Tpeak	T0	Tpeak
GLS (%)	−17.8±3.2	−14.1±3.0	−17.6±3.0	−15.2±2.8	−17.4±3.0	−15.0±2.8	0.002c	HD-CONT versus HD-PER	1.31 (0.23 to 2.39)c	0.01c
								HD-CONT versus HD-PRE	1.34 (0.27 to 2.42)c	0.01c
								HD-PER versus HD-PRE	−0.03 (−1.04 to 1.10)	0.99

CI, confidence interval; GLS, global longitudinal strain; HD_-CONT, standard hemodialysis; HD_-PER, hemodialysis with intradialytic exercise; HD_-PRE, hemodialysis preceded by exercise.

^aP value for the effect of condition generated by the generalized linear mixed model (using a Poisson regression model) for regional wall motion abnormalities and the linear mixed-effects regression model for delta change (e.g., difference from immediately before hemodialysis onset [T₀] to peak stress of hemodialysis [T_peak]) of other variables.

^bBetween-session difference in regional wall motion abnormalities and change from T₀ to T_peak in other variables, P value generated by the generalized linear mixed model (using a Poisson regression model) for regional wall motion abnormalities and the linear mixed-effects regression model for delta change (e.g., difference between T_peak and T₀) of other variables.

^cStatistically significant result (P < 0.05).

Secondary Outcomes

When compared with HD-_CONT, both HD-_PRE and HD-_PER attenuated the decline in GLS observed during hemodialysis from T₀ to T_peak (estimated difference, 1.34%; 95% CI, 0.27 to 2.42; P = 0.01; 1.31%; 95% CI, 0.23 to 2.39; P = 0.01; respectively; Table 4). Changes over the course of hemodialysis in cardiac loading condition indices were comparable in the three sessions (condition effect: delta end-diastolic volume: P = 0.93 and delta σ_es: P = 0.68). There was no condition by sequence interaction on delta changes in GLS (P = 0.10), end-diastolic volume (P = 0.26), and σes (P = 0.49).

Hemodynamics and WBV during Hemodialysis

Figure 3 illustrates the kinetics of delta changes from baseline (Δ_b) in hemodynamics parameters from T₀ to T_peak. Significant time by condition interactions was observed for Δ_bheart rate (P = 0.02, Δ_bcardiac output (P = 0.03) and Δ_bsystolic BP (P = 0.04), with a significantly greater increase (P < 0.001) in HD-_PER at 60 minutes compared with HD-_CONT. There were no differences between HD-_CONT and HD-_PRE. Systolic BP progressively decreased during hemodialysis (systolic BP time effect P = 0.03), whereas no change was obtained for diastolic BP (P = 0.38) and mean BP (P = 0.23). Cardiac output progressively decreased during all sessions (time effect: P = 0.009), while stroke volume (time effect: P = 0.13) and heart rate (time effect: P = 0.25) remained unchanged. It should be noted that no differences were observed between conditions at T₀ for all variables (P = 0.10 in each case for data in absolute terms). There was no condition by sequence interaction on Δ_bheart rate (P = 0.98), Δ_bstroke volume (P = 0.73), Δ_bcardiac output (P = 0.76), Δ_bsystolic BP (P = 0.84), Δ_bdiastolic BP (P = 0.48), and Δ_bmean BP (P = 0.67). No association of differences between HD-_CONT versus HD-_PRE or HD-_PER in RWMAs with hemodynamic variables (expressed as delta changes between T_peak and T₀) was found (P = 0.42). Finally, there was a trend toward fewer intradialytic hypotension episodes per 100 hours of hemodialysis in HD-_PRE and HD-_PER compared with HD-_CONT (HD-_CONT=103±125, HD-_PER=75±35, HD-_PRE=65±70, P = 0.13). No condition by sequence interaction was found on intradialytic hypotension (P = 0.16).

Figure 3.

Kinetics of hemodynamics parameters. Delta changes over the hemodialysis procedure (%) for heart rate (A), stroke volume (B), cardiac output (C), systolic BP (D), diastolic BP (E) and mean BP (F). HD-_CONT (dotted line), HD-_PER (solid line), and HD-_PRE (dashed line). Gray box indicates 30 minutes of exercise during hemodialysis. Estimated difference of change between HD-_PER versus HD-_CONT and HD-_PRE: *P < 0.05; ***P < 0.001.

Figure 4 shows changes in WBV measured at different shear rates in the three conditions. WBV was not different at T₀ in all three sessions at all shear rates (P = 0.46). WBV increased from T₀ to T_peak during HD-_CONT over the full range of shear rates, whereas no changes were noticed for HD-_PER (time effect P = 0.59) as well as for HD-_PRE, except at 225 s⁻¹ (P < 0.05). There was no condition by time by sequence by shear rate interaction on WBV (P = 0.21). Significant relationships of differences between HD-_CONT with HD-_PER in RWMAs with WBV changes (i.e., T_peak−T₀) were found at high (225 s⁻¹: Exp [β]=1.65, P = 0.006; 90 s⁻¹: Exp [β]=1.28, P = 0.04) but not medium (45 s⁻¹: Exp [β]=1.02, P = 0.93) or low (2.25 s⁻¹: Exp [β]=1.01, P = 0.23) shear rate. No relationships were however observed for HD-_CONT compared with HD-_PRE regardless of shear rates (225 s⁻¹: Exp [β]=1.08, P = 0.42; 90 s⁻¹: Exp [β]=0.99, P = 0.87; 45 s⁻¹: Exp [β]=1.01, P = 0.99; 2.25 s⁻¹: Exp [β]=1.02, P = 0.98).

Figure 4.

Changes in WBV measured at different shear rates. WBV at T₀ (dashed line) and T_peak (solid line) in HD-_PRE (A), HD-_CONT (B), and HD-_PER (C). Estimated difference of change between time (T₀ and T_peak): *P < 0.05; **P < 0.01; ***P < 0.001. WBV, whole blood viscosity.

Discussion

Standard hemodialysis is associated with disturbances in myocardial segmental perfusion,^4,6 leading to acute RWMAs and transient LV dysfunction,^5,7 consistent with our findings during the HD-_CONT condition. These recurrent cardiac disturbances contribute to increased long-term cardiac risks and mortality,²⁵ underscoring the need for effective countermeasures. IDE has emerged as an effective nonpharmacologic strategy to mitigate hemodialysis-induced myocardial stunning. Our results during the HD-_PER condition align with two previous exploratory studies conducted on a small number of patients with ESKD, which also documented a reduction in RWMAs prevalence when IDE was applied acutely compared with standard hemodialysis.^10,11 Although IDE confers cardioprotective effects, its implementation in routine hemodialysis care is hindered by several challenges (see Clinical Implications below). Predialysis exercise may represent a viable and promising alternative. The salient, yet never demonstrated, finding from this study is that exercise performed before hemodialysis provides cardioprotection that is comparable with that of IDE. Indeed, both predialysis and IDE significantly reduced RWMAs in comparison with standard hemodialysis, with no significant difference between the two modalities, thus demonstrating similar cardioprotective benefits (Table 4).

The precise mechanisms responsible for the reduction in hemodialysis-induced RWMAs by exercise are as yet unclear. The rapid elimination of fluids during hemodialysis through ultrafiltration has been associated with hemodynamic instability and intradialytic hypotension,^26,27 which in turn could have an adverse effect on myocardial perfusion.⁷ IDE has been shown to transiently enhance central hemodynamic parameters (i.e., heart rate, cardiac output) as well as BP and may reduce the incidence of intradialytic hypotension,^23,28,29 agreeing with our own findings during the HD-_PER condition. However, whether these hemodynamic improvements, albeit transient, directly mediate IDE-induced cardioprotection remains unclear. In our comparison of HD-_CONT and HD-_PER, we founded no significant associations between changes in RWMAs and hemodynamics variables. Furthermore, the hemodynamic profiles were nearly identical between the two conditions during the 2 hours preceding RWMAs assessment (Figure 3). Although these observations do not support a hemodynamic mechanism underlying IDE-induced cardioprotection, they do not definitively exclude it either, given the limitations of the study design. By implementing exercise before hemodialysis, we were able to maintain a hemodynamic profile strictly comparable with that of standard hemodialysis throughout the dialysis session (Figure 3). Importantly, we observed a cardioprotective effect with predialysis exercise and similarly found no association between RWMAs and hemodynamic changes when comparing HD-_CONT and HD-_PRE. Some researchers have proposed that the exercise-induced reduction in RWMAs may be attributed to enhanced coronary perfusion through coronary dilation induced by increased parietal shear stress.²⁸ However, it is unlikely that coronary dilation persists for several hours after a short-duration, moderate-intensity exercise session, as was the case in this study—particularly during the HD-_PRE condition. Our findings therefore strongly suggest that the cardioprotective effects observed in both HD-_PRE and HD-_PER are predominantly mediated by mechanisms other than transient alterations in central and/or systemic hemodynamic. Buchanan et al. reported during standard hemodialysis a decline in LV myocardial function without any changes in epicardial coronary artery flows by magnetic resonance imaging,⁶ reinforcing the hypothesis that altered microcirculatory blood flow is the dominant factor of hemodialysis-induced RWMAs.

WBV, a critical determinant of microcirculatory blood flow,³⁰ has been shown to increase significantly during hemodialysis across all shear rates,^31,32 a phenomenon we also observed in the HD-_CONT condition (Figure 4). In pathologic contexts, any increase in WBV can impair coronary perfusion and exacerbate ischemic injury.^33,34 However, in contrast to HD-_CONT, WBV remained unchanged from T₀ to T_peak during HD-_PER. Notably, significant relationships were found between changes in RWMAs and delta (i.e., T₀−T_peak) WBV at high shear rates when comparing HD-_CONT and HD-_PER. The mechanisms underlying the exercise-induced hemorheologic effect remain unclear but might be linked to exercise-induced hemodilution. An attenuation of the ultrafiltration-induced blood volume reduction at the end of hemodialysis has indeed been reported by previous studies in the situation with IDE compared with standard hemodialysis.^35,36 It can therefore be postulated that in the HD-_PER condition, exercise would have mitigated the reduction in blood volume resulting from ultrafiltration, thereby preventing the increase in WBV. This would subsequently assist in the prevention of myocardial stunning by improving the perfusion of the coronary microcirculation in our patients free from epicardial coronary artery disease. However, the role of WBV in exercise-induced cardioprotection is speculative, as our study was not specifically design to test this hypothesis. Notably, although WBV did not increased from T₀ to T_peak in HD-_PRE, no relationships were seen between changes in RWMAs and delta WBV across all shear rates when comparing HD-_CONT and HD-_PRE. It is noteworthy that the cardioprotective effect of exercise was of comparable magnitude in both HD-_PER and HD-_PRE conditions.

Taken together, our result supports the hypothesis that other mechanisms are likely to contribute significantly to the observed cardioprotective effects. Among these, exercise-induced ischemic/nonischemic preconditioning is a plausible candidate. Preclinical and clinical evidence suggests that brief episodes of myocardial ischemia induced by coronary artery occlusion before prolonged occlusion (i.e., ischemic preconditioning) or remote ischemic preconditioning by transient limb occlusion using cuff pressure confer some cardioprotection, reducing the size of the subsequent infarction and/or the risk of harmful ventricular arrhythmia or improving postischemic recovery (i.e., improved myocardial contractile performance).^37–39 Exercise is a form of preconditioning that can mediate protection similar to ischemic or remote ischemic preconditioning against myocardial ischemic events,⁴⁰ as seen at peak-stress hemodialysis. The mechanisms driving this exercise-induced cardioprotection are not yet fully elucidated but may involve, at least in part, humoral circulating factors modulating coronary artery vascular function and myocardial signaling.⁴¹ Of note, in a pilot study, Salerno et al. showed that remote ischemic preconditioning was able to reduce hemodialysis-induced RWMAs and preserve GLS at peak-stress hemodialysis, in line with our own data in both HD-_PRE and HD-_PER conditions.⁴² Furthermore, an increase in red blood cell deformability has been observed following exercise and remote ischemic preconditioning interventions.^43,44 Red blood deformability has a strong effect on WBV measured at high shear rate.⁴⁵ Moreover, the ability of red blood cells to pass through capillaries effects on tissue oxygenation.⁴⁵ Interestingly, Zhao et al. in patients undergoing coronary angiography recently demonstrated that application of remote ischemic preconditioning is able to increase coronary blood flow and improve coronary microcirculation function.⁴⁶ It is therefore tempting to speculate that acute exercise, before or during hemodialysis, can precondition the heart, thereby limiting the functional disorders that might have been observed during more severe ischemia insults, such at the peak stress of hemodialysis. Nevertheless, the experimental design of this study does not permit the hypothesis to be verified, and further studies will be necessary.

Our main finding that predialysis exercise is cardioprotective and provides benefits comparable in magnitude with IDE has important clinical implications. Although IDE has demonstrated efficacy in mitigating hemodialysis-induced myocardial stunning, its implementation into routine clinical practice may be limited by several practical and patient-related barriers. Many patients undergoing hemodialysis report substantial fatigue due to both their underlying condition and the dialysis process itself, limiting their ability or willingness to engage in physical activity during treatment.⁴⁷ The physical constraints imposed by vascular access (i.e., catheters or arteriovenous fistulas) can restrict movement, making exercise potentially uncomfortable or impractical. Moreover, IDE raises safety concerns as hemodialysis can induce BP fluctuations and even severe hypotensive episodes, making IDE potentially risky unless closely supervised. These barriers and risks may discourage both patients and health care providers from pursuing IDE routinely. Logistical constraints within dialysis units—such as lack of specific equipment—further hinder the widespread adoption of IDE. In addition, psychologic barriers must be considered, as some patients may experience fear or anxiety about exercising while connected to the dialysis machine. By contrast, exercise performed before dialysis can circumvent many of these limitations. It is however not without its own limitations. Patients may encounter logistical difficulties, such as the need to arrive earlier than usual, which can be a barrier to regular participation. Furthermore, some individuals may feel unwell, be fasting, or lack motivation before treatment, making exercise less feasible or appealing.¹³ Implementing and supervising predialysis exercise also requires additional resources and coordination within clinical settings. Despite these obstacles, its demonstrated cardioprotective effect suggests that predialysis exercise could still represent a viable and more accessible alternative for many patients. Finally, reducing RWMAs and limiting the decline in GLS during hemodialysis through exercise interventions is clinically meaningful, as both parameters are linked to adverse outcomes and an increased risk of mortality.^48,49

The first limitation of this exploratory study is its small sample size. Nevertheless, the exercise effects we observed, which were consistent with findings from previous exploratory studies,^10,11 and further supported by our own larger-scale investigation¹² lend a certain robustness to our findings. Using a cross-over design with N=25, an α-level of 0.05, and a within-subject correlation of 0.5, a post hoc power analysis revealed 67% power to detect a medium effect size (d=0.51) for HD-_PRE versus HD-_CONT and 76% power for HD-_PER versus HD-_CONT (d=0.55), assuming two-tailed paired t tests. Second, the intervention took place after the long interdialytic interval, when patients tend to be more fluid overloaded and have a higher risk of hypotension and ischemia.^50,51 Further work is needed to clarify the effect of exercise before and during hemodialysis in a midweek session. Third, patients with heart failure or coronary artery disease, who are at risk during exercise, were excluded for safety reasons. Fourth, our recruitment rate for women was only 32%. These factors limit the generalizability of the results to the broader ESKD population. Unfortunately, the limitations of our ultrasound technique prevented assessment of global and regional coronary blood flow. Future research, using functional cardiac magnetic resonance imaging or positron emission tomography, is warranted to investigate the effect of exercise on these variables and their link to improvements in regional LV myocardial function.

In stable ESKD patients, exercise performed before hemodialysis provides cardioprotective benefits that are comparable with those observed with IDE. These benefits seem to be attributable to mechanisms that are independent of hemodynamic factors. This observation suggests the involvement of alternative pathways, such as exercise-induced preconditioning.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at XXX.

Author Contributions

Formal analysis: Antoine Grandperrin, Christophe Hédon, Matthieu Josse, Claire Maufrais, Stéphane Nottin, Philippe Obert.

Investigation: Antoine Grandperrin, Christophe Hédon, Matthieu Josse, Claire Maufrais, Stéphane Nottin, Philippe Obert.

Resources: Philippe Connes, Jean-Paul Cristol, Claire Maufrais, Philippe Obert, Laure Patrier, François Roubille, Cécile Turc-Baron.

Supervision: Jean-Paul Cristol, Claire Maufrais, Philippe Obert.

Validation: Jean-Paul Cristol, Claire Maufrais, Philippe Obert.

Visualization: Matthieu Josse, Claire Maufrais, Philippe Obert.

Writing – original draft: Matthieu Josse, Claire Maufrais, Philippe Obert.

Writing – review & editing: Matthieu Josse, Claire Maufrais, Philippe Obert.

Funding

C. Maufrais: Société Française de Cardiologie (SFC21EXCRO).

Data Availability Statements

Anonymized data created for the study are or will be available in a persistent repository upon publication. Clinical Trial Data. Figshare. Data are openly available in repository Figshare at https://doi.org/10.6084/m9.figshare.25700007.v1.