Effect of Eccentric Cycling on Oxygen Uptake and Hemodynamics in Patients with Chronic Obstructive Pulmonary Disease: A Randomized Controlled Crossover Trial Open Access

During eccentric cycling exercise (ECC), a muscle is lengthened by an applied force greater than the force produced by the muscle itself [1]. Although the mechanisms are not yet fully understood, ECC has been shown to be a promising alternative to the widely used traditional concentric cycling exercise (CON) in patients with cardiopulmonary limitations, due to lower metabolic and cardiorespiratory demand despite producing an increased force [2, 3]. ECC could allow patients with severe exercise limitation to achieve higher, more beneficial exercise intensities with less dyspnea perception and lower load on the right ventricle [4]. Downhill walking also involves a high eccentric component of the quadriceps muscle and has been shown to induce muscle fatigue at lower ventilatory demand than level walking [5, 6]. However, to eliminate discrepancies between various modes of eccentric exercise, this article only focuses on eccentric cycling.

Chronic obstructive pulmonary disease (COPD) is the most common lung disease worldwide with an estimated prevalence of around 10% among people aged 30–79 years [7]. The Global Burden of Disease Study reported 3.3 million deaths in 2019 attributed to COPD, making it the third leading cause of death globally [8, 9]. Risk factors for the development of COPD include smoking, air pollution, biomass smoke, occupational exposures, infections, or genetics [10]. The clinical presentation of COPD is characterized by chronic respiratory symptoms, e.g., dyspnea, productive cough, and exercise intolerance caused by chronic airflow obstruction and airway inflammation [11]. Dyspnea is the most prevalent symptom in COPD and causes major disability in advanced stages of the disease [12]. It is widely supported that breathlessness in COPD patients is caused by a mismatch between increased inspiratory neural drive, driven by pulmonary gas exchange abnormalities and acid-base imbalance, and insufficient respiratory function, which is exacerbated in exercise due to dynamic hyperinflation [13].

According to current guidelines, spirometry indicating non-fully reversible airflow obstruction (FEV₁/FVC ≤0.7 post-bronchodilation) confirms COPD in the relevant clinical context [14]. Progressive disease is further classified according to the severity of airflow obstruction and the frequency of previous exacerbations [14‒16]. The etiology of exercise intolerance in COPD is multifactorial. Airflow obstruction is caused by chronic inflammation of the airways leading to narrowed or obstructed airways and destruction of lung parenchyma, which in return promotes gas trapping, hyperinflation and gas exchange abnormalities [14, 17]. In advanced disease, pulmonary hypertension and right heart failure are frequent, worsen with exercise and are associated with poorer prognosis [18, 19]. While COPD is primarily a pulmonary disease, muscular dysfunction is an independent factor in the poor health status of patients [20]. The etiology of this dysfunction is multifactorial and not fully understood. Contributing factors include hypoxia, hypercapnia, systemic inflammation, oxidative stress, drugs, nutritional depletion, and hormonal imbalances [20]. COPD patients spend more time in sedentary activities, which further promotes muscle dysfunction and may explain the effectiveness of pulmonary rehabilitation programs [21‒25]. Thus, it is particularly important for patients with COPD to exercise as guided rehabilitation programs after an exacerbation may improve quality of life and reduce hospital admissions and mortality [26, 27].

Therefore, the aim of the study was to investigate the difference in oxygen uptake (⁠$V˙O2$⁠) as a global measure of exercise capacity and additional cardiopulmonary exercise parameters as well as hemodynamic variables, including echocardiographic assessment of right heart function, at identical submaximal workloads in ECC vs. CON with COPD as a basis for future training and rehabilitation studies.

This was a randomized controlled crossover trial conducted between November 2023 and May 2024 at the University Hospital Zurich, with participants recruited from COPD outpatient clinics. Data are reported according to the CONSORT 2010 statement to randomized crossover trials (see CONSORT Checklist in online suppl. material; for all online suppl. material, see https://doi.org/10.1159/000545787) [28]. Randomization was achieved using a software-based block randomization method with variable block lengths, generated randomly. Allocation to treatment sequences was concealed and executed by independent researchers. Following data collection, all participant data were coded to ensure the confidentiality of the source data. We included patients of all sexes between 18 and 80 years, diagnosed with COPD in accordance with current clinical guidelines [14], who were on a stable medication regimen for >4 weeks. Patients with severe daytime hypoxemia under ambient air (PaO₂ <7.3 kPa) were excluded. Other exclusion criteria were the inability to follow the procedures, pregnancy, enrolment in another clinical trial with active treatment and other clinically significant concomitant diseases.

All participants provided written informed consent and the study was conducted in compliance with the Declaration of Helsinki. The study was approved by the Local Ethics Committee (KEK 2021-0132) and was registered on clinicaltrials.gov (NCT05185856).

Interventions were conducted on separate days to eliminate potential carryover effects. Participants performed two submaximal, standardized and identical stepwise incremental cycling tests, each of the three steps involving 3- to 5-min cycling intervals (totaling 9–15 min) at a cadence of 55–65 revolutions per minute. The interventions included two exercise tests, one using a traditional concentric ergometer (Ergoselect 200; Ergoline GmbH) and another using a recumbent eccentric ergometer with upright upper body position (Cyclus 2 Recumbent; RBM elektronik-automation GmbH), in randomized order. The participants underwent a familiarization session on the eccentric ergometer to practice proper pedaling technique and reduce the risk of muscle soreness. The initial intensity was set between 20 and 30 W, according to the patient’s fitness levels and was increased by 10 W per step, aiming to reach up to 60% of maximal exercise capacity at end-exercise. Due to the study design, comparing ECC to CON at identical work-rates was more important than maximizing the absolute intensity and risking early exhaustion.

During the exercise tests, participants breathed through a mouthpiece connected to the flow sensor of a metabolic unit (Ergostick; Geratherm Medical), which was calibrated before each test. The unit measured $V˙O2$ and additional ventilatory parameters such as tidal volume (Vt), breathing frequency (Bf), minute ventilation (⁠$V˙E$⁠), CO₂ output (⁠$V˙CO2$⁠), and respiratory exchange ratio, and derived variables on a breath-by-breath basis. The physiologic dead space fraction was calculated according to the Enghoff modification of the Bohr equation (PaCO₂ – P_ECO₂)/PaCO₂, using arterial and mixed-expiratory partial pressure of carbon dioxide [29]. Arterial oxygen saturation was continuously monitored using finger clip pulse oximetry. The $V˙O2$ efficiency slope was determined by the regression slope (a) in the equation $V˙O2=alogV˙E+b$ [30].

Echocardiography was performed at rest and during exercise, within the last minute of each incremental step. All echocardiographic recordings were obtained using a real-time sector scanner (CX 50; Philips) equipped with integrated color, continuous wave, and pulsed wave Doppler systems, following guidelines set by the American Society of Echocardiography. Systolic pulmonary arterial pressure (sPAP) was estimated by calculating the maximal pressure gradient of tricuspid regurgitation using the continuous wave Doppler and the modified Bernoulli equation, ΔPressure = 4 × TRVmax², where TRV represents the tricuspid regurgitation velocity, excluding right atrial pressure. The sPAP/Cardiac output_slope was calculated as (sPAP_{peak exercise} − sPAP_rest)/(Cardiac output_{peak exercise} − Cardiac output_rest).

Heart rate was continuously monitored using a 12-lead ECG, and blood pressure was measured with an automated arm cuff. Arterial blood samples were drawn from the radial artery at rest and during the final 30 s of exercise for immediate blood gas analysis (ABL90 FLEX; Radiometer). Post-exercise, patient-reported outcomes, including the Borg CR10 scale for perceived leg fatigue and dyspnea, were recorded.

The sample size was determined based on the primary outcome measure, $V˙O2$⁠. An aimed sample size of 16 patients was calculated to detect an assumed minimal clinically important difference of 30 ± 30 mL/min [31], with a significance level of 0.05 and 80% power. Physiological values were averaged and compared over the last 30 s of each step to evaluate the main outcomes between ECC and CON. Ergospirometry data were displayed as mean ± SD. A linear mixed model was used that included intervention type (CON vs. ECC), period (according to the crossover design), and intervention-period interaction as fixed effects, and patients as a random intercept. This approach controlled for carryover effects and period effects inherent in crossover trials. The necessity of retaining the intervention-period interaction term in the model was assessed. Model assumptions, such as normal distribution, were assessed by visual inspection of normality and homogeneity of residuals and random effects. The model also accounted for missing values, and analyses were conducted on an intention-to-treat basis; however, missing values were not imputed. Therefore, in case of incomplete date sets, only available data were analyzed. Pre-defined baseline characteristics, age, sex and medication intake were included in the model to control for confounding. Model comparisons were used to develop the final model.

When assumptions for parametric methods were not met, a non-parametric analysis was performed through Wilcoxon signed rank test for paired data, reporting median and quartiles. A 95% confidence interval excluding the null effect and a p value <0.05 was considered statistically significant. All statistical analyses were assessed using RStudio software version 4.1.0 (Posit PBC).

For the parametric analysis, model assumptions of homogeneity and normality of the residuals and random effects were met, and neither carryover nor period effects were identified. The patient flow is shown in Figure 1. We included 17 patients with COPD (6 female, mean age 67 ± 7 years, FEV₁ 1.33 ± 0.34 L, DLCO 50 ± 14% predicted), of whom 15 completed both periods according to the protocol. One patient could not complete the first period due to dyspnea and was unable to participate in period 2 due to an exacerbation. Another patient was not able to complete the eccentric period due to knee-pain (see Fig. 1). In Table 1, the baseline characteristics are listed.

At end-exercise with identical workload (41 ± 3 W), $V˙O2$ in CON vs. ECC was 729 ± 102 mL/min vs. 596 ± 172 mL/min resulting in a significantly lower mean delta $V˙O2$ of 122 mL/min (95% CI: 47–213 mL/min, p = 0.005) in ECC compared to CON (Fig. 2; Table 2). Furthermore, we identified a lower delta $V˙E$ of 5.7 L/min (95% CI: 1.6–10.0 L/min, p = 0.012) in ECC compared to CON at end-exercise. This was mainly caused by a lower Vt, whereas Bf showed no significant difference.

In alignment with $V˙O2$⁠, mean VO₂/kg was significantly lower in ECC (7.9 ± 2.1 mL/kg/min) compared to CON (9.8 ± 1.3 mL/kg/min) resulting in a mean difference of −1.9 mL/kg/min (95% CI: −3.2 to −0.7 mL/kg/min, p = 0.004), as shown in Table 3. However, perceived dyspnea and leg fatigue were similar for both interventions.

There was no difference in dead space fraction, $V˙E/V˙O2$ or oxygen uptake efficiency slope. Similarly, there was no difference in arterial blood gas parameters (Table 2).

In the non-parametric analysis of echocardiography data (Table 4), median sPAP tended to be lower in ECC vs. CON (p = 0.063). There was also a trend toward lower heart rate and sPAP/CO-slope in ECC vs. CON. Stroke volume, cardiac output, TAPSE, and TAPSE/sPAP were similar. Diastolic blood pressure at maximal exercise was 16 mm Hg higher in ECC than in CON (95% CI: 7–25 mm Hg, p = 0.003), while systolic blood pressure did not differ significantly.

In this randomized controlled crossover trial in 17 patients with moderate to severe COPD (mean FEV₁ 48%), we found that $V˙O2$ and $V˙E$ were significantly lower during ECC compared to traditional CON cycling exercise. Analysis of pulmonary hemodynamics by stress echocardiography revealed a trend toward a lower burden of the pulmonary circulation during ECC cycling.

Delta $V˙O2$ was 25% lower in ECC compared to CON and this effect was significant across all three measured identical submaximal work intensities (see online suppl. Table 1 online for step 1 and step 2 data). This is in line with existing data [32‒35]. Based on the findings of Lechauve et al. [36], describing more hyperpneic breathing patterns in healthy subjects during ECC vs. CON, Nahmias et al. [34] investigated 14 patients with severe COPD performing ECC and CON in a randomized crossover trial. They found poorer ventilatory efficiency at highest identical $V˙E$ and $V˙O2$ in ECC compared to CON, which differs from our study design aimed at submaximal work rates in guided rehabilitation programs with increasing interval durations. At identical submaximal training work rates, we found a significantly lower $V˙E$ related to the lower Vt in ECC compared to CON, whereas Bf and ventilatory efficiency (⁠$V˙E/V˙O2$⁠) showed no difference between ECC and CON. Thus, we believe that the benefits of reduced metabolic demand in ECC reduced the risk of early exhaustion in these COPD patients as it counter-acted an exercise-induced hyperpneic breathing pattern. It may therefore be that patients with advanced COPD, who are prone to dynamic hyperinflation during exercise, may particularly benefit from ECC, which is supported by a study by Rocha Vieira et al. [32] that found ECC to be feasible in severe COPD.

In contrast to previously published studies, subjective dyspnea and leg fatigue as measured by Borg Scale did not differ between ECC and CON. During ECC, some participants exhibited difficulty adapting to the unfamiliar braking of the pedals against the motor-force. Possible physiological factors contributing to this effect include the requirement of an unknown coordination pattern, which in a short-term exposure may lead to discomfort and subsequently bias perceived exertion. Thus, we hypothesize that a longer and standardized familiarization at a lower intensity might have improved these results. Furthermore, endurance in this deconditioned population is limited by muscular dysfunction. Therefore, patients might have exercised at intensities where they were subjectively not able to distinguish between exercise modalities anymore, even though oxygen consumption was lower.

To our knowledge, this was the first randomized controlled crossover trial to examine right-heart strain between ECC and CON in COPD patients using baseline and stress echocardiography in combination with ergospirometry at identical workloads. Many patients had severe pulmonary emphysema, making echocardiographic measurements during short exercise steps difficult. Due to missing echocardiographic data, the analysis may have been underpowered to address differences in echocardiographic parameters. Nevertheless, sPAP data may indicate a tendency toward a reduced right ventricular load during ECC compared to CON, as the median sPAP in ECC was 11 mm Hg lower than in CON (p = 0.063). This favorable exercise-hemodynamic effect has also been shown in patients with pulmonary vascular disease [4]. The observed isolated increase in diastolic blood pressure during ECC has been reported previously by our group [4]. We found no similar results or physiological mechanism explaining this finding in the literature. It may be attributable to the different mode of exercise in a recumbent position compared to traditional upright concentric exercise.

Data on eccentric exercise in patients with COPD show certain heterogeneity within the literature, not only for cycling [37]. Side effects from ECC (e.g., muscle soreness) are reported more frequently in COPD patients than in healthy subjects, suggesting that a slow increase of training intensity is appropriate in this population [38]. However, with correct instructions and a slow increase in intensity, even patients with severe COPD showed no side effects after 5 weeks [32]. In small studies, patients with COPD were able to exercise at workloads 3–5 times higher in ECC than CON, at subjectively lower dyspnea and fatigue [32, 39]. Still, the physiological effects of ECC vs. CON in patients with COPD are incompletely understood.

Guided long-term ECC training programs have been shown to be superior to CON in increasing $V˙O2max$⁠, 6-min walking distance and knee extensor strength as reported by a systematic review and meta-analysis of studies with healthy patients and cardiopulmonary diseases [2]. These results confirm that ECC is a promising tool for pulmonary rehabilitation, especially for patients with advanced lung disease, as these patients can exercise their muscles at a higher intensity and are less likely to give up prematurely due to cardiopulmonary limitations. As previously described by our group, choosing an appropriate intensity is a challenge because patient-reported or objectively assessed fitness levels do not necessarily correlate with how well they cope with ECC [4]. Based on our experience, we therefore suggest a low intensity of around 20% of peak work rate for a prolonged familiarization, allowing patients to get used to the movement patterns before starting guided rehabilitation programs. However, to investigate whether this assumption is valid, further clinical studies are needed.

Choosing optimal work rates for COPD patients was challenging, as recent ergospirometry data were often lacking. Therefore, in these severely limited patients, exercise intensity might have been chosen too high or too low, which could have influenced our results. As patients already had two study visits including exercise, additional baseline ergospirometry or prolonged familiarization were not feasible in the present setting in these severely ill COPD patients. However, the impact on the primary and secondary outcomes is negligible due to the crossover trial design. Although the difference in body position between ECC and CON could have affected the results, we chose a recumbent position for ECC, prioritizing patient safety as this exercise was new to them [40]. The need for a special eccentric recumbent ergometer also limits the applicability of the training modality in clinical practice.

This randomized controlled crossover trial in patients with predominantly severe COPD showed that ECC vs. CON submaximal identical load cycling exercise was associated with lower $V˙O2$ and $V˙E$⁠. A reduced right heart strain could not be significantly demonstrated, possibly due to low sample size and large standard deviation, warranting further investigation. Whether this group of severely limited COPD patients could benefit from ECC vs. CON in long-term training studies by increasing their muscle force and $V˙O2$ and which training intensities are best to achieve these goals remains to be studied.

Written informed consent was obtained by all study participants. The study protocol was reviewed and approved by the Cantonal Ethics Committee Zurich (KEK-ZH), Approval No. KEK 2021-0132, and was registered on clinicaltrials.gov (NCT05185856).

Silvia Ulrich and Esther Irene Schwarz were both a member of the journal’s Editorial Board at the time of submission.

The Swiss National Foundation (SNF 3200B_197706/1) financially supported this study.

All authors meet criteria for authorship as recommended by the International Committee of Medical Journal Editors and critical revised the manuscript for important intellectual content. A.K. and J.M. contributed to drafting of the manuscript and statistical analysis, had full access to all of the data in the study, and take responsibility for the integrity of the data and the accuracy of the data analysis. J.M. and S.U. contributed to study concept and design. A.K., S.R.S., S.U., and J.M. contributed to acquisition, analysis, or interpretation of data. A.K., S.R.S., M.L., S.U., and J.M. provided administrative, technical, or material support. S.U. supervised the study.

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further enquiries can be directed to the corresponding author.