Intercept of minute ventilation vs. carbon dioxide output relationship in chronic obstructive pulmonary disease: utility as an index of ventilatory inefficiency

Background : Ventilatory inefficiency is known to be a contributor to exercise intolerance in chronic obstructive pulmonary disease (COPD). The intercept of the minute ventilation (V̇ E ) vs. carbon dioxide output (V̇CO 2 ) plot is a key ventilator inefficiency parameter. However, its relationships with lung hyperinflation (LH) and airflow limitation are not known. This study aimed to evaluate the correlations between the V̇ E /V̇CO 2 intercept and LH in COPD to determine its utility as an index of functional impairment. Methods: We conducted a retrospective analysis of data from 53 COPD patients and 14 healthy controls performed incremental cardiopulmonary exercise tests and resting pulmonary function. Ventilatory inefficiency was represented by parameters reflecting the V̇ E /V̇CO 2 nadir and slope (linear region), and intercept of the V̇ E /V̇CO 2 plot. Their correlations with measures of LH and airflow limitation were evaluated. Results: Compared to the control, the slope (30.58±3.62) and intercept (4.85±1.11) higher in COPD stages1-2 , leading to a higher nadir (31.47±4.47) ( p <0.05). Despite an even higher intercept in COPD stages3-4 (7.16±1.41), the slope diminished with disease progression (from 30.58±3.62 in COPD stages1-2 to 28.36±4.58 in COPD stages3-4 ). Compared to the V̇ E /V̇CO 2 nadir and V̇ E /V̇CO 2 slope, the intercept was better correlated with peak V̇ E /maximal voluntary ventilation (MVV) ( r =0.489, p <0.001) and peak V̇O 2 /watt ( r =0.354, p =0.003). The intercept was also significantly correlated with RV/TLC ( r =0.588, p <0.001), IC/TLC ( r =-0.574, p <0.001), peak V T /TLC ( r =-0.585, p <0.001); and airflow limitation forced expiratory volume in 1s (FEV

disease (COPD) [1]. The limitations in activity and dyspnea are multifactorial. The development of lung hyperinflation (LH) plays an important role in the pathophysiology of dyspnea and exercise intolerance [2]. Static hyperinflation which is caused by destructions of pulmonary parenchyma and loss of lung elastic recoil is characterized by increased functional residual capacity (FRC) and reduced inspiratory capacity (IC) [3]. Dynamic hyperinflation which occurs when the expiratory time becomes insufficient to allow the lung to achieve full exhalation yields an increased end-expiratory lung volume (EELV) during exercise [4]. LH increases ventilatory workload and decreases inspiratory muscle pressure generating capacity, despite some compensatory mechanisms [3]. The diminished ventilatory capacity coupled with the increased ventilatory demand during exercise yields exercise intolerance.
The minute ventilation (V̇E) vs. carbon dioxide production (VĊO 2 ) relationship is a measure of the ventilatory efficiency at removing CO 2 produced by the body. Early in exercise, V̇E/VĊO 2 decreases with a decrease in dead space ventilation (V D )/tidal volume (V T ) ratio. The V̇E/VĊO 2 nadir is typically reached just before ventilation starts to increase to compensate for lactic acidosis at the respiratory compensation point [5]. The V̇E/VĊO 2 nadir was found to be highly reproducible in healthy subjects [6] and COPD patients [7]. However, the V̇E/VĊO 2 nadir might underestimate ventilatory efficiency if the descending curve is prematurely interrupted by lactic acidosis or an excessively short test duration [8]. On the other hand, V̇E/VĊO 2 ratio might higher than the nadir as the hyperventilation response to late-exercise acidosis in patients who are able to can exercise beyond the respiratory compensation point [9].
The V̇E/VĊO 2 slope has been used to assess disease progression and to identify the presence of comorbidities [10][11][12][13][14][15]. However, in many patients with moderate-to-severe COPD, concomitant increases in the partial pressure of carbon dioxide (PaCO 2 ) and mechanical constraints will predictably flatten the V̇E/VĊO 2 curve. In these patients, the V̇E/VĊO 2 slope might, paradoxically, decrease as the disease evolves if CO 2 retention during exercise worsens. It is plausible that the V̇E/VĊO 2 nadir might be stable while the slope and intercept change in opposite directions despite COPD progression [9].
Theoretically, the y-intercept of the V̇E/VĊO 2 plot (intercept = V̇E when VĊO 2 = 0, that is, in the absence of pulmonary gas exchange) corresponds to the basal V̇E that contributes to the wasted V D [16]. By definition, the intercept cannot be constrained by dynamic mechanics (unlike the slope) or the test duration (unlike the nadir). The increased intercept in COPD patients theoretically results from dead space (when metabolic demand is null) and might result from an altered breathing strategy (increased breathing frequency to compensate for reduced V T secondary to greater mechanical constraints) and/or a progressive ventilation-perfusion mismatch in COPD patients [1]. Thus, the V̇E/VĊO 2 intercept increases with greater disease severity in COPD patients, and it seems to be a particularly useful index for ventilatory inefficiency across the continuum of COPD severity [9].
However, the clinical implications of the V̇E/VĊO 2 intercept and its association with LH in COPD have not been formally examined.
This study aimed to evaluate the relationship between the V̇E/VĊO 2 intercept and LH and airflow limitation in patients with COPD. We hypothesized that the V̇E/VĊO 2 intercept correlated well with measures of both LH and airflow limitation and could be used as a particularly useful index for ventilatory inefficiency in COPD.

Methods: Study participants
This study was a retrospective analysis of data collected during incremental cardiopulmonary exercises from ethically-approved research studies on COPD at the Respiratory Investigation Unit, Beijing Friendship Hospital, Capital Medical University (Beijing, China). Participants were males and females aged ≥ 40 years with body mass index (BMI) of 18-35 kg/m 2 . The patients were current or exsmokers (smoking history ≥ 10 pack-years) and had a well-established diagnosis of COPD [17] without asthma or other pulmonary diseases. Patients were required to have had no exacerbation in the preceding 6 weeks. Control subjects with no smoking history were in the same age range and they had no major orthopedic, neuromuscular, cardiac or metabolic diseases, to allow them to safely undertake the incremental exercise tests.

Pulmonary Function Tests
Each subject underwent resting spirometry (MasterScreen Body, CareFusion, Hoechberg, Germany), including inspiratory capacity (IC) assessment. Body plethysmography was performed to measure residual volume (RV), total lung capacity (TLC) and diffusing capacity of the lungs for carbon monoxide (DL CO ). Patients took 400 µg albuterol by inhalation 20 min before testing. All pulmonary function tests fulfilled the American Thoracic Society /European Respiratory Society guidelines [18].

Cardiopulmonary Exercise Test (CPET)
Symptom-limited incremental exercise testing was performed on an electronically braked cycle ergometer (ViaSprint, CareFusion, Hoechberg, Germany) with a pedaling rate of 60/min. After 3 min of rest and 3 min of unloaded pedaling, the work rate (WR) was increased by 5-15 W/min in a ramp fashion (5 W/min if FEV 1 < 1.0 L and 10 W/min if FEV 1 ≥ 1.0 L for the COPD patients; 15 W/min for the controls, with repetition at 20 W/min if the peak WR was ≥ 200 W). Participants were asked to continue to exercise to the limit of tolerance, marked by the inability (despite encouragement) to maintain pedaling frequency or intolerable shortness of breath. Any participant with chest pain suggestive of ischemia, ventricular tachycardia and blood pressure (BP) ≥ 240/130 mmHg was prevented further exercise. Patients were continuously monitored with a 12-lead electrocardiogram and blood pressure by sphygmomanometer every 2 min.

Data Collection
Respiratory gas exchange (V̇E, VȮ 2 , and VĊO 2 ) and V T were measured breath-by-breath throughout the exercise testing. Serial measurements of these parameters were averaged at 30-s intervals.
Arterial oxygen saturation was measured noninvasively by pulse oximetry (SpO 2 ; %). The V̇E/VĊO 2 nadir and peak were the lowest and the mean of the last 30-s of data, respectively [19]. The slope of the V̇E/VĊO 2 relationship was determined based on the V̇E vs.VĊO 2 plot (V̇E on the y-axis and VĊO 2 on the x-axis). A linear regression line was determined based on these data points [19,20] from the start of loading exercise to the nadir. The V̇E/VĊO 2 intercept was calculated by extrapolating the regression line to VĊO 2 = 0. Maximal voluntary ventilation (MVV) was calculated as FEV 1 × 4 0. Peak V T /TLC, was used as measures dynamic LH during exercise while IC/TLC and RV/TLC were used as static LH [21].

Statistical analysis
Values are reported as mean ± SD unless otherwise stated. P-value < 0.05 was considered significant in all analyses. Intraclass correlation coefficients were used to determine the level of betweeninvestigator agreement in the calculation of the slope and intercept. Between-group comparisons were performed using one-way analysis of variance (ANOVA) with LSD post-hoc testing of significant variables. Pearson's correlation coefficient (r) was used to assess the correlations between the ventilatory inefficiency parameters (V̇E/VĊO 2 intercept, slope and nadir) and ventilatory capacity (peak V̇E/MVV, peak VȮ 2 /watt), static or dynamic hyperinflation measures (RV/TLC, IC/TLC and peak V T /TLC), and airflow limitation (FEV 1% predicted and FEV 1 /FVC). All analyses were performed with IBM SPSS Statistics 20.0 (Chicago, USA).

Participant characteristics and resting spirometric measurements
As shown in Table 1, the COPD stages1−2 (n = 35), COPD stages3−4 (n = 18) and control (n = 14) were well matched in terms of age, weight, and BMI. The resting spirometric measures of the participants presented in Table 1. There are the expected decreases in FEV 1 % predicted, FEV 1 /FVC, IC % predicted, IC/TLC, DL CO % predicted and the expected increases in RV%, RV/TLC, FRC/TLC from COPD stages1−2 to COPD stages3−4 (p < 0.01). Table 1 shows demographics and selected resting pulmonary function variables in control and COPD patients.  Table 2 shows the exercise variables at peak exercise in the control and COPD patients categorized by GOLD.  Figure 1 shows measures of ventilatory inefficiency in control and COPD patients. Figure 2 shows VĖ/VĊO2 intercept in correlation with peak V̇E/MVV and peak VȮ 2 /watt in the entire study group.

Correlation of ventilatory inefficiency with lung hyperinflation and airflow limitation
The relationships between the measures of ventilatory inefficiency and lung hyperinflation and airflow  Table 3 shows the correlations between the ventilatory inefficiency parameters and measures of lung hyperinflation and airflow limitation. Figure 3 shows V̇E/VĊO 2 intercept significant correlations with peak V T /TLC, IC/TLC, RV/TLC, and FEV 1% predicted in the entire study group.

Discussion
The main finding of this study was that V̇E/VĊO 2 intercept was consistently correlated with worsening LH and increasing airflow limitation in COPD. V̇E/VĊO 2 intercept could be a useful index for ventilatory inefficiency during incremental exercise in COPD.
V̇E/VĊO 2 relationship was analyzed according to the V̇E/VĊO 2 ratio vs. time plot [22]. For healthy subjects who can tolerate high levels of exercise, the V̇E/VĊO 2 nadir and V̇E/VĊO 2 ratio at the anaerobic threshold were usually very similar [6]. Abnormalities in the V̇E/VĊO 2 relationship were present across the spectrum of COPD severity. The V̇E/VĊO 2 nadir showed superior test-retest reliability compared to the V̇E/VĊO 2 slope in COPD patients [23]. Increases both in V̇E/VĊO 2 nadir and slope were associated with lower maximal exercise capacity in COPD patients [24,25]. A retrospective study with a large range of resting pulmonary function (FEV 1 = 12-148% predicted) showed an increased V̇E/VĊO 2 slope in mild-moderate COPD but a decreased slope in advanced stage in comparison to control. As for V̇E/VĊO 2 nadir, there was no significant difference in different stages.
However, the V̇E/VĊO 2 intercept was higher across all stages of COPD [9]. In our study, compared to control, COPD stages1−2 had a higher slope and nadir, while patients with more advanced stages (COPD stages3−4 ) had a lower slope and a stable nadir (i.e., with no significant change compared to COPD stages1−2 ). The V̇E/VĊO 2 intercept increased from COPD stages1−2 to COPD stages3−4 . In advancedstage COPD, the stable V̇E/VĊO 2 nadir likely reflected the opposite changes in the V̇E/VĊO 2 slope and intercept.
There was mounting evidence that ventilatory inefficiency parameters were powerful prognostic predictors in COPD patients with comorbidity. A retrospective study in 145 COPD patients undergoing surgery for non-small cell cancer showed that VĖ/VĊO2 slope > 34 predicted mortality after lung resection surgery [26]. As for the VĖ/VĊO2 nadir, Neder et al. reported that the nadir > 34 in combination with resting hyperinflation predicted mortality in COPD [27]. Importantly, a series of studies demonstrated that the VĖ/VĊO2 intercept (cutoff values ranging from 2.64-4.07 L/min) might discriminate COPD from heart failure [28,29].
Ventilatory inefficiency increases ventilatory demand and exercise capacity limitation due to expiratory flow limitation that enhances dynamic hyperinflation. Two other independent studies showed correlations between the V̇E/VĊO 2 nadir and emphysema severity on high-resolution computed tomography scans in COPD patients with largely preserved FEV 1 [30,31]. Static LH caused by reduction of elastic recoil due to emphysema in COPD and development of expiratory flow limitation promoted progressive air trapping with an increase in the EELV and a decrease in IC. RV was also increased in emphysema/COPD because of both loss of elastic recoil and premature closure of the small airways [32][33][34]. In expiratory flow-limited patients, EELV was a continuous dynamic variable, which depended on expiratory duration and breathing pattern. DH referred to this temporary and variable increase in EELV. DH was a consequence as ventilation increases and expiratory duration decreases, there was not enough time to allow EELV to decline to its baseline resting value during exercise [35].
Studies reported both static hyperinflation and the degree of dynamic lung hyperinflation were associated with the development of dyspnea and exercise intolerance in COPD patients [36,37].
Assuming stability of TLC, the resting IC and inspiratory reserve (IRV) showed the operating position of V T relative to TLC. The smaller the resting IC, the shorter the exercise time before V T reached plateau and dyspnea abruptly escalates [38]. A four-year longitudinal study reported that significant reductions in peak VȮ 2 and V̇E were related to a decrease in resting IC [39]. Both IC/TLC and RV/TLC in patients with COPD reflected not only the degree of lung static hyperinflation but also the functional reserve. IC/TLC was also found to be a valuable and independent predictor of all-cause and respiratory mortality in COPD compared with that of the BODE (body mass index, airflow obstruction, dyspnea, exercise performance) index [40]. The present study showed V̇E/VĊO 2 intercept exhibited better correlated with rest IC/TLC (r=-0.574, p < 0.001) and RV/TLC (r = 0.588, p < 0.001) than V̇E/VĊO 2 nadir .with peak IC/TLC (r=-0.350, p = 0.004) and RV/TLC (r = 0.431, p < 0.001) while V̇E/VĊO 2 slope had no correlation with static LH parameters.
The EELV progressively increases while IC decreases were associated with dyspnea and exercise intolerance in COPD during exercise [41]. Serial measurements of IC to detect its changes had been reported to be a classic way to identify dynamic hyperinflation [36,37,42]. However, the study participates had to be familiar with the maneuvers, and IC measurements also had to be standardized by researchers [43]. Nevertheless, dynamic IC measurement was not recommended for ramp-pattern protocols where V T cannot steadily proceed to perform IC maneuver. However, the ramp-pattern protocol was a widely used for incremental test [43]. Elevated EELV can substantially constrain the expansion of V T at higher exercise intensities. It followed that COPD patients reached a V T plateau and a similar minimal inspiratory reserve volume. Chuang et al. investigated peak V T /TLC as a convenient new marker of DH and the cutoff value was 0.27 [44]. The present results showed among ventilatory inefficiency parameters (slope, nadir and intercept), only V̇E/VĊO 2 intercept exhibited better correlated with peak V T /TLC (r=-0.585, p < 0.001) than V̇E/VĊO 2 nadir with peak V T /TLC (r=-0.503, p < 0.001) and V̇E/VĊO 2 slope with peak V T /TLC (r=-0.148, p = 0.232). To our knowledge, this is the first study to describe the relationship between ventilatory inefficiency and DH.
A limitation of our study is the modest number of subjects. We believe that the increased ventilatory inefficiency associated with LH might be more pronounced in patients with more advanced COPD.
However, in the absence of a true criterion test for ventilatory inefficiency during exercise, we relied on a cluster of variables that were indirect markers of pulmonary gas-exchange disturbances. We also recognize that variables related to disease phenotypes and test factors (e.g., duration) affect the different strategies to reflect ventilatory inefficiency.

Availability of data and materials
The datasets used and analyzed in the present study will be available from the corresponding author on reasonable request.

Authors' contributions
Haoyan Wang and Bo Xu participated in the conception, the design and coordination of the study.
Fang Lin collected data, interpreted the patient data and drafted the manuscript. Min Cao participated in its coordination and helped to draft the manuscript. Shan Nie, Ranran Zhao, Wei Yuan Yunxiao Li, Chunting Tan participated in collecting data. All authors read and approved the final manuscript.

Ethics approval and consent to participate
The study was approved by the ethics committee of the Beijing Friendship Hospital and the written informed consent was obtained from every participant.

Consent for publication
Not applicable.