Estimating exercise Pa_CO2 in patients with heart failure with preserved ejection fraction

Abstract

Patients with heart failure with preserved ejection fraction (HFpEF) exhibit cardiopulmonary abnormalities that could affect the predictability of exercise $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ from the Jones corrected partial pressure of end-tidal CO₂ (PJ_CO2) equation (PJ_CO2 = 5.5 + 0.9 × $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ − 2.1 × V_T). Since the dead space to tidal volume (V_D/V_T) calculation also includes $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ measurements, estimates of V_D/V_T from PJ_CO2 may also be affected. Because using noninvasive estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T could save patient discomfort, time, and cost, we examined whether partial pressure of end-tidal CO₂ ($\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$) and PJ_CO2 can be used to estimate $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T in 13 patients with HFpEF. $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ was measured from expired gases measured simultaneously with radial arterial blood gases at rest, constant-load (20 W), and peak exercise. V_D/V_T[art] was calculated using the Enghoff modification of the Bohr equation, and estimates of V_D/V_T were calculated using $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ (V_D/V_T[ET]) and PJ_CO2 (V_D/V_T[J]) in place of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$. $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ was similar to $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest (−1.46 ± 2.63, P = 0.112) and peak exercise (0.66 ± 2.56, P = 0.392), but overestimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at 20 W (−2.09 ± 2.55, P = 0.020). PJ_CO2 was similar to $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest (−1.29 ± 2.57, P = 0.119) and 20 W (−1.06 ± 2.29, P = 0.154), but underestimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at peak exercise (1.90 ± 2.13, P = 0.009). V_D/V_T[ET] was similar to V_D/V_T[art] at rest (−0.01 ± 0.03, P = 0.127) and peak exercise (0.01 ± 0.04, P = 0.210), but overestimated V_D/V_T[art] at 20 W (−0.02 ± 0.03, P = 0.025). Although V_D/V_T[J] was similar to V_D/V_T[art] at rest (−0.01 ± 0.03, P = 0.156) and 20 W (−0.01 ± 0.03, P = 0.133), V_D/V_T[J] underestimated V_D/V_T[art] at peak exercise (0.03 ± 0.04, P = 0.013). Exercise $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and V_D/V_T[ET] provides better estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T[art] than PJ_CO2 and V_D/V_T[J] does at peak exercise. Thus, estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T should only be used if sampling arterial blood during CPET is not feasible.

NEW & NOTEWORTHY $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ provides a better estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ than PJ_CO2 at peak exercise, and V_D/V_T[ET] provides a better estimate of V_D/V_T[art] than V_D/V_T[J] at peak exercise. Although we reported significant correlations, we did not find an identity between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, nor did we find an identity between V_D/V_T[art] and estimates of V_D/V_T[art]. Thus, caution should be taken and estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T should only be used if sampling arterial blood during CPET is not feasible.

INTRODUCTION

During a cardiopulmonary exercise test (CPET), measuring arterial blood gases provides useful information about gas exchange efficiency. To avoid the risks associated with arterial puncture, efforts have been made to estimate arterial CO₂ ($\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$) noninvasively. One method uses end-tidal CO₂ ($\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$) in place of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ (1), and a second method estimates $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ from a prediction equation developed by Jones et al. (1) that utilizes $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and tidal volume (V_T)

$${\text{P}\text{J}}_{{\text{CO}}_{\text{2}}}=\text{ 5}.\text{5}+0.{\text{9 × P}\text{ET}}_{{\text{CO}}_{\text{2}}}-\text{2}.{\text{1 × V}}_{\text{T}}$$ (1)

where V_T is in liters.

Studies in younger individuals have shown that $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ is a good index of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest, and PJ_CO2 corrects for the overestimation of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ by $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ during exercise (1–6). In contrast, we (4) and others (1) have shown that in older individuals PJ_CO2 significantly underestimates $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ during exercise, whereas $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ provides a better estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$. These contrasting findings are likely explained by age-related changes in pulmonary function that increase ventilation-perfusion (V/Q) mismatching (7), which alters the relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$.

The relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ is also altered by cardiopulmonary abnormalities (8). In particular, heart failure with preserved ejection fraction (HFpEF) is associated with many cardiopulmonary abnormalities including pulmonary congestion (9, 10), elevated cardiac filling pressures and pulmonary hypertension (11, 12), impaired pulmonary diffusing capacity (13, 14), impaired pulmonary vascular recruitment/distension (14–16), and right ventricle-pulmonary artery uncoupling (17–19). These cardiopulmonary abnormalities could increase V/Q mismatching and thus reduce the predictability of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ from PJ_CO2 in patients with HFpEF, particularly during exercise when cardiopulmonary abnormalities are potentially exacerbated and the magnitude of V/Q mismatching is likely to worsen. In addition to V/Q mismatch, peripheral extraction of oxygen delivered to exercising muscle is impaired with patients with HFpEF (20). Impaired extraction of oxygen at the exercising muscle would reduce the normal elevation in mixed venous Pco₂, which may also alter the relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ during exercise in patients with HFpEF.

An additional variable important in studies of gas exchange is the determination of the magnitude of V/Q mismatch, which can be estimated by calculating the physiologic dead space to tidal volume (V_D/V_T) ratio (21). The standard method of calculating V_D/V_T (V_D/V_T[art]) utilizes the Enghoff modification of the Bohr equation, which includes measurements of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, partial pressure of mixed expired CO₂ ($\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$), and a correction for apparatus mechanical dead space volume (V_DM) (21)

$${\text{V}}_{\text{D}}/{\text{V}}_{\text{T}[\text{art}]}=\left[\left({\text{Pa}}_{\text{CO}}{}_{{}_{\text{2}}}\pm {\text{P}\text{E}}_{{\text{CO}}_{\text{2}}}\right)/{\text{Pa}}_{\text{CO}}{}_{{}_{\text{2}}}\right]\pm \left({\text{V}}_{\text{DM}}/{\text{V}}_{\text{T}}\right)$$ (2)

Although noninvasive estimates of V_D/V_T are included in commercial CPET systems (22), the accuracy of noninvasive V_D/V_T estimates is critically dependent on the accuracy in estimating $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ (21). The accuracy of an estimated V_D/V_T value is important, particularly since obtaining arterial blood samples is frequently not feasible and because an abnormal V_D/V_T response often suggests the presence of underlying pulmonary vascular derangements; a pathophysiological feature that is becoming increasingly prevalent in the HFpEF population (12, 23). Indeed, HFpEF is associated with many cardiopulmonary abnormalities that could result in V/Q mismatch thereby increasing V_D/V_T, particularly during exercise. Moreover, V_D/V_T is a major contributor to exercise intolerance and an elevated ventilatory response in HFpEF (24), which are both widely used as prognostic indicators for heart failure. Since HFpEF is a common clinical population in which exertional dyspnea is frequently documented, and thus in whom CPET is regularly performed (11), the use of noninvasive estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T could save patient discomfort, time, and cost.

Therefore, the purpose of this study was to 1) determine the usefulness of $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and PJ_CO2 to estimate $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and 2) determine if $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ or PJ_CO2 could be used to predict V_D/V_T[art] (V_D/V_T[ET] and V_D/V_T[J], respectively) by substituting either variable for $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ in Eq. 2, in patients with HFpEF. We hypothesized that when compared with PJ_CO2 and V_T/V_T[J], $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and V_D/V_T[ET] would provide better estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T[art] respectively, in patients with HFpEF.

METHODS

Participants

We evaluated 13 patients with HFpEF. Participants were eligible to participate if they had signs and symptoms of heart failure based on Framingham criteria (25), an ejection fraction ≥50%, and evidence of volume overload (e.g., pulmonary edema) confirmed by an elevated pulmonary capillary wedge pressure (PCWP) at rest of >15 mmHg and/or with exercise ≥25 mmHg. Participants with valvular or congenital heart disease, restrictive or infiltrative cardiomyopathy, acute myocarditis, New York Heart Association (NYHA) Class IV chronic heart failure or unstable chronic heart failure, a documented ejection fraction <50% at any time, manifest/provocable ischemic heart disease, or severe obstructive lung disease [forced expiratory volume in 1 s (FEV₁) <40% predicted] were excluded. Before all testing, the purpose and protocols of all testing procedures were disclosed in detail and written and informed consent was obtained. The experimental procedures were reviewed and approved by the UT Southwestern Institutional Review Board (Ref. No: STU2019-0617).

Study Design

Participants visited the laboratory on two occasions separated by at least 48 h. During the first visit, participants underwent preparticipation health screening and performed pulmonary function testing according to ATS/ERS guidelines (26) and a maximal CPET to determine participant eligibility. During the second visit, participants performed a 6-min submaximal constant-load cycling test followed by a rapid incremental cycling test to exhaustion with temperature-corrected arterial blood gas samples (radial artery catheter) collected at rest, constant-load cycling, and peak exercise.

Submaximal Constant-Load Cycling Test

Following resting measurements, participants cycled at 20 W for 6 min on a recumbent cycle ergometer (Lode BV, Groningen, the Netherlands). This exercise work rate (WR) was set to represent the hemodynamic demands of activities of daily living. Heart rate and rhythm were monitored continuously using a 12-lead electrocardiogram, and blood pressure was monitored via arterial catheter waveforms. Gas exchange was measured to quantify minute ventilation (V̇e), V̇o₂, carbon dioxide elimination (V̇co₂), respiratory exchange ratio (RER), breathing frequency (F_b), V_T, and $\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$ using a customized breath-by-breath measurement system (Beck Integrative Physiological System, BIPS; KCBeck, Physiological Consulting, Liberty, UT) integrated with a mass spectrometer (Perkin-Elmer 1100). $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ was sampled via a lateral port located on the mouthpiece.

Rapid Maximal Incremental Cycling Test

Following the submaximal constant-load cycling test, participants performed a rapid maximal incremental cycling test on the same cycle ergometer to determine peak exercise capacity. After a brief rest period, participants started cycling at an initial WR equivalent to 90% of V̇o_2peak that was determined during the CPET performed on the first visit. The WR increased by 5 or 10 W each minute until the participant reached volitional exhaustion or symptom limitation. Heart rate and gas exchange variables were measured as aforementioned for the submaximal constant-load cycling test.

Arterial Catheterization

All participants underwent radial artery catheterization using a modified Seldinger technique with ultrasound guidance before testing during the second visit. Arterial placement was confirmed via observation of arterial pressure waveforms. During the same visit, a Swan-Ganz catheter was placed in the pulmonary artery via brachial or antecubital vein access under fluoroscopic guidance and was used to measure central blood temperature. Immediately following the elimination of sample-line dead space (i.e., 5 mL waste), 2 mL of arterial blood was withdrawn using a syringe with dry lithium heparin for gases and electrolytes for blood gas analysis. These blood gas samples were withdrawn over a 10-s period to reduce the fluctuations of blood gas tensions over a given respiratory cycle during the same time-frame in which $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and gas exchange measurements were made. Arterial blood gases were measured at rest, during the final minute of constant-load exercise, and during the final minute at peak exercise. The samples were immediately placed in an ice bath and were subsequently analyzed (ABL90 FLEX blood gas analyzer, Radiometer). Reference gases and commercial standards were used to calibrate the blood-gas analyzer before all testing.

Data Analysis

PJ_CO2 was derived from the prediction equation developed by Jones et al. (1): PJ_CO2 = 5.5 + 0.9 × $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ − 2.1 × V_T, where V_T is in liters. V_D/V_T[art] was calculated using the Enghoff modification of the Bohr equation (21): V_D/V_T[art] = [($\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ – $\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$)/$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$] – (V_DM/V_T). V_D/V_T estimated from $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ was calculated as: V_D/V_T[ET] = [($\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ – $\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$)/$\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$] – (V_DM/V_T). V_D/V_T estimated from PJ_CO2 was calculated as: V_D/V_T[J] = [(PJ_CO2 – $\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$)/PJ_CO2] – (V_DM/V_T). V_DM was measured to equal 0.18 L. $\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$ was calculated as: 863 × V̇co₂ (STPD)/ V̇e (BTPS).

Statistical Analysis

All data were analyzed using SPSS 22.0 (SPSS, Chicago, IL). Paired t-tests were used to determine whether significant group differences existed between measured $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and estimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ (i.e., $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and PJ_CO2), as well as measured V_D/V_T[art] and estimated V_D/V_T (V_D/V_T[ET] and V_D/V_T[J]) at rest, during constant-load exercise, and at peak exercise. The association between measured and estimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, and measured and estimated V_D/V_T was analyzed using Pearson correlation coefficients. Bland–Altman plots and 95% limits of agreement (LoA) were presented as an indicator of typical measurement error. Also, intraclass correlation coefficients (ICC) were calculated to determine the level of agreement between measured and estimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, and measured and estimated V_D/V_T at rest, during constant-load exercise, and at peak exercise. ICC values were reported in accordance with Bland and Altman (27), where <0.2 is considered poor agreement, 0.21–0.40 as fair agreement, 0.41–0.60 as moderate agreement, 0.61–0.80 as good agreement, and 0.81–1.00 as very good agreement. Statistical significance was defined as P < 0.05 and all data are presented as means ± SD.

RESULTS

Participant Characteristics

Participant characteristics, including spirometric and pulmonary diffusing capacity variables are displayed in Table 1. All participants were middle or older age and had a preserved ejection fraction. Of the 13 participants included in the study, 12 were considered to have obesity based on body mass index (BMI). All participants at the time of the study were currently not smoking. Twelve out of 13 participants had never smoked, whereas the only participant with a history of smoking stopped 30 years prior (∼29 pack·yr history). Five out of the 13 participants had a diagnosis of obstructive sleep apnea. Cardiorespiratory measures obtained at rest, submaximal constant-load exercise, and at peak exercise are shown in Table 2.

Table 1.

Participant characteristics

	Means ± SD
Demographics
Age, yr	69 ± 6
Sex, male/female	5/8
Height, cm	170 ± 10
Weight, kg	110 ± 14
BMI, kg/m2	38.4 ± 7.1
Left ventricular ejection fraction, %	54 ± 2
Pulmonary function
FVC, %pred	87 ± 16
FEV1, %pred	86 ± 17
FEV1/FVC, %	76 ± 11
DLCO, %pred	71 ± 20
DLCO/VA, %pred	108 ± 29

BMI, body mass index; DLCO, diffusing capacity for carbon monoxide; FEV₁, forced expired volume in 1 s; FVC, forced vital capacity; V_A, alveolar volume.

Table 2.

Cardiorespiratory parameters at rest and during exercise

	Means ± SD
Rest
V̇E, L/min	12.44 ± 2.64
VT, L	0.75 ± 0.15
Fb, breaths/min	17.11 ± 4.06
V̇o2, L/min	0.33 ± 0.07
V̇co2, L/min	0.26 ± 0.06
$\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$, Torr	18.1 ± 2.9
RER	0.79 ± 0.05
PCWP, mmHg	5.9 ± 2.4
Constant-load exercise at 20 W
V̇E, L/min	31.31 ± 3.98
VT, L	1.13 ± 0.25
Fb, breaths/min	28.98 ± 6.37
V̇o2, L/min	0.88 ± 0.17
V̇co2, L/min	0.77 ± 0.16
$\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$, Torr	21.2 ± 2.8
RER	0.87 ± 0.05
PCWP, mmHg	20.5 ± 6.6
Peak exercise
WR, W	79 ± 29
V̇E, L/min	61.34 ± 18.78
VT, L	1.50 ± 0.59
Fb, breaths/min	43.56 ± 11.24
V̇o2, L/min	1.30 ± 0.41
V̇o2, %predicted	71 ± 17
V̇o2, mL/min/kg	11.85 ± 3.59
V̇co2, L/min	1.44 ± 0.57
$\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$, Torr	20.2 ± 3.4
RER	1.10 ± 0.11
PCWP, mmHg	32.6 ± 6.7

F_b, breathing frequency; PCWP, pulmonary capillary wedge pressure; $\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$, partial pressure of mixed expired CO₂; RER, respiratory exchange ratio; V̇E, minute ventilation; V̇co₂, carbon dioxide elimination; V̇o₂, oxygen consumption; V_T, tidal volume; WR, work rate.

$\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and Estimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$

$\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ was not significantly different from $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ (Fig. 1, A and B) at rest (−1.46 ± 2.63, P = 0.112) or peak exercise (0.66 ± 2.56, P = 0.392); however, $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ overestimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at 20 W (−2.09 ± 2.55, P = 0.020). $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ were significantly correlated at rest (Fig. 2A), 20 W (Fig. 2B), and peak exercise (Fig. 2C). $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ showed very good agreement with $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest (ICC = 0.824), 20 W (ICC = 0.862), and peak exercise (ICC = 0.953). Bland–Altman analyses show the agreement between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ at rest (Fig. 3A), 20 W (Fig. 3B), and peak exercise (Fig. 3C).

Figure 1.

A: means ± SD values for $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, PJ_CO2, and $\text{P}{\text{ET}}_{\text{C}{\text{O}}_2}$ at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13); B: mean differences between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and PJ_CO2 (P_a-JCO₂), and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ (P_a-ETCO₂) at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13). Data in both A and B were analyzed using paired t tests. *Significant difference between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and PJ_CO2; #significant difference between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$; †significant difference between P_a-ETCO₂ and zero; Ωsignificant difference between P_a-JCO₂ and zero. n, Number of subjects; $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, partial pressure of arterial CO₂; $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$, partial pressure of end-tidal CO₂; PJ_CO2, Jones corrected partial pressure of end-tidal CO₂.

Figure 2.

Relationship between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ (A–C) and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and PJ_CO2 (D–F) at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13). Relationships were analyzed using Pearson correlation coefficients. Dashed line: line of identity; solid line: linear regression. n, Number of subjects; $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, partial pressure of arterial CO₂; $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$, partial pressure of end-tidal CO₂; PJ_CO2, Jones corrected partial pressure of end-tidal CO₂.

Figure 3.

Bland–Altman plots (and 95% LoA) of the average of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ vs. their difference (A–C) and the average of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and PJ_CO2 vs. their difference (D–F) at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13). LoA, limits of agreement; n, number of subjects; $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, partial pressure of arterial CO₂; $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$, partial pressure of end-tidal CO₂; PJ_CO2, Jones corrected partial pressure of end-tidal CO₂.

PJ_CO2 was not significantly different from $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ (Fig. 1, A and B) at rest (−1.29 ± 2.57, P = 0.119) or 20 W (−1.06 ± 2.29, P = 0.154). However, PJ_CO2 underestimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at peak exercise (1.90 ± 2.13, P = 0.009). Also, PJ_CO2 and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ were significantly correlated at rest (Fig. 2D), 20 W (Fig. 2E), and peak exercise (Fig. 2F). PJ_CO2 showed very good agreement with $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest (ICC = 0.806), 20 W (ICC = 0.940), and peak exercise (ICC = 0.959). Bland–Altman analyses show the agreement between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and PJ_CO2 at rest (Fig. 3D), 20 W (Fig. 3E), and peak exercise (Fig. 3F).

V_D/V_T and Estimated V_D/V_T

V_D/V_T[ET] was similar to V_D/V_T[art] (Fig. 5, A and B) at rest (−0.01 ± 0.03, P = 0.127) and peak exercise (0.01 ± 0.04, P = 0.210); however, V_D/V_T[ET] overestimated V_D/V_T[art] at 20 W (−0.02 ± 0.03, P = 0.025). V_D/V_T[ET] and V_D/V_T[art] were significantly correlated at rest (Fig. 5A), 20 W (Fig. 5B), and peak exercise (Fig. 5A). V_D/V_T[ET] showed very good agreement with V_D/V_T[art] at rest (ICC = 0.962), 20 W (ICC = 0.957), and peak exercise (ICC = 0.931). Bland–Altman analyses show the agreement between V_D/V_T[art] and V_D/V_T[ET] at rest (Fig. 6A), 20 W (Fig. 6B), and peak exercise (Fig. 6C).

Although V_D/V_T[J] was similar to V_D/V_T[art] at (Fig. 4, A and B) rest (−0.01 ± 0.03, P = 0.156) and 20 W (−0.01 ± 0.03, P = 0.133), V_D/V_T[J] underestimated V_D/V_T[art] at peak exercise (0.03 ± 0.04, P = 0.013). V_D/V_T[J] and V_D/V_T[art] were significantly correlated at rest (Fig. 5D), 20 W (Fig. 5E), and peak exercise (Fig. 5F). V_D/V_T[J] showed very good agreement with V_D/V_T[art] at rest (ICC = 0.963), 20 W (ICC = 0.959), and peak exercise (ICC = 0.924). Bland–Altman analyses show the agreement between V_D/V_T[art] and V_D/V_T[J] at rest (Fig. 6D), 20 W (Fig. 6E), and peak exercise (Fig. 6F).

Figure 4.

A: means ± SD values for V_D/V_T[art], V_D/V_T[J], and V_D/V_T[ET] at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13); B: mean differences between V_D/V_T[art] and V_D/V_T[J] (V_D/V_T[art-J]), and V_D/V_T[art] and V_D/V_T[ET] (V_D/V_T[art-ET]) at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13). Data in both A and B were analyzed using paired t tests. *Significant difference between V_D/V_T[art] and V_D/V_T[J]; #significant difference between V_D/V_T[art] and V_D/V_T[ET]; †significant difference between V_D/V_T[art-ET] and zero; Ωsignificant difference between V_D/V_T[art-J] and zero. n, Number of subjects; V_D/V_T, dead space to tidal volume ratio.

Figure 5.

Relationship between V_D/V_T[art] and V_D/V_T[ET] (A–C) and V_D/V_T[art] and V_D/V_T[J] (D–F) at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13). Relationships were analyzed using Pearson correlation coefficients. Dashed line: line of identity; solid line: linear regression. n, Number of subjects; V_D/V_T, dead space to tidal volume ratio.

Figure 6.

Bland–Altman plot of the average of V_D/V_T[art] and V_D/V_T[ET] vs. their difference (A–C) and the average of V_D/V_T[art] and V_D/V_T[J] vs. their difference (D–F) at rest (n = 13), 20 W (n = 12), and peak exercise (n = 13). LoA, limits of agreement; n, number of subjects; V_D/V_T, dead space to tidal volume ratio.

DISCUSSION

Consistent with our original hypothesis, these are the first data to demonstrate that although $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and PJ_CO2 provide a reasonable estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest, $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ overestimates $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ during submaximal exercise but provides a better estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ than PJ_CO2 does at peak exercise in patients with HFpEF. Similar observations were made when estimating V_D/V_T[art] using $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and PJ_CO2 in place of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ in the calculation, such that V_D/V_T[ET] overestimated V_D/V_T[art] during submaximal exercise but provided a better estimate of V_D/V_T[art] compared with V_D/V_T[J] at peak exercise.

HFpEF is associated with cardiopulmonary abnormalities that may impact the relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ (9, 10, 13–20). Our results agreed with previous studies in those with normal pulmonary function (1–6), such that $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and PJ_CO2 provided reasonable estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest in patients with HFpEF. Our findings also showed that $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and PJ_CO2 slightly overestimated $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$. These findings contrast with previous studies in healthy individuals where estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ have been shown to slightly underestimate measured $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at rest, which can be ascribed to the dilution of gas from poorly perfused alveoli in the apical regions of the lung due to gravitational forces acting upon pulmonary blood flow (28, 29). However, we must emphasize that the P_a-ETCO₂ (−1.46 ± 2.63 Torr) and P_a-JCO₂ (−1.29 ± 2.57 Torr) differences observed at rest in the present study were very small and well within previously published ranges (1–6). Thus, based on the findings of this study we would argue that the relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ is not significantly impacted at rest in patients with HFpEF. The fact that many compensated patients with HFpEF appear to exhibit normal resting hemodynamics, and that hemodynamic derangements seem to only develop during the stress of exercise supports this suggestion (11).

During exercise, $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ has been shown to increase to a greater extent than $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ in young healthy individuals (6). This is because the partial pressure of alveolar CO₂ ($\text{P}{\text{A}}_{{\text{CO}}_2}$) fluctuates cyclically with breathing, and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ during exercise is greater than the average $\text{P}{\text{A}}_{{\text{CO}}_2}$ over the complete breathing cycle (30), particularly when CO₂ production and delivery to the lungs and V_T are increased, and alveolar air is continuously decreasing throughout exhalation (1). Cardiac output and pulmonary blood flow are also increased during exercise, resulting in improved V/Q matching (particularly in the apical regions of the lung) and increased $\text{P}{\text{A}}_{{\text{CO}}_2}$ (and thus, $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$) (29). The prediction equation developed by Jones et al. (i.e., PJ_CO2) serves to correct for the overestimation of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ by $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ during exercise (1). However, Jones et al. (1) studied young healthy individuals and acknowledged that in patients with abnormal cardiopulmonary function, the use of this prediction equation could be misleading.

In keeping with Jones et al. (1), the results of this study is somewhat contrast to that of previous studies in younger individuals (1–6). Indeed, we demonstrated that although PJ_CO2 provides a reasonable estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ during submaximal exercise, $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ provides a better estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ than PJ_CO2 does at peak exercise in patients with HFpEF. These findings agree with our previous work in older individuals (4), as well as work by others in patients with congestive heart failure (8), who reported that estimating $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ from PJ_CO2 appeared unreliable when measured at peak exercise, irrespective of whether patients also presented with signs of obstructive lung disease (i.e., FEV₁ less than 80% of predicted). Moreover, our findings also agree with Liu et al. (31) who compared the P_a-ETCO₂ difference during exercise in chronic obstructive pulmonary disease (COPD) patients with that in healthy individuals. Although Liu et al (31) did not report PJ_CO2, the difference between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ was significantly less in those with COPD compared with healthy individuals at peak exercise, but the difference between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ in COPD was similar (1.19 ± 2.04 Torr) to what we observed in patients with HFpEF in this study (0.66 ± 2.56 Torr).

The discordance between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and PJ_CO2 during exercise has not been demonstrated before in patients with HFpEF. Based on our results, the reason as to why the predictability of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ by PJ_CO2 is affected cannot be determined. However, it is well known that cardiopulmonary abnormities that increase V/Q mismatching can alter the relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$. Studies have shown that $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ values exceed $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ values during exercise under conditions of V/Q mismatching (32–35). Indeed, individuals with unevenly distributed ventilation and perfusion exhibit lung units where the amount of ventilation is high relative to the amount of blood flow (28, 36). Expired gas coming from these units contains a Pco₂ (i.e., $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$) that is less than $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$. This is because the gas coming from areas with uneven V/Q matching dilutes gas coming from other areas of lung in which ventilation and perfusion are more closely matched, so that when measured during expiration, $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ is lower than $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$. A lower $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ relative to $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ is often observed in patients with pulmonary hypertension due to heterogenous pulmonary blood flow, resulting in a greater relative distribution of high V/Q lung units (37). It could be that a similar mechanism of V/Q mismatch contributed to the slightly lower $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ (relative to $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$) values at peak exercise reported in the patients with HFpEF in this study. The fact that HFpEF is associated with a number of cardiopulmonary abnormalities such as pulmonary congestion (9, 10), elevated cardiac filling pressures and pulmonary hypertension (11, 12). impaired pulmonary diffusing capacity (13, 14), impaired pulmonary vascular recruitment/distension (14–16), and right ventricle-pulmonary artery uncoupling (17–19), all of which could increase V/Q mismatching, supports this suggestion.

Moreover, it is also possible that the relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ could be influenced by mechanical limitations that prevent V_T expansion. Indeed, patients with heart failure typically exercise with a smaller V_T (38). A smaller exercise V_T could attenuate the rise in $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and potentially result in a smaller difference between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$. The fact that V_T at peak exercise (1.13 ± 0.25 L) was similar to that reported in previous studies of mechanical ventilatory limitation in heart failure (1.70 ± 0.20 L), but smaller compared with that reported in healthy individuals (2.43 ± 0.44 L) (3) supports the contention that a smaller V_T could, in part, explain why $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ provides a better estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ than what PJ_CO2 does during exercise in patients with HFpEF. In addition to mechanical limitations imposed on V_T expansion, Jones et al. (1) demonstrated that that magnitude of difference between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ could also be affected by CO₂ production and delivery to the lung. The cohort of patients tested in this study exhibited a reduced exercise capacity, evidenced by a reduced V̇o_2peak expressed as a percent predicted value. As such, these patients likely did not achieve a V̇co₂ at peak exercise that was sufficient to increase $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ beyond $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, to the extent that has been shown previously in healthy individuals.

In light of that stated in the previous paragraph, it may be argued that the prediction equation developed by Jones et al. (PJ_CO2) overcorrects in situations where $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ does not rise to a significantly higher value than $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ during exercise, as may occur in disease states that exhibit cardiopulmonary abnormalities that have the propensity to result in uneven matching of V/Q, mechanical constraints that limit V_T expansion, and a reduced exercise capacity. These findings confirm the need to directly measure $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ during exercise, if one is to accurately assess pulmonary gas exchange efficiency in those with HFpEF, or other cardiac or pulmonary diseases.

V_D/V_T is a measure of physiological dead space and provides a valuable estimate of the degree of V/Q matching (21). With respect to estimating V_D/V_T[art], using $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ or PJ_CO2 as substitutes for $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ in the V_D/V_T calculation showed that V_D/V_T[ET] and V_D/V_T[J] provided reasonable estimates of V_D/V_T[art] at rest, whereas V_D/V_T[J] significantly underestimated V_D/V_T[art] at peak exercise. These findings are in keeping with that of previous studies in respiratory disease (39) and congestive heart failure (8) where V_D/V_T estimated from PJ_CO2 significantly underestimated V_D/V_T measured with $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ at peak exercise. In addition, Vicken and Cosio (40) compared measured V_D/V_T with V_D/V_T estimated from $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ during submaximal constant-load exercise in patients with various respiratory disorders. Although Vicken and Cosio (40) did not report on V_D/V_T estimated from PJ_CO2, V_D/V_T estimated from $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ was similar to measured V_D/V_T; a finding that is similar to that observed in this study. Notably, in this study, the measured V_D/V_T[art] and estimated (V_D/V_T[ET] and V_D/V_T[J]) V_D/V_T values may only differ in the source of the $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ variable in the Enghoff modification of the Bohr equation; this is provided that $\text{P}{\text{E}}_{\text{C}{\text{O}}_2}$ is not affected/diluted by any underlying pulmonary vascular abnormalities. Given that we did not find an identity between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ (as evidenced by relatively large LoA), this may increase the risk of failing to identify possible underlying pulmonary vascular derangements as V_D/V_T[J] appears to underestimate V_D/V_T[art] in those with cardiopulmonary abnormalities, particularly if V_D/V_T is estimated with PJ_CO2.

In addition to demonstrating the usefulness of noninvasive estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T in patients with HFpEF, we reported an abnormal increase in V_D/V_T[art] from rest to peak exercise (rest: 0.30 ± 0.07 vs. peak exercise: 0.35 ± 0.05), in conjunction with a positive P_a-ETCO₂ difference at peak exercise. Taken together, these findings suggest that V/Q mismatching worsened during exercise in patients with HFpEF. The finding that V/Q mismatch worsened, as reflected by a progressive increase in V_D/V_T[art] during exercise, has not been previously demonstrated in patients with HFpEF. The reason as to why V_D/V_T[art] increased during exercise in the present study is unclear, but it is likely due to the development of high V/Q lung units during exercise, which has previously been demonstrated in patients with heart failure and pulmonary hypertension (37, 41). However, it should be noted that patients with HFpEF have not yet had their V/Q relationships analyzed by means of the multiple inert gas elimination technique (42), which remains the only way to quantify the distribution of V/Q relationships throughout the lung. Investigation into this question is beyond the scope of this study but represents an interesting avenue for further research.

Methodological Considerations

We must acknowledge that participants exercised in the recumbent position in this study. Thus, the findings of this study cannot be generalized to patients that perform invasive exercise testing in the supine position, which can alter V/Q relationships in the lung (and the relationship between $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$) by removing gravitational forces acting upon pulmonary blood flow. Furthermore, the decision to perform exercise testing in the recumbent position was made purely to optimize patient comfort and to minimize logistical issues throughout the day of testing. Being positioned in the recumbent position meant that the arm that was catheterized could be placed securely on a portable table next to the participant, which ensured that the catheters remained in place. Also, having the patients in the recumbent position meant that resting periods (before, and in between, exercise bouts) could be taken while remaining on the recumbent cycle ergometer.

Conclusions

Our findings demonstrate that $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ provides a better estimate of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ than PJ_CO2 does at peak exercise in patients with HFpEF. Similar observations were made when estimating V_D/V_T[art] using $\text{P}\text{E}{\text{T}}_{\text{C}{\text{O}}_2}$ and PJ_CO2 in place of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ in the calculation, such that V_D/V_T[ET] provided a better estimate of V_D/V_T[art] compared with V_D/V_T[J] at peak exercise. Although we reported significant correlations, we did not find an identity between $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$, nor did we find an identity between V_D/V_T[art] and estimates of V_D/V_T[art], which were evidenced by relatively large LoA. Therefore, from our data, we emphasize that caution should be taken when estimating individual data, and estimates of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ and V_D/V_T should only be used if sampling arterial blood during CPET is not feasible in the patient with HFpEF. Sampling of arterial blood and direct measurement of $\text{P}{\text{a}}_{\text{C}\text{O}}{}_{{}_2}$ remains the gold-standard clinical practice for calculating V_D/V_T.

GRANTS

B. N. Balmain is supported by an American Physiological Society (APS) Postdoctoral Fellowship. This research was supported by the National Institutes of Health (Grant No.: 1P01HL137630) and Texas Health Presbyterian Hospital Dallas.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.N.B. and T.G.B. conceived and designed research; B.N.B. performed experiments; B.N.B. analyzed data; B.N.B., A.R.T., J.P.M., S.S., B.D.L., L.S.H., and T.G.B. interpreted results of experiments; B.N.B. prepared figures; B.N.B. drafted manuscript; B.N.B., A.R.T., J.P.M., S.S., B.D.L., L.S.H., and T.G.B. edited and revised manuscript; B.N.B., A.R.T., J.P.M., S.S., B.D.L., L.S.H., and T.G.B. approved final version of manuscript.