Reliability of Peak Exercise Testing in Patients with Heart Failure with Preserved Ejection Fraction

INTRODUCTION

Non-invasive assessment of exercise intolerance using cardiopulmonary exercise testing (CPET) provides crucial diagnostic data for the evaluation and treatment of patients with heart failure (HF), and is commonly used as a clinical endpoint in HF therapeutic trials.^1–3 CPET guidelines have been developed and regularly updated by expert working groups in order to standardize both testing procedures and interpretation of exercise data.^4,5 These guidelines commonly recommend that at least two CPETs be conducted: the first for habituation/learning purposes; the second for acquisition of baseline data.^5,6 The stated purpose of this multiple test approach is to reduce test variability. However, this adds significantly to participant burden, and study time and cost. Furthermore, although the reliability of CPET has been examined in symptomatic HF and reduced EF (HFrEF),⁴ there are relatively few data verifying the reliability of peak exercise tests in patients ≥55 years, women, and patients with severely reduced exercise function who might benefit most by a habituation session. The latter patients are typical for the entity of HF with preserved EF (HFpEF), in which very few reliability data have been reported.⁷ Importantly, the impact of a placebo run-in period on CPET variables in this population, relevant to pharmacological interventions, has not previously been investigated. The purpose of this study was to determine the reliability of a treadmill exercise test with a placebo run-in period for assessment of peak and submaximal indices of exercise tolerance in women and men ≥55 years with HFpEF utilizing participant screening, well-trained testing personnel, and adherence to stringent CPET criteria.

METHODS

The study protocol was approved by the Wake Forest University Medical Center Institutional Review Board, and written informed consent was obtained from all participants. This report is from the screening visit (test 1) and baseline visit (test 2) of a randomized, double-blind, placebo controlled pharmacological clinical trial whose primary outcome is exercise capacity. Participant eligibility was assessed at an initial screening visit, which included medical history, physical examination, blood labs, echocardiogram to document ejection fraction and evidence of diastolic dysfunction, and peak CPET (test 1). Approximately one week later, participants returned for a second screening visit. If the initial screening visit blood labs were within inclusion criteria, participants entered into a two week placebo run-in period. At the subsequent baseline visit, if compliance to the placebo run-in by pill count was at least 80%, participants completed all baseline study assessments, including a second peak CPET (test 2). Patients had NYHA class II–III of at least 1 month duration with preserved resting ejection fraction, echocardiographic evidence of diastolic dysfunction (early diastolic lateral mitral annular velocity ≥9 cm/sec.), and no evidence of significant anemia or coronary artery, valvular, infiltrative, pericardial, pulmonary, or renal disease. Additionally, only participants with evidence of severe exercise intolerance (exercise time of at least 2 minutes but not greater than 12 minutes for women or 14 minutes for men) were enrolled. Potential participants were initially screened for inclusion by retrospective review of clinic visits and hospital discharge records at the Wake Forest University Medical Center. Of the resultant 273 preliminarily eligible participants, 66 were scheduled for an initial screening visit, and 52 met the study criteria and were enrolled.

The multi-step screening, education, and quality control CPET protocol is outlined in Appendix 1. Briefly, exercise testing was performed on a treadmill using the modified Naughton protocol and conducted by the same trained, experienced individual with master's degree in exercise physiology. Prior to each test, participants were screened for any recent acute illness or injury or change in status and were given detailed, standardized, verbal instructions. During the test, participants were encouraged to walk for as long as possible and, once fatigued, to attempt the subsequent stage, resulting in a peak, symptom-limited exhaustive effort. Complete calibration of gas concentrations and flow were performed before each test. Metabolic gas exchange was measured continuously during exercise and averaged over 15-second intervals (Medgraphics Ultima, Medical Graphics Corp., St. Paul, Minnesota). Peak VO₂ was defined as the average of the 2 highest VO₂ values for a given 15-second interval within the last 90 seconds of exercise. Ventilatory anaerobic threshold (VAT) was calculated using standard methods.⁸ The VE/VCO₂ slope was determined by measuring the slope across the entire course of exercise.⁹ Oxygen uptake efficiency slope (OUES) was determined measuring the slope of VO₂ (ml/min) and log₁₀VE (L/min) across the entire course of exercise.¹⁰ All tests were evaluated by the exercise physiologist for validity based on a combination of symptoms (dyspnea on exertion and/or fatigue as primary endpoints) and physiological measures (respiratory exchange ratio, heart rate, minute ventilation, respiratory rate) and confirmed if necessary by a cardiologist. To avoid bias, the tester intentionally did not view the results of test 1 immediately before conducting test 2. Additionally, validation of the gas exchange unit was conducted prior to each testing day by performing a 2-stage, incremental exercise test in a known, constant test subject. In order to be validated, VO₂ and VCO₂ needed to be within 2 standard deviations of the usual, known value at each workload.

Means are expressed using either ± standard deviation or with 10th and 90th percentiles. Within-subject variability from test 1 to test 2 was quantified by the within-subject absolute change (i.e., either increase or decrease from test 1 to test 2). A paired t-test was used to test for a statistical difference between the test 1 and test 2 values. Bland-Altman plot was used to visually assess whether the magnitude of within-subject variability in peak VO₂ varied with the magnitude of the measurements.¹¹ Reliability of the CPET outcome measures between the screening visit and baseline visit was tested by Pearson correlations and intraclass correlations. One-sided 95% confidence intervals of Pearson correlations were made using Fisher's z transformation. For each variable, the coefficient of variation (CV) was defined as the (within-subject standard deviation/within-subject mean) × 100%. Significance was set at p<0.05.

RESULTS

Participants were predominantly white women with a history of hypertension and NYHA functional class II symptoms (Table 1). Test 1 and test 2 were separated by an average of 23 ± 13 days. Compliance by pill count to the placebo run-in period was excellent (96%). No adverse events occurred during or after any peak CPET. In three participants, the screening visit CPET was not considered to be a valid peak test and was repeated two weeks later. The reasons were: 1) abnormal gait and coordination resulting in early fatigue and test termination in two participants, and 2) test ended primarily due to knee pain in one participant. The two participants with abnormal gait and coordination were coached on proper mechanics immediately following the screening visit CPET with good results. The participant with initial knee pain returned following a bout of NSAIDS and completed the repeated CPET with no pain. In each case, the repeated screening visit CPET was considered to be valid and was used in this analysis.

Peak VO₂ and VAT were nearly identical on the 2 tests. Within-subject peak VO₂ demonstrated high agreement between tests as shown by Bland-Altman analysis (Figure 1). Reliability of CPET parameters was excellent (Table 2), and was similar across age and gender (Tables 3 and 4). Exercise time increased significantly from test 1 to test 2. Likewise, VE/VCO₂ slope was significantly greater in test 2 than test 1 (Table 5).

DISCUSSION

Few data are available regarding the reliability of CPET in women, and specifically in patients ≥55 years of age with HFpEF despite the fact that HFpEF occurs in approximately 50% or more of HF patients in the community, and the proportion is higher among women and individuals >55 years.^12–14 Since HFpEF patients are predominantly ≥55 years and female, they may be expected to be less familiar with exercise testing, to have greater variability, and to have a greater need for “learning” or practice sessions. Moreover, the impact of a placebo run-in period on CPET variables in this population, relevant to pharmacological interventions, has not previously been investigated. Accordingly, in this study we investigated the reliability of CPET-derived variables specifically in 52 patients ≥55 years with HFpEF who underwent 2 identical CPETs approximately 3 weeks apart with a placebo run-in period between tests. We utilized a multi-step screening, education, and quality control protocol to optimize reliability. The major new findings of this study were 1) No change in the mean value or significant within-subject variability in peak VO₂ or VAT between tests, suggesting no significant learning or placebo effect, 2) Good correlations in all CPET parameters between tests, with no difference across age and gender. Thus, in patients with HFpEF ≥55 years of age, there may be no need to perform more than one CPET to obtain valid, baseline peak VO₂ and other CPET results.

Numerous studies have shown that peak VO₂ is a key outcome for assessing the prognosis of patients with HF, and is commonly used as a clinical endpoint in HF therapeutic trials, including those with HFpEF.^1,2,15 Current cardiology guidelines recommend that at least two CPETs be conducted based on prior work demonstrating that peak VO₂ can vary significantly over serial tests.⁵ For example, Elborn et al.¹⁶ performed three consecutive treadmill tests separated by two weeks on 30 subjects with HFrEF. The mean peak VO₂ improved significantly from the first to the second test (14.1 vs. 14.9 mL/kg/min, p<0.005), while the average within-subject CV for the three tests was 6%. The authors concluded that a single baseline test was insufficient and suggested that the performance of at least two tests be conducted for clinical research applications. In contrast, Bensimhon et al.⁴ performed two baseline peak treadmill tests on 398 men and women with symptomatic HFrEF and found that there was no significant change in peak VO₂ (15.2 ± 5.0 vs. 15.2 ± 5.0 mL/kg/min; p=0.78). Although within-subject variability of peak VO₂ was significant, the authors concluded that in multi-center trials, there is no need to perform more than 1 baseline CPET. In the only study to examine HFpEF to date, Marburger et al.⁷ examined the reproducibility of a cycle ergometer CPET in 5 HFpEF patients and 4 HFrEF patients. The authors concluded that the reliability of peak testing was good in both HFpEF and HFrEF, and that only one test was necessary. However, the small number of patients with HFpEF significantly limits the widespread applicability of results. In the present study with 52 HFpEF patients, for peak VO₂, there was good correlation between tests (r = 0.85; ICC = 0.855), the overall group means on test 1 and test 2 were essentially identical, the percentage of subjects who decreased or increased on test 2 were similar, Bland-Altman plots demonstrated high agreement between the 2 tests, and the within-subject variability was low (average CV <5%). Taken together, these findings further suggest that a single baseline test is sufficient to measure peak VO₂ for the evaluation of therapy or assessment of prognosis.

A large body of research has also evolved regarding CPET variables other than peak VO₂. Responses such as VAT, VE/VCO₂ slope, OUES, and exercise time have been used with greater frequency to classify functional limitations and to stratify risk in patients with HF.⁵ VAT and OUES are independent of effort and protocol and, similar to peak VO₂, have been shown to be highly reproducible.^4,10 In this study, we demonstrate that VAT and OUES are also reproducible with repeat testing as OUES and VAT had good correlation between tests and nearly identical mean values on test 1 and 2. As seen previously in HFrEF,^4,16,17 we found that the mean exercise time increased from test 1 to test 2, with 54% of subjects increasing on test 2. Additionally, mean values for VE/VCO₂ slope increased from test 1 to test 2, and there were a greater number of subjects who increased on test 2 than decreased. The reasons for discordant findings for VE/VCO₂ slope are uncertain; however, may be related to daily variations in ventilatory oscillation. Guazzi et al.¹⁸ reported that the prevalence of abnormal ventilatory oscillation in patients with HFpEF approximates 30%, which could affect VE/CO₂.

To achieve these results and be able to conduct a single, valid baseline CPET in longitudinal studies, there are several important aspects to consider (see appendix). These include: standardized participant instructions, screening for intercurrent acute illness or injury or change in condition, frequent calibration and validation of the gas exchange unit, strong encouragement to reach a peak level, and evaluation of reasons for termination and physiological responses. CPET that do not meet these criteria will likely have reduced reliability. Although inter-observer variability was not assessed in the present study given that each test was conducted by the same experienced individual, high reproducibility is obtainable with appropriate training of laboratory personnel.^4,7 These results challenge conventional wisdom that serial baseline testing is required in clinical trials with exercise capacity outcomes, and suggest that any incremental gain be balanced with the increase in participant burden and cost.

Supplementary Material