This retrospective observational study included patients with HF referred to the Cardiac Rehabilitation Unit of Unidade Local de Saúde de Santo António (Porto, Portugal) between January 2019 and June 2022. We included all patients with HF according to the criteria of the European Society of Cardiology (ESC) irrespective of left ventricle ejection fraction (LVEF).2 We excluded patients with paced rhythm at the time of exercise testing, except for the ones under cardiac-resynchronization therapy and in sinus rhythm; for these patients, atrial sensing and biventricular pacing had to be present during the exercise phase. Electronic health records were reviewed for clinical characteristics (including comorbidities, clinical presentation, medications, and echocardiographic data) and CPET report data. Our hospital Human Research Committee approved this retrospective chart review and waived the requirement for informed consent (Ref: 2021.187(154-DEFI/160-CE)).

Exercise capacity was evaluated through a maximal CPET testing on a treadmill (Medisoft, Model 870C). The testing protocol was chosen according to the patient's physical activity level (modified Naughton, Bruce, modified Bruce, or Gardner-Skinner). The CPET was conducted under medical supervision, with electrocardiography continually monitored throughout the protocol. Using a stationary metabolic cart system (Geratherm® Respiratory Ergostik, under BLUE CHERRY®), respiratory gas exchange measurements were obtained breath-by-breath and the average values were recorded every 30 seconds. Heart rate and blood pressure were recorded at regular intervals throughout the test. During CPET testing, patients were strongly encouraged to achieve maximum exercise capacity, and a respiratory exchange ratio >1.10. Maximum effort was considered when peak RER was equal or above 1.05. Peak VO2 was determined as the highest VO2 achieved during exercise. The VT1 was determined using the nadir of the VE/VO2 versus time. The second VT (VT2) was not systematically reported and therefore not considered for the analysis. The VTs were determined by a single observer. VO2 and HR were determined at VT1 as well as at 40%, 70% and 85% of peak VO2 (according to the training zones identified in the ESC guidelines).6 HRR was calculated as the difference between the peak and resting HR (obtained from the resting electrocardiogram) and expressed as %HRR at those exercise periods. We used the EI domains depicted in the ESC guidelines for comparison of the different EI indices. VT1 was considered to correspond to the moderate intensity domain.1,16,19

Statistical analysis was performed using IBM SPSS Statistics, Version 27.0 (Armonk, NY: IBM Corp; 2020). In the descriptive analysis, normally distributed continuous variables were characterized using the mean and standard deviation (SD), whereas for non-normally distributed variables, median and interquartile range (IQR) were used. For group comparison, Chi-square test was used for categorical variables, and t-Student or Mann-Whitney U tests were used for continuous variables depending on distribution normality. For the whole study population, we compared the EI classification (ESC EI domains) according to the different physiological measures at VT1 and assessed the rate of correspondence to the moderate intensity domain. Consistency between % Peak HR and %HRR, and % Peak VO2 at 40%, 70% and 85% (i.e., the transition point to moderate, high and very high intensity according to the ESC guidelines6) was also assessed.

We conducted two subgroup analyses and stratified patients according to their LVEF in two groups: preserved (LVEF ≥50%) and reduced (LVEF <50%). The correspondence rates between the different EI measures at VT1 and the ESC exercise intensity domains were compared between these groups. A similar analysis was performed by dividing patients according to their physical fitness and overall exercise capacity in two groups defined by the median peak VO2. Four sensitivity analyses were performed excluding patients on atrial fibrillation at the time of CPET, patients who were not on beta-blockers, patients with peak RER <1.05 and <1.10; and including only the patients subject to the mostly used exercise protocol: modified Naughton. The significance level was set at 0.05 (two-tailed).

Of a total of 259 CPETs performed during the study period on HF patients, 88 were excluded as they had been repeated on the same patients; eight other cases were excluded because the patients were on paced rhythm at the time of CPET. Therefore, 163 patients were included in the analysis (Figure 1). 105 (64%) were male and the mean age was 61±1 years. Regarding the HF phenotype, most of the patients, 107 (66%), had reduced LVEF, 28 (17%) had mildly reduced LVEF (41%≤LVEF≤49%) and 28 (17%) had preserved/recovered LVEF. Ischemic heart disease was the cause of HF in 62 (38%) patients. Mean LVEF was 36±12% and 71% of patients were in New York Heart Association (NYHA) Functional Class II. The main cardiovascular risk factors were dyslipidemia (66%) and hypertension (64%), and almost half of the population was diabetic (45%). A history of atrial fibrillation was present in 39 (24%) individuals, but only 10 (6%) had this rhythm at the time of CPET. Most patients were on beta-blockers (93%), 26 (16%) were on amiodarone and 9 (5.5%) on digoxin. Forty-one (25%) patients had implanted cardiac devices (pacemakers, cardiac-resynchronization therapy devices and implantable cardioverter-defibrillators). A detailed demographic and clinical characterization of the study population is displayed in Table 1.

Figure 1.

Study population flowchart. CPET: cardiopulmonary exercise test.

(0.15MB).

In the 163 CPETs included in the analysis, the most common protocol was the modified Naughton in 116 (71%) patients, followed by modified Bruce in 25 (15%). Most of the patients were in sinus rhythm during the test (94%). Mean exercise duration was 09 min 25 s±03 min 32 s and mean RER was 1.06±0.12. The main reason to stop the test was fatigue, in 130 (79.8%) patients. Although perception of maximal exertion was documented for all patients, peak RER ≥1.05 was attained in 91 (56%) cases. Mean peak HR was 121±21 beats per minute (bpm) and mean peak VO2 was 18.3±5.4 mL/kg/min. Mean HRR was 47±18 bpm. At VT1, mean HR was 96±21 bpm and mean VO2 was 10.0±4.4 mL/kg/min (41.6±20.1% of predicted peak VO2). Detailed CPET data can be found in Tables 2 and 3.

VT1 was observed at 82±10% of peak HR, corresponding to the ESC high intensity training zone. However, the same threshold was found at 54±25% of HRR and 54±17% of peak VO2, corresponding to the moderate intensity training zone (Table 4).

Considering VT1 as the moderate intensity training zone, and using % Peak VO2, 65% of the patients were classified in the correct EI domain at VT1. % Peak HR was the parameter with the worst concordance (25%) because it tended to overestimate the EI by placing 75% of patients at a higher training zone at VT1. %HRR showed a better correspondence than % Peak HR but was worse than % Peak VO2, classifying 46% of patients appropriately (Table 4 and Figure 2).

Figure 2.

Exercise intensity distribution at VT1 according to the different studied physiological parameters (% Peak VO2, % Peak HR, %HRR) and the ESC guidelines classification. ESC: European Society of Cardiology; HR: heart rate; HRR: heart rate reserve; VO2: oxygen uptake; VT1: first ventilatory threshold.

(0.22MB).

Considering the % Peak VO2 cut-offs listed in the ESC guidelines (40% as the transition point to moderate intensity, 70% to high intensity, and 85% to very high intensity) as the baseline measure of EI, its consistency with the other recommended indices is presented in Table 5. At 40% of peak VO2, consistency is high when using % Peak HR (74% of patients classified in moderate intensity exercise domain) but drops to 3.2% when using %HRR. Similar results are found at 70% of peak VO2, where 81% of patients are correctly classified in the high intensity domain according to % Peak HR but only 5.7% are assigned to this category according to the %HRR. In contrast, at 85% of peak VO2, consistency drops for % Peak HR (only 47% of patients adequately classified in the very high intensity domain), and %HRR consistency remains low (13%).

We divided our study population according to LVEF: 135 patients with reduced LVEF and 28 patients with preserved LVEF. Patients with preserved LVEF were older (65.0±13.3 vs. 60.3±10.4 years; p=0.043), had lower prevalence of males (43% vs. 69%; p=0.009) and had higher prevalence of atrial fibrillation history (50% vs. 19%; p<0.001) – Supplementary Table 1. There were no significant differences regarding maximal exercise capacity between HF subgroups (% predicted peak VO2; 73.5±19.4 vs. 77.3±19.9; p=0.351). Correspondence between the recommended indices of EI at VT1 with the moderate intensity domain was numerically superior for patients with reduced LVEF. Classification of patients according to % Peak VO2 had a 67% correspondence to the moderate intensity domain in reduced LVEF patients compared to 54% in preserved LVEF patients. There were also differences when other indices were used, such as % Peak HR: 26% correspondence in reduced LVEF group vs. 20% in the preserved LVEF group and according to %HRR: 49% in reduced LVEF group vs. 30% in the preserved LVEF group (Table 6).

The other subgroup analysis performed considered overall physical fitness. HF patients were stratified into two groups according to median peak VO2 (17.4 mL/kg/min). In general, patients with higher exercise capacity (peak VO2 above the median) were younger (57.9±11.0 vs. 64.4±10.2 years; p<0.001), more likely to be male (73% vs. 56%; p=0.019), had higher prevalence of reduced LVEF (92% had LVEF <50% vs. 74% with LVEF ≥50%; p=0.019), had a lower baseline NT-pro-BNP (286±634 vs. 726±1721 pg/mL; p<0.001); history of atrial fibrillation (15% vs. 34%; p=0.004) and chronic kidney disease (6% vs. 24%; p=0.002) were less prevalent and hemoglobin concentration was higher (14.7±1.4 vs. 13.6±1.8 g/dL; p<0.001) – Supplementary Table 2. The correspondence of the recommended indices of EI at VT1 with the moderate intensity domain was superior for patients with higher exercise capacity. The classification of patients according to % Peak VO2 had 73% correspondence to the moderate intensity domain in patients with higher exercise capacity compared to 56% in the other group. The differences remained for the other indices, such as % Peak HR: 42% concordance in the higher exercise capacity group vs. 7% in the other group; and according to %HRR, 55% in the higher exercise capacity group vs. 37% in the other group (Table 7).

To assess possible bias due to certain population subgroups, four sensitivity analyses were performed. In the first test, we excluded patients with atrial fibrillation at the time of the CPET (10 patients). Then we performed the same analysis excluding patients not on beta-blockers (BB) (12 patients). Afterwards, we assessed the influence of underperforming patients on the results, excluding patients that attained a peak RER <1.05, and hence submaximal effort (72 patients), and excluding patients that did not attained a maximal effort, defined by peak RER <1.10 (60 patients). Finally, we performed sensitivity analysis only including patients subject to the most used exercise protocol: modified Naughton (116 patients). Correspondence of the recommended indices of EI at VT1 with the ESC guidelines training zones was not significantly different in these subgroups, despite patients with submaximal and maximal effort demonstrating slightly better concordance (Supplementary Tables 3–7).

Our study has several limitations that need to be acknowledged. First, the retrospective nature and the single-center design of the study limit the generalizability of our results to other patient groups. Besides, the usage of a treadmill exercise protocols might not be directly comparable to cycle-ergometer-based protocols described in other studies. However, contrasting with other European populations, our population is less familiar with cycling-based exercises and might preform worse on those protocols. Second, the low percentage of patients (17%) who had preserved LVEF in our cohort limits the external validity of our data regarding HFpEF; a larger sample will be needed to confirm the results of our sub-analysis. Third, there is a chance that the peak performance measures are underestimated in this cohort. The mean peak RER was 1.06, lower than in some of the previously cited studies.15,19 Also, only 56% of patients reached a peak RER ≥1.05 (and only 37% reached a peak RER ≥1.1), which suggests that a significant proportion of patients did not reach maximal effort. This fact can be partially explained by the clinical characteristics of the studied patients, which also included patients with respiratory, vascular and musculoskeletal comorbidities. Because of this proportion of patients not achieving a (near-) maximal effort during CPET, especially among those with HF and reduced ejection fraction, indices of peak exercise capacity are limited. Another factor hampering the utility of these indices is the influence of the ramp rate during the test at the work peak; a continuous incremental ramp protocol individualized for the functional capacity of each patient instead of a standard staged intermittent one as we used may have helped to achieve maximal effort in more patients. However, previous published data showed a sensitive prognosis assessment of peak VO2 in patients with HF in all RER subgroups.22 The overall results of our study were replicated after performing sensitivity analyses excluding the patients with a peak RER <1.05 and <1.10. Fourth, we did not analyze the VT2 as few patients had it reported. Finally, women are underrepresented in our cohort (36%), which limits the generalizability of our results to this population.