Ventilatory compensation during the incremental exercise test is inversely correlated with air trapping in COPD

Background: Air trapping and gas exchange abnormalities are major causes of exercise limitation in chronic obstructive pulmonary disease (COPD). During incremental cardiopulmonary exercise testing, actual nadir values of ventilatory equivalents for carbon dioxide (V E/VCO 2) and oxygen (V E/VO 2) may be difficult to identify in COPD patients because of limited ventilatory compensation capacity. Therefore, we aimed in this exploratory study to detect a possible correlation between the magnitude of ventilation augmentation, as manifested by increments in ventilatory equivalents from nadir to peak exercise values and air trapping, detected with static testing. Methods: In this observational study, we studied data obtained previously from 20 COPD patients who, during routine follow-up, underwent a symptom-limited incremental exercise test and in whom a plethysmography was obtained concurrently. Air trapping at rest was assessed by measurement of the residual volume (RV) to total lung capacity (TLC) ratio (RV/TLC). Gas exchange data collected during the symptom-limited incremental cardiopulmonary exercise test allowed determination of the nadir and peak exercise values of V E/VCO 2 and V E/VO 2, thus enabling calculation of the difference between peak exrcise value and nadir values of V E/VCO 2 and V E/VO 2, designated ΔV E/VCO 2 and ΔV E/VO 2, respectively. Results: We found a statistically significant inverse correlation between both ΔV E/VCO 2 (r = -0. 5058, 95% CI -0.7750 to -0.08149, p = 0.0234) and ΔV E/VO 2 (r = -0.5588, 95% CI -0.8029 to -0.1545, p = 0.0104) and the degree of air trapping (RV/TLC). There was no correlation between ΔV E/VCO 2 and forced expiratory volume in the first second, or body mass index. Conclusions: The ventilatory equivalents increment to compensate for acidosis during incremental exercise testing was inversely correlated with air trapping (RV/TLC).

Amendments from Version 1 - Figure 1 has been deleted. -Values of peak VO2 (% of predicted VO2max) and RER slightly changed after correcting data of one patient (patient no. 12) thus affecting values described in Table 1 and Table 3. This did not affect results and conclusions of our study.
- Table 2 includes now the main correlations studied between the ventilatory equivalents increments (i.e., delta VE/VCO2, delta VE/VO2) and RV/TLC, in first two rows.
- Table 2, values of correlations of RV/TLC with nadir VE/VCO2 and nadir VE/VO2 were updated and corrected for minor errors that do not change the essence of our findings.
- Table 3, the mean value of Nadir VE/VCO2 in patients who had an increment in VE/VCO2 has been corrected to the value of 37.2+5.5 with a p=0.6849. (The previous calculation did not show a significant difference either.
-Updated study raw data was uploaded -We added data of VE/VCO2 slope and VE/VCO2 line intercept as suggested, providing further insight into our cohort.

Introduction
Chronic obstructive pulmonary disease (COPD) patients often demonstrate significant exercise limitation, chiefly resulting from gas exchange abnormalities and ventilation-perfusion mismatching 1 . This situation is often compounded by hyperinflation and air trapping with dynamic hyperinflation during exercise and gradual reduction of inspiratory capacity 2 . To evaluate exercise capacity and determine the degree of exercise limitation and its mechanisms, the incremental exercise test is often applied, during which several ventilatory events occur and draw close attention.
During incremental cardiopulmonary exercise testing, the ratio of minute ventilation (V E ) to carbon dioxide output (VCO 2 ) and to oxygen consumption (VO 2 ), also known as ventilatory equivalent for carbon dioxide (V E /VCO 2 ) and oxygen (V E /VO 2 ), respectively, serve to evaluate ventilatory efficiency. The ventilatory threshold, a term coined in the context of gas exchange measurements and represents to a great extent the VO 2 at anaerobic threshold, heralds the onset of the isocapnic buffering phase, in which there is an increased contribution of anaerobic metabolism to provide the energy required for the increasing demands of exercise, accompanied by increased CO 2 production with corresponding ventilation increase.
With the accumulation of lactic acid and H + protons, as the bicarbonate reserves are decreased beyond the isocapnic buffering phase, an augmented ventilatory response starts at the ventilatory compensation point, which is disproportionate to the degree of CO 2 production, leading to V E /VCO 2 surge towards peak exercise, a phenomenon termed the ventilatory compensation phase.
The V E /VCO 2 is often increased as a result of ventilationperfusion mismatching 1,3 . The nadir V E /VCO 2 value during incremental exercise is inversely related to the degree of exercise limitation 4 , as well as the severity of airway obstruction, as determined by the values of forced expiratory volume in first second (FEV 1 ) 5,6 . The nadir value is considered to reflect ventilatory efficiency as higher values obtained at the early stages of the incremental exercise test result from frequently encountered hyperventilation. Interestingly, marked hyperinflation and reduced ventilatory capacity tend to reduce ventilatory equivalents as ventilation may become constrained. Both baseline and peak exercise V E /VCO 2 values were found to be lower in patients with more severe airway obstruction as V E is decreased relative to VCO 2 in these patients 5,6 .
Although COPD patients may terminate exercising at an early stage during the incremental exercise test due to airflow obstruction and air trapping, suboptimal effort and deconditioning, the latter of which may be associated with somewhat earlier development of metabolic acidosis, may also lead to early termination of the exercise test. Regardless of the reason causing decreased exercise performance, the ventilatory compensation phase may not be demonstrated. Ventilatory equivalents during incremental exercise may be affected by the abovementioned processes in different directions. In this exploratory study we sought to confirm the presumption that the ability to achieve ventilatory compensation in response to acidosis is related to the degree of air trapping, with the hypothesis that the ability to augment ventilation beyond the ventilatory compensation point in COPD patients is inversely correlated with the degree of air trapping.

Study design and subjects
A retrospective analysis of data obtained from the medical records of COPD patients who underwent incremental cardiopulmonary exercise testing in the pulmonary function laboratory of the Institute of Pulmonary Medicine in the Hadassah Medical Center, Jerusalem, Israel between June 2010 and August 2016, and in whom whole body plethysmography was performed concurrently, as part of their routine clinical evaluation. The number of patients was determined by the availability of incremental exercise test and plethysmography data that was performed concurrently. Subjects were previously diagnosed with COPD based on post-bronchodilator spirometry showing an FEV 1 to forced vital capacity (FVC) ratio (FEV 1 /FVC) of ≤0.7, regardless of the value of FEV1 1 (according to 2017 GOLD guidelines). Comorbidities such as obesity, cardiac disease and use of beta blockers were noted, but were not a cause to exclude patients from the study. The research ethics board (REB) of the Hadassah Medical Organization (protocol No. 0040-16-HMO) approved data collection and waived the need for informed consent, as this research was retrospective, did not affect patient management, or involve collecting biospecimens.

Procedures
Pulmonary function tests and exercise tests were performed by technicians in the presence of a physician during exercise tests. Whole-body plethysmography was performed using a commercially available body plethysmograph (Elite series, MedGraphics). Spirometry performance and slow vital capacity determination were followed by assessment of lung volumes, which was performed by direct measurement of thoracic gas volume (TGV), from which the residual volume (RV), total lung capacity (TLC) and RV/TLC ratios could be calculated as the primary method to estimate gas trapping. Additionally, we determined the ratio of inspiratory capacity (IC) to total lung capacity, IC/TLC, which has been proposed to reflect lung expansion as a result of reduced lung recoil in emphysema 7 . Using the same lung function system, transfer factor of the lung for carbon monoxide (TL CO ) was measured by the single breath method.
Patients performed incremental symptom-limited cycle ergometry connected to a metabolic system with cycle ergometer (Ultima CardiO 2 , MedGraphics). For each patient, baseline measurements were obtained during the resting period of 1 minute, after which the patient would start cycling at a constant rate of 50-60 rpm. Following a 1-2 minute warm-up period of unloaded exercise and with VO 2 and VCO 2 reaching a plateau, a ramp protocol was started with a workload increase of 15-25 watt per minute, depending on the predicted maximal VO 2 and the general state of the patient. The test was terminated at exhaustion, when the patient could not keep the required cycling pace or asked for exercise termination. Routine precordial 12-lead electrocardiogram monitoring, continuous measurements of V E , VO 2 , VCO 2 (averaged every 10 seconds), heart rate and finger arterial pulse oximetry were recorded. The peak VO 2 and the ventilatory threshold were both noted, as well as nadir ventilatory equivalents (V E /VCO 2 and V E /VO 2 ). Values of V E /VCO 2 and V E /VO 2 obtained at the termination of loaded exercise (peak exercise) enabled calculation of the difference between peak exercise value and nadir values of V E /VCO 2 and V E /VO 2 (designated ΔV E /VCO 2 and ΔV E /VO 2 , respectively).

Statistical analysis
Statistical analysis was performed using the GraphPad Prism 3.0 program software and Spearman's test was used for correlation analysis, with calculation of the Pearson correlation coefficient (r) between measured parameters. Correlations were considered of statistical importance if two-tailed p value was <0.05. Unpaired t-test was used to compare between groups, and a two-tailed p value < 0.05 was considered statistically significant

Standards of reporting
This manuscript is compliant with the STROBE guideline.

Results
In total, 20 patients were included in our analysis with a mean ± SD age of 63 ± 10 years (range 37-77). None of the patients had cardiac failure. Two patients were receiving beta blockers. Table 1 summarizes clinical data of these patients. We found a statistically significant inverse correlation between both ΔV E /VCO 2 (r = -0.5058, 95% CI -0.7750 to -0.08149, p = 0.0234) and ΔV E /VO 2 (r = -0.5588, 95% CI -0.8029 to -0.1545, p = 0.0104), with the degree of air trapping as estimated by the RV/TLC ratio (Table 2). In this cohort, there was a significant correlation between Nadir VE/VCO2 and VE/VCO2 slope (r=0.82, 95% CI 0.59 to 0.92, p<0.0001).
In contrast to increments in ventilatory equivalents, there was no correlation between the nadir values of the ventilatory equivalents and RV/TLC (Table 2). Interestingly, inspiratory capacity to total lung capacity ratio (IC/TLC), which reflects pulmonary expansion in emphysema, did not show a correlation with either ΔV E /VCO 2 or ΔV E /VO 2 ( Table 2).
We examined possible correlation between ΔV E /VCO 2 and other static parameters of pulmonary function or findings on exercise testing (summarized in Table 2). Notably, we found no significant correlation between neither ΔV E /VCO 2 nor nadir V E /VCO 2 and peak VO 2 (% of predicted VO 2 max) or FEV 1 (% of predicted). There was a correlation between ΔV E /VCO 2 and peakVO 2 (in ml/kg/min) which just achieved a significance at p= 0.0462, ( Table 2). As expected 1 , nadir values of V E /VCO 2 correlated with TL CO (r = -0.5281, 95% CI -0.7923 to -0.09708, p = 0.0201); however, we did not find a correlation between ΔV E /VCO 2 and TL CO (Table 2).
In an attempt to identify factors related to the lack of increment of V E /VCO 2 besides those related to air trapping, we compared values of parameters related to static pulmonary functions and those related to performance of exercise in patients with and without increment in V E /VCO 2 (i.e., ΔV E /VCO 2 = 0) (Table 3). We noted a statistically important difference in peak RER, VO2 (in ml/kg/min), peak WR; However, FEV 1 , IC/TLC, TL CO , peak VO 2 (as % of predicted VO 2 ), and nadir V E /VCO 2 were not different between these two groups. RV/TLC had a statistically important difference between the two groups. However, RV/TLC as well as peak VO 2 adjusted to body weight are inherently derived from the parameters that the correlation study had examined and therefore, this difference is not surprising. There was also a significant difference in the values of peak exercise end-tidal PCO 2 between the two groups with lower values of end-tidal PCO 2 in patients who had ΔV E /VCO 2 >0. Interestingly, patients who did not have a V E /VCO 2 increment at peak exercise, had a trend towards a higher peak exercise end-tidal PCO 2 than their baseline values (37.3±8.1 vs 32.5±6.7) although the difference did not achieve a statistically important difference (p=0.1651). Moreover, the value of V E /VCO 2 y-abcissa intercept (i.e., V E intercept) in patients who did not have an increment of V E /VCO 2 (i.e., ΔV E /VCO 2 =0) was significantly higher than in those who had a ΔV E /VCO 2 >0. Variables that could be affected by ventilatory mechanical constraints during exercise, the breathing reserve and peak tidal volume to vital capacity ratio (Vt/VC) were not different between the two groups.

Discussion
Inefficient ventilation in patients with COPD and particularly emphysema is often evidenced by elevated nadir V E /VCO 2 . The elevation of nadir V E /VCO 2 may be related to the extent and severity of emphysematous changes in the lungs. It has been shown that in patients with mild to moderate airflow decrease, the values of nadir V E /VCO 2 correlated with the percentage of low attenuation areas on computerized tomography 8,9 as well as the decrease in pulmonary perfusion, as estimated by the inert gas rebreathing method 9 . Interestingly, in smokers without COPD, the degree of CO diffusion decrease correlated with elevation of nadir V E /VCO 2 during the incremental exercise test 10 , again pointing to a strong relation between impaired ventilation efficiency and gas exchange ability.
Notably, significantly increased ventilatory constraints, as occurs in marked air trapping, decrease V E , thus leading to lower nadir V E /VCO 2 in more severe airway obstruction 2,6 .
In this study we examined the ability of COPD patients during incremental exercise testing to increase ventilation towards peak exercise, focusing on the increase in V E /VCO 2 and V E /VO 2 that occur following the ventilatory compensation point. We quantified this phenomenon measuring the difference between the values of V E /VCO 2 and V E /VO 2 at nadir and peak exercise obtained post the ventilatory compensation point, which we termed ΔV E /VCO 2 and ΔV E /VO 2 , respectively. We found that the ability to increase ventilation during incremental exercise testing in response to metabolic acidosis, as normally occurs in higher exercise intensities beyond the ventilatory compensation point, was diminished in COPD patients and in correlation with the severity of air trapping as represented by RV/TLC obtained at rest. Furthermore, this may implicate a difficulty in coping with the ventilatory demands of metabolic acidosis resulting from states such as acute renal failure or septic shock, thus ΔV E /VCO 2 (and seemingly ΔV E /VO 2 ) may allow assessment of the ability of COPD patients to withstand such metabolic challenges. Limited ventilatory compensation is clinically significant as it has been shown that people with COPD have higher arterial PCO 2 and lower pH in response to blood lactate elevation during exercise, compared to people without COPD 11 . In our cohort, peak end-tidal PCO 2 in patients without V E /VCO 2 increment had a non-significant trend towards higher values than baseline end-tidal PCO 2 thus pointing to a limited ability to compensate for acidosis. Our findings also show that patients who did not have an increment in V E /VCO 2 had higher average V E intercept value of the V E /VCO 2 slope than patients with ΔV E /VCO 2 >0, thus suggesting a role for the presence of higher dead space breathing 5 in patients who do not increase ventilatory equivalents and produce ventilatory compensation.
One way in which pulmonary hyperinflation can be expressed is by the increased ratio of RV/TLC, which reflects air trapping and is associated with airway narrowing. Alternatively, IC/TLC may also be decreased because of lung expansion associated with emphysematous changes and reduced lung recoil. A recent study showed that either increased RV/TLC or decreased IC/TLC or both simultaneously can be present, with airway narrowing on imaging being associated with increased RV/TLC and emphysema associated with deceased IC/TLC 7 . Therefore, it seems that the lack of increments in ventilatory equivalents in our study is correlated with air trapping but not to the mere presence of emphysema.
One may argue that the limited rise of ventilatory equivalents beyond the ventilatory threshold during the incremental exercise test may reflect sub-optimal exercise performance; however, we think that the lower respiratory exchange ratio at peak exercise may represent another manifestation of the decreased ventilatory response.
Although both ΔV E /VO 2 and ΔV E /VCO 2 seem to show similar correlation to RV/TLC and utility in estimating the ventilation response to acidosis under these conditions, a further study with larger numbers of patients may help distinguish the different roles of these parameters.
This study has several limitations. The retrospective design and the small number of patients included are important limitations, in particular when comparing two subgroups. However, these limitations may be of lesser significance when searching for correlations. Measured increment of ventilatory equivalents in a significant number of patients was nil, with a value =0. This may point to an important limitation of this measurement. Future studies should be designed to utilize a lower ramp than the ramps that were used in our patients studied retrospectively, to allow time for patients to adjust and generate ventilatory compensation, something that may be more difficult to do exercising at high work rates. Some patients were obese, which may have caused mild restriction, somewhat contributing to air trapping. However, we didn't find a correlation between body mass index (BMI) and either ΔV E /VCO 2 or ΔV E /VO 2 . Unfortunately, dynamic hyperinflation, a well-known phenomenon that is associated with ventilatory constraints during exercise in COPD, was not assessed by performing repeated inspiratory capacity maneuvers during exercise and therefore, we cannot describe the relation between dynamic hyperinflation (i.e. changes in IC/TLC) and ventilatory equivalents increment in this cohort. However, dynamic hyperinflation developing during exercise is associated with changes in Vt/VC at peak exercise with lower value than the expected portion of 50-60% at peak incremental exercise 12 as dynamic hyperinflation develops with increasing work rate during exercise performance 2 . The average Vt/VC at peak exercise of all patients in our cohort was 46±8%, without a statistically meaningful difference in peak exercise Vt/VC between those who had V E /VCO 2 increment and those who didn't (Table 3). Therefore, it is unlikely that development of dynamic hyperinflation can differentiate between ventilatory equivalent increment patterns during exercise in our patients.
In conclusion, ventilation augmentation during incremental exercise testing at exercise intensities beyond the ventilatory compensation point in COPD patients was diminished in correlation with the severity of air trapping. Future studies will confirm the clinical usefulness of measuring ΔV E /VCO 2 and ΔV E /VO 2 , including their potential role as serially measured physiologic parameters to assess effects of interventions (e.g. pulmonary rehabilitation) in COPD.

References
This project contains the following underlying data: -nadir to peak VE-VCO2 during exercise in COPD.tab (demographic information and parameter measurements for each patient).
Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication). 1.

5.
6. They were interested in air trapping and the effect on exercise performance, specifically trying to better understand the changes in Ventilatory Equivalents for Oxygen and Carbon Dioxide during incremental exercise tests on a cycle ergometer (nadir value to peak value). Further, they relate these changes to resting pulmonary function testing. The study was retrospective. They conclude that "ventilatory equivalents increment to compensate for acidosis during incremental exercise testing was inversely correlated with air trapping (RV/TLC) and may be a candidate prognostic biomarker." The study is well written.

Major Issues
The study would have been greatly assisted by obtaining inspiratory capacity (IC maneuvers) during exercise to document dynamic hyperinflation.
The study is retrospective, is a very small group, and was collected across 7 years (2010-2016, about 2-3 subjects per year), meaning that there is great potential for changes in devices, calibration, technician training, etc. that may have affected the PFT and CPET measurements.
The GOLD III and IV subjects are very poorly represented (Gold III (20%) and Gold IV (5%)), i.e. most subjects (75% were Gold I or II). Per the Neder and Thirapatarapong references in your paper, Gold I-II have lower ETCO2 and higher VE/VCO2 , and reflecting less efficient ventilation GOLD III-IV have higher ETCO2 and lower VE/VCO2 . reflecting mechanical limitation This paper is also hampered by so many of the changes in VE/VCO2 (10) and VE/VO2 (7) from peak-nadir were nil, i.e. = "0" There is no analysis of VE/VCO2 slope or intercept reported in the paper or data spreadsheet.
No blood gases (to document acidosis or lactate levels) or end tidal CO2 values to document 6. 7. 1.

9.
No blood gases (to document acidosis or lactate levels) or end tidal CO2 values to document changes in ventilatory efficiency. Importantly, there is no discussion of the determinants of VE/VCO2, i.e. VE/VCO2 i.e., = k /( ) * ( 1-) and how these two variables interact in the PaCO2 VD/VT current study.
COPD patients are less likely to protect their arterial pH and arterial CO2 during exercise (Casaburi, et al, Am Rev Respir Disease, 143: 9-18, 1991 ) See figure # 8. There is no discussion of this important feature of COPD patients in the manuscript.

Minor Issues
Introduction, page 3, First Paragraph, line 2. "Effort" should likely be exercise or activity.
Introduction, page 3, First Paragraph, line 3+4. Here you mention gas exchange abnormalities and ventilation perfusion mismatching, but in paragraph 4, lines 3 you also discuss deconditioning. All three mechanisms appear to be important and should be mentioned in both places.
Introduction, page 3, Third Paragraph, line 6 and 14. Reference 5 and 6 would both seem to contribute to the concepts here, and should both be quoted.
Methods, Paragraph 1, Study Design and Subjects, lines 1-5. How are the authors confident that all subjects with COPD and full PFTs studied in the laboratory were identified for the current manuscript?
Methods, Paragraph 1+2, Procedures. Were there any changes in PFT or CPET equipment, calibration, or technician protocols during the 7 years of data collection that might affect spirometry, lung volumes, gas transfer or CPET variables?
Methods, Paragraph 2, Procedures. 15-25 watts appears to be a very large ramp increment in patients with moderate to severe COPD. Please explain why these very large numbers were chosen, and how that may have affected the findings in the study (i.e. short ramp times may decrease time for ventilatory compensation of metabolic acidosis). Table 1 -Consider including peak VO2/kg/kg/min, peak WR, peak RER, MMVV-VE (L/min), and end tidal CO2 values for the various subjects (and in the accompanying spreadsheet). The spreadsheet accompanying the article is very helpful and the authors should be A large number of subjects (~ 70%) had a VE/MVV commended for making the data available. below 85% or > 15 liters/minute (MVV-VE) at end exercise, indicating a lack of ventilatory limitation in most of these subjects.  Table 3 addresses it, it would appear that the authors have neglected to mention this as a limitation of the entire study in the discussion.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed. We would like to thank Prof. Stringer for his valuable comments. This a retrospective study with small number of patients, and we certainly acknowledge the inherent significant limitations, which are referred to in the Discussion section We consider this work as a preliminary/exploratory study, and as that it has well acknowledged limitations and implications. We added the term "exploratory study" in the abstract and the introduction section to allude to that fact.
We would like to address the concerns raised by the Reviewer.
Re Major comments: 1. At the time of performance of these CPET studies, we did not routinely perform repeated inspiratory capacity maneuvers during exercise to allow detecting dynamic hyperinflation, because of lack of both software add-on and experience in performing theses maneuvers. Notably, dynamic hyperinflation is associated with changes in tidal volume/vital capacity ratio (Vt/VC) towards peak exercise with lower ratios of Vt/VC as dynamic hyperinflation is more evident during exercising at higher workloads. All in all there was a very little decrease in Vt/VC obtained at peak exercise, with average values of 46% 8% (Table 1). + We now added to Table 3 a comparison of Vt/VC values of patients who had an increment in VE/VCO2 as opposed to patients who did not (deltaVE/VCO2=0). We found that the peak exercise Vt/VC did not differ between the two groups. Despite the limitations of this method, we conclude that dynamic hyperinflation probably did not play a key role in preventing VE/VCO2 increment from nadir values to peak exercise in our patients. The last paragraph in the Results section was altered to describe the added data and findings, and also description of findings investigating possible ventilatory mechanical constraints. Additionally, the Discussion alluded to the possible use of Vt/VC as a way to evaluate mechanical constraints on ventilation (paragraph before last).
2. Throughout the years 2010-2016 we used the same CPET and plethysmography devices described in the methods and applied the same calibration methods as recommended by the manufacturer.
3. As patients have more severe obstructive abnormality (GOLD III-IV more than GOLD I-II), they have lower ventilation, with reduced VE/VCO2 and VE/VO2 values. This has been referred to in the Introduction section, paragraph 3. Certainly, if we had bigger number of patients and all GOLD classes of patients were represented equally, we may have been able to deduce important differences between these classes. Data on end-tidal PCO2 has now been provided.
4. The fact that many patients had no increment in ventilatory equivalents from nadir to peak exercise (i.e., deltaVE/VCO2=0) is a limitation of the measurement. As mentioned in the study limitations paragraph, this may as well be explained by suboptimal effort of the patients or the ventilatory limitations of these patients. We referred to this issue now in the Limitations paragraph.
5. Data of VE/VCO2 slope was added. There was a significant correlation between Nadir VE/VCO2 and VE/VCO2 slope (R=0.82 95%CI 0.59-0.92, p<0.0001). Data of VE/VCO2 intercept was also added. (Represented in all 3 Tables). 6. We do not take arterial blood gases during routine clinical CPET performance to keep the test acceptable for our patients. This may be considered in a future prospective study if approved by ethical board. Alternatively, there are minimally invasive methods to allow blood lactate measurements which we do not have available in our clinical service. Regarding end-tidal CO2, we added the end-tidal PCO2 values at rest and at peak exercise. (See further, response to comment 7). 7. The reviewer brings our attention to the findings from a previous study which shows that COPD patients have higher arterial PCO2 and lower pH in response to blood lactate elevation during exercise, compared to people without COPD. This is somewhat expected, as with the presence of ventilatory constraints, these patients are not able to develop sufficient ventilatory compensation during exercise. In this study we did not perform these measurements; However, end-tidal PCO2 did not decrease towards peak exercise, in the patients who did not have VE/VCO2 increments, thus probably in agreement with the previously published findings highlighted by the reviewer. We added a reference to the suggested paper in our discussion. Note that at the end of the discussion (paragraph 2), we refer to the possible implication of limited ability of COPD patients of coping with (paragraph 2), we refer to the possible implication of limited ability of COPD patients of coping with metabolic acidosis from other reasons (e.g., diabetic acidosis) when they are not able to generate ventilatory compensation.
Re Minor Comments: 1. We agree with the reviewer and "effort" was changed to "exercise" as suggested.
2. We agree with the reviewer regarding the importance of deconditioning as a reason for reduction in exercise capacity in COPD patients. However, the opening sentences were concentrated around describing the pulmonary mechanisms in exercise limitation in these patients. We do refer to deconditioning in the last paragraph of the Introduction.
4. Citations of References 5 & 6 were joined twice in the Introduction (Paragraph 3) as suggested.
5. To identify patients and include them in the study, we looked into the CPET system software and identified all the patients that underwent CPET between the years 2010-2016 (on the same CPET device) and had a background diagnosis of COPD. Then from those patients we included only the patients that underwent body plethysmography concurrently in our laboratory, on the same day of CPET or within few days close to the date of CPET performance.
6. As stated above, there were no changes in the equipment during the 7 years. 7. Ramp increment is routinely chosen to allow for predicted WR to be achieved within 8 minutes of exercise. Only one patient exercised at a ramp of 25W/min. All other patients had a ramp of 15-20W/min. We agree with the reviewer that lower ramps may be warranted for more debilitated patients, which we often offer our patients who report low activity level. However, most patients do not handle a longer period of loaded exercise, because of other factors such as claustrophobia and mental fatigue.
Regarding the effect of short ramp times on ventilatory compensation, we agree with the reviewer that higher ramps may shorten the time interval of ventilatory compensation and make it harder to evaluate the ventilation increment during the respiratory compensation phase but we doubt that this might affect the findings of our study. Perhaps, a future prospective evaluation, with exercise tests performed at lower ramps may allow a better evaluation of ventilation increment towards peak exercise. A note on that is added in study limitations. 8. We considered the suggestions of the Reviewer regarding the expanding of Table 1 to include data already shared with the readers in the accompanying research data file which is already available at data sharing site. Regarding peak VO2 in ml/min/kg, peak WR, peak RER, and MVV-VE, were added to Table 1 in the body of the article, and also in Table 3 which shows comparison between those who did not have VE/VCO2 increment and those who did. Interestingly, VO2 in ml/kg/min was found to be different between the two groups, unlike VO2 expressed as % of predicted VO2max.
9. The absence of ventilatory equivalent increment (i.e., delta VE/VCO2=0) has now been addressed in the study limitations paragraph.

Gabriele Valli
Emergency Department, San Giovanni Addolorata Hospital, Rome, Italy In the present study, the authors investigate a possible correlation between the magnitude of ventilation augmentation (manifested by increments in ventilatory equivalents from nadir to peak effort) and the lung air trapping indexes measured at rest. The authors analyzed retrospectively data from 20 symptom-limited incremental exercise test performed by 20 COPD patients. Patients underwent to CPET as part of a routine evaluation. The differences between nadir to peak exercise of VE/CO were correlated to air trapping parameters measured at rest in a plethysmography. A positive correlation was found between nadir to peak exercise VE/CO and VE/VO and the degree of air trapping expressed as RV/TLC. The authors concluded that ventilatory equivalents increment (above anaerobic threshold) could be inversely correlated with air trapping and it may be a candidate prognostic biomarker.
Although the paper is well written I have some comments to do: The authors choose to calculate ventilatory equivalent subtracting the values at peak exercise to the values measured at the nadir, assuming that the reduced increment of ventilatory equivalent after the minimum values could be mainly referred to a reduced ventilatory capacity of the subject due to air trapping measured at rest. That could be right but it is not completely correct. The simple correlation measured by the authors does not completely clarify the reasons for the interruption of the test by patients, above or below the respiratory compensation point (RCP). Some studies had demonstrated a reduced ventilatory compensation capacity in patients with a more severe emphysema i.e Paoletti , 2011 nerveless it should take in account that several metabolic et al alterations occur after the anaerobic threshold that are independent to the ventilatory mechanics and that are more closely related to physical conditioning of the patient. After the anaerobic threshold, peak ventilatory response could be related to several different causes: the exercise intensity and the power increments, the anaerobic reserve of the subject, the muscular O extraction capacity, the amount of metabolic acidosis, the increase in catecholamine levels, the developed or not of dynamic hyperinflation, the ventilatory chemiosensitivity of the carotid bodies, some cardiovascular alteration as an opening of PFO or the develop of pulmonary hypertension, even the individual motivation to complete the exercise and the subject propensity to perform that specific exercise. It is difficult to understand which is the real cause of exercise interruption and this could significantly affect the amount of nadir to peak ventilatory equivalent difference. The difference between nadir to peak ventilatory equivalents does not seems to be a good estimate of patient's ventilatory compensation during the exercise. 3.
It should be noted that, in the regression analysis there are half of the patients that did not increase ventilatory equivalents, which means that they stop the exercise at RCP or immediately after. Table  3 show a significant difference in peak RER between the patients who have not shown increments in VE/CO and the ones who have continued to increase ventilation, suggesting that the two groups reach different intensity exercise domain and probably are different populations (i.e. RC/TLC significantly different). A more complete table, that includes Watt Max, VO peak in ml/kg, VO , VE/MVV and FEV /FVC of the two groups could be helpful. Unfortunately it is impossible to know why the subject who showed no increase in VE/VCO , interrupt the exercise immediately after the reach of RCP. To put all the results in the same regression appear as an hazard, there are too many unpredictable sources of error. The authors must provide consistent evidence that the observed correlation is not only apparent. The estimated coefficients of the linear regression and the relative interval of confidence should be calculated. An estimation of the standard error of regression coefficient R should be provide together with the 95% confidence band for R. A residual analysis could help to understand how good is the correlation and if the assumption of normality of the distribution is preserved.
The variable chosen by the authors to represent the exercise ventilatory compensation of the patients is not the one universally accepted and utilized for this propose. The authors should clarify why they have not utilized, for instance, the VE/CO . In a recent paper Gargiulo P. and coworkers demonstrated that the analysis of VE/CO and its intercept on the abscissa (VE ), that represent the ventilatory requirements when pulmonary gas exchange is null, represents better the ventilatory efficiency (Gargiulo P et al. 2014 ). In COPD patients VE correlate with the VD and Neder and colleagues recently showed that increases in VE better reflected the progression of functional impairment form mild to end-stage of COPD (Neder JA et al. 2015 ). I suggest to test also these parameters in you analysis that are more clearly related to pulmonary efficiency and that are less dependent from the individual motivation of the patients to tolerate the effort.
The authors conclude the abstract suggesting that the reduced ventilatory increment in VE/VCO from rest to peak exercise could be a prognostic biomarker. This assumption is not clarified in the text, no data was provided to support this statement and the assumption remains speculative. Minor comments:

Abstract
Line 4, p 1_ "During incremental cardiopulmonary exercise testing, ventilatory equivalents for carbon dioxide….. may be difficult to identify in COPD patients because of limited ventilatory compensation . This is not correct. Ventilatory equivalents are always identifiable, in COPD patients could be capacity" difficult to identify the respiratory compensation point, the point at which it could be observed an hyperventilatory response to late-exercise acidosis.
. DV /VCO abbreviation is usually utilized to indicate VE/CO Line 19, p 1_ "DV /VCO and DV /VO " and could be confounding, I suggest to change it with a new acronym that specifies that it is the slope difference between the peak and nadir values.
the statement is speculative and not Line 28, p 1_ "…and may be candidate prognostic biomarker" slope the statement is speculative and not Line 28, p 1_ "…and may be candidate prognostic biomarker" investigated in the study.

Introduction
"The ventilatory threshold (VT)…. ". I suggest to utilize a different abbreviation (i.e. AT or Line 26, p 3 _ LT), "VT" could be confused with Tidal Volume.

Methods
Lines 1-10, p 4 _ The authors "Patients performed incremental symptom-limited cycle ergometer…" mentioned a 1-2 minute of warm-up period at 0 watt. 0 watts assumes that an ergometer with pedal assistance has been used. What kind of cycle ergometer was used? did you mean a period of free pedaling at the minimum of watts? Have you estimated the amount of the minimum load of your cycle-ergometer? 1-2 min of warm up could be correct with a real 0 watts ergometer but it could be a too short period if the minimum load provide by the ergometer was 20W or greater.
Line 10, p4_ Can you please better define "exhaustion"? " the test was terminated at exhaustion"

Results
The table should be integrated with the absolute value of VO in ml/kg, VO in ml/kg, Table 1. Watt Max and RER at peak exercise. I suggest to put SpO rather than the desaturation at peak effort. An explication of how the authors calculated BR should be added to Methods. The percentage of tests in which the respiratory compensation point was identifiable should be added somewhere. DVE/CO should be change with some other abbreviation.
Lines 5-16, p4 _ a p = 0.0234 looks " DV /VCO (r= -0.5058, 95% CI -0,7750 to -0.08149, p =0.0234)…" quite significantly, nerveless the adaptation of the regression appear very low, it seems that R is not more than 0.25. I suggest to add to your plot the regression coefficients with the respective standard errors. It could be useful analyzed the residual plot in order to increase the adaptation.
Lines 19-20, p4. "Notably, we found no significant correlation between neither DVE/VCO nor nadir . Do you mean the absolute values or % predicted? VE/VCO and peak VO or FEV . " . The table shows a significant difference in RER peak between the two population analyzed. The Table 3 patients with a significant increase in probably achieve a greater effort intensity. As the entire DVE/VCO correlation depends on the patients who sustain the effort beyond their VEVCO , it is likely that the correlation is no longer detectable in the more severe patients. I would like to know the differences between the two groups in VO expressed as ml/kg, Watt Max, BR%, Heart Rate Reserve (220-age), VO at peak exercise, FEV in mL and FEV /VC.

Discussion
Lines 17-19, p 5. "In this study we examined the ability of …. Focusing on the increase in the V /VCO . This is not correct. You have not measured any slope. You measure just a difference. and V /VO " Lines 14-19. P 6.
. This statement is entirely speculative, beyond "Therefore the inability to compensate…" the scope of the study and not supported by any data that can be highlighted in the paper. the scope of the study and not supported by any data that can be highlighted in the paper.
You are comparing a static "Therefore, it seems that the lack of increment…." measurement with a dynamic alteration that occurs in the moment of maximum stress. The problem is that you have not measured any indexes of dynamic hyperinflation. Your conclusion cannot be proven.
Lines 34-35, p. 6 "however, we think that the lower respiratory exchange ratio at peak exercise Unfortunately it does not stand represents another manifestation of the decreased ventilatory response". out easily if the patient has stopped due to actual exhaustion or to other reasons. Especially in a retrospective study.
Lines 49-50, p. 6 " You should take "Some patients were obese, which may have caused mild restriction… in account that BMI could largely affect the VCO body storage and, as a consequence, the RCP detection or VE/VCO values. Is there any differences between the two groups (No Nadir to Peak increment vs patients with nadir to peak increment) in BMI? Can you provide this data?
No competing interests were disclosed.

Competing Interests:
Reviewer Expertise: Exercise Physiology -Hypoxia -Ventilatory Response to Exercise and High Altitude -Critical Care and Emergency Medicine I confirm that I have read this submission and believe that I have an appropriate level of expertise to state that I do not consider it to be of an acceptable scientific standard, for reasons outlined above.
Author Response 04 Mar 2020 , Hadassah-Hebrew University Medical Center, Jerusalem, Israel Samir Nusair Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com