Abstract
BACKGROUND: Exercise capacity assessed by cardiopulmonary exercise testing is usually measured by peak oxygen consumption (V̇O2). However, not uncommonly, patients achieve a relatively higher work load (peak work) compared to their peak V̇O2. In these situations it is difficult to know which parameter to use in assessing exercise capacity. The purpose of this study was to determine whether there are distinguishing physiological characteristics of patients with discordance between percent-of-predicted peak work versus peak V̇O2, in order to understand how to use these measurements in interpreting exercise capacity.
METHODS: We conducted a retrospective study of 172 cardiopulmonary exercise tests performed at our institution between 2003 and 2010.
RESULTS: The subjects in the higher peak work group demonstrated higher ventilatory efficiency (lower slope of minute ventilation to carbon dioxide production) and lung function (FEV1 and FVC), a greater breathing reserve (higher breathing reserve, lower V̇E/maximal voluntary ventilation), and achieved a higher maximal heart rate. Subjects in the higher maximum V̇O2 group were heavier, had lower ventilatory efficiency, and had a reduced breathing reserve. Multivariate logistic regression analysis showed that the predominant independent factors associated with group assignment were body mass index, breathing reserve, and peak heart rate. The subjects with higher percent-of-predicted peak work than peak V̇O2 had a lower body mass index, a greater breathing reserve, and a higher peak heart rate.
CONCLUSIONS: The observation that there are distinguishing physiological features between those who have a higher peak work and those who have higher peak V̇O2 provides insight into the underlying processes determining maximal exercise capacity.
Introduction
Cardiopulmonary exercise testing is a comprehensive approach to assess exercise capacity and limitations to exercise. Exercise capacity is usually interpreted in terms of peak oxygen consumption (V̇O2).1–7 Indeed, peak V̇O2 is a robust predictor of outcomes in many cardiopulmonary diseases such as COPD, congestive heart failure, cystic fibrosis, and pulmonary hypertension.1–3,8,9 However, we have observed that there are some patients who achieve higher percent-of-predicted peak work than percent-of-predicted peak V̇O2, and vice versa. In these cases it is unclear whether to use peak work or peak V̇O2 as the parameter by which to judge the patient's exercise capacity. As we pointed out in an editorial,10 this fundamental question was not addressed by a 2003 statement on cardiopulmonary exercise testing published by the American Thoracic Society/American College of Chest Physicians.2 In the majority of exercising subjects this distinction is unimportant, because the relationship between work and V̇O2 is linear; thus, the percentage of the predicted value achieved for either parameter is similar.7,11,12 However, as described by Wasserman and colleagues,13 a difference in V̇O2 may be observed between expected and observed values when calculating the expected V̇O2 based on the maximum work rate achieved. The general goal of the current study was to evaluate the exercise characteristics that distinguish patients with discordant values of peak work and peak V̇O2, to better understand the underlying processes that determine maximal exercise capacity. We specifically expected that patients achieving higher work relative to V̇O2 would demonstrate better cardiopulmonary fitness and/or ability to tolerate the discomfort of a high level of exercise.
QUICK LOOK
Current knowledge
Cardiopulmonary exercise testing is a comprehensive approach to assess exercise capacity and limitation. Exercise capacity is commonly assessed as peak oxygen consumption.
What this paper contributes to our knowledge
In patients with a lower body mass index, higher breathing reserve, and greater peak heart rate, peak work may be greater than peak oxygen consumption. Including both peak oxygen consumption and peak work in the interpretation of the exercise test can detect underlying physiological characteristics that reflect important aspects of exercise.
Methods
Study Design
We retrospectively analyzed all cardiopulmonary exercise tests performed at the University of Vermont/Fletcher Allen Health Care, a tertiary care academic medical center, from 2003 to 2010. The study was approved by the University of Vermont Committee on Human Research in the Medical Sciences (study 10-232). A total of 172 tests were included.
All studies were performed according to standard guidelines.2 We conducted exercise testing by using a 1 min step protocol to exhaustion on a bicycle ergometer system (MGC Diagnostics, St Paul, Minnesota), using work increments of 15 watts/min or 30 watts/min, estimated to bring the subject to maximal work load in approximately 10 min, as recommended by the American Thoracic Society/American College of Chest Physicians2 and Wasserman and colleagues.13 We measured baseline spirometry and maximal voluntary ventilation before exercise (or, estimated the latter as FEV1 × 40). We continuously monitored cardiovascular parameters (heart rate, blood pressure, electrocardiogram), ventilatory parameters (breathing frequency, tidal volume, SpO2), and metabolic parameters (CO2 production [V̇CO2], V̇O2). Of note, we did not consistently record exercise flow-volume loops during the time frame of this study, so we do not report data on ventilatory limitation or changes in end-expiratory lung volume. We measured dyspnea with the modified Borg scale.14 The following factors were calculated:
Slope of V̇O2 versus work: (final V̇O2 − initial V̇O2)/final work − initial work
Respiratory exchange ratio: V̇CO2/V̇O2
Expired minute ventilation (V̇E): breathing frequency × tidal volume
Breathing reserve: maximum V̇E/maximal voluntary ventilation, or maximal voluntary ventilation − maximum V̇E
Ventilatory equivalent for oxygen: V̇E/V̇O2
Ventilatory equivalent for carbon dioxide: V̇E/V̇CO2
Slope of V̇E vs V̇CO2
Oxygen pulse: V̇O2/heart rate
Anaerobic threshold: By inspection of the plot of V̇CO2 versus work and of V̇E/V̇O2 versus work for their respective inflection points in the slope
We used predicted values for work and V̇O2 based on modified Jones criteria.12 We defined a discordance between predicted peak work and predicted peak V̇O2 as any absolute difference in the percent-of-predicted between these values of > 15 (eg, peak work = 110% of predicted, peak V̇O2 = 85% of predicted). This cutoff was derived empirically from exploring cutoffs of 10%, 15%, and 20%. A cutoff of 10% was too small and within the range of experimental accuracy for peak V̇O2 and peak work. A cutoff of 20% was overly sensitive and yielded a sample size of 11 in the smallest group, which we felt was too small to allow valid statistical analysis. We also assessed whether the results would appear to be substantially altered by using the normative equations of Wasserman and colleagues for work and V̇O2, after adjusting for body weight.13 We noted that the Wasserman criteria resulted in a more unbalanced distribution of subjects in the predicted work versus predicted V̇O2 categories, which did not seem consistent with our clinical impression of the prevalence of this discrepancy. Accordingly, we used the Jones predicted equations,12 with a discordance difference of > 15%.
Based on this difference, we divided the subjects into 3 groups: those with percent-of-predicted peak V̇O2 greater than percent-of-predicted peak work (peak V̇O2 > peak work), those with percent-of-predicted peak work greater than percent-of-predicted peak V̇O2 (peak work > peak V̇O2), and those with equivalent percent-of-predicted peak V̇O2 and percent-of-predicted peak work (peak V̇O2 = peak work).
Data Analysis
We analyzed all data for distribution, and expressed the data as mean ± SD. We assessed differences across the 3 groups by analysis of variance, with differences between any 2 groups analyzed with the Tukey honest significant difference test. In addition we performed a backwards, stepwise multivariate logistic regression analysis to determine which exercise parameters identified as significant between the peak work > peak V̇O2 and peak V̇O2 > peak work groups were independently associated with group assignment. Two-tailed P values < .05 were taken as indicating statistical significance. All data were analyzed using statistics software (JMP 9.0, SAS Institute, Cary, North Carolina).
Results
A total of 172 tests were included for analysis. Patients were referred for testing for a variety of reasons, the most common of which were unexplained dyspnea (n = 90, 52%) and COPD, the latter mostly for evaluation prior to pulmonary resection for lung cancer (n = 28, 16%) or pulmonary rehabilitation (n = 25, 14%). There were 2 cases of cardiac disease (cardiomyopathy and diastolic dysfunction). Other diagnoses included cough, chest pain, interstitial lung disease, sarcoidosis, scoliosis, tracheobronchomalacia, and vocal cord dysfunction. All subjects were instructed to exercise to exhaustion. The most common reasons for stopping exercise were leg fatigue (42%), shortness of breath (28%), and a combination of equal leg fatigue and shortness of breath (21%). None of the subjects had any particular experience with bicycle ergometry, which might have given them an unseen advantage in terms of motor efficiency.
The Figure displays the distribution of groups based on the 15% cutoff between percent-of-predicted peak V̇O2 versus peak work. The majority of subjects (n = 92) were in the peak work = peak V̇O2 group. As seen in Table 1, there were distinct differences in the groups' demographics. Subjects who achieved higher percent-of-predicted peak work than percent-of-predicted maximum V̇O2 were younger, had a lower body mass index (BMI), and had a more even distribution of males versus females than those who achieved a higher percent-of-predicted peak V̇O2 than percent-of-predicted peak work. There were no overall differences in diagnostic category distribution between the 3 groups (P = .57). Both the cardiac subjects fell within the peak V̇O2 = peak work group, as did the majority (62%) of the COPD subjects and half (49%) of the unexplained dyspnea subjects. The remaining COPD and unexplained dyspnea subjects were mainly in the peak V̇O2 > peak work group rather than the peak work > peak V̇O2 group (32 vs 6%, and 34 vs 17%, respectively).
Percent-of-predicted peak oxygen consumption (V̇O2) versus percent-of-predicted peak work. The solid line is the line of identity, and the 2 dashed lines represent the 15% differences above and below the line of identity, thus defining 2 groups: peak V̇O2 > peak work (above the upper dashed line), and peak work > peak V̇O2 (below the lower dashed line).
Subject Demographics and Physiological Characteristics
As shown in Table 1, there were also differences in exercise variables between the groups. All subjects had similar baseline V̇O2 at rest. Subjects who achieved a higher percent-of-predicted peak V̇O2 than percent-of-predicted peak work had a higher V̇O2/work slope, as at least partly expected, due to their having a lower work rate increment during exercise.15 Subjects with peak V̇O2 > peak work were more likely female, heavier, had lower ventilatory efficiency (higher V̇E/V̇CO2 slope), and a reduced breathing reserve. The subjects in the peak work > peak V̇O2 group demonstrated higher ventilatory efficiency, lung function (FEV1 and FVC), and breathing reserve (higher breathing reserve, lower V̇E/maximal voluntary ventilation). The subjects in this group also achieved a higher peak respiratory exchange ratio and heart rate. Multivariate logistic regression analysis showed that the independent factors associated with having a higher percent-of-predicted peak work than peak V̇O2 were lower BMI, higher breathing reserve, and greater peak heart rate (Table 2).
Factors Associated With Being in the Peak V̇O2 > Peak Work Group Rather Than the Peak Work > Peak V̇O2 Group, Via Multivariate Logistic Regression Analysis*
Discussion
In the current study we sought to determine which physiological characteristics distinguish patients who have discordance between peak work and peak V̇O2. We found that those who can perform relatively higher work relative to their maximal V̇O2 were less heavy, had a greater breathing reserve and a higher peak heart rate. Meanwhile, those who consumed more oxygen relative to their work achieved were heavier, with less breathing reserve and a lower peak heart rate. Our data support the concept that lower weight (ie, mean BMI 26 vs 30 kg/m2), better ventilatory function, and ability to exercise to a higher heart rate are associated with the ability to perform higher work loads for a given V̇O2.
Most clinicians and researchers alike classify exercise tolerance on the basis of maximal V̇O2 achieved (eg,7,16–18). Ideally, one would want to demonstrate maximal exercise capacity on the basis of achieving a plateau in maximum V̇O2, suggesting that the limit of aerobic energy production has been reached.19–21 However, the visualization of such a plateau is highly variable among patients and typically not seen in most clinical cardiopulmonary exercise tests22 (only 22% of subjects in the current study achieved such a visual plateau); for this reason, we used the term “peak” V̇O2, which is more accurate and tends to confer the same information.20 Classifying maximal exercise capacity on the basis of peak V̇O2 makes sense, as the majority of work performed during physical activity utilizes energy from aerobic metabolism.5,7 In fact, the terms “work capacity” and “maximal V̇O2” are often used interchangeably (eg, in Oren et al7). Indeed, maximum V̇O2 or peak V̇O2 is a robust measure of outcomes and influence of interventions in a wide variety of cardiopulmonary diseases.1,3,8 Yet the functional aerobic capacity reflected in the peak V̇O2 does not necessarily reflect the total work performed. This so-called “V̇O2 difference,” as coined by Wasserman,13 may be due to many factors, such as failure to reach steady state, obesity, cardiovascular disease, or improper ergometer calibration.13 While some authors use power output or work capacity as a reflection of exercise tolerance (eg, Jones and Killian5; Katz et al23), most studies continue to use peak V̇O2 as the primary measure of exercise capacity.
We view the cause of this discrepancy between peak V̇O2 and peak work to fall into the following categories. First, some work is achieved on the basis of anaerobic energy production (ie, that above the anaerobic threshold), and use of peak V̇O2 only will underestimate the total work or energy expenditure achieved.24 Second, total work, at least in endurance athletes, is linked to other variables besides peak V̇O2, such as fractional utilization of peak V̇O2 (ie, the percent of peak V̇O2 that can be sustained during exercise) and mechanical exercise efficiency.19
Indeed, mechanical efficiency is a critical factor to consider. For example, subjects with obesity are reported to have a higher V̇O2 at rest, but a lower work capacity for a given V̇O2, reflecting their lower gross mechanical efficiency.25 Indeed, the obese subjects in our study had a higher V̇O2 at rest (350 ± 104 mL/min vs 298 ± 91 mL/min, P = .002). In addition, the obese subjects had a higher V̇O2/work slope, even when isolated to the subgroup of subjects with peak V̇O2 > peak work (obese 13.7 ± 3.7 mL/min/watt, non-obese 12.2 ± 2.5 mL/min/watt, P = .08). One may consider that the subjects in this study with peak work > peak V̇O2 were more familiar with bicycle exercise than the subjects in the peak V̇O2 > peak work group, thus allowing them a potential biomechanical advantage by more efficient pedaling to generate work.26 However, we do not think this factor was relevant, because there was no overall difference in the use of or experience with bicycle ergometry in any of the 3 groups.
There are also other factors to consider. Muscle fiber type, specifically the relative amount of type I (slow twitch) and type II (fast twitch) fibers, might be associated with greater V̇O2/work slope, as seen in the peak V̇O2 > peak work group in our study, but data are conflicting,27,28 and we have no direct information on muscle fiber type in this study. Biochemical inefficiency may also result in peak V̇O2 > peak work, such as in McArdle disease, where there is a higher V̇O2 for a given amount of work, due to altered substrate utilization and a greater cardiovascular response to exercise.29 Both mechanical and biochemical factors may explain why different levels of work are achieved for the same peak V̇O2 depending on the exercise protocol. For example, total work achieved is higher with a greater work increment protocol, as seen here.15 In addition, the work involved in an incremental exercise test is 30% greater than that achieved for a constant load test for the same level of V̇O2.30 Meanwhile, an important factor that may result in peak work > peak V̇O2 is circulatory impairment. For example, in congestive heart failure, V̇O2 is limited for a given work load, due to the inability of cardiac output to fully compensate for increased oxygen extraction by the tissues, yielding a lower V̇O2 versus work relationship.31 We believe that circulatory impairment was an unlikely contributing factor in this study, because there was no clear discordance of cardiovascular disease in any of the subject groups. Finally, psychological factors, such as anxiety, fear, motivation, and perception of breathlessness or fatigue, clearly play a role in limiting exercise capacity.5,32,33
Since the amount of work for any given V̇O2 can vary due to all the reasons discussed above, we suggest that overall exercise tolerance be judged on the basis of the peak work achieved. This is in accordance with the general concept of rating impairment versus disability,34 wherein the former reflects physiological function, while the latter reflects the ability to perform work in the context of personal and environmental factors: a more global measure. It is important to emphasize, however, that peak V̇O2 remains an important measure of cardiovascular conditioning,35 preoperative risk in lung resection surgery,36 and a predictor of future cardiopulmonary outcomes,1–3,8,9 and such relationships have not been found for direct measures of work.
Our findings suggest that a lower BMI, greater breathing reserve, and higher peak heart rate are important independent factors associated with the ability to achieve relatively higher peak work than peak V̇O2. A lower BMI would allow a greater amount of work to be done for a given level of V̇O2, because of the improved mechanical efficiency of exercise.25 A greater breathing reserve would be consistent with the ability to handle the metabolic acidosis associated with exercise above the anaerobic threshold.13 The greater heart rate seen in subjects achieving higher peak work than peak V̇O2 most likely reflects these subjects' ability, effort, and motivation to exercise to physiological cardiovascular limitation.37 The higher respiratory exchange ratio in this group is consistent with this as well.
Limitations
This was a single center, retrospective review using bicycle ergometry and a step exercise protocol. How our results would compare to a different population using different equipment or protocols is unclear; however, since our hospital is a large, tertiary care facility that cares for a wide variety of patients, we believe our results would be broadly applicable. Importantly, even though the majority of subjects with COPD were within the peak V̇O2 = peak work group, there were more subjects with COPD in the peak V̇O2 > peak work group, which might have been the cause of the lower breathing reserve associated with this group and thus confounded our findings. Our definition of equivalence of peak work and peak V̇O2 was arbitrary, but seemed to best reflect our clinical impression of prevalence and importance.
We were unable to assess the contribution of psychological factors involved in exercise, such as motivation, anxiety, or thresholds for pain, although we did find no significant differences in the change in Borg dyspnea scores. Interestingly, this lack of differences may reflect reduced sensitivity to dyspnea among the subjects achieving high work loads.38 This reduced sensation of dyspnea is supported by the lower change in Borg dyspnea score adjusted for the amount of work performed among subjects in the peak work > peak V̇O2 group. In addition, although we suggest that peak work better reflects global exercise capacity than does peak V̇O2, we have not validated this concept by correlating this finding with any clinically meaningful outcomes, such as response to therapy or relationship to job or athletic performance. However, total work load is thought to be a critical factor in improving cardiopulmonary conditioning.35
It is important to note that there was discordance between the exercise protocols used in the peak work > peak V̇O2 group and the peak V̇O2 > peak work group. Namely, the group achieving higher work than V̇O2 exercised in higher work load increments (30 watts/min) than the group achieving higher V̇O2 than work (15 watts/min). These increments were chosen by the exercise technologist on the basis of the subject's self-reported exercise capacity, in order to create a protocol that would allow the subject to reach peak work capacity in approximately 10 min.13 The difference in these increments may have contributed to why different relationships of peak work versus peak V̇O2 were achieved.15 However, the altered rate of V̇O2 for different levels of work is usually seen for work rates above the anaerobic threshold,13,15 and there were no significant differences in slope of V̇O2-work above the anaerobic threshold in any of the 3 groups (see Table 1). In addition, in the study by Hansen and colleagues15 higher work rate increments resulted in shorter total exercise time among the same healthy subjects who were tested under different work rate protocols. We found just the opposite, which was that subjects who had the higher work rate increment (the majority of the peak work > peak V̇O2 group) had longer total exercise times (see Table 1). This suggests that other factors besides the exercise protocol are important in determining peak work versus peak V̇O2. Nevertheless, given this aspect of our study, and the finding of only modest odds ratios from the multivariate logistic regression analysis, one must interpret the findings with caution and realize that they may be of marginal clinical importance. In addition, the fact that the exercise increments were chosen on the basis of subject self-reported exercise capacity may have resulted in a selection bias that favored the more fit subjects exercising on the higher work increment protocol, thus creating a self-fulfilling situation of achieving higher peak work relative to peak V̇O2.
Conclusions
Subjects who had higher peak work than peak V̇O2 had lower BMI, higher breathing reserve, and higher peak heart rate. While the work protocol might have primarily determined this, the fact that differences in BMI, breathing reserve, and peak heart rate differentiate the peak work and peak V̇O2 groups suggests that underlying physiological characteristics reflect important aspects of exercise capacity that go beyond the peak V̇O2. The clinical corollary of this is that encouraging weight loss, treating any reversible lung disease, and recommending regular, strenuous exercise are appropriate strategies to help these patients improve their exercise capacity. When a cardiopulmonary exercise test is ordered to assess overall exercise tolerance, we suggest measuring both peak work and peak V̇O2, to provide the most comprehensive interpretation of the results.
Acknowledgment
We thank Peter Callas PhD for reviewing our statistical methods.
Footnotes
- Correspondence: David A Kaminsky MD, Pulmonary and Critical Care, University of Vermont College of Medicine, Given D-213, 89 Beaumont Avenue, Burlington, VT 05405. E-mail: david.kaminsky{at}uvm.edu.
Dr Knyazhitskiy presented a version of this paper at the International Conference of the American Thoracic Society, held May 13–18, 2011, in Denver, Colorado.
The authors have disclosed no conflicts of interest.
- Copyright © 2014 by Daedalus Enterprises
