Exercise Ventilation in COPD: Influence of Systolic Heart Failure

ABSTRACT

Systolic heart failure is a common and disabling co-morbidity of chronic obstructive pulmonary disease (COPD) which may increase exercise ventilation due to heightened neural drive and/or impaired pulmonary gas exchange efficiency. The influence of heart failure on exercise ventilation, however, remains poorly characterized in COPD. In a prospective study, 98 patients with moderate to very severe COPD [41 with coexisting heart failure; ‘overlap’ (left ventricular ejection fraction < 50%)] underwent an incremental cardiopulmonary exercise test (CPET). Compared to COPD, overlap had lower peak exercise capacity despite higher FEV₁. Overlap showed lower operating lung volumes, greater ventilatory inefficiency and larger decrements in end-tidal CO₂ (PETCO₂) (P < 0.05). These results were consistent with those found in FEV₁-matched patients. Larger areas under receiver operating characteristic curves to discriminate overlap from COPD were found for ventilation (E)-CO₂ output CO₂) intercept, E-CO₂ slope, peak E/CO₂ ratio and peak PETCO₂. Multiple logistic regression analysis revealed that CO₂ intercept ≤ 3.5 L/minute [odds ratios (95% CI) = 7.69 (2.61–22.65), P < 0.001] plus E-CO₂ slope ≥ 34 [2.18 (0.73–6.50), P = 0.14] or peak E/CO₂ ratio ≥ 37 [5.35 (1.96–14.59), P = 0.001] plus peak PETCO₂ ≤ 31 mmHg [5.73 (1.42–23.15), P = 0.01] were indicative of overlapping. Heart failure increases the ventilatory response to metabolic demand in COPD. Variables reflecting excessive ventilation might prove useful to assist clinical interpretation of CPET responses in COPD patients presenting heart failure as co-morbidity.

KEYWORDS:

• Carbon dioxide
• cardiopulmonary exercise testing
• dyspnoea
• exertion
• gas exchange

Introduction

Exercise intolerance is characteristically multi-factorial in chronic obstructive pulmonary disease (COPD) (1). However, mechanical ventilatory constraints assume a more prominent role as disease progresses as suggested, for instance, by a low sub-maximal ventilation (E) to CO₂ output (CO₂) slope, a high E–CO₂ intercept and increased end-exercise end-tidal pressure for CO₂ (PETCO₂) (2–6).

Systolic heart failure, in particular, is a common and disabling co-morbidity of COPD (hereafter termed ‘overlap’) (7,8) with a potential to influence on these key exercise responses. In fact, heart failure may increase exercise ventilation in COPD due to heightened neural drive and/or poor pulmonary gas exchange efficiency (9,10). Thus, COPD patients with coexisting heart failure may present with increased E–CO₂ slope and a low E–CO₂ intercept (2,11,12). End-tidal pressure for CO₂ (PETCO₂) might be lower in overlap secondary to alveolar hyperventilation (i.e., low ‘mean’ alveolar PCO₂) and/or increased ventilation of poorly perfused alveoli (9,13). Unfortunately, the few retrospective studies which looked at the influence of heart failure on ventilatory responses to exercise in COPD compared COPD versus heart failure (2) or heart failure versus overlap (11,12), i.e., they did not contrast COPD patients with and without heart failure. Moreover, those studies did not control for heterogeneities in disease diagnosis, treatment and clinical stability at the time of testing (2,11,12). The influence of heart failure on exercise ventilation in COPD, therefore, remains to be adequately characterized in a prospective study involving stable patients under standardized treatment for both COPD and heart failure.

We therefore set such a study in which we systematically investigated the influence of heart failure on cardiopulmonary exercise test (CPET) responses reflecting excessive ventilation and/or impaired gas exchange efficiency in COPD (2–6). We reasoned that if those abnormalities were more frequently found in overlap than COPD, they might eventually prove useful to assess the contribution of heart failure to exercise intolerance, to determine the effects of interventions aimed to improve ventilatory efficiency (e.g., sildenafil, whole-body and inspiratory muscle training) (14–16), and to assist on risk stratification [as excessive ventilation is a negative prognostic marker in both heart failure (17) and COPD (18)].

Methods

Subjects

One-hundred and fifty-three patients were consecutively referred for cardiopulmonary assessment jointly performed by the same respirologists and cardiologists in 3 outpatient clinics dedicated to the care of COPD and/or heart failure (Brazil and Canada) from 01 April 2012 to 30 November 2014. Disease treatment was optimized carefully before study entry and patients underwent CPET only after an agreement had been reached between respirologists and cardiologists regarding disease stability. As outlined in Figure 1, 98 sedentary males with COPD [post-bronchodilator FEV₁/forced vital capacity (FVC) ratio < 0.70] and a previous history of cigarette smoking (at least 10 pack-years) underwent CPET. Forty-one patients presenting with echocardiographic left ventricular ejection fraction ≤ 45% in combination with symptoms indicative of heart failure (such as orthopnoea, paroxysmal nocturnal dyspnoea, fatigue, peripheral oedema, nocturia more than twice a night, or any combination of these symptoms) comprised the overlap group (19). Regarding heart failure treatment, 24 patients were on the maximal recommended doses of carvedilol (50 mg/day if weight <85 kg and 100 mg/day if weight >85 kg) with the remaining patients receiving at least 25 mg/day. Daily mode doses for the most frequently used medications in each class were: captopril (150 mg), losartan (125 mg), spironolactone (50 mg) and furosemide (120 mg). Participants were asked to report chronic dyspnoea [modified (1–5) Medical Research Council (MRC) scale score 2–4] and intolerance to physical exertion (NYHA class 2 or 3). Main exclusion criteria included: long-term O₂ therapy, recent (within a year) rehabilitation, osteomuscular limitation to cycling, type I or non-controlled type II diabetes mellitus, and peripheral arterial disease associated with claudication. All subjects gave written informed consent and the study was approved by the appropriate Medical Ethics Committee (Sao Paulo Hospital, Hospital de Clinicas of Porto Alegre and Kingston General Hospital).

Figure 1. Study flow chart.

Measurements

Lung function

Spirometry, transfer factor and static lung volumes were evaluated (1085 ELITE D™, Medical Graphics). Resting blood gases were obtained from samples from the radial artery.

Cardiopulmonary exercise testing

Patients performed a symptom-limited incremental CPET on an electronically braked cycle ergometer using the Vmax229d System (SensorMedics). The rate of work rate increment was individually selected according to reported exercise tolerance (typically 5–10 W). Oxygen uptake (O₂, L/minute), carbon dioxide output (CO₂, L/minute), respiratory exchange ratio (RER, CO₂/O₂), minute ventilation (E, L/minute), end-tidal partial pressure for carbon dioxide (PETCO₂, mmHg), tidal volume (VT, L) and breathing frequency (f, cycles/minute) were averaged at 20-second intervals. Peak E was also expressed relative to the estimated maximal voluntary ventilation [MVV (L/minute) = FEV₁ × 35]. E/CO₂ nadir was the lowest test data point. Slope and intercept of the linear E–CO₂ relationship were determined by linear regression (5,6). Serial inspiratory capacity (IC, L) manoeuvres tracked the operating lung volumes (1). Lactate threshold was estimated by the gas exchange method (V-slope). Arterial oxygen saturation was measured non-invasively by pulse oximetry (SpO₂, %). Breathlessness and leg effort scores were rated according to the 10-point Borg category ratio scale.

Statistical analysis

Values are reported as mean ± standard deviation (SD) unless otherwise specified (IBM® SPPS® Statistics version 22.0.0.0). Overlap and COPD patients were contrasted by non-paired t, Mann–Whitney test or a χ² test for differences in proportions. Receiver operating characteristic (ROC) curve analyses selected the optimal threshold values to differentiate overlap from COPD based on CPET responses. Univariate and multivariate binary logistic regression analyses evaluated the discriminative value of CPET variables using ‘overlap’ as the outcome. Results were expressed as odds ratios with the corresponding 95% confidence intervals.

Results

Subject resting characteristics

Subjects were well matched for general characteristics, including age and body mass index (P > 0.05; Table 1). Overlap patients, however, reported more dyspnoea (mMRC scores) and physical limitation (NYHA scores) on daily life. As previously reported (7,8), there was a greater burden of co-morbidities in overlap compared to COPD. Treatment for COPD did not differ between the groups. COPD patients showed a typical obstructive ventilatory defect with concomitant air trapping and lung hyperinflation, i.e., reduced FEV₁ and FEV₁/FVC with increased residual volume (RV) and RV/total lung capacity (TLC). As expected from coexisting heart failure, those abnormalities were lessened in overlap due to a restrictive component (as suggested by reduced FVC and preserved TLC) (P < 0.05). Resting IC, however, did not differ between the groups due to greater reductions in expiratory reserve volume in overlap (data not shown). Despite lower transfer factor values, overlap patients had better arterial oxygenation and lower PaCO₂ than COPD. Similar results were found in FEV₁-matched COPD and overlap patients (Table 1).

Table 1. Subject resting characteristics.

	All patients		FEV1-matched patients
Variables	COPD (n = 57)	Overlap (n = 41)	COPD (n = 35)	Overlap (n = 28)
Demographic/anthropometric
Age, years	66.3 ± 9.4	66.7 ± 8.8	67.3 ± 9.0	66.8 ± 8.8
Height, cm	169.9 ± 9.0	167.6 ± 7.5	169.9 ± 9.0	167.6 ± 7.5
Weight, kg	76.4 ± 21.6	74.0 ± 13.5	79.4 ± 23.8	76.6 ± 13.7
Body mass index, kg/m^2	26.2 ± 5.8	26.3 ± 4.0	27.1 ± 6.3	26.8 ± 4.2
GOLD stage (I/II/III/IV)	5/24/22/6	9/19/10/3	3/18/12/2	4/12/10/2
mMRC dyspnoea score (≥2)	44.7	85.3^*	38.5	81.7^*
NYHA functional class (2–3)	47.4	82.9^*	42.6	75.9^*
Ischemic dilated cardiomyopathy, %	—	58.5	—	55.5
Ejection fraction, %	66.0 ± 4.3	35.3 ± 7.4^*	65.8 ± 4.5	36.0 ± 7.6^*
Smoking, %	49.2 ± 29.5	60.0 ± 33.4	51.2 ± 34.4	51.6 ± 30.1
Main laboratory results
Hb, g/dL	13.9 ± 1.8	13.5 ± 1.4	14.0 ± 1.2	13.7 ± 1.5
Na^+, mMol/L	139 ± 2.9	139 ± 3.1	140 ± 3.0	140 ± 3.3
Creatinine, mg/dL	0.88 ± 0.27	1.19 ± 0.38^*	0.92 ± 0.29	1.20 ± 0.36^*
Main co-morbidities
Hypertension	68.4	85.4	63.1	82.2
Type II diabetes	22.8	34.1^*	18.5	30.4^*
Coronary artery disease	17.5	58.5^*	15.0	55.5^*
Chronic kidney failure	28.1	43.9^*	25.8	40.5^*
Main medications
Long-acting β2 agonists	100	85.4	100	82.2
Long-acting anti-muscarinics	75.4	69.4	75.4	65.3
Inhaled steroids	93.0	83.0	90.5	81.5
Diuretics	24.3	100^*	20.8	100^*
Selective β-blockers	3.5	85.4^*	3.5	81.5^*
ACE inhibitors	47.4	100^*	44.5	100^*
Lung function
FVC, %pred	96.8 ± 15.9	71.0 ± 14.7^*	92.4 ± 14.7	78.0 ± 15.3^*
FEV1, L	1.57 ± 0.54	1.77 ± 0.62^*	1.78 ± 0.45	1.84 ± 0.43
FEV1, %pred	53.7 ± 16.1	60.5 ± 17.7^*	61.9 ± 10.3	61.7 ± 10.5
FEV1/FVC	0.45 ± 0.10	0.58 ± 0.12^*	0.49 ± 0.10	0.51 ± 0.10
TLC, %pred	112.6 ± 17.2	98.3 ± 20.3^*	109.47 ± 15.3	92.1 ± 20.1^*
RV, %pred	160.2 ± 37.1	134.5 ± 37.3^*	150.2 ± 30.8	126.3 ± 32.6^*
RV/TLC	50.5 ± 7.2	43.0 ± 19.6^*	49.7 ± 7.0	38.6 ± 21.8^*
IC, % pred	72.3 ± 13.5	69.5 ± 12.8	73.1 ± 11.5	68.3 ± 14.7
TLCO, %pred	59.2 ± 17.3	46.4 ± 16.3^*	60.2 ± 15.3	51.9 ± 13.8^*
PaO2, mmHg	73.6 ± 9.2	77.9 ± 7.2^*	73.1 ± 8.1	79.0 ± 7.0^*
PaCO2, mmHg	37.6 ± 4.1	35.4 ± 4.0^*	38.0 ± 3.5	35.1 ± 3.8^*
SaO2, %	93.7 ± 2.6	95.0 ± 1.5^*	93.9 ± 2.4	95.2 ± 1.1^*

^*P < 0.05 (COPD vs. overlap). Values are mean ± SD or frequency (%).

mMRC, modified Medical Research Council scale; NYHA, New York Heart Association; GOLD, Global Initiative for COPD; ACE, angiotensin-converting enzyme; FVC, forced expiratory volume; FEV₁, forced expiratory volume in 1 s; TLC, total lung capacity; RV, residual volume; IC, inspiratory capacity; TL_CO, transfer factor; PaO₂, arterial partial pressure for oxygen; PaCO₂, arterial partial pressure for carbon dioxide; SaO₂, arterial oxygen saturation.

Metabolic and cardiovascular responses to exercise

Peak work rate and peak O₂ were significantly lower in overlap compared to COPD. Moreover, overlap showed considerable evidence of impaired submaximal aerobic metabolism as suggested by shallower O₂–work rate slope and lower estimated lactate threshold (i.e., lower gas exchange threshold, P > 0.05; Table 2). As expected from pharmacological β-blocking (Table 1), overlap patients showed significantly lower peak heart rates. Similar results were found contrasting FEV₁-matched COPD and overlap patients (Table 2).

Table 2. Physiological and sensory responses to incremental CPET.

	All patients		FEV1-matched patients
Variables	COPD (n = 57)	Overlap (n = 41)	COPD (n = 35)	Overlap (n = 28)
Exercise time, minute	10.4 ± 2.3	9.8 ± 3.7	10.6 ± 3.0	10.1 ± 2.9
Power
Peak WR, W	79 ± 31	56 ± 22^*	87 ± 31	60 ± 22^*
Peak WR, %pred	69.4 ± 18.7	51.3 ± 15.3^*	72.3 ± 21.0	54.9 ± 14.7^*
Metabolic responses
Peak O2, %pred	74.4 ± 18.6	64.3 ± 15.4^*	78.58 ± 18.57	66.20 ± 13.70^*
Peak O2, mL/minute	1361 ± 450	1048 ± 287^*	1460 ± 446	1098 ± 271^*
Peak RER	1.04 ± 0.83	1.08 ± 0.10^*	1.05 ± 0.76	1.10 ± 0.10^*
O2 at the GET, mL/minute†	916 ± 264	759 ± 183^*	1016 ± 185	774 ± 180^*
O2-WR slope, mL/minute/W	11.2 ± 1.7	9.7 ± 1.8^*	11.2 ± 1.6	9.9 ± 2.0^*
Cardiovascular responses
Peak HR, bpm	131 ± 19	113 ± 21^*	130 ± 18	115 ± 21^*
Peak O2 pulse, mL/beat	10.3 ± 3.2	9.3 ± 2.4	10.1 ± 2.9	9.5 ± 2.8
Peak systolic blood pressure, mmHg	158 ± 35	155 ± 30	160 ± 33	158 ± 29
Peak diastolic blood pressure, mmHg	82 ± 20	78 ± 18	81 ± 19	77 ± 15
Ventilatory responses
E, L/minute	47.8 ± 15.0	46.0 ± 15.0	52.8 ± 14.2	48.7 ± 11.9
VT, L	1.49 ± 0.42	1.47 ± 0.38	1.60 ± 0.37	1.54 ± 0.29
f, rpm	32 ± 5	31 ± 6	33 ± 5	33 ± 4
E/MVV	0.84 ± 0.18	0.74 ± 0.19^*	0.80 ± 0.15	0.75 ± 0.19^*
E-CO2 slope	28.9 ± 6.0	36.0 ± 6.6^*	29.8 ± 4.5	36.4 ± 6.0^*
E-CO2 intercept, L/minute	6.0 ± 2.9	2.9 ± 2.7^*	5.3 ± 2.5	2.7 ± 2.7^*
E/CO2 peak	34 ± 4	40 ± 7^*	35 ± 3	41 ± 8^*
Peak-rest IC, L	−0.62 ± 0.37	−0.31 ± 0.38^*	−0.54 ± 0.29	−0.26 ± 0.19^*
Peak EILV/TLC	0.93 ± 0.04	0.89 ± 0.06^*	0.90 ± 0.05	0.86 ± 0.064^*
Gas exchange responses
Peak PETCO2, mmHg	38 ± 7	32 ± 6^*	36 ± 5	32 ± 5^*
Peak-rest PETCO2, mmHg	5 ± 4	2 ± 6^*	4 ± 4	1 ± 6^*
Peak SpO2, %	91 ± 6	93 ± 6^*	93 ± 4	94 ± 4
Peak-rest SpO2, %	−3.6 ± 4.4	−1.7 ± 3.5^*	−3.9 ± 3.6	−1.6 ± 3.2^*
Sensory responses
Peak dyspnoea score	7 (4–10)	5.5 (0–9)^*	7.4 (4–10)	5.8 (0–9)^*
Peak leg effort score	5.4 (1–10)	6.8 (2–10)^*	5.3 (1–10)	7.4 (3–10)^*
Peak leg effort − dyspnoea scores	−2 (−7 to 4)	1 (−4 to 9)^*	−1.5 (−6 to 5)	1.5 (−5 to 8)^*
Peak leg effort ≥ dyspnoea scores	18 (35.2)	35 (85.3)^*	12 (34.3)	24 (85.3)^*

^*P < 0.05 (COPD vs. overlap). Values are mean ± SD, frequency (N) or median (range).

^†Identified in 21, 24, 14, 18 patients, respectively.

WR, work rate; O₂, oxygen uptake; RER, respiratory exchange ratio; GET, gas exchange threshold; HR, heart rate; E, Ventilation; MVV, maximal voluntary ventilation; CO₂, carbon dioxide output; IC, inspiratory capacity; EILV, end-inspiratory lung volume; TLC, total lung capacity; PET, end-tidal partial pressure; SpO₂, oxygen saturation by pulse oximetry.

Ventilatory and gas exchange responses to exercise

Compared to COPD, overlap patients had greater peak ventilatory reserve (lower peak E/MVV ratio) despite greater ventilatory inefficiency (higher peak E/CO₂ ratio and E-CO₂ slope) (Table 2). In addition, lower E-CO₂ intercept was found in overlap when considering both the whole group and FEV₁-matched sub-groups (Table 2). Overlap also showed smaller decrements in IC from rest to peak exercise, i.e., exercise-induced dynamic hyperinflation. These findings were associated with larger decrements in peak PETCO₂ and attenuated exercise-induced O₂ desaturation (Table 2). As anticipated, dyspnoea was the main exercise-limiting symptom in COPD. Conversely, leg effort ratings were higher than dyspnoea scores at exercise termination in overlap. Similar results were found when contrasting FEV₁-matched COPD and overlap patients (Table 2).

Figure 2 shows the comparative analysis among ROC curves with higher AUCs to discriminate overlap from COPD (E–CO₂ slope and intercept, peak E/CO₂ ratio and peak PETCO₂). Based on the suggested optimal thresholds for these and other key physiological variables, odds ratios for overlapping prediction were calculated by univariate logistic regression analysis. As shown in Table 3, particularly high odds ratios were found for E-CO₂ intercept ≤3.5 L/minute, E-CO₂ slope ≥ 34, peak E/CO₂ ratio ≥ 37 and peak PETCO₂ ≤ 31 mmHg (P < 0.0001). Multivariate logistic regression models for overlapping prediction were then obtained for variables derived from the linear E-CO₂ plot (i.e., slope and intercept) (‘model 1’, Table 3) or discrete end-exercise variables (E/CO₂ ratio and PETCO₂) (‘model 2’, Table 3). Attenuated O₂ desaturation (peak SpO₂ ≤ 89%) did not significantly add predictive value to those variables (P > 0.05). For instance, Figure 3 (left panels) shows that only 8/50 (16%) subjects with E-CO₂ intercept > 3.5 L/minute and E-CO₂ slope < 34 were overlap patients. Moreover, whereas 7/41 patients (17%) with peak E/CO₂ < 37 and peak PETCO₂ ≥ 31 mmHg were overlap, 16/19 (84.3%) subjects with higher peak E/CO₂ and lower PETCO₂ had overlap (P < 0.001). Matching patients by FEV₁ produced similar results (Figure 3, right panels).

Figure 2. Receiver operating characteristic (ROC) curve analysis to discriminate overlap from COPD. Circles depict the optimal thresholds according to individual variables. ^*P < 0.05. †P < 0.05 versus peak PETCO₂. E, ventilation; CO₂, carbon dioxide output; PET, end-tidal partial pressure.

Table 3. Univariate and multivariate logistic regression modelling for overlap prediction

	Odds ratio (95% CI)	P	χ^2	−2 Log LR
Univariate analysis
Ventilatory
E-CO2 slope ≥ 34	5.85 (2.39–14.32)	0.0001	16.4	116.8
E-CO2 intercept ≤ 3.5 L/minute	11.35 (4.35–29.61)	<0.0001	29.2	104.0
E/CO2 peak ≥ 37	8.25 (3.19–21.30)	<0.0001	22.4	110.8
E/CO2 nadir ≥ 34	1.12 (1.03–1.22)	0.002	8.9	124.2
Gas exchange
Peak PETCO2 ≤ 31 mmHg	11.52 (3.07–43.17)	<0.0001	18.0	115.2
Peak-rest PETCO2 ≤ 0 mmHg	6.02 (2.10–17.19)	0.0003	12.8	120.4
Peak SpO2 ≤ 89%	3.15 (1.13–8.76)	0.02	5.4	127.8
Peak-rest SpO2 ≥ 4%	2.07 (0.84–5.11)	0.10	2.6	135.1
Multivariate analysis
Model 1			31.1	102.1
E-CO2 intercept ≤ 3.5 L/minute	7.69 (2.61–22.65)	0.0002
E-CO2 slope ≥ 34	2.18 (0.73–6.50)	0.14
Peak SpO2 ≤ 89%	—	0.42
Model 2			29.6	103.5
E/CO2 peak ≥ 37	5.35 (1.96–14.59)	0.0010
Peak PETCO2 ≤ 31 mmHg	5.73 (1.42–23.15)	0.0142
Peak SpO2 ≤ 89%	—	0.50

CI, confidence interval; LR, likelihood ratio; E, ventilation; CO₂, carbon dioxide output; PET, end-tidal partial pressure; SpO₂, oxygen saturation by pulse oximetry.

Figure 3. Ventilation CO₂ output (CO₂) slope versus intercept (upper panels) and end-tidal partial pressure for CO₂ (PETCO₂) versus peak E/CO₂ ratio in COPD and overlap patients. Dotted lines indicate the optimal thresholds for overlap discrimination as established by the ROC curve analysis (Figure 2).

Discussion

This is the first study to prospectively investigate the influence of heart failure with reduced left ventricular ejection fraction on exercise ventilation in COPD. Results indicate that lower maximal exercise capacity in COPD-heart failure overlap was associated with attenuated exercise-induced dynamic hyperinflation, greater ventilatory inefficiency and larger decrements in peak PETCO₂. A low E-CO₂ intercept (≤3.5 L/minute) coupled with a high E-CO₂ slope (≥34) or a high peak E/CO₂ ratio (≥37) plus a low peak PETCO₂ (≤31 mmHg) were more frequently found in overlap than COPD even when between-group differences in FEV₁ were taken into consideration. Thus, CPET responses reflecting a heightened ventilatory response to metabolic demand might prove useful in different clinical scenarios involving overlap patients, e.g., to assess the contribution of heart failure to exercise intolerance, to determine the effects of interventions and to assist in prognostic stratification.

There is growing recognition that identification and treatment of co-morbidities are highly relevant to clinical management of COPD. Heart failure in COPD (‘overlap’), in particular, has been associated with increased morbidity, poor quality of life and greater utilization of health care resources. Moreover, overlap frequently compounds with other systemic co-morbidities (Table 1) thus contributing to poor prognosis (7,8,20). Considering that exercise intolerance is the hallmark of both COPD and heart failure (21), exercise-based evaluations such as CPET are likely to provide relevant information for clinical management of overlap patients.

In this context, the most consistent abnormality found in overlap was an excessive ventilatory response to metabolic demand, expressed as relatively higher E-CO₂ slope, lower E-CO₂ intercept or higher peak E/CO₂ ratio (Table 2; Figures 2 and 3). Physiological dead space (i.e., dead space/VT ratio) and neural ventilatory drive, the main determinants of exercise ventilation (22), are typically increased in patients with COPD or heart failure (3–6,9–11). Increased physiological dead space in these patients might be due to elevated dead space (e.g., augmented ventilation/perfusion mismatch, larger emphysematous areas) and/or lowered VT secondary to mechanical constraints. Overlap patients showed slightly, but significantly, lower PaCO₂ values at rest (Table 1) which might chronically reduce CO₂ set-point. Interestingly, similar responses to CPET as described herein in overlap might also be found in patients with coexisting COPD and pulmonary hypertension (23) thereby suggesting that heightened pulmonary arterial pressures may also play a role in increasing exercise ventilation in overlap. Moreover, increased activation of group III-IV skeletal muscle afferents could be involved in the heightened ventilatory responses to exercise (24). It is thus conceivable that negative interactions among those abnormalities may underlie greater ventilatory inefficiency in overlap compared to COPD alone.

Overlap patients had lower exercising PETCO₂ responses compared to COPD (Table 2). PETCO₂ is biased to represent poorly ventilated lung units with longer time constants (and higher alveolar CO₂ pressures) in patients with COPD (9). In contrast, low PaCO₂ (i.e., alveolar hyperventilation), shallow and fast breathing pattern, more homogenous lung emptying, and poor pulmonary perfusion secondary to impaired cardiac output can reduce exercising PETCO₂ in overlap (18,25). Heightened ventilation of areas with better preserved ventilation/perfusion matching might also explain why overlap presented with higher resting and exercising SpO₂ despite relatively lower transfer factor and, possibly, lower mixed venous O₂ saturation. Additional studies with direct measurements of arterial blood gases are warranted to determine at which extent high dead space/VT ratio versus low CO₂ set-point might mechanistically explain high ventilatory inefficiency and low PETCO₂ in overlap.

A particularly interesting finding of the present study was the lower inspiratory capacity from rest to peak exercise in overlap compared to COPD (Table 2). It thus seems that heart failure, similar to other conditions associated with a restrictive pattern (such as obesity) (26), may have a deflating effect in COPD. This rather unexpected effect of heart failure allowed overlap patients to increase VT despite their significantly lower TLC (Table 2). Consequently, lower operating lung volumes may have allowed overlap patients to meet the increased ventilatory demands of exercise while delaying the reaching of a critical inspiratory reserve volume associated with limiting dyspnoea (1). Although it could be argued that the total (resistive plus elastic) work of breathing – and the attendant inspiratory neural drive – would be greater in overlap patients (27), it is noteworthy that breathlessness assumed a secondary role (to leg effort) as limiting symptom in overlap. Thus, attenuated volume constraints and preservation of ventilatory demand/capacity ratio may explain lower dyspnoea/ventilation ratio in overlap compared to COPD. These findings were confirmed in a sub-group of FEV₁-matched patients and thus cannot be ascribed solely to between-group differences in airflow limitation (at least as indicated by FEV₁).

Another key finding of the current study was the shift from dyspnoea to leg effort as the major limiting symptom to exercise in overlap compared to COPD (Table 2). Thus, although the primary central insult differs in the isolated diseases, structural and functional aberrations appear to conflate at the periphery (i.e., skeletal muscle) and may play a dominant role in determining physical capacity in overlap (21,28). Impaired cardiac output and poor muscle blood flow (28), unfavourable metabolic and hormonal milieu, increased proportion of less efficiency type II fibres, intrinsic muscle abnormalities and severe deconditioning might be mechanistically linked to those findings (29). From a practical perspective, targeting the peripheral skeletal muscles might be particularly relevant for improving exercise intolerance in this patient population.

There are some clinical scenarios in which we envisage our data might eventually prove useful. Our results suggest that heart failure likely contributes to exercise pathophysiology in COPD in the presence of markedly impaired ventilatory efficiency (E-CO₂ intercept ≤ 3.5 L/minute, E-CO₂ slope ≥ 34, peak E/CO₂ ratio ≥ 37) and/or reduced peak PETCO₂ (≤31 mmHg). Although those CPET responses clearly lack sensitivity and specificity to separate overlap from COPD (Figure 3), their presence might suggest the need of additional cardiologic investigations in patients with a high pre-test likelihood of heart failure (e.g., previous myocardial infarction, suspicious imaging findings, and out-of-proportion (to FEV₁) exercise intolerance). These variables might also be useful targets to evaluate the efficacy of ‘cardiovascular’ interventions in overlap patients, e.g., cardiac resynchronization therapy or further β-blockage to improve ventilatory inefficiency. Considering the relevance of poor ventilatory efficiency and reduced PETCO₂ in predicting poor prognosis in heart failure (17,25), it is conceivable that they may also prove useful to predict survival in overlap. In fact, Alencar et al. recently provided preliminary evidence that E-CO₂ slope ≥ 34 is an independent predictor of mortality in overlap (20).

Our study presents with some important methodological particularities compared to previous studies which looked at exercise responses in overlap. Patients were seen prospectively by the same respirologists and cardiologists with their treatment being optimized carefully before CPET. In contrast, previous retrospective studies did not control for heterogeneities in COPD and heart failure treatment. For instance, overlap patients followed by cardiologists are less likely to receive adequate bronchodilator treatment; conversely, respirologists are frequently reluctant to prescribe β-blockers (7,8). Thus, the observed differences in overlap versus COPD cannot be ascribed to insufficient treatment for either disease. In fact, pharmacological β-blocking precluded us to test peak heart rate or O₂ pulse to discriminate COPD from overlap. Whether coexisting COPD and heart failure with preserved left ventricular ejection (30) similarly impact on CPET responses remains to be investigated. Our results might not be applicable to end-stage COPD with severe resting hypercapnia and/or hypoxaemia. Nevertheless, severely disabled patients are usually unable to tolerate physical exertion and thus are not routinely submitted to incremental CPET.

In conclusion, heart failure with reduced left ventricular ejection fraction had discernible effects on selected physiological responses to exertion in patients with COPD. Thus, CPET provided a cluster of ventilatory and gas exchange variables that were more frequently found in overlap than COPD vis-à-vis an unusually high ventilatory response to metabolic demand (low E-CO₂ intercept, high E-CO₂ slope and high peak E/CO₂ ratio) and low peak PETCO₂ values (reflecting alveolar hyperventilation and/or lung perfusion abnormalities). These variables might prove eventually useful in practice to assist the clinical interpretation of CPET responses in COPD patients presenting heart failure with reduced left ventricular ejection fraction as co-morbidity.