Exercise ventilatory inefficiency in mild to end-stage COPD
- J. Alberto Neder1,2,
- Flavio F. Arbex2,
- Maria Clara N. Alencar2,
- Conor D.J. O’Donnell1,
- Julia Cory1,
- Kathy A. Webb1 and
- Denis E. O’Donnell1⇑
- 1Respiratory Investigation Unit and Laboratory of Clinical Exercise Physiology, Queen’s University and Kingston General Hospital, Kingston, ON, Canada
- 2Clinical Exercise Physiology Unit (SEFICE), Division of Respirology, Federal University of Sao Paulo, Sao Paulo, Brazil
- Denis E. O’Donnell, 102 Stuart Street, Kingston, K7L 2V6, ON, Canada. E-mail: odonnel@queensu.ca
Abstract
Ventilatory inefficiency during exercise is a key pathophysiological feature of chronic obstructive pulmonary disease. Currently, it is unknown how this physiological marker relates to clinically relevant outcomes as resting ventilatory impairment progresses across disease stages.
Slope and intercept of the linear region of the ventilation–carbon dioxide output relationship and the ratio between these variables, at the lowest point (nadir), were contrasted in 316 patients with Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages 1–4 (forced expiratory volume in 1 s, ranging from 148% pred to 12% pred) and 69 aged- and gender-matched controls,
Compared to controls, slope and intercept were higher in GOLD stages 1 and 2, leading to higher nadirs (p<0.05). Despite even larger intercepts in GOLD stages 3 and 4, slopes diminished as disease evolved (from mean±sd 35±6 in GOLD stage 1 to 24±5 in GOLD stage 3, p<0.05). As a result, there were no significant differences in nadirs among patient groups. Higher intercepts, across all stages (p<0.01), and to a lesser extent lower slopes in GOLD stages 2–4 (p<0.05), were related to greater mechanical constraints, worsening pulmonary gas exchange, higher dyspnoea scores, and poorer exercise capacity.
Increases in the ventilation intercept best indicate the progression of exercise ventilatory inefficiency across the whole spectrum of chronic obstructive pulmonary disease severity.
Abstract
Exercise ventilatory inefficiency relates to dyspnoea and exercise intolerance across whole COPD severity spectrum http://ow.ly/CpWLj
Introduction
Minute ventilation (V′E) during moderate exercise (i.e. in the absence of metabolic acidosis) is tightly coupled to the rate of carbon dioxide production, washed-out by the lungs (V′CO2) [1, 2]. In fact, V′E increases almost exactly enough to keep arterial carbon dioxide tension (PaCO2) constant, despite increasing V′E [3, 4]. As cogently pointed out by Whipp andWard [1] and Whipp [5], this can occur only if there is a proportional matching between two decreasing ratios: dead space (VD) to tidal volume (VT) and V′E/V′CO2. The hyperbolic V′E/V′CO2 behaviour towards its lowest value (nadir) is a necessary consequence of a positive y-intercept in the linear V′E–V′CO2 relationship under isocapnic conditions [5]. In cardio-respiratory disease, the V′E–V′CO2 slope and V′E/V′CO2 nadir increase (i.e. ventilatory inefficiency worsens) reflecting a higher VD/VT and/or a lower PaCO2 set-point 6; provided the ventilatory response is not constrained by abnormal lung mechanics. This explains why these findings may be useful to assess disease severity when mechanical abnormalities do not play a major role in limiting exercise capacity, e.g. heart failure and pulmonary arterial hypertension [7, 8].
However, a more complex scenario emerges in a highly-heterogeneous condition, such as chronic obstructive pulmonary disease (COPD). The V′E–V′CO2 relationship has been used to assess disease progression [9–11], identify the presence of comorbidities [12–15] and to evaluate the effect of therapeutic interventions [16–19] in such patients. This is physiologically justified by evidence of COPD progression being associated with increased ventilation-perfusion mismatch and lower exercise tolerance, leading to poorer ventilatory efficiency [20]. However, concomitant increases in PaCO2 and mechanical constraints are expected to flatten the ventilatory response in these patients, i.e. the V′E–V′CO2 slope might, paradoxically, decrease as the disease evolves and mechanical abnormalities worsen (fig. 1a) [22–25]. The V′E/V′CO2 nadir may also not accurately reflect the extent of ventilatory inefficiency as slope and intercept may change in opposite directions to potentially cancel each other out, i.e. the nadir would become stable despite COPD progression (fig. 1b) [5, 24, 25]. As a further complicating issue, a high nadir may be a result of too short a test, in which the decreasing profile may be “amputated” by premature exercise cessation in unfit and severely limited patients (fig. 1b) [6, 26]. Consequently, it remains unknown how best to express ventilatory inefficiency as the mechanical constraints, gas exchange disturbances and exercise intolerance worsen from the mild to the end-stage of COPD.
The primary aim of this study, therefore, was to provide a comprehensive frame of reference to interpret the measures of ventilatory inefficiency as COPD evolves. Secondly, we wished to examine associations between those measures with ventilatory capacity and patient centred outcomes (dyspnoea and exercise tolerance), across the continuum of COPD severity. We reasoned that these data would provide a more sound physiological background to clinically interpret the relationship in this patient population [9–19].
Methods
Subjects
This study involved a retrospective analysis of incremental cycle cardiopulmonary exercise studies performed as screening for several ethically-approved research studies on COPD at the Respiratory Investigation Unit, Queen’s University and Kingston General Hospital (Kingston, ON, Canada). The Queen’s University and Affiliated Teaching Hospitals Research Ethics Board approved the use of these anonymous data sets and waived the need for patient informed consent (DMED-1659-13). Patients had a well-established diagnosis of COPD [27], with no evidence of asthma or any other lung disease. Patients were required to have been free of any exacerbation in the preceding 6 weeks. Controls had no major respiratory, cardiovascular or metabolic diseases that could interfere with the study’s results. Lack of orthopaedic, neuromuscular, cardiac and metabolic conditions, which could preclude the subject from safely undertaking incremental, exercise tests, were part of the inclusion criteria of the study.
Procedures
Lung Function Tests
Spirometry (including inspiratory capacity (IC) measurements), body plethysmography (residual volume (RV), total lung capacity (TLC), and airways resistance (Raw)), and diffusing capacity of the lung for carbon monoxide (DLCO) were performed using automated testing equipment (2130 spirometer with 6200 Autobox DL or V6200 Autobox; SensorMedics, Yorba Linda, CA, USA). All short-acting and long-acting bronchodilators were withdrawn for at least 4 h and 12–24 h, respectively. As patients had taken part in studies with different bronchodilators, spirometric Global Initiative for Chronic Obstructive Lung Disease (GOLD) stages were defined according to pre-bronchodilator forced expiratory volume in 1 s (FEV1) [27].
Exercise Tests
Symptom-limited, incremental, exercise testing was performed on an electronically braked cycle ergometer using the Vmax229d Cardiopulmonary Exercise Testing System (SensorMedics). The rate of work rate increment was individually selected according to reported exercise tolerance and resting functional impairment (5–10 W·min−1 in GOLD stages 3 and 4, 10–15 W·min−1 in GOLD stages 1 and 2, and 5–20 W·min−1in healthy subjects). V′E (L·min−1), V′CO2 (L·min−1), oxygen uptake (V′O2) (L·min−1), end-tidal carbon dioxide tension (PETCO2) (mmHg), respiratory frequency (fR) (breaths·min−1) and VT (L) were averaged at 30 s intervals. Arterial oxygen saturation was measured by pulse oximetry (SpO2) (%) noninvasively. Breathlessness was rated according to the 10-point Borg category-ratio scale [28]. The linear phase of the V′E–V′CO2 relationship was determined in the V′E–V′CO2 plot (V′E is y-axis and V′CO2 is x-axis). Linear regression was then applied to these data points [21, 26]. Nadir and the peak V′E/V′CO2ratio (V′E/V′CO2peak) were the lowest and the last 30-s average data point, respectively [21]. End-inspiratory lung volume (EILV) (L) was calculated as end-expiratory lung volume (EELV), taken from IC, plus VT and related to TLC [20]. Peak V′E response (V′Epeak) was also expressed relative to estimated maximal voluntary ventilation (MVV) (L·min−1), which was equal to FEV1×35. Due to the expected differences in maximal exercise capacity among the COPD stages, EILV/TLC (×100) and dyspnoea scores were corrected for the V′Epeak. For the same reasons V′Epeak and MVV were corrected for the peak metabolic stress (V′O2peak).
Statistical Analysis
Values are reported as means±sd, unless otherwise stated. A p-value <0.05 was considered significant in all analyses. Intraclass correlation coefficients determined the level of between-investigator agreement in the calculation of slope and intercept. Comparisons across subgroups were performed using ANOVA with post hoc testing of significant variables carried out using t-tests with Bonferroni adjustment for multiple comparisons. Chi-squared analysis tested the association between categorical variables. Pearson’s R tested the correlation between continuous variables.
Results
Subject characteristics
316 patients distributed across all GOLD stages and 69 controls were evaluated. Patients and controls were well matched for age (67.7±6.2 years versus 68.4±7.2 years), sex (179 (56.6%) out of 316 versus 37 (53.7%) out of 69 males) and body mass index (26.9±4.8 kg·m−2 versus 26.0±3.1 kg·m−2) (p>0.05). Table 1 shows the expected decreases in IC, IC/TLC and DLCO and increases in RV, TLC and Raw as airway obstruction gradually worsened from GOLD stage 1 to 4 (p<0.01).
Physiological and perceptual responses according to GOLD stages
Peak exercise capacity was progressively reduced from healthy controls to very-severe COPD (table 2 and fig. 2a). These findings were associated with gradual worsening of the mechanical ventilatory constraints (fig. 2b and c). Peak-rest changes in PETCO2 increased (fig. 2d) and SpO2 decreased (fig. 2e) as airway obstruction increased. Peak-rest changes in PETCO2 did not correlate with fR either in controls or patients (r=0.04 and 0.12, respectively, p>0.05). Ventilation-corrected dyspnoea scores increased from GOLD stage 1 to 4 (fig. 2f). Exercise test duration was lower in GOLD stages 4 and 3 when compared to GOLD stages 2 and 1, and controls (table 2) (p<0.05).
Ventilatory inefficiency across GOLD stages
Between observers, the intraclass correlation coefficients for intercept and slope estimation were excellent (r=0.98 and 0.97, respectively, p<0.01) with r2 values for the relationship ranging from 0.92 to 0.99. Compared to controls, all patient groups had higher intercepts; in contrast, slopes were increased in GOLD stages 1 and 2, similar in GOLD stage 3 and lower in GOLD stage 4 (actual values being: 3.0±1.1 L·min−1, 4.2±1.3 L·min−1, 5.8±2.6 L·min−1, 7.2±2.8 L·min−1 and 8.3±3.0 L·min−1 for the intercepts and 27±3, 35±6, 31±5, 27±5 and 24±5 for the slopes, respectively) (fig. 3 shows representative patients and fig. 4a and b show the mean data).
There was a negative correlation between intercept and slope across the patient groups (fig. 5a). Intragroup prevalence of high intercepts and slopes (>95% upper confidence limit observed in controls, 5 L·min−1 and 32, respectively) varied in opposite directions in GOLD stages 1–4 (fig. 6). These results were consistent with those found in a between sex, comparative analysis, i.e. the relative proportion of high intercepts and slopes in each stage did not differ in males versus females (p>0.05).
Nadirs were greater than slopes in the great majority of subjects (96.8%) (fig. 5b). Nadir values were reached at peak exercise in all patients from GOLD stages 3 and 4, and in the majority of those in GOLD stages 1 and 2 (74.5% and 87.3%, respectively) (fig. 3). Significant correlations between nadir versus intercept and slope were found only in controls and GOLD stage 1 patients (r=0.61 and 0.59, respectively, p<0.05). As depicted in figure 4c nadirs were increased to a similar extent in all patient groups compared to controls. As expected from similar nadirs and progressively lower slopes, nadir–slope differences markedly increased from GOLD stages 1 to 4 (figs 4d and 5b). Despite a significant positive association between intercept and nadir–slope differences, the scatter around the correlation line increased at higher values (fig. 5b).
Functional and subjective correlates of ventilatory inefficiency
The putative relationships between different measures of ventilatory inefficiency and clinically relevant outcomes were assessed in patients separated by COPD severity. Increases in intercept, slope and nadir were all similarly related to lower DLCO (r=−0.62, 0.64 and −0.61, respectively), lower V′O2peak (r=−0.58, 0.50 and −0.53) and greater dyspnoea/V′E ratios (r=0.65, −0.63, and 0.60, respectively) in GOLD stage 1 patients (p<0.01). Interestingly, all ventilatory inefficiency parameters were better related to dyspnoea scores than FEV1 (r=0.18, p>0.05) in this group. Higher dyspnoea/V′Eratio scores were found in those with slopes >32 compared to their counterparts (0.14±0.03 versus 0.08±0.04 Borg score units·L−1·min−1) (p<0.01).
Increases in intercepts related consistently better to DLCO and V′O2peak than decreases in slope in GOLD stages 2–4 patients (r=−0.69 versus 0.42 in GOLD stage 2, r=−0.72 versus 0.45 in GOLD stage 3, and r=−0.70 versus 0.38 in GOLD stage 4; p<0.01 for intercept and p<0.05 for slope correlations). Particularly high dyspnoea/V′Eratio scores were found in those with very high intercepts (>8 L·min−1) and pronouncedly shallow slopes (<24) compared to their counterparts (0.28±0.08 versus 0.14±0.07 Borg score units·L·min−1) (p<0.01). Similar results were found in relation to pulmonary gas exchange impairment (peak-rest decreases in SpO2 and increases in PETCO2) and mechanical ventilatory constraints (higher (V′E/MVV)/V′O2 and (EILV/TLC)/V′E); r values ranged from 0.62 to 0.73 for the intercept (p<0.01) and from −0.48 to −0.60 for the slope (p<0.05). There were no significant correlations between nadir and any of these variables in GOLD stages 2–4. Conversely, only the nadir was related to exercise-test duration across the whole sample (r=−0.63, p<0.01).
Discussion
This is the first study to systematically investigate the effect of COPD severity on measures of exercise ventilatory inefficiency (intercept, slope and nadir) in a large sample of males and females with mild to end-stage disease. Our main results can be summarised as follow: 1) significant ventilatory inefficiency was found in mild, GOLD stage 1, COPD patients when compared to controls, a finding related to exertional breathlessness and poorer exercise capacity; 2) higher ventilation intercepts and, to a less extent, lower slopes related to greater mechanical constraints, worsening pulmonary gas exchange, higher dyspnoea scores and exercise intolerance; and 3) changes in the nadir, however, failed to predict those abnormalities in GOLD stages 2–4. These results indicate that exercise ventilatory inefficiency is a physiological marker related to clinically relevant end-points in mild to end-stage COPD. A hitherto neglected variable, the V′E–V′CO2 intercept [1], emerged as a particularly useful index for ventilatory inefficiency, across the continuum of COPD severity.
There has been renewed interest in exercise ventilatory inefficiency as a physiological marker in patients with COPD [9–19]. However, it is rather surprising that no previous study has systematically looked at this topic in a sufficiently large number of patients with varied degrees of disease severity. In fact, some smaller studies evaluated patients with advanced emphysema [9] or mid-stage disease [22, 23]. The largest previous study did not evaluate GOLD stage 1 patients and restricted its analysis to V′E–V′CO2peak [10]. Therefore, the prevailing view on the topic remains largely influenced by the cardiovascular literature, i.e. high V′E–V′CO2 slope and V′E/V′CO2 nadir indicate worsening ventilatory inefficiency, regardless of COPD severity [9, 10, 21, 26].
However, our main results demonstrate that this approach is likely to lead to substantial misinterpretation of ventilatory inefficiency in COPD. Apart from GOLD stage 1 patients in whom both slope and nadir were higher, compared to controls, our results would indicate a paradoxical “improvement” (slope) or an apparent stability (nadir) in ventilatory inefficiency in GOLD stages 2–4 (fig. 4b and c). However, lower V′E–V′CO2 slopes in more advanced disease, were explained by worsening mechanical constraints (fig. 2c) [9, 23] and, probably, an increase in carbon dioxide set-point [29]. Similar V′E/V′CO2nadirs across patient groups (fig. 4c) reflected the opposite changes in V′E–V′CO2 slope and intercept (figs 5a and 6c) and shorter test durations in more severe patients.
In contrast, we found that the expected increase in “wasted” ventilation as COPD evolved from GOLD stage 1 to 4 [20] was better reflected by progressive increases in intercept (fig. 4a). This interpretation is in line with some [25, 30], but not all [31, 32], experimental studies in normal subjects, in which increases in series (anatomical) VD upshifted the V′E–V′CO2 relationship, and PETCO2, with no appreciable change in slope. As pointed out by Gargulio et al. [25], V′E in the absence of pulmonary gas exchange (i.e. V′E intercept with null V′CO2) theoretically equals to VD. In fact, resting V′E, the closest biologically plausible correlate of the intercept, does shift upwards as COPD progresses, a finding that is likely to reflect the worsening of ventilation-perfusion mismatch [20]. The key limitation of the V′E–V′CO2 intercept, to constitute a mathematical extrapolation, is paradoxically its main advantage in COPD; by definition, the intercept cannot be constrained by dynamic mechanics (as the slope) or test duration (as the nadir). We interpret the overlap of intercept values among GOLD stages (fig. 4a) as a reflection of FEV1 being a poor predictor of the ventilatory response in individual patients [20]. For instance, the ventilatory drive in COPD is variably influenced by afferent information from the working limbs [33] and peripheral chemoreceptors [16] at a given FEV1.
It is also noteworthy that V′E/V′CO2 nadir exceeds the slope to a greater extent if the subjects stop exercising “prematurely” [4, 6, 26], e.g. due to limiting dyspnoea. In fact, we found an inverse relationship between nadir and test duration in patients. Moreover, the nadir–slope differences not equal to the intercept, particularly in more advanced COPD (fig. 5b). These data indicate that too short a test resulted in V′E/V′CO2 nadir overestimating the slope in these patients. Of note, nadir–slope difference may decrease after beneficial interventions, if higher exercise tolerance (leading to lower nadir in more severe patients) is associated with improved ventilatory response (higher slope). Prospective trials are warranted to assess the usefulness of nadir–slope difference as an outcome in COPD.
The present results hold other important clinical implications for the interpretation of the V′E–V′CO2 relationship in COPD. First and foremost, it is rather remarkable that GOLD stage 1 patients, with largely preserved FEV1, showed significant impairment in exercise ventilatory efficiency, a finding closely related to exertional breathlessness. These data suggest that ventilatory inefficiency might constitute a useful physiological marker of disease severity in this subpopulation [11]. It is also conceivable that GOLD stages 2–4 patients, showing high intercepts and lower slopes, are at greater risk of a negative outcome. However, some patients with pulmonary artery pressure values greater than expected by the degree of hypoxaemia [14, 15], do present with high slopes; for these patients, such findings may still represent a negative prognostic marker. Similar considerations apply to directional changes induced by interventions aimed at decreasing VD/VT [16, 18, 19]. Slope, intercept and nadir are all expected to decrease in GOLD stage 1 patients. However, if the ventilatory constraints are also alleviated by the intervention, lower intercepts and higher slopes can be anticipated in more severe patients. Highly variable, and even neutral, effects may occur in the nadir. Another clinical application of the intercept relates to its potential to suggest COPD in heart failure patients and vice versa [24]. For instance, null or negative intercept values, a not so infrequent finding in heart failure [13, 24], were found in <1% of our patients (fig. 5). Whether higher intercepts and slower slopes in patients with coexistent COPD would impact in the prognostic relevance of ventilatory inefficiency in heart failure [7], deserves further investigation. Finally, the controversy on nadir or V′E–V′CO2peak being the preferred measure of ventilatory inefficiency [7, 21, 26] is not clinically pertinent to COPD as these variables differ in only a minority of less severe patients.
The present study has, naturally, some limitations. As a noninvasive, clinical physiology study, involving a large number of patients, our mechanistic extrapolations are necessarily limited [34, 35]. For instance, it remains speculative whether dynamic decreases in carbon dioxide set-point and/or increases in physiological VD/VT would steepen the slope in particular patients [25]. However, the calculation of changes in VD/VT during rapidly progressive exercise is, notoriously, inaccurate, particularly in COPD [36]. In the absence of a true criterion test for exercise ventilatory inefficiency, we relied on a cluster of variables that are indirect markers of pulmonary gas-exchange disturbances. In this context, exercise induced elevations in PETCO2 reflect either true carbon dioxide retention or delayed lung emptying in COPD [37, 38]. Considering that PETCO2 grossly underestimates alveolar carbon dioxide tension in these patients [36], our assumptions, which are increases in PETCO2indicate more severe gas exchange disturbances in COPD, still hold true. It should be noted that we might have underestimated the role of hypoxaemia in modulating ventilatory inefficiency, as we did not evaluate overtly hypoxaemic patients. We also recognise that the modulating effects of disease phenotypes (particularly emphysema) [9] and test modality [39], on the different strategies to express ventilatory inefficiency, require better characterisation in COPD.
In conclusion, a hitherto neglected variable, the V′E–V′CO2 intercept [1], better expressed the progressive worsening on exercise ventilatory inefficiency across the continuum of COPD severity (GOLD stages 1–4). Compared to commonly used measures, i.e. slope and nadir, the intercept correlated better with key outcomes of clinical relevance, i.e. dyspnoea and exercise tolerance, regardless of the stage of disease. High intercepts in patients with largely preserved spirometry (GOLD stage 1), suggest coexistent ventilation-perfusion abnormalities that may explain persistent symptoms (dyspnoea and activity restriction) and may prompt further investigations. This variable may, therefore, assist in clinical phenotyping and deserves to be prospectively tested as a prognostic index in longitudinal studies and a physiological end-point in interventional trials in COPD patients.
Footnotes
Conflict of interest: None declared.
- Received July 24, 2014.
- Accepted September 4, 2014.
- Copyright ©ERS 2015