Effects of exercise on thoracic blood volumes, lung fluid accumulation, and pulmonary diffusing capacity in heart failure with preserved ejection fraction

Abstract

Patients with heart failure with preserved ejection fraction (HFpEF) experience symptoms of exertional dyspnea that may be related to lung fluid accumulation during exercise. A computed tomography (CT)-based method was used to measure exercise-induced changes in extravascular lung fluid content and thoracic blood volumes and to determine the effect of lung fluid on lung diffusing capacity for carbon monoxide (DL_CO) in stable subjects with HFpEF and healthy controls. Nine subjects with HFpEF (age = 68 ± 8 yr; body mass index = 32.1 ± 2.6 kg/m²) and eight healthy controls (62 ± 9 yr, 23.8 ± 2.4 kg/m²) performed triplicate rebreathe DL_CO/DL_NO (lung diffusing capacity for nitric oxide) tests in a supine position at rest and duplicate measurements during two 5-min submaximal exercise stages (15W and 35W) and recovery. Subjects subsequently performed a 5-min exercise bout (35W) inside a CT scanner, and extravascular lung fluid content and thoracic blood volumes were quantified at rest and immediately following exercise from thoracic and contrast perfusion scans, respectively. Subjects with HFpEF had a higher lung fluid content at rest compared with controls (means ± SD, HFpEF: 14.4 ± 1.7%, control: 12.8 ± 1.7%, P = 0.043) and a higher lung fluid content following exercise (15.2 ± 2.0% vs. 12.6 ± 1.5%, P = 0.009). Higher lung fluid content was associated with a lower DL_CO and alveolar-capillary membrane conductance (D_m) in subjects with HFpEF (DL_CO: R = −0.57, P = 0.022, D_m: R = −0.61, P = 0.012) but not in controls. Pulmonary blood volume was not altered by exercise and was similar between groups. Submaximal exercise elicited a greater accumulation of lung fluid in subjects with HFpEF compared with in controls, and lung fluid content was negatively correlated with lung diffusing capacity and alveolar-capillary membrane conductance in subjects with HFpEF.

INTRODUCTION

Pulmonary congestion causes symptoms of dyspnea and exercise intolerance in patients with heart failure (HF). There is evidence of persistent subclinical interstitial edema even in stable patients with HF that worsens with exercise, is related to lung function, and may adversely impact hospital readmission rates (8, 9, 24, 26). Patients with HF and a preserved ejection fraction (HFpEF) are particularly prone to exertional dyspnea and have very limited therapeutic options to treat the underlying disease (3). Our group has recently shown that exercise-induced increases in pulmonary capillary wedge pressure and signs of lung fluid accumulation are related to dyspnea in HFpEF (22, 24). Although increased pulmonary capillary hydrostatic pressure is likely a primary factor for increased lung fluid in HFpEF during exercise, it remains unclear if this is caused solely by increased left atrial pressure or also by centralization of blood volume, altered recruitment/distension of pulmonary capillaries, or ventilation/perfusion mismatch. High pulmonary capillary hydrostatic pressure induces fluid movement into the interstitial and alveolar spaces in the lung, which unless removed by pulmonary lymphatics may impede gas transfer (7, 12, 17).

Computed tomography (CT) methods can be used to quantify thoracic, pulmonary, and cardiac blood volumes, as well as extravascular lung fluid content (1, 9). The functional impact of thoracic blood volume and extravascular lung fluid changes on gas transfer can be discerned by simultaneously measuring lung diffusing capacity for carbon monoxide (DL_CO) and nitric oxide (DL_NO), which enables calculation of alveolar-capillary membrane conductance (D_m) and pulmonary capillary blood volume (V_c). In the present study, we hypothesized that lung fluid accumulation with exercise would be greater in HFpEF, and that this would be associated with abnormalities in lung diffusing capacity. To test this hypothesis, we performed CT imaging, gas exchange, and DL_CO/DL_NO measurements at rest and immediately following exercise in subjects with HFpEF and control subjects to determine the impact of thoracic blood volumes and extravascular lung fluid on lung diffusing capacity, gas exchange, and dyspnea during exercise.

METHODS

Subjects.

Nine subjects with HFpEF [age = 68 ± 8 yr, body mass index (BMI) = 32.1 ± 2.6 kg/m²] and eight healthy controls (age = 62 ± 9 yr, BMI = 23.8 ± 2.4 kg/m²) were recruited to participate in two study sessions performed on the same day. This study was powered for an anticipated change in lung fluid with exercise of 35 HU in mean lung density (detailed in Extravascular lung fluid) in HF (based on previous data using this technique in HFrEF) (8) and no change in controls, as previously reported (19). The power calculation resulted in a sample size of eight subjects per group. Subjects were excluded if they currently smoked, were pacemaker dependent, or had abnormal kidney function (16). The study protocols were approved by the Mayo Clinic Institutional Review Board in accordance with the declaration of Helsinki, and all participants gave written informed consent.

Study protocol.

During the initial study session, each subject underwent a baseline screening exam that included blood sampling and pulmonary function testing. Fresh whole blood was collected via venipuncture for determination of hemoglobin (Sysmex XN9000, Lincolnshire, Illinois) and renal (estimated glomerular filtration rate, creatinine) and cardiac [serum NH₂-terminal prohormone of brain natriuretic peptide (NT-proBNP)] biomarkers (Roche Diagnostics, Indianapolis, IN). Echocardiography and invasive hemodynamics (rest and exercise) were assessed clinically in the supine position as part of the HFpEF diagnostic protocol, and data from medical records were collected and reported to characterize the HFpEF group. All subjects were studied on chronic medications (medications were not withheld during the study). Forced vital capacity tests were conducted using a body plethysmograph system (MGC Diagnostics, Saint Paul, MN) according to American Thoracic Society/European Respiratory Society guidelines (11). Following the screening exam, subjects performed triplicate rebreathe DL_CO/DL_NO maneuvers while supine at rest to establish baseline values and ensure they could correctly perform the maneuver. Subsequently, subjects completed an exercise protocol in a supine position, which consisted of 2 min of rest, two 5-min submaximal stages of cycling exercise (Magnetrainer, 3D Innovations, Greeley, CO) separated by 2 min of rest, and a 5-min recovery phase (Fig. 1). Pulmonary gas exchange was measured continuously throughout the rest and exercise protocol, and DL_CO/DL_NO measurements were performed 4 min into each exercise stage and the recovery phase. Blood pressure measurements were obtained using a manual blood pressure cuff (WelchAllyn 767, Skaneateles, NY) at rest and midway through each exercise stage and the recovery phase. Similarly, measurements of rate of perceived exertion (RPE) and dyspnea were taken during each stage using the Borg scale. Cardiac rate and rhythm were monitored continuously via 12-lead electrocardiography (GE CASE V6, Milwaukee, WI), and peripheral estimate of oxygen saturation was measured from forehead pulse oximetry (Masimo Radical-7, Irvine, CA).

Fig. 1.

Timeline of the study protocol. During the first session, lung diffusing capacity for carbon monoxide (DL_CO) tests were performed in triplicate at rest and twice per exercise stage and recovery. Breath-by-breath gas exchange (GX) was recorded continuously and averaged over the last 2 min of the rest stage and the 30 s before the DL_CO maneuver during each exercise stage and recovery. Subjective measures of rate of perceived exertion (RPE) and dyspnea were recorded following the DL_CO test at each stage. During the computed tomography (CT) session, thoracic and contrast perfusion CT scans were performed before and after a 5-min bout of exercise.

During a second session, subjects reported to a dedicated CT clinical innovation center for full chest and lung perfusion scans that were performed at rest and immediately after 5 min of supine cycle exercise at a workload matched to the second stage of exercise in the DL_CO session (Fig. 1).

Gas exchange and lung diffusing capacity.

Gas exchange measures were obtained while subjects breathed on a mouthpiece connected to a pneumotach to measure the flow signal and a sample line to measure breath-by-breath expired gases (MGC Diagnostics, Saint Paul, MN). Breath-by-breath pulmonary gas exchange was measured continuously throughout the experimental procedures. Measurements of oxygen consumption (V̇o₂), carbon dioxide production (V̇co₂), tidal volume, breathing frequency, and end-tidal CO₂ (${\text{P}\mathrm{E}\mathrm{T}}_{\text{C}{\text{O}}_2}$) and calculations of minute ventilation (V̇e), respiratory exchange ratio (RER), and ventilatory equivalent for CO₂ (V̇e/V̇co₂) were averaged over the 2 min at rest before the start of exercise and the last 30 s before the DL_CO maneuver of each exercise and recovery stage (Fig. 1).

Simultaneous measurements of DL_CO and DL_NO were obtained using a rebreathe maneuver, and cardiac output was determined using the acetylene technique, as described previously in detail (6, 25). Briefly, a switching valve (Hans Rudolph, Shawnee, KS) was used to switch subjects at end expiration from room air to breathing a mixture of gases (35% O₂, 0.6% C₂H₂, 0.3% C¹⁸O, 40 ppm NO, 9% He, and balance N₂) during paced breathing at a rate of 32 breaths/min for 8–10 breaths from a bag containing ∼75% of their inspiratory capacity. Gases in the bag were sampled using a mass spectrometer (Marquette 1100 Medical Gas Analyzer, Perkin-Elmer, St. Louis, MO) and a nitric oxide analyzer (Sievers Instruments, Boulder, CO). Repeat lung diffusing capacity assessments were separated by at least 4 min, and exhaled partial pressure of C¹⁸O was monitored to ensure adequate elimination of the test gas (18). Analysis, including calculation of DL_CO and DL_NO, was performed using custom software (25). Alveolar-capillary membrane conductance (D_m) and pulmonary capillary blood volume (V_c) were calculated according to European Respiratory Society guidelines (28).

Thoracic blood volumes.

Subjects were scanned feet first, and imaging was performed on the same scanner (SOMATOM Force, Siemens, Erlangen, Germany). The Medspira Breath Hold System (Medspira, Minneapolis, MN) was used to ensure that subjects maintained a consistent lung volume throughout the scan protocol. Thoracic CT scans were used to align the ECG-gated contrast perfusion scan. Scanning techniques and calculations have been described previously in detail (9). Briefly, from the thoracic scan, a region of interest was chosen that included the superior vena cava, pulmonary artery, pulmonary vein, and aorta. Iodinated contrast (0.33 mL/kg body wt; mean contrast volume = 26.8 mL) was infused intravenously at the antecubital fossa, and scanning was initiated. A scan (11 slices/scan, 5 mm/slice) was taken during each diastole over a period of 30 s. Cardiac output was calculated as mass of contrast injected (116.7 mg iodine/kg body wt) divided by the area under curve from the pulmonary artery after converting CT attenuation to contrast concentration. The volume of blood within a specific compartment was subsequently calculated by multiplying the blood flow (cardiac output) by the time it takes for a given amount of contrast to move through that compartment (mean transit time, MTT).

Extravascular lung fluid.

The presence of extravascular lung fluid was determined from thoracic CT scans. Lung tissue was segmented from adjacent tissues and large blood vessels using MATLAB built-in active contour algorithms (Mathworks, Natick, MA). Only voxels within the attenuation range of air (−1,000 HU) and water (0 HU) were included in the analysis. The mean lung density was calculated from the attenuation distribution within the segmented areas. Extravascular lung fluid was quantified as fluid content (FC) from the mean lung density (MLD) as FC = (MLD+1,000)/10 (1).

Statistical analysis.

All statistical tests were performed using SPSS (IBM SPSS Statistics 25, Chicago, IL). Shapiro–Wilk tests were used to assess normality of the data. For normally distributed demographic variables, between-groups differences were compared using a t test with a statistical significance level of P < 0.05 and reported as means ± standard deviation. A Mann–Whitney U test was used to compare non-normally distributed demographic variables. NT-proBNP is reported as median (range). A repeated-measures ANOVA was used to determine between-groups (HF vs. control) differences, within-group (rest, stage 1, stage 2, recovery) differences, and interactions and reported as means ± standard deviation. Sphericity of the data was determined using Maulchy’s test, and degrees of freedom were adjusted using the Greenhouse–Geisser correction where appropriate. Blood volumes were compared using a repeated-measures ANOVA used to determine between-groups (HF vs. control) differences, within-group (pre- and postexercise) differences, and interactions and reported as means ± standard deviation. Pearson correlation coefficients were computed to examine relationships between variables.

Lung diffusing capacity tests were removed for two subjects (one HFpEF and one control) due to equipment malfunction. A recovery stage was added to the protocol after the first control subject completed the study, and resting DL variables for this subject were included for baseline comparison between groups only. Four subjects (two controls and two HFpEF) had incomplete CT perfusion data (missing pre- and/or postexercise) due to technical challenges and were not included in thoracic blood volume analyses. Thoracic CT scans were successfully acquired pre- and postexercise in all subjects.

RESULTS

Subjects.

Demographics and medication use at the time of study are presented in Table 1. The groups were similar in proportion of females, age, and height. As is typically observed, subjects with HFpEF displayed higher weight, body mass index, and body surface area compared with controls. Pulmonary function values (FVC and FEV₁) were lower in subjects with HFpEF compared with in controls. The HFpEF group had elevated pulmonary capillary wedge pressure at rest (≥15 mmHg) and during exercise (≥25 mmHg) (4).

Table 1.

Subject characteristics

	Control, n = 8	HFpEF, n = 9	P Value
Subject characteristics
% Female	63	56	0.788
Age, yr	62 ± 9	68 ± 8	0.173
Height, cm	169 ± 8	169 ± 9	0.914
Weight, kg	69 ± 12	92 ± 14	0.002
BMI, kg/m2	23.8 ± 2.4	32.1 ± 2.6	0.000
BSA, m2	1.80 ± 0.21	2.08 ± 0.20	0.010
NT-proBNP, pg/mL	58 (25–221)	183 (34–893)	0.074
Hemoglobin, g/dL	13.6 ± 0.6	12.9 ± 1.4	0.248
Creatinine, mg/dL	0.9 ± 0.1	1.0 ± 0.2	0.376
eGFR, mL/min/m2	77.5 ± 9.6	68.7 ± 16.7	0.208
Medications
β Blocker, %	0	11	0.362
Diuretics, %	0	33	0.081
Resting Hemodynamics
Right atrial pressure, mmHg	-	11 ± 4
Pulmonary artery pressure, mmHg	-	27 ± 9
Pulmonary capillary wedge pressure, mmHg	-	19 ± 7
Exercise Hemodynamics
Right atrial pressure, mmHg	-	17 ± 5
Pulmonary artery pressure, mmHg	-	46 ± 10
Pulmonary capillary wedge pressure, mmHg	-	37 ± 8
Echocardiography
Left ventricular ejection fraction, %	-	60 ± 6
Left atrial, volume index, mL/m2	-	33 ± 12
E/e′ ratio	-	12 ± 5
Pulmonary function
Baseline DLCO, mL·min−1·mmHg−1	18.4 ± 4.6	14.3 ± 3.9	0.087
FVC, l	4.3 ± 1.4	3.1 ± 0.8	0.086
FVC, % predicted	107 ± 16	85 ± 16	0.017
FEV1, l	3.5 ± 1.2	2.3 ± 0.6	0.046
FEV1, % predicted	115 ± 19	86 ± 21	0.013
FEV1/FVC	81.3 ± 2.7	74.9 ± 10.1	0.098

Values are reported as means ± SD. BMI, body mass index; BSA, body surface area; DL_CO, lung diffusing capacity for carbon monoxide; eGFR, estimated glomerular filtration rate, values of >90 were considered to be 90; E/e′, ratio of early mitral inflow velocity to early diastolic mitral annular velocity; FEV₁, forced expiratory volume in 1 s; FVC, forced vital capacity; HFpEF, heart failure with preserved ejection fraction; NT-proBNP, serum NH₂-terminal prohormone of brain natriuretic peptide, reported as median (range), values of <25 were considered to be 25, a Mann–Whitney U test was used to calculate P value.

Thoracic blood volumes.

Thoracic, pulmonary, and cardiac blood volumes (normalized to body surface area) were not different between subjects with HFpEF and controls at rest or following 5 min of cycling exercise (Fig. 2). There was no significant difference between groups in the change in pulmonary blood volume following exercise. However, there was a significant interaction in the change in thoracic (P = 0.046) and cardiac (P = 0.023) blood volumes immediately following exercise: the control group displayed 27% and 68% increases in thoracic and cardiac blood volumes, respectively, whereas the corresponding blood volumes in the thorax and heart were unchanged in the HFpEF group immediately following exercise.

Fig. 2.

Thoracic, pulmonary, and cardiac blood volumes before (left) and after (right) 5-min supine cycling exercise. Significant P values for group × stage interaction are shown. Includes data from five controls and eight subjects with HFpEF. HFpEF, heart failure with preserved ejection fraction.

Extravascular lung fluid.

The CT attenuation distribution for each group (10 HU bin averages) is shown in Fig. 3. At rest, the HFpEF distribution is right-shifted, indicating greater mean lung density, with a larger spread (wider standard deviation), indicating the presence of more extravascular lung fluid. Expressed as fluid content (%), extravascular lung fluid was higher in the HFpEF group (means ± SD, HFpEF: 14.4 ± 1.7%, control: 12.8 ± 1.7%, P = 0.043) at rest.

Fig. 3.

Computed tomography (CT) attenuation distribution for heart failure with preserved ejection fraction (HFpEF) and control groups. The CT attenuation value for each voxel was summed for 10 HU bins (excluding voxels with attenuation values greater than zero). A curve with a less negative mean lung density or greater spread indicates greater extravascular lung fluid content. Standard error of the mean is shown at rest (left) but removed in the pre- and postexercise figure (right) for clarity. Includes data from eight controls and nine subjects with HFpEF.

Immediately after exercise, the CT attenuation distribution shifted further to the right, and the spread increased in subjects with HFpEF, whereas it remained unchanged in controls (Fig. 3). As such, the differences in lung fluid content became greater between subjects with HFpEF and control subjects following exercise (15.2 ± 2.0% vs. 12.6 ± 1.5%, P = 0.009). Compared with the respective baseline values, there was a trend to greater increase in fluid content in the HFpEF group (+0.8 ± 1.2% vs. −0.1 ± 0.5%, P = 0.052).

Gas exchange and lung diffusing capacity.

Oxygen consumption was similar between groups for absolute V̇o₂, but when normalized to body mass, V̇o₂ was lower in subjects with HFpEF at rest and both exercise stages compared with in controls. Absolute V̇co₂ was higher in HFpEF during recovery, whereas RER was greater in HFpEF at stage 2 of exercise and recovery. There was a significant interaction (group × time) in breathing frequency and V̇e/V̇co₂ such that subjects with HFpEF had a greater increase in breathing frequency with exercise and V̇e/V̇co₂ was lower in subjects with HFpEF at rest and decreased less than in controls with exercise (Table 2).

Table 2.

CT imaging variables, pulmonary gas exchange, cardiovascular variables, and subjective measures of exertion at rest, during exercise, and in recovery

	Rest		Exercise (Stage 2)		Recovery		ANOVA, P Value
	Control	HFpEF	Control	HFpEF	Control	HFpEF	Group	Time	Group × Time
CT variables
Fluid content, %	12.8 ± 1.8	14.4 ± 1.7*	-	-	12.6 ± 1.5	15.2 ± 2.0*	0.021	0.146	0.051
Thoracic blood volume, l/m2	0.43 ± 0.16	0.47 ± 0.15	-	-	0.55 ± 0.16	0.43 ± 0.16	0.626	0.320	0.046
Pulmonary blood volume, l/m2	0.22 ± 0.12	0.23 ± 0.12	-	-	0.19 ± 0.11	0.18 ± 0.10	0.953	0.117	0.603
Cardiac blood volume, l/m2	0.22 ± 0.07	0.24 ± 0.09	-	-	0.36 ± 0.11	0.25 ± 0.11	0.367	0.011	0.023
Pulmonary gas exchange
V̇o2, mL/min	284 ± 43	285 ± 33	791 ± 145	791 ± 109	264 ± 48	321 ± 50	0.200	<0.001	0.504
V̇o2, mL·min−1·kg−1	4.2 ± 0.5	3.1 ± 0.4*	11.6 ± 2.0	8.8 ± 1.9*	3.8 ± 0.4	3.5 ± 0.6	0.002	<0.001	0.071
V̇co2, mL/min	222 ± 42	233 ± 45	669 ± 109	753 ± 133	268 ± 38	383 ± 76*	0.010	<0.001	0.246
Respiratory exchange ratio	0.78 ± 0.10	0.82 ± 0.11	0.85 ± 0.09	0.95 ± 0.08*	1.02 ± 0.10	1.18 ± 0.11*	0.015	<0.001	0.450
Ventilation, L/min	8.6 ± 1.8	8.0 ± 1.8	20.9 ± 4.5	23.4 ± 4.4	9.8 ± 1.6	13.5 ± 3.3	0.056	<0.001	0.071
Breathing frequency, breaths/min	16.2 ± 2.6	13.5 ± 5.0	21.9 ± 2.4	23.6 ± 6.1	16.9 ± 2.9	19.0 ± 6.2	0.826	<0.001	0.021
Tidal volume, mL	566 ± 132	660 ± 171	991 ± 251	1071 ± 278	590 ± 91	763 ± 177	0.070	<0.001	0.611
${\text{P}\mathrm{E}\mathrm{T}}_{\text{C}{\text{O}}_2}$, mmHg	37.8 ± 3.3	39.6 ± 3.3	39.5 ± 3.2	39.7 ± 2.8	37.6 ± 2.1	37.6 ± 2.9	0.821	<0.001	0.326
V̇e/V̇co2	38.5 ± 2.7	34.7 ± 4.8	31.1 ± 2.8	31.2 ± 3.0	36.4 ± 2.5	35.0 ± 2.9	0.308	<0.001	0.009
Cardiovascular variables
Heart rate, beats/min	60.9 ± 7.7	69.7 ± 10.3	80.4 ± 8.5	90.2 ± 15.4	67.1 ± 15.0	78.0 ± 13.8	0.100	<0.001	0.939
Mean arterial pressure, mmHg	86.6 ± 9.0	91.2 ± 10.4	93.1 ± 7.9	101.5 ± 11.4	86.2 ± 5.1	93.4 ± 10.5	0.261	<0.001	0.337
Stroke volume, mL	70.9 ± 19.3	52.4 ± 12.3	97.9 ± 22.2	83.7 ± 12.9	76.8 ± 13.8	77.7 ± 17.4	0.158	<0.001	0.283
Cardiac output, L/min	4.1 ± 1.0	3.6 ± 0.9	7.6 ± 1.4	7.4 ± 1.5	5.0 ± 0.8	5.7 ± 1.0	0.981	<0.001	0.135
Cardiac index, L·min−1·m−2	2.3 ± 0.5	1.7 ± 0.4	4.2 ± 0.5	3.6 ± 0.7	2.8 ± 0.5	2.8 ± 0.5	0.159	<0.001	0.17
$\text{S}{\text{p}}_{{\text{O}}_2}$, %	99.3 ± 0.9	98.2 ± 1.4	97.8 ± 1.7	97.8 ± 1.7	98.7 ± 0.8	97.6 ± 5.2	0.475	0.120	0.531
Subjective measures of exertion
RPE	6.1 ± 0.2	6.2 ± 0.4	9.5 ± 2.3	11.7 ± 3.1	8.0 ± 2.4	9.2 ± 2.2	0.534	<0.001	0.831
Dyspnea	0.1 ± 0.2	0.3 ± 0.4	2.1 ± 1.3	2.8 ± 1.6	1.5 ± 1.3	1.6 ± 1.4	0.814	<0.001	0.718

Data are presented as means ± standard deviation. HFpEF, heart failure with preserved ejection fraction; ${\text{P}\mathrm{E}\mathrm{T}}_{\text{C}{\text{O}}_2}$, end-tidal partial pressure of CO₂; RPE, rate of perceived exertion; $\text{S}{\text{p}}_{{\text{O}}_2}$, peripheral estimate of oxygen saturation; V̇co₂, carbon dioxide production; V̇e, minute ventilation V̇o₂, oxygen consumption. ANOVA results include the P value of between-subjects effect (group), within-subjects effect (time), and the interaction (group × time). Subject numbers for each outcome; fluid content, n = 8 controls, 9 HFpEF; blood volumes, n = 5 controls, 8 HFpEF; pulmonary gas exchange, n = 7 controls, 9 HFpEF; cardiovascular variables, n = 7 controls, 9 HFpEF, subjective measures of exertion, n = 7 controls, 9 HFpEF.

^*post hoc difference between groups.

There was a trend toward a lower DL_CO in HFpEF at rest (HFpEF: 14.3 ± 3.9, control: 18.4 ± 4.6 mL·min⁻¹·mmHg⁻¹, P = 0.087), due to a significantly lower V_c (HFpEF: 60.0 ± 9.2, control: 79.7 ± 22.0 mL, P = 0.038). DL_CO, D_m, and V_c were not different between groups during either exercise stage or recovery (Fig. 4).

Fig. 4.

Lung diffusing capacity for carbon monoxide (DL_CO, top), alveolar-capillary membrane conductance (D_m, middle), and pulmonary capillary blood volume (V_c, bottom) at rest (pre), stage 1 (15W), stage 2 (35W), and recovery (post). Includes data from six controls and eight subjects with heart failure with preserved ejection fraction (HFpEF). Error bars represent standard error of the mean.

Relationship between lung diffusing capacity and lung fluid.

In the HFpEF group, there was a significant negative correlation between fluid content and both DL_CO and D_m (Fig. 5). In controls, there was no relationship between fluid content and D_m. Fluid content was not correlated with V_c in either group.

Fig. 5.

Lung diffusing capacity for carbon monoxide (DL_CO, top), alveolar-capillary membrane conductance (D_m, middle), and pulmonary capillary blood volume (V_c, bottom) and fluid content pre- and postexercise. Includes data from five controls and eight subjects with heart failure with preserved ejection fraction (HFpEF).

DISCUSSION

This study used a robust CT-based method to quantify blood distribution, pulmonary congestion, and lung fluid changes with exercise in stable subjects with HFpEF and the functional impact of lung fluid accumulation on lung diffusing capacity. Immediately following exercise, the increase in cardiac blood volume was reduced in HFpEF, in keeping with a reduction in the myocardial compliance reserve. Despite the lesser increase in cardiac blood volume, there was a greater increase in lung fluid following exercise in the HFpEF group, reflected by the rightward-shifted CT attenuation distribution and greater spread, compared with controls at rest, and immediately following submaximal exercise. These slight changes in lung fluid content were negatively correlated with DL_CO and D_m in this stable, early-stage HFpEF cohort, suggesting that modest changes in lung fluid distribution with exercise may contribute to gas conductance impairments in HFpEF.

Subjects with HFpEF displayed higher levels of lung fluid at baseline, consistent with previous findings in compensated HF with reduced ejection fraction (9, 26) and HFpEF (20). It is important to differentiate the current results from patients with earlier stage HFpEF from those observed in patients with severe, advanced HF, where remodeling of pulmonary vessels and reduced filtration keep the lungs dry (10, 13, 27). The mechanisms governing lung fluid transit across the spectrum of HF remain unclear. For example, downregulation of pulmonary β-receptors may impair lung fluid clearance mechanisms (5), and/or elevated systemic venous pressures may cause a back pressure and impair pulmonary lymph drainage (24).

Because pulmonary blood volumes were not different between groups at rest or immediately following exercise, lung congestion in HFpEF appears to be related to extravascular lung fluid and not excessive centralization of blood volume. Although invasive hemodynamics were not assessed simultaneously, it is well known that patients with HFpEF display marked elevation in both right and left heart filling pressures during low-intensity exercise (4, 21). Indeed, the subjects with HFpEF in this study had high pulmonary vascular pressures, particularly during exercise, assessed by right heart catheterization. The blunted increases in cardiac blood volume in the HFpEF group compared with in the control group in the present study despite what were likely much higher filling pressures emphasizes the profound abnormalities in diastolic cardiac compliance in HFpEF (2), which likely contributed to increased fluid filtration by increasing pulmonary capillary hydrostatic pressures.

In this study, lung fluid increased following short-duration submaximal exercise in subjects with HFpEF, aligning with previous findings of exercise-induced extravascular lung water in patients with HFpEF with normal pulmonary capillary wedge pressures at rest but large increases during exercise (24). High pulmonary capillary hydrostatic pressure causes passive fluid movement into the extravascular lung spaces; however, vascular permeability may also be increased in response to mechanical stretch of the endothelium (14), impeding gas transfer from alveoli to pulmonary capillaries, and causing a reduction in alveolar-capillary membrane conductance. In line with this, higher lung fluid content was associated with a lower D_m in the HFpEF group, which suggests that even modest increases in lung fluid have an impact on alveolar-capillary gas conductance, which may limit exercise tolerance.

Interestingly, higher levels of lung fluid content were not associated with a reduced alveolar-capillary membrane conductance in controls. Slight decreases in alveolar-capillary membrane conductance have been shown in healthy individuals following rapid fluid loading (7). Under normal conditions, lung fluid first develops in the interstitial spaces and is removed from the gas exchange interface rapidly by lymphatics, and therefore does not impair gas diffusion. The lack of a relationship may be due to the limited range of fluid content and fluid content changes induced by the relatively low-intensity exercise intervention used in this study, or it may reflect superior lymphatic fluid clearance in controls. Importantly, the values presented in this study are in the healthy range compared with previous studies using CT techniques (∼15% fluid content) (9, 15). Additional studies are needed to determine whether lung fluid shifts within the healthy range are clinically meaningful.

The differences in DL_CO, D_m, and V_c between groups in this study are smaller than those previously reported (23)—a finding that may reflect the relatively healthy cohort of subjects with HFpEF who were stable, willing, and able to participate in a voluntary exercise study. Nonetheless, although DL_CO increased similarly between groups, there was a trend of a lower DL_CO in HFpEF despite similar cardiac output, indicating that pulmonary capillary recruitment and distension patterns may differ between groups. Taken together, recruitment and distension patterns of pulmonary capillaries during exercise may differ in HFpEF, and this may be related to elevated pulmonary capillary hydrostatic pressures and the development of lung fluid.

Physiological and methodological considerations.

The sample size for this study is small; however, the robust methods used enabled the detection of subtle changes in the distribution of blood volume, lung fluid content, and lung diffusing capacity and were adequately powered to detect group differences. Blood volumes and extravascular lung fluid measurements were obtained immediately following exercise (<1 min) and not during exercise to minimize movement artifact and to enable the CT table to be repositioned during cycling to ensure sufficient leg space within the CT bore to pedal, but this applies similarly to subjects with HFpEF and control subjects, which would limit bias. While the exercise intervention was performed in the supine position to control for positional effects between lung diffusing capacity testing and the CT imaging, this may alter the distribution of blood and where fluid accumulates in the lungs compared with upright exercise. Finally, the CT method of quantifying extravascular lung fluid shifts cannot discriminate between lung fluid and blood within the pulmonary vessels. However, in this study, pulmonary blood volumes were measured using a separate perfusion technique, and no differences were observed between groups before or after exercise. For this reason, the differences in fluid content between groups are likely due to extravascular lung fluid and not changes in pulmonary blood volume.

Conclusions.

This study assessed pulmonary congestion using CT techniques before and immediately following exercise and provides evidence of resting and exercise-induced extravascular lung fluid accumulation in stable subjects with HFpEF. Greater lung fluid accumulation was associated with lower lung diffusing capacity and alveolar-capillary membrane conductance in subjects with HFpEF. These data provide new insights into the pathophysiology of pulmonary congestion development during exercise in HFpEF and suggest that lung fluid regulation may be an important therapeutic target for this population.

Perspectives and Significance

Pulmonary congestion is a major cause of hospitalization and death in heart failure; however, it remains poorly defined and may include shifts in extravascular lung fluid and thoracic blood volumes. Patients with heart failure with preserved ejection fraction (HFpEF) have greater extravascular lung fluid at rest, and this further increases with exercise, which relates to impaired gas transfer across the alveolar-capillary membrane. In addition, although patients with HFpEF have evidence of high pulmonary vascular pressures, pulmonary blood volume was not greater in HFpEF. CT techniques enable quantification of pulmonary congestion, which may be important for optimizing therapies for individual patients within the heterogeneous HFpEF population.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute (NHLBI) Grant R01 HL71478 to B. D. Johnson. The American Heart Association Postdoctoral Fellowship (AHA19POST34450022) and a Career Development Award in Cardiovascular Disease Research Honoring Dr. Earl H. Wood from Mayo Clinic was granted to G. M. Stewart. The NHLBI Grants R01 HL128526 and U10 HL110262 supported B. A. Borlaug. Mayo Clinic Graduate School of Biomedical Sciences supported C. C. Fermoyle.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

C.C.F., G.M.S., B.A.B., and B.D.J. conceived and designed research; C.C.F. performed experiments; C.C.F., G.M.S., and B.D.J. analyzed data; C.C.F., G.M.S., and B.D.J. interpreted results of experiments; C.C.F. prepared figures; C.C.F. drafted manuscript; C.C.F., G.M.S., B.A.B., and B.D.J. edited and revised manuscript; C.C.F., G.M.S., B.A.B., and B.D.J. approved final version of manuscript.