Pulmonary gas exchange in relation to exercise pulmonary hypertension in patients with heart failure with preserved ejection fraction

Abstract

Background

Exercise pulmonary hypertension, defined as a mean pulmonary arterial pressure (mPAP)/cardiac output $\left(\dot{Q}c\right)$ slope >3 WU during exercise, is common in patients with heart failure with preserved ejection fraction (HFpEF). However, the pulmonary gas exchange-related effects of an exaggerated exercise pulmonary hypertension (EePH) response are not well defined, especially in relation to dyspnoea on exertion and exercise intolerance.

Methods

48 HFpEF patients underwent invasive (pulmonary and radial artery catheters) constant-load (20 W) and maximal incremental cycle testing. Haemodynamic measurements (mPAP and $\dot{Q}c$), arterial blood and expired gases, and ratings of perceived breathlessness (Borg 0–10 scale) were obtained. The $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope was calculated from rest to 20 W. Those with a $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope ⩾4.2 (median) were classified as HFpEF+EePH (n=24) and those with a $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope <4.2 were classified as HFpEF (without EePH) (n=24). The alveolar–arterial oxygen tension difference, dead space to tidal volume ratio (Bohr equation) and the minute ventilation to carbon dioxide production slope (from rest to 20 W) were calculated.

Results

Arterial oxygen tension was lower (p=0.03) and dead space to tidal volume ratio was higher (p=0.03) at peak exercise in HFpEF+EePH than in HFpEF. The alveolar–arterial oxygen tension difference was similar at peak exercise between groups (p=0.14); however, patients with HFpEF+EePH achieved the peak alveolar–arterial oxygen tension difference at a lower peak work rate (p<0.01). The minute ventilation to carbon dioxide production slope was higher in HFpEF+EePH than in HFpEF (p=0.01). Perceived breathlessness was ⩾1 unit higher at 20 W and peak oxygen uptake was lower (p<0.01) in HFpEF+EePH than in HFpEF.

Conclusions

These data suggest that EePH contributes to pulmonary gas exchange impairments during exercise by causing a ventilation/perfusion mismatch that provokes both ventilatory inefficiency and hypoxaemia, both of which seem to contribute to dyspnoea on exertion and exercise intolerance in patients with HFpEF.

Shareable abstract (@ERSpublications)

Pulmonary gas exchange impairment, secondary to abnormal pulmonary arterial haemodynamics, contributes to dyspnoea and exercise intolerance in patients with HFpEF https://bit.ly/3AcW0r6

GRAPHICAL ABSTRACT

Overview of the study. EePH: exaggerated exercise pulmonary hypertension; HFpEF: heart failure with preserved ejection fraction; mPAP: mean pulmonary arterial pressure; $P_{\mathrm{A}-\mathrm{a}{\mathrm{O}}_2}$: alveolar–arterial oxygen tension difference; PAWP: pulmonary arterial wedge pressure; PVR: pulmonary vascular resistance; $\dot{Q}c$: cardiac output; RAP: right atrial pressure; TPG: transpulmonary gradient; TMP: transmural pressure; $V_{\mathrm{D}}/V_{\mathrm{T}}$: dead space to tidal volume ratio; $\dot{V}/\dot{Q}$: ventilation–perfusion; ${\dot{V}}_{\mathrm{E}}$: minute volume; ${\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$: carbon dioxide elimination.

Introduction

Heart failure with preserved ejection fraction (HFpEF) has emerged as the dominant form of heart failure. The pathophysiology is complex, which makes the management of this condition particularly challenging. Exercise pulmonary hypertension (ePH), including that due to left heart disease, is defined as a mean pulmonary arterial pressure (mPAP)/cardiac output ($\dot{Q}c$) slope >3 WU between rest and exercise [1–3]. It is common in patients with HFpEF [4–10]. Although investigations into the cardiac-related consequences of ePH in HFpEF are ongoing, the pulmonary-related consequences of ePH in HFpEF are less well defined. Nevertheless, ePH could alter the distribution of, and/or limit, pulmonary blood flow and contribute to pulmonary gas exchange impairments such as ventilation/perfusion ($\dot{V}/\dot{Q}$) mismatch (e.g. elevated dead space to tidal volume ratio ($V_{\mathrm{D}}/V_{\mathrm{T}}$) and alveolar–arterial oxygen tension difference $\left.\left(P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}\right)\right)$ and ventilatory inefficiency (elevated slope of the relationship between minute ventilation $\left({\dot{V}}_{\mathrm{E}}\right)$ and carbon dioxide elimination (${\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$), i.e. the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope) [11].

Prior studies have assessed pulmonary gas exchange during exercise in patients with HFpEF and ePH. However, these studies have been limited by 1) including patients across a wide range of ejection fractions [10]; 2) performing experiments in the supine position [4, 10, 12], which markedly alters central haemodynamics and $\dot{V}/\dot{Q}$ matching; 3) examining $V_{\mathrm{D}}/V_{\mathrm{T}}$ without accounting for mechanical dead space [10, 12], which can confound its interpretation [13, 14]; and 4) examining the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ ratio at peak exercise [12, 15, 16], which is influenced by an elevated ventilatory response independent of pulmonary gas exchange impairments [17]. Therefore, the conclusions that can be drawn from the reported data regarding the effect of ePH on pulmonary gas exchange during exercise in patients with HFpEF are limited.

Even less is known about whether potential gas exchange impairments relate to the severity of exertional symptoms (dyspnoea on exertion (DOE) and exercise intolerance). Because ePH is associated with worse clinical outcomes (e.g. more frequent hospitalisations and reduced survival) [18], such information on the pulmonary gas exchange response to exercise could have a significant public health outcome by allowing for a better understanding of the (patho)physiological mechanism(s) that may influence DOE and exercise intolerance in patients with HFpEF. This is especially important given that there is a critical need to develop targeted therapeutic interventions with the intention of managing patients’ symptoms more effectively.

Therefore, the purpose of the present study was to test the hypothesis that patients with a higher $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope (i.e. an exaggerated ePH (EePH) response) would demonstrate worse $\dot{V}/\dot{Q}$ mismatch and ventilatory inefficiency, higher DOE and lower exercise tolerance compared with patients with a lower $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope. To test this hypothesis, we evaluated invasive central haemodynamics, arterial and mixed venous blood gases, expired gases, DOE and peak exercise capacity during upright exercise in a cohort of patients with HFpEF.

Methods

This was a retrospective analysis of previously collected data. Although some of these data have been published elsewhere [13, 14, 19–21], we repeat only the methods and data essential to the new findings presented herein.

Participants

We evaluated 48 patients with HFpEF who were enrolled in our larger ongoing study (registered at ClinicalTrials.gov: NCT04068844). Patients with HFpEF were included if they were over the age of 55 years, had signs and symptoms of heart failure based on Framingham criteria [22], an ejection fraction ⩾50% and evidence of pulmonary congestion confirmed by hospitalisation requiring intravenous diuretics, pulmonary oedema by chest X-ray, elevated N-terminal pro-brain natriuretic peptide (>900 pg·mL⁻¹) or a pulmonary arterial wedge pressure (PAWP) of ⩾25 mmHg at peak exercise or increase in PAWP ⩾15 mmHg from rest to peak exercise. Participants were excluded if they had severe valvular heart disease, congenital heart disease, left bundle branch block, known restrictive or infiltrative cardiomyopathy, acute myocarditis, New York Heart Association Class IV chronic heart failure or chronic heart failure that cannot be stabilised on medical therapy, a prior ejection fraction <50%, manifest/provocable ischaemic heart disease, stage 4 or greater chronic kidney disease, significant obstructive lung disease or if they regularly used phosphodiesterase inhibitors (e.g. sildenafil). Prior to all testing, written and informed consent was obtained. The experimental procedures were reviewed and approved by the University of Texas Southwestern Medical Center institutional review board (reference number STU2019–0617).

Study design

All patients visited the laboratory on two occasions. During the first visit, patients underwent pre-participation health screening, a maximal cardiopulmonary exercise test (CPET) to determine participant eligibility and a body composition scan (via dual-energy X-ray absorptiometry), and performed pulmonary function testing according to American Thoracic Society/European Respiratory Society (ERS) guidelines [23]. During the second visit, patients underwent pulmonary artery (PA) and radial artery catheterisations and performed an invasive CPET, comprising a 6-min constant-load (20 W) test and a rapid (~3 min) maximal incremental test on an upright cycle ergometer (Lode BV, Groningen, the Netherlands), as described previously [13, 19, 24]. Also as described previously [13, 19, 24], invasive haemodynamic, arterial and mixed venous blood gas, cardiorespiratory and breathing mechanics measurements were acquired at rest, during the final minute of constant-load exercise and at peak exercise. Breathlessness and exertional measurements, including ratings of perceived breathlessness (RPB) (using a Borg 0–10 scale), ratings of perceived unpleasantness (RPU) (Borg 0–10) and ratings of perceived exertion (RPE) (Borg 6–10), were acquired during the final minute of constant-load exercise [19].

Derived parameters

$\dot{Q}c$ was determined by direct Fick ($\dot{Q}c={\dot{V}}_{{\mathrm{O}}_2}/{\mathrm{C}}_{\mathrm{a}-{\mathrm{v}\mathrm{O}}_2}$ difference), where ${\dot{V}}_{{\mathrm{O}}_2}$ represents oxygen uptake and ${\mathrm{C}}_{\mathrm{a}-{\mathrm{v}\mathrm{O}}_2}$ represents the difference between arterial $\left({\mathrm{C}}_{{\mathrm{a}\mathrm{O}}_2}\right)$ and mixed venous $\left({\mathrm{C}}_{{\mathrm{v}\mathrm{O}}_2}\right){\mathrm{O}}_2$ content. Stroke volume (SV) was determined as the quotient of $\dot{Q}c$ and heart rate (HR). mPAP and PAWP were measured during a brief end-expiratory pause (~3–5 s) at rest and constant-load exercise. The tracings were reviewed in real-time and repeated if there was a Valsalva manoeuvre. mPAP and PAWP were obtained and averaged in triplicate. Peak mPAP and PAWP were measured as the average of spontaneous end-expiratory breaths. Pulmonary vascular resistance (PVR) was calculated as $(\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}–\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P})/\dot{Q}c$ and pulmonary arterial compliance was calculated as SV/(PA systolic pressure–PA diastolic pressure). In keeping with ERS guidelines [1–3] and others [25], the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ and $\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P}/\dot{Q}c$ slopes were calculated from the relationship of mPAP and PAWP (measured at end-expiration), respectively, to $\dot{Q}c$ measured from rest to constant-load exercise. The transpulmonary pressure gradient (TPG) was calculated as mPAP–PAWP, and transmural pressure (TMP) was calculated as PAWP–right atrial pressure (RAP). Furthermore, the $V_{\mathrm{D}}/V_{\mathrm{T}}$ was calculated using the Enghoff modification of the Bohr equation, correcting for mechanical dead space [14, 20, 26], which equalled 0.18 L [14]. The alveolar ${\mathrm{O}}_2$ tension ($P_{{\mathrm{A}\mathrm{O}}_2}$) was calculated using the alveolar gas equation and the difference between alveolar and arterial ${\mathrm{O}}_2$ tension (${\mathrm{P}\mathrm{a}\mathrm{O}}_2$) was calculated as $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}=P_{{\mathrm{A}\mathrm{O}}_2}-P_{{\mathrm{a}\mathrm{O}}_2}.{\mathrm{O}}_2$ delivery was calculated as $\dot{Q}c\times {\mathrm{C}}_{{\mathrm{a}\mathrm{O}}_2}$. The ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope was calculated from the relationship between ${\dot{V}}_{\mathrm{E}}$ and ${\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ measured from rest to submaximal exercise [27, 28], and alveolar ventilation (${\dot{V}}_{\mathrm{A}}$) was calculated as ${\dot{V}}_{\mathrm{E}}\times \left(1-V_{\mathrm{D}}/V_{\mathrm{T}}\right)$. End-expiratory lung volume (EELV) was calculated from inspiratory capacity (IC) and total lung capacity (TLC) measured during body plethysmography (EELV=TLC–IC), and end-inspiratory lung volume (EILV) was calculated as $\text{EELV}+V_T$.

Categorisation of patients

After the invasive CPET, all patients were categorised based on their $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope [25]. Similar to others [29], the median $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope value was used to delineate patients with and without EePH (figure 1a). The $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ median slope was 4.2; those patients with a $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope ⩾4.2 were classified as HFpEF+EePH (n=24) and those patients with a $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope <4.2 were classified as HFpEF (n=24).

FIGURE 1.

a) Graphical representation of grouping strategy for patients with heart failure with preserved ejection fraction (HFpEF). b) Mean values and individual data points for the mean pulmonary arterial pressure (mPAP)/cardiac output ($\dot{Q}c$) slope (measured from rest to constant-load exercise) in patients with HFpEF and patients with HFpEF+exaggerated exercise pulmonary hypertension (EePH). c) Mean values and individual data points for the pulmonary arterial wedge pressure $(\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P})/\dot{Q}c$ slope (measured from rest to constant-load exercise) in patients with HFpEF and patients with HFpEF+EePH. d) Scatterplot showing the relationship between the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope (the relationship between mPAP and $\dot{Q}c$ measured from rest to submaximal exercise) and the $\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P}/\dot{Q}c$ slope (the relationship between PAWP and $\dot{Q}c$ measured from rest to constant-load exercise) in patients with HFpEF and patients with HFpEF+EePH. The dashed lines indicate the linear regression.

Statistical analysis

Data were analysed using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA). A two-way ANOVA with repeated measures was performed to determine differences in outcome measures between groups (HFpEF versus HFpEF+EePH) and across time (rest, constant-load and peak exercise). Where significant interactions were detected, post hoc comparisons using Bonferroni adjustments were performed. Independent t-tests were performed to determine differences between participant characteristics, DOE and slope variables (i.e. $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$, $\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P}/\dot{Q}c$ and ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$). Relationships between variables were assessed using Pearson’s correlation coefficient analyses [30]. Statistical significance was defined as p<0.05. All data are presented as mean±sd, where appropriate.

Results

Participant characteristics

Participant characteristics are displayed in table 1. All participants were not smoking at the time of the study. A history of smoking was reported in 10 patients with HFpEF+EePH (~28 pack-years) and seven patients with HFpEF (~18 pack-years). One patient with HFpEF+EePH had a prior lobectomy. Patients with HFpEF+EePH were older. Whereas both groups had a similar magnitude of obesity (i.e. body mass index) and adiposity (i.e. body fat %, fat mass and visceral fat), patients with HFpEF+EePH had a lower lean muscle mass. Although spirometry and lung volumes were similar between groups, diffusing capacity of the lung for carbon monoxide (D_LCO) was lower in patients with HFpEF+EePH. Nevertheless, pulmonary function was above the lower limits of normal in most patients.

TABLE 1.

Participant characteristics

	HFpEF	HFpEF+EePH
Demographics
Age (years)	68±6	74±7*
Sex (n)
Men	11	6
Women	13	18
Height (cm)	168±9	165±10
Weight (kg)	111±18	100±15*
BMI (kg·m−2)	39.3±6.2	37.0±7.2
Comorbidities (%)
Hypertension	95	95
Diabetes mellitus	50	50
Atrial fibrillation	23	45
Obstructive lung disease	14	14
Obstructive sleep apnoea	82	64
Medications (%)
β-blocker	45	64
Calcium channel blocker	27	9
Angiotensin-converting enzyme inhibitor	73	59
Loop diuretic	73	86
Thiazide diuretic	23	27
Aldosterone agonist	18	36
Body composition
Body fat (%)	46±8	47±9
Total fat mass (kg)	50.2±13.7	47.9±14.3
Lean body mass (kg)	53.9±7.7	48.4±7*
Visceral adipose tissue (kg)	6.3±3.3	6.1±2.4
Pulmonary function
FVC (L)	3.22±0.90	2.66±0.79*
FVC (% predicted)	91±11	87±18
FEV1 (L)	2.42±0.75	1.90±0.54**
FEV1 (% predicted)	91±16	84±18
FEV1/FVC (%)	74±8	73±10
MVV (L·min−1)	90.0±30.5	74.3±22.1
MVV (% predicted)	88±18	81±19
TLC (L)	5.49±1.29	4.92±1.20
TLC (% predicted)	99±14	94±15
FRC (L)	2.59±0.71	2.61±0.94
FRC (% predicted)	103±22	104±21
RV (L)	2.14±0.57	2.11±0.59
RV (% predicted)	97±24	95±19
DLCO (mL·min−1·mmHg−1)	20.3±5.6	15.7±2.8**
DLCO (% predicted)	80±20	69±17
KCO (mL·min−1·mmHg−1·L−1)	4.4±0.7	4.1±0.8
KCO (% predicted)	107±25	110±27

Data are presented as mean±SD unless otherwise indicated. BMI: body mass index; D_LCO: diffusing capacity of the lung for carbon monoxide; EePH: exaggerated exercise pulmonary hypertension; FEV₁: forced expired volume in 1 s; FRC: functional residual capacity; FVC: forced vital capacity; HFpEF: heart failure with preserved ejection fraction; K_CO: carbon monoxide transfer coefficient; MVV: maximal voluntary ventilation; RV: residual volume; TLC: total lung capacity.

^*:p<0.05 versus HFpEF

^**:p<0.01 versus HFpEF.

Haemodynamic responses

The $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ and $\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P}/\dot{Q}c$ slopes were higher in patients with HFpEF+EePH (figure 1b, c; p<0.01). PA systolic and diastolic pressure, mPAP, PAWP, RAP and TMP all increased from rest to constant-load exercise and thereafter to peak exercise in both groups (figure 2a–d, g, l; all p<0.01), whereas PVR decreased from rest to constant-load exercise and continued to decrease to peak exercise in both groups (figure 2e; all p<0.05). Pulmonary arterial compliance (PAC) decreased from rest to constant-load exercise (figure 2f; both p<0.01), but did not significantly change thereafter to peak exercise in either group (both p>0.05). TPG increased from rest to constant-load exercise (figure 2h; both p<0.01), but did not significantly change thereafter to peak exercise in either group (both p>0.05). At rest, PA diastolic pressure, mPAP, PVR, PAC, RAP and TPG were similar between groups (all p>0.05), whereas PA systolic pressure, PAWP and TMP were higher in patients with HFpEF+EePH than in patients with HFpEF (all p<0.05). During constant-load exercise, PA systolic pressure, mPAP, PAWP, PVR and TMP were higher, and PAC was lower, in patients with HFpEF+EePH than in patients with HFpEF (all p<0.01). Although both groups had similar PA systolic and diastolic pressure, mPAP, PAWP, PVR, RAP, TPG and TMP at peak exercise (all p>0.05), patients with HFpEF+EePH achieved these similar peak haemodynamic values at a much lower peak work rate (table 2; p<0.01). Despite the lower peak work rate in patients with HFpEF+EePH, these patients also had a lower PAC than patients with HFpEF (p=0.01). Moreover, HR, SV and $\dot{Q}c$ all increased from rest to constant-load exercise in both groups (figure 3; all p<0.01). While HR and $\dot{Q}c$ increased thereafter to peak exercise in both groups (both p<0.01), SV did not change significantly from constant-load exercise to peak exercise in patients with HFpEF+EePH (p=0.89) or patients with HFpEF (p=0.12). HR, SV and $\dot{Q}c$ were lower in patients with HFpEF+EePH than in patients with HFpEF at peak exercise (all p<0.01).

FIGURE 2.

Mean±sd values for a) pulmonary artery systolic pressure (PA_sys), b) diastolic pressure (PA_dia), c) mean pulmonary arterial pressure (mPAP), d) pulmonary arterial wedge pressure (PAWP), e) pulmonary vascular resistance (PVR), f) pulmonary arterial compliance (PAC), g) right atrial pressure (RAP), h) transpulmonary gradient (TPG) and i) transmural pressure (TMP) at rest and during exercise (20 W and peak exercise) in patients with heart failure with preserved ejection fraction (HFpEF) and patients with HFpEF+exaggerated exercise pulmonary hypertension (EePH). ^# p<0.05 versus HFpEF; ^¶: p<0.05 versus rest; ⁺: p<0.05 versus constant-load exercise.

TABLE 2.

Cardiorespiratory, breathing mechanics and breathlessness measurements

Measure	Rest		Constant-load exercise		Peak exercise
	HFpEF	HFpEF+EePH	HFpEF	HFpEF+EePH	HFpEF	HFpEF+EePH
Work rate (W)	0	0	20	20	90±33¶,+	65±25#,¶,+
${\dot{\mathit{V}}}_{\mathbf{E}}$ (L·min−1)	12.5±2.7	11.9±2.9	29.1±5.1¶	27.1±5.1¶	66.2±19.8¶,+	48.4±13.1#,¶,+
${\dot{\mathit{V}}}_{\mathbf{E}}$ (% MVV)	15.2±5.3	16.8±4.7	36.8±15.1¶	39.6±10.9¶	75.3±17.0¶,+	66.7±13.1¶,+
${\dot{\mathit{V}}}_{\mathbf{E}}$ (% max)	19.2±4.8	25.7±6.8#	47.8±15.4¶	59.2±14.5#,¶	–	–
${\dot{\mathit{V}}}_{\mathbf{A}}$ (L·min−1)	4.8±1.7	4.2±1.6	14.8±3.2¶	12.5±3.0#,¶	37.4±14.3¶,+	24.1±8.8#,¶,+
${\dot{\mathit{V}}}_{{\mathbf{O}}_2}$ (L·min−1)	0.27±0.06	0.23±0.05#	0.85±0.17¶	0.70±0.14#,¶	1.53±0.45¶,+	1.08±0.22#,¶,+
${\dot{\mathit{V}}}_{{\mathbf{O}}_2}$ (mL·kg−1·min−1)	2.5±0.5	2.3±0.5	7.6±1.0¶	7.1±0.9#,¶	13.9±4.1¶,+	10.9±1.9#,¶,+
${\dot{\mathit{V}}}_{{\mathbf{O}}_2}$ (% pred)	–	–	–	–	83±20	71±12#
${\dot{\mathit{V}}}_{{\mathbf{O}}_2}$ (% max)	18±4	21±4#	59±14	66±10#	–	–
${\dot{V}}_{\mathbf{C}{\mathbf{O}}_2}$ (L·min−1)	0.21±0.05	0.19±0.05	0.70±0.14¶	0.60±0.14#,¶	1.68±0.56¶,+	1.10±0.32#,¶,+
RER	0.77±0.06	0.82±0.09	0.82±0.04¶	0.85±0.07	1.10±0.14¶,+	1.02±0.14¶,+
VT (L)	0.80±0.23	0.71±0.16	1.20±0.25¶	1.03±0.21#,¶	1.64±0.52¶,+	1.23±0.41#,¶,+
Fb (breaths·min−1)	16.4±3.5	17.2±4.1	25.0±5.2¶	27.3±7.1¶	40.9±5.9¶,+	40.8±10.9¶,+
IC (L)	2.82±0.78	2.18±0.47#	2.64±0.78¶	1.93±0.59#,¶	2.34±0.70¶,+	1.80±0.62#,¶,+
EILV (L)	3.61±0.82	3.58±1.04	4.16±0.86¶	4.11±1.02¶	4.92±1.15¶,+	4.43±1.10¶,+
EILV (% TLC)	65±7	71±6#	75±7¶	83±6#,¶	87±3¶,+	89±3¶,+
EELV (L)	2.74±0.68	2.80±0.94	2.92±0.67¶	3.02±0.84¶	3.21±0.73¶,+	3.16±0.88¶,+
EELV (% TLC)	50±7	55±7#	52±6¶	61±7#,¶	58±6¶,+	64±7#,¶,+
${\mathbf{C}}_{{\mathbf{a}\mathbf{O}}_2}$ (VOl %)	18.6±2.7	17.1±2.1#	18.9±2.7¶	17.4±2.2#,¶	19.6±3.1¶,+	17.8±2.2#,¶,+
${\mathbf{C}}_{{\mathbf{v}\mathbf{O}}_2}$ (vol %)	12.5±2.5	11.2±1.5#	9.4±2.5¶	7.4±1.2#,¶	7.5±1.9¶,+	5.9±1.3#,¶,+
${\mathbf{C}}_{\mathbf{a}-{\mathbf{v}\mathbf{O}}_2}$ (%)	6.1±0.8	5.8±1.1	9.5±1.0¶	10.0±2.0¶	11.9±1.8¶,+	11.8±2.1¶,+
${\mathbf{D}}_{{\mathbf{O}}_2}$ (mL·min−1)	830±210	690±230#	1590±430¶	1240±300#,¶	2440±700¶,+	1650±400#,¶,+
Lactate (mM)	1.31±0.85	1.20±0.72	1.92±0.73¶	2.13±0.96¶	7.23±2.7¶,+	5.32±1.37#,¶,+
RPB (Borg 0–10)	0.9±1.0	1.0±1.1	2.6±1.7¶	3.7±2.0¶	–	–
RPU (Borg 0–10)	0.9±1.2	0.8±0.9	2.4±2.0¶	3.4±2.2#	–	–
RPE (Borg 6–20)	–	–	11.2±2.6	12.2±2.3	–	–

Data are presented as mean±SD. ${\mathrm{C}}_{\mathrm{a}{\mathrm{O}}_2}$: arterial ${\mathrm{O}}_2$ content; ${\mathrm{C}}_{\mathrm{a}-\mathrm{v}{\mathrm{O}}_2}$: arterial–venous ${\mathrm{O}}_2$ content difference; ${\mathrm{C}}_{{\mathrm{v}\mathrm{O}}_2}$: mixed venous ${\mathrm{O}}_2$ content; ${\mathrm{D}}_{{\mathrm{O}}_2}$: oxygen delivery; EELV: end-expiratory lung volume; EILV: end-inspiratory lung volume; F_b: breathing frequency; HFpEF: heart failure with preserved ejection fraction; IC: inspiratory capacity; RER: respiratory exchange ratio; RPB: rating of perceived breathlessness; RPE: rating of perceived exertion; RPU: rating of perceived unpleasantness of breathlessness; ${\dot{V}}_{\mathrm{A}}$: alveolar ventilation; ${\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}:{\mathrm{C}\mathrm{O}}_2$ elimination; ${\dot{V}}_{\mathrm{E}}$: minute ventilation; ${\dot{V}}_{{\mathrm{O}}_2}:{\mathrm{O}}_2$ uptake; V_T: tidal volume.

^#:p<0.05 versus HFpEF

^¶:p<0.05 versus rest

^+:p<0.05 versus constant-load exercise.

FIGURE 3.

Mean±sd values for a) heart rate (HR), a) stroke volume (SV) and c) cardiac output ($\dot{Q}c$) at rest and during exercise (20 W and peak exercise) in patients with heart failure with preserved ejection fraction (HFpEF) and patients with HFpEF+exaggerated exercise pulmonary hypertension (EePH). ^#: p<0.05 versus HFpEF; ^¶: p<0.05 versus rest; ⁺: p<0.05 versus constant-load exercise.

Pulmonary gas exchange

$P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ (figure 4a) increased from rest to constant-load exercise in patients with HFpEF+EePH (p=0.03), whereas $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ did not change significantly in patients with HFpEF (p=0.40). Thereafter, $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ increased to peak exercise in both groups (both p<0.01). $V_{\mathrm{D}}/V_{\mathrm{T}}$ (figure 4b) decreased from rest to constant-load exercise in both patients with HFpEF+EePH (p=0.03) and with HFpEF (p<0.01) but remained relatively stable thereafter to peak exercise in both groups (both p>0.05). At rest, $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ and $V_{\mathrm{D}}/V_{\mathrm{T}}$ were similar between groups (all p>0.05). During constant-load exercise, all these parameters were higher in patients with HFpEF+EePH than in patients with HFpEF (all p<0.01). Although both groups had a similar $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ at peak exercise (p=0.14), patients with HFpEF+EePH achieved the similar peak $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ at a much lower peak work rate (table 2; p<0.01). In addition, $V_{\mathrm{D}}/V_{\mathrm{T}}$ was higher in patients with HFpEF+ EePH than in patients with HFpEF at peak exercise (p=0.03). The ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope (figure 4c) was also higher in patients with HFpEF+EePH than in patients with HFpEF (p<0.01).

FIGURE 4.

Mean±sd values for a) alveolar–arterial oxygen tension difference ($P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$) and b) dead space to tidal volume ratio ($V_{\mathrm{D}}/V_{\mathrm{T}}$) at rest and during exercise (20 W and peak exercise) in patients with heart failure with preserved ejection fraction (HFpEF) and patients with HFpEF+exaggerated exercise pulmonary hypertension (EePH). c) Mean±sd values for the minute ventilation $\left({\dot{V}}_{\mathrm{E}}\right)/\mathrm{c}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{o}\mathrm{x}\mathrm{i}\mathrm{d}\mathrm{e}$ elimination (${\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$) (measured from rest to constant-load exercise) slope in patients with HFpEF and patients with HFpEF+EePH. ^#: p<0.05 versus HFpEF; ^¶: p<0.05 versus rest; ⁺: p<0.05 versus constant-load exercise.

$P_{{\mathrm{A}\mathrm{O}}_2}$ (figure 5a) did not significantly change from rest to constant-load exercise in either patients with HFpEF+EePH (p=0.71) or with HFpEF (p=0.36) but it increased thereafter to peak exercise in both groups (both p<0.01). While $P_{{\mathrm{a}\mathrm{O}}_2}$ (figure 5b) did not significantly change from rest to constant-load exercise in either group (both p>0.05), it increased thereafter to peak exercise in patients with HFpEF (p<0.01) but not in patients with HFpEF+EePH (p=0.87). Thus, $P_{{\mathrm{a}\mathrm{O}}_2}$ was lower in patients with HFpEF+EePH than in patients with HFpEF at peak exercise (p=0.03). Haemoglobin ${\mathrm{O}}_2$ saturation (${\mathrm{H}\mathrm{b}}_{{\mathrm{O}}_2}\mathrm{\%}$) (figure 5c) did not significantly change across all conditions in patients with HFpEF (all p>0.05); however, it tended to decrease to peak exercise in patients with HFpEF+EePH, such that ${\mathrm{H}\mathrm{b}}_{{\mathrm{O}}_2}\%$ was lower at peak exercise than at rest (p=0.03). $P_{{\mathrm{a}\mathrm{C}\mathrm{O}}_2}$ (figure 5d) increased from rest to constant-load exercise and decreased thereafter to peak exercise in both groups (all p<0.05) and was regulated at similar levels between groups across all conditions (all p>0.05).

FIGURE 5.

Mean±sd values for a) alveolar ${\mathrm{O}}_2$ tension ($P_{{\mathrm{A}\mathrm{O}}_2}$), b) arterial ${\mathrm{O}}_2$ tension ($P_{{\mathrm{a}\mathrm{O}}_2}$), c) haemoglobin ${\mathrm{O}}_2$ saturation $\left({\mathrm{H}\mathrm{b}}_{{\mathrm{O}}_2}\right)$ and d) arterial ${\mathrm{C}\mathrm{O}}_2$ tension $\left({\mathrm{P}\mathrm{a}}_{{\mathrm{C}\mathrm{O}}_2}\right)$ at rest and during exercise (20 W and peak exercise) in patients with heart failure with preserved ejection fraction (HFpEF) and patients with HFpEF+exaggerated exercise pulmonary hypertension (EePH). ^#: p<0.05 versus HFpEF; ^¶: p<0.05 versus rest; ⁺: p<0.05 versus constant-load exercise.

Cardiorespiratory, breathing mechanics and breathlessness responses

All cardiorespiratory, breathing mechanics and breathlessness responses are displayed in table 2. All cardiorespiratory parameters increased from rest to constant-load exercise (all p<0.05) and thereafter to peak exercise in both groups (all p<0.05). Similarly, EELV and EILV (absolute volume and as a %TLC) increased from rest to constant-load exercise (all p<0.05) and thereafter to peak exercise in both groups (all p<0.05). IC decreased in both groups from rest to constant-load exercise and decreased further to peak exercise (all p<0.01), which is consistent with the increase in EELV from rest to constant-load exercise and to peak exercise. EILV and EELV as %TLC were greater in patients with HFpEF+EePH than in patients with HFpEF at rest and constant-load exercise (all p<0.01). Whereas both groups reached a similar EILV (%TLC) at peak exercise (p=0.71), EELV (%TLC) was greater in patients with HFpEF+EePH than in patients with HFpEF (p=0.01). The absolute EELV and EILV values (in L) were similar between groups at rest, constant-load exercise and peak exercise (all p>0.05). Peak work rate and ${\dot{V}}_{{\mathrm{O}}_2}$ (L·min⁻¹, mL·kg⁻¹·min⁻¹ and % predicted) were lower in patients with HFpEF+EePH than in patients with HFpEF (all p<0.01). Although RPB (p=0.21), RPU (p=0.12) and RPE (p=0.21) during constant-load exercise were not statistically different between groups, these parameters were higher in patients with HFpEF+EePH by ⩾1 unit, which is considered clinically significant [31].

Correlations

The $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope correlated with the $\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P}/\dot{Q}c$ slope (r=0.91, p<0.01; figure 1d), the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope (r=0.50, p<0.01; figure 6a) and $P_{{\mathrm{a}\mathrm{O}}_2}$ (r= −0.43, p<0.01; figure 6b). We also found that the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope correlated with RPB (r=0.36, p=0.01; figure 6c) and ${\dot{V}}_{{\mathrm{O}}_2\text{peak}}$ (r= −0.48, p<0.01; figure 6d). Furthermore, the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope correlated with RPB (r=0.44, p<0.01) and ${\dot{V}}_{{\mathrm{O}}_2\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}$ (r=0.47, p<0.01), and $P_{{\mathrm{a}\mathrm{O}}_2}$ correlated with RPB (r= −0.42, p<0.01) and ${\dot{V}}_{{\mathrm{O}}_2\text{peak}}$ (r=0.37, p=0.02). During constant-load exercise, the $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ (r=0.68, p<0.01), $V_{\mathrm{D}}/V_{\mathrm{T}}$ (r=0.385, p<0.01) and ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope (r=0.30, p=0.04) correlated with mPAP, but $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ (r= −0.01, p=0.94), $V_{\mathrm{D}}/V_{\mathrm{T}}$ (r= −0.19, p=0.21) and ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope (r= −0.21, p=0.17) did not correlate with $\dot{Q}c$. At peak exercise, the $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ correlated with mPAP (r=0.30, p=0.04) but not $\dot{Q}c$ (r= −0.06, p=0.67), and $V_D/V_T$ correlated with $\dot{Q}c$ (r= −0.57, p<0.01) but not mPAP (r=0.09, p=0.55).

FIGURE 6.

Scatterplots showing the relationships between the mean pulmonary arterial pressure (mPAP)/cardiac output ($\dot{Q}c$) (measured from rest to constant-load exercise) slope and the a) minute ventilation (${\dot{V}}_{\mathrm{E}}$)/carbon dioxide elimination $\left({\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}\right)$ (measured from rest to constant-load exercise) slope, b) arterial ${\mathrm{O}}_2$ tension (${\mathrm{P}\mathrm{a}\mathrm{O}}_2$), c) rating of perceived breathlessness (RPB) and d) peak oxygen uptake (${\dot{V}}_{{\mathrm{O}}_2\text{peak}}$) in patients with heart failure with preserved ejection fraction (HFpEF) and patients with HFpEF+exaggerated exercise pulmonary hypertension (EePH) during constant-load exercise. The dashed lines indicate the linear regression.

Discussion

The novel findings from this study are: 1) $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$, $V_{\mathrm{D}}/V_{\mathrm{T}}$ and the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope were all higher during exercise in patients with HFpEF+EePH than in patients with HFpEF; 2) $P_{{\mathrm{a}\mathrm{O}}_2}$ (and ${\mathrm{H}\mathrm{b}}_{{\mathrm{O}}_2}\%$) did not increase during exercise despite an increase in $P_{{\mathrm{A}\mathrm{O}}_2}$ in patients with HFpEF+EePH; 3) the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope correlated with the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope, $P_{{\mathrm{a}\mathrm{O}}_2},$, RPB and ${\dot{V}}_{{\mathrm{O}}_2\text{peak}}$; 4) the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope and $P_{{\mathrm{a}\mathrm{O}}_2}$ correlated with RPB and peak oxygen uptake (${\dot{V}}_{{\mathrm{O}}_2\text{peak}}$); 5) $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$, $V_{\mathrm{D}}/V_{\mathrm{T}}$ and the ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope during exercise correlated mostly with mPAP, rather than $\dot{Q}c$; and 6) patients with HFpEF+EePH demonstrated greater mechanical ventilatory limitations than patients with HFpEF.

Patients with HFpEF+EePH had a higher $V_{\mathrm{D}}/V_{\mathrm{T}}$ and ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope during exercise than patients with HFpEF. These findings suggest that EePH increased $\dot{V}/\dot{Q}$ mismatch and ventilatory inefficiency, resulting in an increased ventilatory requirement to minimise changes in $P_{\mathrm{a}{\mathrm{C}\mathrm{O}}_2}$. Related to this, $P_{\mathrm{a}{\mathrm{C}\mathrm{O}}_2}$ was maintained at similar levels between groups, albeit at the expense of a higher ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope in patients with HFpEF+EePH. At present, the mechanisms governing the increase in $\dot{V}/\dot{Q}$ mismatch and ventilatory inefficiency during exercise in patients with HFpEF+EePH are unclear. However, it is possible that in patients with HFpEF+EePH, $\dot{V}/\dot{Q}$ mismatch reflects areas of reduced pulmonary blood flow, either because of an altered distribution of perfusion associated with a higher mPAP (e.g. during constant-load exercise) and/or limited perfusion associated with a lower $\dot{Q}c$ (e.g. at peak exercise). Indeed, an altered distribution of, and/or a decrease in, pulmonary blood flow could limit the ability to perfuse well-ventilated areas of the lungs and, in the face of an increasing ${\dot{V}}_{\text{A}}$ during exercise, give rise to high $\dot{V}/\dot{Q}$ areas (and/or dead space) and lead to ventilatory inefficiency [11].

Moreover, patients with HFpEF+EePH also demonstrated a higher $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ during constant-load exercise and achieved a similar peak at a much lower peak work rate than patients with HFpEF, which further supports the development of $\dot{V}/\dot{Q}$ mismatch [32]. In patients with HFpEF+EePH, the widening of $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ was caused by a smaller (or lack of) increase in $P_{{\mathrm{a}\mathrm{O}}_2}$ despite an increase in $P_{{\mathrm{A}\mathrm{O}}_2}$ at peak exercise. Why $P_{{\mathrm{a}\mathrm{O}}_2}$ did not also increase at peak exercise in patients with HFpEF+EePH is unclear. However, the following reasons may be proposed. First, blood flow through the lung could be redistributed from areas with higher vascular resistance to areas with lower vascular resistance [33]. Such a redistribution of blood flow could cause the lower resistance areas to receive more than their usual amount of blood flow, resulting in a modest reduction in their $\dot{V}/\dot{Q}$ ratio [33]. Second, patients with pulmonary hypertension have a diffusion limitation, evidenced by a decrease in D_LCO [34]. Patients with HFpEF+EePH had a lower D_LCO than patients with HFpEF at rest in the present study, but whether a diffusion limitation manifested during exercise and impaired ${\mathrm{O}}_2$ uptake in the lung cannot be determined [35]. It is also unclear whether the diffusion limitation observed in patients with HFpEF+EePH is due to an inherent diffusion issue at the alveolar–capillary interface and/or $\dot{V}/\dot{Q}$ mismatch. Third, a lower ${\mathrm{C}}_{{\mathrm{v}\mathrm{O}}_2}$ could lower $P_{{\mathrm{a}\mathrm{O}}_2}$ independently of, or contribute to that caused by, $\dot{V}/\dot{Q}$ mismatch and/or diffusion limitation [36]. Fourth, the development of pulmonary oedema could also prevent the rise in $P_{{\mathrm{a}\mathrm{O}}_2}$; however, this seems unlikely because we have previously shown that these patients do not develop pulmonary oedema after an acute bout of exercise in the upright position [24]. When taken together, it is certainly possible that one or a combination of the above-mentioned mechanisms could contribute to the decrease (or lack of increase) in $P_{{\mathrm{a}\mathrm{O}}_2}$ during exercise in patients with HFpEF+EePH.

Patients with HFpEF+EePH displayed specific pathophysiological features during exercise that differ from, and are more severe than, what is observed in patients with HFpEF (without EePH). Indeed, our findings suggest that the manifestation of EePH can impair pulmonary gas exchange in a complex manner. The pulmonary gas exchange impairment appears to be characterised by a $\dot{V}/\dot{Q}$ mismatch that causes both ventilatory inefficiency (high $\dot{V}/\dot{Q}$ areas and/or dead space) and hypoxaemia (low $\dot{V}/\dot{Q}$ areas and/or shunt), the latter of which could be amplified by a diffusion limitation and/or a lower ${\mathrm{C}}_{{\mathrm{v}\mathrm{O}}_2}$ content during exercise. To sort through these mechanisms, one would need to employ the multiple inert gas elimination technique (MIGET). Patients with HFpEF have not yet had their $\dot{V}/\dot{Q}$ relationships analysed by means of the MIGET, which remains the only way to separately measure the distribution of ventilation and perfusion, and quantify $\dot{V}/\dot{Q}$ relationships (low and high $\dot{V}/\dot{Q}$ ratios), shunt, dead space and diffusion limitation [37]. The ability to simultaneously quantify and differentiate these mechanisms of pulmonary gas exchange would provide a new level of understanding of respiratory pathophysiology in this disease, which is especially important for the development of new therapeutic strategies and ultimately guiding therapeutic decision making.

Changes in $V_{\mathrm{D}}/V_{\mathrm{T}}$, $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ and ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope were mostly driven by changes in pulmonary arterial haemodynamics, rather than changes in $\dot{Q}c$, which is a particularly novel finding of our study. Indeed, it is possible that the lower ${\mathrm{O}}_2$ delivery caused an earlier onset of metabolic acidosis and further contributed to an increased ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope. These possibilities likely explain some of the variability observed in the relationships between the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope with ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ and $P_{{\mathrm{a}\mathrm{O}}_2}$. We also demonstrated that the magnitude of the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope, ventilatory inefficiency and hypoxaemia were linked with DOE and exercise intolerance. The variability in these relationships likely reflects the fact that DOE and exercise intolerance can be influenced by a myriad of (patho)physiological mechanisms [38], independent of pulmonary gas exchange impairments. Furthermore, EePH could also be evidence of a physiological derangement(s) that exists at rest (e.g. loss of capillary surface area, pulmonary venous remodelling due to chronic exposure to elevated PAWP, pulmonary vascular stiffness), but is not detectable without provocation (e.g. exercise). Nevertheless, these findings provide new insight into the effect of EePH on pulmonary gas exchange as it relates to exertional symptoms experienced during upright exercise. These findings also suggest that the pulmonary vasculature could be an important therapeutic target to not only potentially improve DOE and exercise intolerance, but also prevent the development of overt pulmonary vascular disease and right heart failure in these patients. Further investigation is warranted to determine how the pulmonary vasculature might be targeted therapeutically to improve clinical and functional outcomes in patients with HFpEF.

Patients with HFpEF+EePH demonstrated greater mechanical ventilatory limitations (e.g. dynamic hyperinflation) than patients with HFpEF, as evidenced by a greater EELV (and small IC) and an EILV encroaching on TLC. At this point, V_T expansion becomes limited and the only option left for the patient is to increase breathing frequency, which is limited by their expiratory flow potential (i.e. age-related decline in maximal expiratory flow) [39, 40]. Indeed, V_T was lower in patients with HFpEF+EePH than in patients with HFpEF. These mechanical ventilatory limitations were even evident during constant-load exercise, which may explain, in part, why some patients often complain of DOE and exercise intolerance at low levels of physical activity. These lower ventilatory reserves were further challenged in patients with HFpEF+EePH during exercise by an increased ventilatory demand (i.e. $V_{\mathrm{D}}/V_{\mathrm{T}}$, $P_{\mathrm{A}-{\mathrm{a}\mathrm{O}}_2}$ and ${\dot{V}}_{\mathrm{E}}/{\dot{V}}_{{\mathrm{C}\mathrm{O}}_2}$ slope), which could have also contributed to the greater DOE and exercise intolerance observed when compared with patients with HFpEF. Assessment of pulmonary function and breathing mechanics measurements made at rest and during exercise are critical in determining mechanical ventilatory limitations and could aid in determining the origin of DOE and exercise limitation. From a mechanistic standpoint, the only way to truly characterise the contribution of mechanical ventilatory limitations to DOE and exercise intolerance is to investigate the effects of unloaded breathing (e.g. with heliox) and see if DOE and exercise intolerance improves in these patients.

According to European Society of Cardiology (ESC)/ERS guidelines [1–3], most of our patients had a high $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope (i.e. >3 WU) and a high $\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P}/\dot{Q}c$ slope (i.e. >2 WU). Furthermore, the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ and $\mathrm{P}\mathrm{A}\mathrm{W}\mathrm{P}/\dot{Q}c$ slopes were strongly correlated (figure 1d), indicating that the increase in mPAP was mostly driven by a backwards transmission of left atrial filling pressure due to left heart filling abnormalities. Thus, assigning patients to a mixed pre- and post-capillary pulmonary hypertension group and a post-capillary pulmonary hypertension group was particularly challenging. Related to this, because 1) we were interested in the pulmonary vascular and gas exchange response to exercise, 2) the more well-accepted ESC/ERS criterion for differentiating between such pulmonary hypertension groups uses supine resting data [1] and 3) the methodologies (e.g. supine versus upright and/or rest versus exercise right heart catheterisation, echocardiography, etc.) and cut-off values vary among different studies [10, 15, 25, 29], we did not use ERS diagnostic criteria for pulmonary hypertension to select the groups. To circumvent these considerations, we used the median $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope to select the groups [29], but we cannot exclude the possibility that other cut-off values might provide a better representation of the prevalence of pulmonary vascular remodelling in this population. However, the purpose of this study was not to identify the prevalence of pulmonary vascular remodelling in HFpEF, nor was it to define a specific (patho)physiological threshold for pulmonary vascular remodelling in this population. It is also likely that some patients in the lower $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/{\dot{Q}}_c$ slope group had some degree of ePH based on ESC/ERS definitions [1–3]. As such, the patients in the present study positioned above or below the median likely do not reflect two distinct phenotypes, given that the magnitude of ePH is likely a reflection of disease severity because pulmonary vascular remodelling occurs in a progressive manner. Nevertheless, grouping patients based on the median $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/{\dot{Q}}_c$ slope permitted appropriate statistical power to make between-group comparisons and provided the opportunity to assess the effects of worsening ePH on pulmonary gas exchange because it related to exertional symptoms in patients with HFpEF. Moreover, one of the major strengths of this study is that all exercise testing was performed in the upright position, which contrasts with other HFpEF and ePH studies that have been performed entirely in the supine position [4, 10, 12]. The latter is not typical of how individuals perform physical activity; therefore, supine exercise may not be an appropriate methodology to inform why patients become symptomatic during activities of daily living (in the upright position).

Conclusion

Our findings suggest that EePH can impair pulmonary gas exchange in a complex manner during exercise in patients with HFpEF by causing a $\dot{V}/\dot{Q}$ mismatch that provokes both ventilatory inefficiency and hypoxaemia, the latter of which could be amplified by a diffusion limitation and/or a lower ${\mathrm{C}}_{{\mathrm{v}\mathrm{O}}_2}$ content during exercise. We also demonstrated that the magnitude of the $\mathrm{m}\mathrm{P}\mathrm{A}\mathrm{P}/\dot{Q}c$ slope, ventilatory inefficiency and hypoxaemia were associated with RPB and ${\dot{V}}_{{\mathrm{O}}_2\text{peak}}$, and that changes in $\dot{V}/\dot{Q}$ mismatch and ventilatory inefficiency were mostly driven by changes in pulmonary arterial haemodynamics, rather than $\dot{Q}c$. This study provides evidence that a pulmonary gas exchange impairment, secondary to abnormal pulmonary arterial haemodynamics, seems to contribute to DOE and exercise intolerance in patients with HFpEF. Therefore, we recommend further investigation into how the pulmonary vasculature might be targeted therapeutically to potentially improve pulmonary gas exchange and, thus, exertional symptoms in patients with HFpEF. Lastly, patients with HFpEF+EePH also demonstrated greater mechanical ventilatory limitations (e.g. dynamic hyperinflation) than patients with HFpEF, which (and in combination with an increased ventilatory demand) could have also contributed to DOE and exercise intolerance.