Addressing Assumptions for the Use of Non-invasive Cardiac Output Measurement Techniques During Exercise in COPD

Abstract

The multifactorial functional limitation of COPD increasingly demonstrates the need for an integrated circulatory assessment. In this study cardiac output (Qc) derived from non-inert (CO₂-RB), inert (N₂O-RB) gas rebreathing approaches and bioimpedance were compared to examine the limitations of currently available non-invasive techniques for exercise Qc determination in patients with chronic lung disease. Thirteen COPD patients (GOLD II-III) completed three constant cycling bouts at 20, 35, and 50% of peak work on two occasions to assess Qc with bioimpedance as well as using CO₂-RB and N₂O-RB for all exercise tests. Results showed significantly lower Qc using the N₂O-RB or end-tidal CO₂-derived Qc compared to the PaCO₂-derived CO₂-RB or the bioimpedance at rest and for all exercise intensities. End-tidal CO₂-derived values are however not statistically different from those obtained using inert-gas rebreathing. This study show that in COPD patients, CO₂-rebreathing Qc values obtained using PaCO₂ contents which account for any gas exchange impairment or inadequate gas mixing are similar to those obtained using thoracic bioimpedance. Alternately, the lower values for N₂O rebreathing derived Qc indicates the inability of this technique to account for gas exchange impairment in the computation of Qc. These findings indicate that the choice of a gas rebreathing technique to measure Qc in patients must be dictated by the ability to include in the derived computations a correction for either gas exchange inadequacies and/or a vascular shunt.

Keywords:

• bioimpedance
• exercise testing
• soluble gas
• ventilation:perfusion

Introduction

Increasing evidence of functional limitations associated with systemic comorbidities in chronic disorders has prompted interest for the assessment of integrated circulatory function. This is particularly true in patients with Chronic Obstructive Pulmonary Disease (COPD) due to the potential influence of lung hyperinflation, pulmonary hypertension as well as cardiovascular co-morbidities on oxygen transport and utilization during exercise.

The direct Fick method for routine cardiac output measurement () is not readily feasible or desirable. A renewed interest is however seen for numerous indicator gas rebreathing, imaging or bioconductance non-invasive techniques (1–5) for use at rest or during exercise. Over the last decade, there have been several reports of at rest and exercise in patients with chronic heart failure, pulmonary hypertension or other abnormalities using inert gas rebreathing techniques with acetylene or nitrous oxide (6–9). Theoretically, the choice of the most appropriate technique for use in the COPD population should be based on the established validation in a similar population against a method of reference, its ability to adequately take into account pulmonary and vascular shunts and the ease of utilization of the technique. However, the extent to which the indicator gas rebreathing techniques addresses the assumptions made with respect to the disease-specific potential incompatibility, particularly during exercise, remains unclear (7, 10–12).

The CO₂-rebreathing technique is based on the Fick principle applied to CO₂ serving as an indicator; arterial CO₂ content is estimated from either end-tidal or arterial CO₂ measurements while mixed venous CO₂ content is derived from end-tidal CO₂ taken during rebreathing. This technique was shown to be valid for exercise determination in patients with lung disease when estimation of the arterial CO₂ content is made using the direct arterial CO₂ (13). It is of particular interest for use in the case of patients with ventilation: perfusion mismatch as it remains unaffected by exercise-induced changes in arterial oxygen content.

The Innocor inert gas rebreathing system received much attention because the derived measurement includes a theoretical correction that can potentially account for a right-to-left anatomical shunt (14). This correction is based on the premise that the obtained soluble gas indicator measurement reflects effective pulmonary blood flow and that can be computed from the sum of the measured pulmonary blood flow and the derived shunt volume. The latter in turn is determined from the discrepancy between the pulse oximetry value detected by the system and an assumed pulmonary end-capillary oxygen saturation of 98% (7). The extent to which the latter assumption affects the measured in the presence of gas exchange abnormalities such as seen in patients with COPD remains to be verified.

Bioconductance is increasingly used to assess in a wide spectrum of populations including COPD due to its ease of utilization and the possibility for continuous impedance measurement. Since the technique is not restricted by gas exchange it also presents the advantage that blood shunting is not a factor and thus that the measured thoracic movement of flux adequately reflects the total ventricular ejection. There is, however, still only few reports comparing bioconductance-derived to values obtained using other techniques under similar exercise condition or disease severity (10, 15).

The goal of the present study was to contrast exercise derived from various gas-rebreathing approaches to that measured using thoracic bioconductance in patients with COPD with the view to understand the ability of non-invasive techniques to appropriately address limitations related to gas exchange impairments. We first compared arterial PCO₂ derived values measured using CO₂-rebreathing to those derived using the end-tidal PCO₂, to assess the influence of inadequate gas mixing on the resulting cardiac output measurement. Thus, we reasoned that in the case of patients presenting ventilation:perfusion abnormalities, end-tidal PCO₂ derived should be lower compared to that derived from PaCO₂ (16, 17). Arterial PCO₂ derived values were also contrasted to the INNOCOR-derived outputs as the latter include a shunt volume component that assumes normal lung diffusion and which, therefore, would adequately reflect a vascular but not a pulmonary shunt component.

Finally, gas rebreathing derived were compared to bio-conductance-derived measurements with the understanding that changes in thoracic biophysical properties will reflect the combined shunt and effective ventricular volume (15–17). We hypothesized that the PaCO₂-derived measured using CO₂-rebreathing would be similar to that measured using bioimpedance, while lower values would be computed when derived from gas-rebreathing approaches.

Methods

Subjects

Thirteen patients with confirmed moderate to severe (GOLD II = 10; GOLD III = 3) patients (1 female; 12 males) COPD were recruited from the Montreal Chest Institute (Table 1). Inclusion criteria were: age > 50 years, post-bronchodilator forced expiratory volume in 1 second (FEV₁) between 30 and 80% predicted and FEV₁/forced vital capacity (FVC) < 70% (Global Initiative for Chronic Obstructive Lung Disease [GOLD] stage II and III). Exclusion criteria were: respiratory exacerbation within the 2-month period preceding the study, history of asthma, significant O₂ desaturation (oxygen pulse saturation [SpO₂] < 85%) at rest or during exercise, presence of another pathology that could influence exercise tolerance or patients receiving oxygen therapy. The study protocol was approved by the institutional research ethics boards and written informed consent was obtained in all patients.

Table 1. Patient characteristics

Variable	COPD (n = 13)
Age (years)	66 ± 1
Height (cm)	168 ± 3
Weight (kg)	74 ± 5
BMI (kg/m^2)	26 ± 2
FEV1 (% predicted)	58 ± 3
RV (% predicted)	149 ± 10
TLC (L)	7.0 ± 0.4
RV/TLC	51 ± 2
FRC (% predicted)	152 ± 10
IC (L)	2.5 ± 0.1
D LCO (% predicted)	55 ± 4
D LCO / VA (ml/mmHg/min/L)	5.5±3.3
PaO2 (mmHg)	76.1 ± 2.7
Pa CO2 (mmHg)	42.5 ± 1.5
Hct (%)	48.1 ± 1.1
pH	7.43 ± 0.01
Peak exercise data
VO2 (LO2·min^-1)	1.2 ± 0.4
Power (watt)	88 ± 29
VE (L·min^-1)	60 ± 17
HR (bpm)	131 ± 29
SpO2	94.5 ± 0.5
PET CO2	32.4 ± 1.7

BMI = body mass index; FEV₁ = forced expiratory volume in one second; RV = residual volume; TLC = total lung capacity; FRC = forced residual volume; IC = inspiratory capacity; D_LCO = diffusing capacity of the lung for carbon monoxide; PaO₂ (mmHg) = O₂ content in arterial blood PaCO₂ (mmHg) = CO₂ content in arterial blood;SpO₂ (%): pulse oximetry derived oxyhemoglobin saturation.

Study design

COPD patients came to the laboratory twice within a 2-week period with a minimum of 48 hours between visits. The initial visit consisted of a full pulmonary function assessment according to American Thoracic Society standards (18); and a peak incremental cycle test followed by familiarization with inert-gas and CO₂-rebreathing techniques. The second visit consisted of measurements during constant load sub-maximal cycling.

Protocols and measurements

Peak exercise capacity

A symptom limited peak incremental cycle test with increases of 10 Watts/min was performed on a stationary electromagnetically braked ergometer (Ergoline) for determination of peak-power (Wpeak) and related physiological parameters including peak oxygen consumption (). Electrocardiogram (ECG), transcutaneous oxyhemoglobin saturation (SpO₂; Radical, Irvine, CA) and expiratory gas fractions (MediSoft, Sorinnes, Belgium) were continuously monitored.

Constant load exercise protocol

Participants cycled at three constant workloads corresponding to 20, 35, and 50% of previously determined Wpeak, intensities chosen to assure metabolic stability. Each workload was performed twice to assess with both rebreathing techniques (inert-gas or CO₂-rebreathing), the order of technique being randomized for each exercise intensity. Exercise bouts were 8–10 minutes in duration with each bout separated by a 10- minute rest period. measurements were obtained between the fourth and tenth minute of each stage of cycling; a minimum of two rebreathing maneuvers were performed during each exercise bout.

A third measurement was performed if values for the same technique differed by >15%, or inadequate equilibrium was detected. However, using impedance cardiography was monitored continuously during each workload. Patients had not received bronchodilators prior to the exercise test. Gas exchange and ventilation measurements were obtained at baseline and during each steady-state cycling bout. An arterialized blood sample was collected during steady-state prior to performing the rebreathing maneuver for determination of PaCO₂ and SaO₂ (GEM Premier 3000, Instrumentation Laboratory, Massachusetts). The use of arterialized PCO₂ to adequately reflect PaCO₂ has been previously established in a large cohort of patients with chronic pulmonary of heart disease (19).

Cardiac output () determination

Inert Gas Rebreathing

This technique was performed using the Innocor system (Innovision, Odense, Denmark) with pulmonary blood flow being measured using a rebreathing bag containing a bolus of N₂O (0.5%), SF₆ (0.1%) and O₂ (28%) diluted with atmospheric air as previously described (6). Pulmonary blood flow was calculated from the rate of uptake of N₂O, during the last three breaths of the rebreathing. Only measurements in which the SF₆ curve indicated complete mixing of gases were included in the analysis.

CO₂ Rebreathing

was determined using the PCO₂ equilibrium technique (20) validated for use in patients with lung diffusion disturbances (13), using PaCO₂ obtained from an arterialized blood sample from a pre-warmed earlobe immediately before the rebreathing manover (21). During rebreathing, participants inhaled and exhaled through a two-way valve into the bag for 10–12 seconds or until PCO₂ equilibrium was obtained. was calculated as / (CCO₂-CaCO₂) where mixed venous CO₂ content (CCO₂) was obtained from the equilibrium PCO₂ after correction for the “downstream difference” (20) and CaCO₂ was derived from the arterialized blood PaCO₂. To examine the impact of ventilation:perfusion inhomogeneity on the measurement, values were also calculated using end-tidal CO₂ (PETCO₂) as is done when the technique is used in healthy subjects in whom lung diffusion is normal.

Thoracic Impedance

The Physioflow (PF-05; Manatec Biomedical; Macheren, France) system detects changes in trans-thoracic impedance through two sets of sensing and signaling electrodes placed on the neck and chest during cardiac ejection to reflect stroke volume. Heart rate is obtained simultaneously to calculate (15). was measured beat-to-beat and stored as 5-second averages in the database for the entire exercise bout duration ensuring to capture the same time period as that used for measurement using the other methods.

Treatment of data

For a given individual, the determined by rebreathing techniques was obtained at each exercise intensity when steady-state was achieved, as the average of at least two rebreathing-derived values. For impedance cardiography, values reflect the average of three, 30-second steady-state recordings at each exercise intensity taken prior to the rebreathing manoeuvres.

Measurement reproducibility

Test-retest reproducibility of exercise cardiac output determination in COPD was assessed by repeated exercise tests within at minimum 48 hours, but at most within a 2-week period. Pearson correlations calculated on the submaximal exercise sets of data indicate r values of 0.76 for CO₂-RB; 0.88 for N₂O RB and of 0.89 for impedance cardiography.

Statistical analysis

values are presented as mean ± standard error of the mean (SEM). Mean comparison between techniques was achieved using a two-way ANOVA (measurement technique × exercise intensity) with repeated measures on the last factor. Where main effects were found (P < 0.05), the Holm–Sidak method of post hoc analyses was performed (22). The critical significance value was considered as P < 0.05. All statistical analyses were carried out using Sigmaplot Version 11 (Systat Software Inc., USA).

Results

Patient characteristics

As can be seen from Table 1, COPD patients showed moderate to severe airflow obstruction with static hyperinflation, as indicated by significantly lower FEV₁ as well as exaggerated RV/TLC, FRC and reduced IC. Values also attest to impairment in lung diffusion as seen from the markedly reduced DLCO and DLCO/VA compared to predicted normal values.

Exercise response

As expected, peak exercise values were markedly reduced compared to predicted normal values (Table 1). Table 2 shows cardiorespiratory parameters at baseline and during steady-state sub-maximal exercise intensities. Results indicate the expected increase in , and heart rate, dynamic hyperinflation as reflected by the reduction of IC at 50%W peak compared to the baseline. SpO₂ was not significantly different from the baseline..

Table 2. Cardiorespiratory responses to cycle exercise

	(% Wpeak)
	Baseline	20	35	50
Power output (watts)	0	16.5 ± 0.5^*	29.5 ± 0.8^*	41.2 ± 1.2^*
(L min^-1)	0.31 ± 0.01	0.56 ± 0.01 ^*	0.67 ± 0.02 ^*	0.79 ± 0.02 ^*
HR bpm	80 ± 5	89 ± 5	97 ± 7	103 ± 6 *
(L min^-1)	16.0 ± 0.9	26.1 ± 1.4 ^*	30.5 ± 1.6 ^*	36.4 ± 1.7 ^*
VT (ml)	1.0 ± 0.1	1.1 ± 0.1	1.3 ± 0.1 *	1.4 ± 0.1 *
Fb (breaths/min)	18.9 ± 1.2	24.3 ± 1.1 *	30.5 ± 1.6 *	28.4 ± 1.6 *
IC (L)	2.6 ± 0.2	2.5 ± 0.2	2.5 ± 0.2	2.3 ± 0.2
SpO2 (%)	94.8 ± 0.5	94.7 ± 0.5	94.8 ± 0.7	94.6 ± 0.5

*Indicates P ≤ 0.05 from baseline.

= oxygen uptake; = ventilation; SpO2 (%): pulse oximetry-derived oxyhemoglobin saturation.

Figure 1 shows measurements determined using gas-rebreathing techniques and thoracic impedance at rest and for each exercise intensity. As expected, a significant increase from baseline was seen with exercise irrespective of the method of measurement. The histograms contrast measurements obtained under each condition using the gas indicator techniques or bioconductance. For the CO₂-rebreathing technique, the is shown for both the PaCO₂ and the PETCO₂ content derived measurements to reflect the influence of ventilation:perfusion inhomogeneity. Comparison of results indicates a consistent pattern of significantly lower measured using the inert-gas approach as compared to the PaCO₂-derived CO₂-rebreathing technique or the bioconductance measures, at rest as well as for all exercise intensities. End-tidal CO₂-derived rebreathing values are also significantly lower than those calculated using direct PaCO₂ under all conditions. End-tidal CO₂-derived rebreathing values are, however, not statistically different from those obtained using inert-gas rebreathing.

Figure 1. Cardiac output in COPD derived using various non-invasive techniques. Results consistently show significantly lower measured using N₂O-RB as compared to the PaCO₂-derived CO₂-RB or the bioimpedance approach, at rest as well as for all exercise intensities (*p < 0.05).

End-tidal CO₂-derived rebreathing values are also significantly lower than those calculated using direct PaCO₂ under all conditions (^§p < 0.05); End-tidal CO₂-derived rebreathing values are however not statistically different from those obtained using N₂O-RB; CO₂-RB(PaCO₂): Cardiac output () derived from non-inert CO₂ gas rebreathing with PaCO₂ correction; CO₂-RB(PETCO₂): Cardiac output () derived from non-inert CO₂-RB gas rebreathing without PaCO₂ correction; N₂O-RB: Cardiac output () derived from inert N₂O gas rebreathing; Impedance: Cardiac output () determined by impedance cardiography.

Discussion

Although the limitations of indicator gas rebreathing techniques in patients with lung disease is to a large extent predictable, this study is to our knowledge, the first to provide an explanation for the limitations of inert gas and end-tidal CO₂ rebreathing techniques.

Results from this study show that in COPD patients, CO₂-rebreathing values obtained using PCO₂ contents derived from direct arterial measurements, and thus accounting for any gas exchange impairment or inadequate gas mixing are similar to those obtained using thoracic bioconductance. These findings indicate that the choice of a gas rebreathing technique to measure in patients must be dictated by the ability to include in the derived computations a correction for either gas exchange inadequacies and/or a vascular shunt.

Pulmonary blood flow versus determination

Indicator gas rebreathing techniques are based on the underlying assumptions that pulmonary and systemic blood flows are equal and that exchange of the gas indicator between the rebreathing bag and lung is adequate (23). These assumptions may not hold true in patients with COPD because indicator alveolar content reflected by the end-tidal value does not account for inadequate gas mixing and gas exchange disturbances potentially leading to an underestimation of pulmonary blood flow. It is thus recommended that when using CO₂ rebreathing in patients with COPD, the CO₂ content used to determine be derived from arterial as opposed to end-tidal PCO₂ (13, 16, 17).

The INNOCOR inert gas rebreathing technique is based on the premise that the obtained data reflect pulmonary blood flow whereas results from the sum of pulmonary blood flow and shunt volume (6, 7). Calculation of the shunt volume is based on the computed difference between the oxyhemoglobin saturation measured by the system during the rebreathing manoeuvre and a 98% assumed end-capillary oxygen saturation. The latter is, however, only valid if vascular perfusion is matched with well-ventilated alveoli with normal lung diffusion such that any discrepancy between the assumed end-capillary and measured oxygen saturation may be ascribed to shunting of vascular origin.

This may not hold in patients with COPD in whom inhomogeneous lung ventilation, impaired lung diffusion and ventilation: perfusion inhomogeneity may all contribute to shunting of pulmonary origin; as may be suspected to be the case in the present study, from the impaired pulmonary diffusion characteristics. The importance of adequate gas mixing and exchange may be seen from Figure 1, showing CO₂ RB (PaCO₂) to be 30–50% higher as compared to values computed from PETCO₂ because determination using P_ETCO₂ does not take into account the combined influence of both alveolar-capillary exchanges and vascular shunts.,. Alternately, the use of arterial as opposed to end-capillary gas content enables to capture the effect of not only the vascular but also the pulmonary gas exchange inadequacies, thus providing a more adequate factor for the computation of from the direct pulmonary blood flow determination. This observation is also consistent with the report from Mahler et al. (1985) in patients with both moderate and severe obstructive airway disease showing lower exercise using end-tidal as compared to PaCO₂ or to the direct Fick measurement (13).

The present results also indicate systematically lower measurements using the INNOCOR system compared to both the PaCO₂ derived CO₂-rebreathing and the bioconductance measures. This observation is consistent with the computational approach in the INNOCOR system with built-in correction for vascular shunts but which does not permit correction alveolar gas exchange impairment. Taken together our observations confirm that the use of gas rebreathing method for determination must allow consideration of shunting of both circulatory and pulmonary origins. Inasmuch as CO₂-rebreathing provides such an opportunity while inert gas rebreathing techniques do not, the former would appear as the rebreathing method of choice.

CO₂-rebreathing remains however technically challenging both from a technician execution as well as patient perspective as it requires patients to maintain metabolic steady state while performing the rebreathing manoeuver. For this reason, measurements in the present study were not obtained at near-maximum intensities but during low and moderate exercise corresponding to 20 and 50% peak , which nevertheless resulted in a significant dynamic hyperinflation.

Data from our laboratory from a subset of healthy control subjects without circulatory or pulmonary impairments, of a similar mean age as that of the COPD group (66 ± 2 yrs), using similar techniques at the same relative exercise intensities show no difference in cardiac output measurements derived using P_ETCO₂ for RBCO₂, N₂O RB and bioimpedance (Figure 2). This observation in healthy subjects is in support of the reasoning presented in this study as it shows that when gas exchange disturbances is not of concern, available non-invasive indicator rebreathing techniques are appropriate to assess exercise cardiac output.

Figure 2. Cardiac output derived using various non-invasive techniques in healthy subjects measured in our laboratory. Comparison of techniques at rest or at each exercise intensities reveals no statistically significant differences.

CO₂-RB(PETCO₂): Cardiac output () derived from non-inert CO₂-RB gas rebreathing without PaCO₂ correction; N₂O-RB: Cardiac output () derived from inert N₂O gas rebreathing. Impedance: Cardiac output () determined by impedance cardiography.

Use of bioconductance

The use of bioimpedance relies on the physical properties of bio-conductance such that shifts in thoracic blood volume consecutive to ventricular ejection will cause variations in thoracic conductance that can be detected between the inferior vena cava taken as the entry point at the level of the xyphoid process and the exit at the carotid trunk captured at the base of the neck. The resulting change in conductance therefore reflects the integrated flow that may be ascribed to the systemic ventricular ejection allowing to derive .

The algorithms used by the Physioflow impedance cardiograph device have been reported to provide acceptable and reliable measures of in healthy individuals at rest and during moderate to maximal-intensity exercise (4, 24). An advantage for use in populations with respiratory limitations is that unlike gas rebreathing techniques it is not restricted by the occurrence of gas exchange disturbances. Two studies have to date reported on the agreement of impedance cardiography against direct Fick for exercise determination in COPD (10, 25). Results from Charloux (2000) showed a strong correlation (r = 0.85) between impedance and direct Fick during dynamic exercise with a mean difference of only 0.29 L/min between the two techniques suggesting that impedance cardiography provided clinically acceptable accuracy.

The patient population used in that study was similar in severity of COPD as patients form from the present study with exercise intensities being in the same moderate range. Results obtained in the present study are also in agreement with other reports of bioimpedance-derived exercise cardiac output (26, 27) in COPD patients with similar disease severity. The present observations that the absolute values of obtained through bio-conductance or PaCO₂-derived CO-rebreathing are of the same magnitude suggest that lung hyperinflation does not appear to affect the application of the algorithms derived for the computation of from the changes in impedance.

Study limitation

The extent to which our observations could be extended to higher exercise intensities is limited since the use of indicator gas technique during high intensity exercise in COPD is not possible. Similarly, the impact to more advanced disease severity on the present observations cannot be inferred from the present observations.

Conclusion

Determination of exercise cardiac output may be of substantial clinical value, especially in patients with chronic diseases such as COPD, as it provides additional information on disease mechanism and related functional status. Although use of a direct Fick method for routine cardiac output measurement () may not be justified, numerous non-invasive techniques are now available for use at rest or exercise. The decision for a center to include cardiac output measurement in routine patient assessment or follow-up depends on the accuracy of the information to be provided and the ease of application. Results from this research shows that in patients with gas exchange impairments, the use of PaCO₂ can provide adequate assessment of CO₂-rebreathing derived QC. However, this technique somewhat complex for routine use as it requires a high level of technical ability and expertise in addition to the obtention of an arterial or at minimum an arterialized blood sample. The use of bioimpedance is appealing as it does not require active patient participation and allows for continuous measurement from rest to maximal exercise.

Declaration of Interest Statement

The authors report no conflict of interest. The authors alone are responsible for the content and writing of the paper.

Acknowledgments

Hélène Perrault and Ruddy Richard contributed equally to this work.