Systemic Determinants of Exercise Intolerance in Patients With Fibrotic Interstitial Lung Disease and Severely Impaired D_LCO

Abstract

BACKGROUND:

The precise mechanisms driving poor exercise tolerance in patients with fibrotic interstitial lung diseases (fibrotic ILDs) showing a severe impairment in single-breath lung diffusing capacity for carbon monoxide (D_LCO < 40% predicted) are not fully understood. Rather than only reflecting impaired O₂ transfer, a severely impaired D_LCO may signal deranged integrative physiologic adjustments to exercise that jointly increase the burden of exertional symptoms in fibrotic ILD.

METHODS:

Sixty-seven subjects (46 with idiopathic pulmonary fibrosis, 24 showing D_LCO < 40%) and 22 controls underwent pulmonary function tests and an incremental cardiopulmonary exercise test with serial measurements of operating lung volumes and 0–10 Borg dyspnea and leg discomfort scores.

RESULTS:

Subjects from the D_LCO < 40% group showed lower spirometric values, more severe restriction, and lower alveolar volume and transfer coefficient compared to controls and participants with less impaired D_LCO (P < .05). Peak work rate was ∼45% (vs controls) and ∼20% (vs D_LCO > 40%) lower in the former group, being associated with lower (and flatter) O₂ pulse, an earlier lactate (anaerobic) threshold, heightened submaximal ventilation, and lower S_pO2. Moreover, critically high inspiratory constrains were reached at lower exercise intensities in the D_LCO < 40% group (P < .05). In association with the greatest leg discomfort scores, they reported the highest dyspnea scores at a given work rate. Between-group differences lessened or disappeared when dyspnea intensity was related to indexes of increased demand-capacity imbalance, that is, decreasing submaximal, dynamic ventilatory reserve, and inspiratory reserve volume/total lung capacity (P > .05).

CONCLUSIONS:

A severely reduced D_LCO in fibrotic ILD signals multiple interconnected derangements (cardiovascular impairment, an early shift to anaerobic metabolism, excess ventilation, inspiratory constraints, and hypoxemia) that ultimately lead to limiting respiratory (dyspnea) and peripheral (leg discomfort) symptoms. D_LCO < 40%, therefore, might help in clinical decision-making to indicate the patient with fibrotic ILD who might derive particular benefit from pharmacologic and non-pharmacologic interventions aimed at lessening these systemic abnormalities.

Introduction

Fibrotic interstitial lung disease (fibrotic ILD) is an umbrella term used to describe patients with ILD who, independent of the specific etiology, at some point in time exhibit a progressive fibrosing phenotype.^1,2 Longitudinal loss of lung function is part of the fibrotic ILD definition, and most patients eventually decrease their daily activities due to poor exercise tolerance.³ The consequence is a vicious circle of progressive disablement⁴ and reduced quality of life.⁵ It is crucial, therefore, to identify pulmonary function testing (PFT) parameters that can be used in clinical practice to predict a high burden of activity-related symptoms and exercise intolerance in these patients.^6,7

In this context, a severely impaired (< 40% predicted) single-breath lung diffusing capacity for carbon monoxide (D_LCO)^8,9 has long been recognized as a marker of disease progression across all fibrotic ILD etiologies.¹⁰ The underlying mechanisms and their connection with clinical outcomes (ie, activity-related symptoms), however, are still not fully understood. Although patients with markedly reduced D_LCO present with worst hypoxemia on exertion,^6,7,11 there is a wide variability on exercise tolerance and breathlessness at a given P_aO2.^12-14 D_LCO is a metric of gas transfer that is strongly influenced by the accessible alveolar volume (V_A),^15,16 which, in turn, depends on the severity of the associated restriction (total lung capacity [TLC] decrement)¹⁰ and potential heterogeneity of ventilation distribution^17,18 decreasing the V_A/TLC ratio.¹⁶ D_LCO does correlate well with the anatomical underpinnings of the disease, that is, fibrosis extension.^6,7,11 Of note, a severely impaired D_LCO has been associated with exercise-related pulmonary hypertension in diffuse parenchymal lung disease;^19-21 moreover, those with more pronounced abnormalities in skeletal muscle structure and function show particularly low D_LCO.^22,23 The corollary is that D_LCO < 40% might signal marked decrements in O₂ delivery to and utilization by the contracting skeletal muscles,²⁴ contributing to increased neuromuscular (contractile) fatigue,²⁵ an early lactate (anaerobic) threshold,¹⁹ and, consequently, a heightened sensation of leg discomfort.⁴ It is conceivable, therefore, that rather than just reflecting impaired O₂ transfer across the alveolar-capillary membrane¹⁶ a severely reduced D_LCO is a marker of multiple interconnected respiratory and non-respiratory derangements that jointly conspire to increase the burden of exertional symptoms and, consequently, exercise intolerance in fibrotic ILD.

To explore this overarching hypothesis, we aimed to contrast the physiological (cardiovascular, skeletal muscle metabolism, ventilatory requirements, operating lung volumes, pulmonary gas exchange) and sensory (dyspnea and leg discomfort) responses to incremental exercise in subjects with fibrotic ILD showing a severely impaired D_LCO versus those with mildly-to-moderately reduced D_LCO (ie, ≥ 40%) and non–fibrotic ILD controls of similar sex and age distribution. We postulated that the lowest exercise tolerance shown by subjects in the former group would be associated with (1) greater cardiovascular impairment, (2) an earlier anaerobic threshold, (3) heightened exertional ventilation, (4) more severe mechanical-inspiratory constraints, and (5) worse exertional hypoxemia, leading to a greater cumulative burden of (6) respiratory (dyspnea) and (7) non-respiratory (leg discomfort) sensations. Confirmation of the study hypothesis would provide a solid rationale for D_LCO as a key PFT parameter to expose the pulmonary and systemic consequences of fibrotic ILD.

QUICK LOOK.

Current knowledge

A severely reduced single-breath lung diffusing capacity for carbon monoxide (D_LCO < 40%) has long been associated with exertional hypoxemia in fibrotic lung diseases (fibrotic ILDs). Rather than only indicating impaired O₂ transfer, however, D_LCO < 40% might serve as a physiological marker of interconnected respiratory and non-respiratory abnormalities that increase the burden of exertional symptoms and, consequently, poor exercise tolerance in fibrotic ILD.

What this paper contributes to our knowledge

D_LCO < 40% was associated with deranged cardiovascular/metabolic, mechanical-ventilatory, and gas exchange adjustments to exercise that increased the severity of respiratory (dyspnea) and peripheral (leg discomfort) symptoms in fibrotic ILD. Since patients showing D_LCO < 40% might derive particular benefit from interventions aimed at lessening these systemic abnormalities, our data provide support for the relevance of D_LCO in helping clinical decision-making in pulmonology.

Methods

Participants

This was a cross-sectional study enrolling 67 participants followed in a multidisciplinary clinic specialized in the care of patients with ILD at Kingston Health Science Center, Queen’s University, Kingston, Ontario, Canada. They were prospectively recruited to take part in ethically approved physiological studies.^26-28 Data and participant informed consent were obtained during the initial screening visits to assess eligibility for these studies, but there is no overlap with the current analysis. All subjects had an ILD in which the fibrotic component might progress over time. In keeping with this fibrotic ILD definition,² subjects had idiopathic pulmonary fibrosis (IPF) (n = 46), non-specific interstitial pneumonia (n = 10), chronic hypersensitivity pneumonitis (n = 4), autoimmune/connective tissue disease–related ILD (n = 3), unclassifiable idiopathic interstitial pneumonia (n = 2), and sarcoidosis (n = 2). As common denominators, they showed (1) no evidence of pulmonary hypertension in a recent (within 6 months) transthoracic echocardiogram, (2) absence of musculoskeletal abnormalities that could limit exercise tolerance, and (3) no change in medication or exacerbations requiring oral or intravenous steroids (or another anti-inflammatory therapy) in the 8 weeks preceding study enrollment. Since smoking is highly prevalent in patients with IPF and other fibrotic ILDs,^1,2 we did not exclude those with visually quantified mild emphysema²⁹ to improve the external validity of our results. To further minimize the confounding effects of extensive emphysema on D_LCO, no subject presented with radiographic abnormalities consistent with combined pulmonary fibrosis and emphysema. Twenty-two age- and sex-matched historical controls (age ≥ 40 and presenting with all respiratory functional data at or above the lower limit of normal) were also enrolled. The current study received ethical approval from the Queen’s University and Affiliated Teaching Hospitals Ethics Board (DMED-01659-13).

Procedures

The Modified 0–4 Medical Research Council (mMRC) classification gradated participants’ activity-related dyspnea. PFTs (including dynamic and static lung volumes and hemoglobin-corrected single-breath D_LCO) were performed according to standard techniques and compared to reference values established by the Global Lung Function Initiative.^30-32 Maximal voluntary ventilation was estimated (MVV_est) as FEV₁ x 40 (L/min). Incremental cardiopulmonary exercise testing (CPET) to symptom limitation (10–20 W increase every 2 min) was conducted on an electronically braked cycle ergometer with dynamic inspiratory capacity (IC_dyn) measurements at each work-rate step.³³ The anaerobic threshold was identified based on the V-slope (CO₂ output [V̇_CO2]-O₂ uptake [V̇_O2]) plot with corroboratory evidence from the standard ventilatory method. The minute ventilation (V̇_E)-V̇_CO2 relationship assessed ventilatory (in)efficiency, herein termed excess ventilation.³⁴ We report the frequency of subjects showing the lowest V̇_E-V̇_CO2 (nadir) ≥ 34 or ≥ 40 since these cutoffs were associated with more severe exertional dyspnea and important patient-centered negative outcomes in COPD.³⁵ O₂ saturation was estimated by S_pO2. The intensity of the ventilatory output relative to its theoretical limits (dynamic ventilatory reserve [VR_dyn]) was calculated as [1−(submaximal V̇_E/MVV)] × 100.³⁶ The severity of mechanical constraints to tidal expansion was calculated based on the tidal volume (V_T)/IC_dyn, end-inspiratory lung volume, and inspiratory reserve volume (IRV)/TLC ratios.³³ Since the rate of increase in V_T as a function of V̇_E may become blunted in the presence of critically high inspiratory constraints,³³ we individually determined the V_T-V̇_E inflection point. Borg leg discomfort and dyspnea ratings (0–10 category ratio scale) were also obtained at each work rate. VR_dyn and dyspnea scores at the highest equivalent work rate across all groups (60 W) were compared to sex- and age-adjusted normative data, as established in our laboratory.^36,37 Scores > 75th percentile indicate very severe dyspnea burden.³⁷

Statistical Analysis

Given multiple outcomes, we were unable to a priori power our study. Based on our previous investigations in fibrotic ILD, however, a sample size of at least 18 subjects with fibrotic ILD in each category and 18 controls was deemed sufficient to detect the minimal clinically important difference of 1 Borg unit in dyspnea at the highest equivalent work rate (60 W) achieved by all participants³⁸ (SD = 1, α = 0.05, and β = 0.80). One-way analysis of variance (ANOVA) with post hoc Bonferroni test and chi-square test (for categorical variables) assessed between-group differences in participant characteristics, resting lung function, and selected CPET variables. Two-way repeated measure ANOVA with Bonferroni post hoc multiple comparisons was used to evaluate the effect of group (fixed factor) on key dependent variables during incremental exercise (repeated factor). Statistical significance was set at P < .05.

Results

General Characteristics and Resting Lung Function

Controls and subjects were well matched by demographic and anthropometric characteristics, though body mass index (BMI) was numerically higher in the fibrotic ILD groups (E-Table 1, see related supplementary materials at http://www.rcjournal.com). As expected, most subjects reported a history of (usually cigarette) smoking. More participants in the D_LCO < 40% group reported severe dyspnea on daily life (mMRC score ≥ 3) compared to the other groups. The former group showed the lowest spirometric values and static lung volumes; whereas approximately half of the D_LCO ≥ 40% group had mild restriction, 3 of 4 participants in the D_LCO < 40% group showed moderate-to-severe decrements in TLC (P < .001). Lower D_LCO in the latter group was associated with lower V_A and V_A/TLC ratio: As D_LCO decreased to a greater extent than V_A, these subjects showed the lowest transfer coefficient (κ_CO) (P < .001) (E-Table 1, see related supplementary materials at http://www.rcjournal.com).

Exercise Tolerance

Peak exercise capacity, either expressed as work rate or V̇_O2, was lower in the D_LCO < 40% group compared to other subjects and controls (both P < .001) (Table 1 and E-Fig. 1A, see related supplementary materials at http://www.rcjournal.com). For instance, the frequency of subjects showing peak work rate and/or peak V̇_O2 below the lower limit of normal was 2-fold greater in the D_LCO < 40% group compared to their counterparts with higher D_LCO (Table 1).

Table 1.

Peak and Submaximal Physiological and Sensory Responses to Incremental Cardiopulmonary Exercise Testing in Controls and Subjects With Fibrotic Interstitial Lung Disease Separated by the Severity of Diffusing Capacity for Carbon Monoxide Impairment

Metabolic and Cardiovascular Responses to Exercise

Despite no between-group differences in V̇_O2-work-rate slope (typically between 9–11 mL/min/W in both groups) (E-Fig. 1A, see related supplementary materials at http://www.rcjournal.com), V̇_CO2 increased at a faster rate than V̇_O2 beginning at early exercise in the D_LCO < 40% group (E-Fig. 1B–1C, see related supplementary materials at http://www.rcjournal.com), that is, the DLCO < 40% group demonstrated a higher respiratory exchange ratio (RER) at a given exercise intensity (E-Fig. 1D, see related supplementary materials at http://www.rcjournal.com). This was also influenced by an earlier anaerobic threshold either as a function of V̇_O2 or work rate (Table 1). Given similar V̇_O2 at a given exercise intensity (E-Fig. 1A, see related supplementary materials at http://www.rcjournal.com) but a steeper heart rate-V̇_O2 slope (Table 1), subjects showed lower submaximal and peak O₂ pulse than controls, a finding particularly pronounced in those showing D_LCO < 40% (E-Fig. 1F, see related supplementary materials at http://www.rcjournal.com) (Table 1). Interestingly, subjects in the latter group showed a consistent upward inflection in heart rate close to peak exercise (E-Fig. 1E), resulting in a downward shift in O₂ pulse (E-Fig. 1F, see related supplementary materials at http://www.rcjournal.com).

Ventilatory and Gas Exchange Responses to Exercise

Subjects from the D_LCO < 40% group showed significantly higher V̇_E at a given work rate (Fig. 1A) and metabolic demand (V̇_CO2) (Fig. 1B). Thus, V̇_E/V̇_CO2 was consistently higher throughout exercise (Fig. 1C), and VR_dyn lower (Fig. 1D), in these subjects compared to the other groups (P < .05). In fact, a significantly higher fraction of subjects showing D_LCO < 40% had moderate (≥ 34) and severe (≥ 40) increases in V̇_E/V̇_CO2 and VR_dyn below the sex- and age-adjusted lower limit of normal³⁶ at the highest equivalent work rate of 60 W (both P < .001) (Table 1). These abnormalities were associated with lower end-tidal partial pressure for CO₂ (P_ETCO2) (Fig. 1E) and S_pO2 (Fig. 1F) throughout exercise in the former group (P < .05).

Fig. 1.

Ventilatory (panels A-B and D) and gas exchange (panels C and E-F) responses to incremental exercise in subjects with fibrotic interstitial lung disease with or without a severely decreased diffusing capacity for carbon monoxide (D_LCO) and sex- and age-matched healthy controls. Triangles are values at the tidal volume inflection point, reflecting critically high inspiratory constraints. Data are mean ± SD. P values are for main effect according to 2-way analysis of variance for repeated measures. Significant (P < .05) Bonferroni adjusted post hoc between-group comparisons are indicated by * = D_LCO ≥ 40% versus controls; † = D_LCO < 40% versus controls; ‡ = D_LCO < 40% versus D_LCO ≥ 40%. D_LCO = diffusing capacity for carbon monoxide; V̇_E = minute ventilation; V̇_CO2 = carbon dioxide output; VR_dyn = dynamic ventilatory reserve; P_ETCO2 = end-tidal partial pressure of carbon dioxide.

Operating Lung Volumes and Breathing Pattern During Exercise

As expected by the restrictive abnormalities (E-Table 1, see related supplementary materials at http://www.rcjournal.com), both groups showed a downward shift in their absolute (L) operating lung volumes but higher end-inspiratory lung volumes as a percentage TLC compared to controls (Figs. 2A-2C and 2D-2F, respectively) (P < .05). The D_LCO < 40% group reached similar end-exercise end-inspiratory lung volume/TLC (Fig. 2D-2F) at significantly lower peak work rate (Table 1) compared to D_LCO ≥ 40% (P < .001). Similar V_T (E-Fig. 2A, see related supplementary materials at http://www.rcjournal.com) in the setting of an IC_dyn ∼0.2–0.3 L lower (Table 1) implied a consistent trend to higher V_T/IC_dyn ratio throughout exercise in the D_LCO < 40% group (E-Fig. 2B, see related supplementary materials at http://www.rcjournal.com). In fact, critically high inspiratory constraints (V_T/IC_dyn 70–80%) occurred at work rates ∼10–15 W lower in this group (Table 1 for V_T/IC_dyn and end-inspiratory lung volume/TLC/peak work-rate ratios). In keeping with greater inspiratory constraints, a faster breathing frequency (P < .001) (E-Fig. 2C-2D, see related supplementary materials at http://www.rcjournal.com) was instrumental to explain the higher V̇_E shown by the D_LCO < 40% group (Fig. 1A–1D). In association with a faster breathing frequency, both the inspiratory (T_I) and expiratory time were significantly lower in the latter group (E-Fig. 2E-2F, see related supplementary materials at http://www.rcjournal.com): Given similar T_I/total respiratory time (E-Fig. 2G, see related supplementary materials at http://www.rcjournal.com), the D_LCO< 40% group showed higher mean inspiratory flows (V_T/T_I) (P < .05) (E-Fig. 2H, see related supplementary materials at http://www.rcjournal.com).

Fig. 2.

Absolute (panels A-C) and relative to total lung capacity (panels E-F) operating lung volumes during incremental exercise in subjects with fibrotic interstitial lung disease with or without a severely decreased diffusing capacity for carbon monoxide (D_LCO) and sex- and age-matched healthy controls. Triangles are values at the tidal volume inflection point, reflecting critically high inspiratory constraints. Data are mean ± SD. P values are for main effect according to 2-way analysis of variance for repeated measures. Significant (P < .05) Bonferroni adjusted post hoc between-group comparisons are indicated by * = D_LCO ≥ 40% versus controls; † = D_LCO < 40% versus controls; ‡ = D_LCO < 40% versus D_LCO ≥ 40%. D_LCO = diffusing capacity for carbon monoxide; EELV = end-expiratory lung volume; TLC = total lung capacity.

Sensory Responses to Exercise

Subjects from the D_LCO < 40% group reported the highest leg discomfort scores during submaximal exercise (E-Fig. 1G-1H, see related supplementary materials at http://www.rcjournal.com), a finding particularly noticeable at higher exercise intensities. For instance, a 3-fold lower variation in work rate in the D_LCO < 40% group (60W–70W) compared to D_LCO ≥ 40% (60W–90W) led to the same increase in mean leg discomfort scores (3–5) (E-Fig. 1G, see related supplementary materials at http://www.rcjournal.com). Given a leftward shift in the submaximal dyspnea–work-rate relationship (Fig. 3A), the D_LCO < 40% group showed significantly steeper dyspnea-work rate and dyspnea-V̇_O2 slopes compared to their counterparts and controls (P = .006) (Table 1). Moreover, a higher fraction of subjects in the D_LCO < 40% group reported dyspnea scores in the very severe range when compared to normative data³⁷ (P = .005) (Table 1). In contrast, dyspnea-V̇_E relationship did not differ between the 2 fibrotic ILD groups (P = .26) (Table 1, Fig. 3B). Interestingly, the differences in dyspnea scores between subjects and controls as exercise progressed were markedly reduced versus decreasing ventilatory reserves (VR_dyn; Fig. 3C), lacking statistical significance against lowering inspiratory reserves (IRV/TLC; Fig. 3D).

Fig. 3.

Borg dyspnea scores as a function of exercise intensity (panel A), ventilatory output (panel B), and submaximal ventilatory (panel C) and volume (panel D) reserves in subjects with fibrotic interstitial lung disease with or without a severely decreased diffusing capacity for carbon monoxide (D_LCO) and sex- and age-matched healthy controls. Triangles are values at the tidal volume inflection point, reflecting critically high inspiratory constraints. Data are mean ± SD. P values are for main effect according to 2-way analysis of variance for repeated measures. Significant (P < .05) Bonferroni adjusted post hoc between-group comparisons are indicated by * = D_LCO ≥ 40% versus controls; † = D_LCO < 40% versus controls; ‡ = D_LCO < 40% versus D_LCO ≥ 40%. D_LCO = diffusing capacity for carbon monoxide; V̇_E = minute ventilation; VR_dyn = dynamic ventilatory reserve; IRV = inspiratory reserve volume; TLC = total lung capacity.

Discussion

The current study shed novel light into the physiological and sensory mechanisms by which a severely reduced single-breath D_LCO (< 40%) indicates more pronounced impairment in exercise tolerance in subjects showing the common phenotype of fibrotic ILD.^1,2 As we a priori postulated, such a severe decrement in gas transfer marked interconnected systemic abnormalities in cardiocirculatory (tachycardia relative to V̇_O2), metabolic (earlier shift to anaerobic metabolism), ventilatory (excess ventilation), lung-mechanical (inspiratory constraints), and pulmonary gas exchange (hypoxemia and hyperventilation) adjustments to exercise that were translated into a higher burden of respiratory (breathlessness) and peripheral (leg discomfort) symptoms. In association with our previous results showing the clinical relevance of a low D_LCO in the initial stages of fibrotic ILD,^27,39 these data expose the key role of D_LCO as a physiological biomarker of activity-related impairment across the spectrum of disease severity.

Resting Functional Abnormalities According to the Severity of D_LCO Impairment in fibrotic ILD

The D_LCO < 40% group showed significantly lower spirometric values and more relevant restrictive abnormalities compared to their counterparts with higher D_LCO (E-Table 1). Statistically lower vital capacity and numerically lower IC, in particular, were secondary to greater decrements in TLC than residual volume and functional residual capacity, respectively. Thus, subjects from the D_LCO < 40% group started exercising with relatively smaller room for V_T expansion,³³ a finding that strongly contributes to exertional dyspnea in these subjects (also see Inspiratory Constraints on Exertion and a Severely Reduced D_LCO in fibrotic ILD).²⁷ Lower resting lung volumes also imply that part of the severe impairment in D_LCO can be ascribed to lower distribution volume, that is, V_A (E-Table 1, see related supplementary materials at http://www.rcjournal.com). Interestingly, lower V_A/TLC ratio indicates a greater degree of ventilation distribution abnormalities, further reducing D_LCO.⁴⁰ V_A/TLC is thought to be influenced by distributive abnormalities at the level of the small airways.⁴¹ Future studies using more specific tests of small airway function⁴² are warranted to link any abnormalities with their putative structural underpinning, eg, architectural distortion, traction bronchiolectasis.⁴³ Out-of-proportion decreases in D_LCO relative to volume loss (that is, lower K_CO) points out impaired gas exchange efficiency and likely pulmonary vascular abnormalities,²⁰ both leading to a high V̇_E/V̇_CO2³⁴ (also see Excess Exertional Ventilation and a Severely Reduced D_LCO in fibrotic ILD). Jointly, lower lung volumes, thickened alveolar-capillary membrane in more severe subjects, disturbed pulmonary blood flow,¹⁰ and ventilation distribution inhomogeneities resulting in ventilation/perfusion (V̇/Q̇) mismatch may have contributed to markedly slowing the rate of CO transfer in the D_LCO < 40% group.

Excess Exertional Ventilation and a Severely Reduced D_LCO in fibrotic ILD

Excess ventilation has been described in a plethora of cardiopulmonary diseases in which breathlessness and exercise intolerance are dominant features.^34,44,45 Both inefficient ventilation (ie, increased wasted ventilation in areas of physiological dead space [VD_phys])⁴⁶ and alveolar hyperventilation (low P_aCO2)^47,48 may explain a high V̇_E/V̇_CO2 in these subjects.^38,41 In the current study, excess ventilation, even at rest, was a noticeable abnormality in the D_LCO < 40% group (Fig. 1C). As mentioned, D_LCO and VD_phys share the commonality of being influenced by V̇/Q̇ mismatch,⁴⁹ particularly the extension of areas of high V̇/Q̇, that is, dead-space effect.⁴⁶ Areas of low V̇/Q̇ (shunt effect), a key determinant of hypoxemia as seen in fibrotic ILD,¹⁸ also contribute to decrease D_LCO and increase Bohr-Enghoff VD_phys.⁴⁶ Fibrotic tissue deposition, architectural distortion, increased axial airway traction, patchy foci of enlarged air spaces, and vascular obliteration/dysfunction¹⁰ common to all fibrotic ILDs^1,2 may provide the structural substrate to V̇/Q̇ disturbances, high VD_phys, and low P_aO2 that contributed to the highest V̇_E/V̇_CO2 shown by the D_LCO < 40% group (Table 1, Fig. 1D).⁵⁰

It should be recognized, however, that the D_LCO < 40% group showed higher RER since the earlier stages of exercise (E-Fig. 1D, see related supplementary materials at http://www.rcjournal.com), that is, hyperventilation well before any compensatory respiratory alkalosis to metabolic acidosis. Although we did not measure the source of carotid chemostimulation on hypoxia, a low P_aO2,⁴⁵ S_pO2 was only mildly reduced at these low exercise intensities, that is, ≥ 90% (Fig. 1F). Thus, other sources of ventilatory (over)stimulation likely contributed to a high V̇_E/V̇_CO2 in the D_LCO < 40% group, eg, increased central chemosensitivity (low CO₂ set point), airway-lung-chest-wall mechano(vagal) receptors overactivity,^51,52 heightened ergo receptor discharge⁴⁵ (also see Heightened Leg Discomfort on Exertion and a Severely Reduced D_LCO in fibrotic ILD), and increased pulmonary vascular pressures.⁵³ It is also noteworthy that a high VD_phys frequently coexists with a low P_aCO2, exposing the challenges faced by the central controller to stabilize mean alveolar P_CO2 when ventilatory output is not readily translated into the expected increase in alveolar ventilation.^43,47 In any case, the lowest (Fig. 1E) and blunted (lower rest-apex difference in Table 1) P_ETCO2 in the D_LCO < 40% group may have resulted from expired P_CO2 dilution due to enlarged VD_phys/impaired perfusion plus alveolar hyperventilation.⁴⁶

Inspiratory Constraints on Exertion and a Severely Reduced D_LCO in fibrotic ILD

In keeping with resting lung function data indicating slightly lower lung volumes (E-Table 1), subjects from the D_LCO < 40% group reached similar V_T/IC_dyn and end-inspiratory lung volume/TLC at significantly lower peak work rates compared to D_LCO ≥ 40% (Table 1). Of note, the V_T inflection point, indicating critical constraints to further lung-chest-wall displacement³³ occurred at similar V̇_E in both subject groups (∼45 L/min) (Fig. 1A) (also see Exertional Dyspnea and a Severely Reduced D_LCO in fibrotic ILD). Given the higher ventilatory response (Fig. 1A-1D), however, this occurred at significantly lower exercise intensities. In the setting of higher operating lung volumes and a closer proximity to TLC (Fig. 2D-2F),⁴⁵ the heightened ventilatory requirements in this group was primarily met by faster breathing frequencies rather than larger V_T (E-Fig. 2C-2D, see related supplementary materials at http://www.rcjournal.com). Tachypnea may have also been influenced by higher chemostimulation when S_pO2 decreased to ≤ 88%, that is, values that are more likely to be associated with substantial carotid body stimulation (P_aO2 < 60 mm Hg).¹¹ Regardless of the underlying mechanisms, faster breathing frequency at a similar duty cycle (E-Fig. 2G, see related supplementary materials at http://www.rcjournal.com) implies a shorter T_I (E-Fig. 2F); consequently, the mean inspiratory flow (V_T/T_I ratio) was higher in the D_LCO < 40% group (E-Fig. 2H, see related supplementary materials at http://www.rcjournal.com). As long postulated by Burdon and colleagues,⁵⁴ shorter T_I and higher V_T/T_I may serve to optimize V_T and breathing frequency, likely to minimize the elastic and resistive work of breathing and, potentially, breathlessness in fibrotic ILD.

Heightened Leg Discomfort on Exertion and a Severely Reduced D_LCO in fibrotic ILD

There is growing evidence that decrements in D_LCO are associated with abnormalities in peripheral (and potentially respiratory) muscle structure and function in fibrotic ILD.^22,23 Higher sense of leg discomfort throughout exercise in the D_LCO < 40% group (E-Fig. 1G–1H, see related supplementary materials at http://www.rcjournal.com) is likely multifactorial. For instance, these subjects reported more severe dyspnea in daily life (E-Table 1, see related supplementary materials at http://www.rcjournal.com); thus, they were likely less active and deconditioned. Lower O₂ delivery secondary to reduced O₂ content likely contributed as well as any impairment in intramuscular blood flow distribution.²⁴ Whether the consistent downward shift in O₂ pulse closer to exercise cessation (E-Fig. 1F, see related supplementary materials at http://www.rcjournal.com) signals impaired stroke volume (secondary to higher pulmonary vascular pressures)²¹ and/or the consequences of more severe hypoxemia (reflex increase in heart rate and/or reduced arterial-venous O₂ gradient) remain unclear, demanding invasive CPET. We recently showed that reduced convective O₂ delivery²⁴ either secondary to hypoxemia and/or lower muscle blood flow is mechanistically linked to heightened neuromuscular (contractile) fatigue²⁵ and perceived sense of fatigability,⁵⁵ both being predictors of poor exercise tolerance in fibrotic ILD. As shown in Table 1, subjects from the D_LCO < 40% group did present with an earlier anaerobic threshold that may have further increased the discharge of overexcited ergo receptors.⁴⁵ It is rather interesting to note the temporal concomitance between an upward shift in leg discomfort with the V_T inflection point (triangles in E-Fig. 1G–1H, see related supplementary materials at http://www.rcjournal.com). Although this may not reflect causality, putative explanations include (1) given the parallel acceleration of dyspnea at this point (Fig. 3), the subjects may have become more aware of any simultaneous uncomfortable sensations; (2) negative hemodynamic consequences of heightened pleural pressure swings and/or blood flow redistribution to the overloaded respiratory muscles⁴⁵ may have further impaired muscle O₂ delivery; and (3) the blunting effect of the respiratory compensation to lactic acidosis on V_T with simultaneous increase in breathing frequency.⁵⁶

Exertional Dyspnea and a Severely Reduced D_LCO in fibrotic ILD

A noticeable finding of the present study was the similar peak dyspnea scores reported by subjects and controls: Given ∼45% (vs controls) and ∼20% (vs D_LCO ≥ 40%) lower peak work rate in the D_LCO < 40% group, dyspnea increased at a faster rate in the latter group (Table 1, Fig. 3A). We, therefore, confirmed a recurrent finding across our studies involving subjects with obstructive (eg, COPD, cystic fibrosis)⁵⁷ and restrictive (ILD)²⁷ abnormalities: the close relationship between metrics of demand-capacity imbalance and exertional dyspnea. For instance, differences in dyspnea intensity between ILD subjects and controls were markedly lessened (vs VR_dyn) or disappeared (vs IRV/TLC) when the symptom was related to metrics of reduced ventilatory and volume reserves, respectively (Fig. 3C-3D). This is consistent with the notion that increased dyspnea in fibrotic ILD largely reflects the heightened awareness of an increased neural respiratory drive⁵² triggered by alterations in pulmonary gas exchange efficiency (reducing VR_dyn) and respiratory mechanics (decreasing IRV/TLC).¹³ In keeping with our recent results in COPD,³⁶ Figure 1D illustrates the advantages of submaximal VR_dyn to expose the seeds of exertional dyspnea since peak VR, the traditional approach to suggest ventilatory limitation, did not differ between subjects and controls. Of note, participants showing D_LCO < 40% were more likely to report dyspnea scores within the very severe range (Table 1), a cutoff that consistently pointed out a higher V̇_E/V̇_CO2, greater inspiratory constraints, and poorer exercise tolerance in COPD.³⁴ The differences in dyspnea trajectory as the ventilatory and volume reserves diminished with exercise progression were also informative. Thus, dyspnea rose linearly as VR_dyn decreased without a discernible break point (Fig. 3C); conversely, it quickly accelerated when the volume reserve to further inspiration reached excessively low values (∼15% from TLC at ∼45 L/min V̇_E in both subject groups) (Fig. 3D).³³ It follows that to avoid the emergence of intolerable dyspnea, patients with fibrotic ILD might benefit from training modalities that are not associated with sustained increases in V̇_E above a certain critical level.

Study Weaknesses and Strengths

Our results are largely descriptive, and we carefully avoided suggesting any cause-effect relationship between D_LCO < 40% and the associated resting and exercise outcomes. Thus, we emphasized that D_LCO < 40% signals (ie, predicts, marks) multiple abnormal responses to exertion that are beyond its most logical correlate: hypoxemia. In this context, they provide strong support for the contentions by Enright⁹ about the relevance of D_LCO in helping clinical decision making in pulmonology. Although we did not measure mouth pressure during the D_LCO maneuvers, Kaminsky and Jarzembosky⁵⁸ reported that variations were not associated with the measured D_LCO value in a real-life study that included subjects with ILD. We opted for the percentage predicted criterion to define a severely impaired D_LCO since Z score–based classifications have not yet been extensively validated.⁵⁹ The suggested cutoff (40%) has been proposed by an European Respiratory Society task force and widely used over almost 2 decades.⁸ To improve the external validity of our results, we did not exclude subjects with a previous smoking history. Importantly, however, emphysema extension was defined as only mild²⁹ by trained radiologists in all subjects. In particular, no subject presented with combined pulmonary fibrosis and emphysema or resting pulmonary hypertension,²¹ decreasing sample heterogeneity. Both fibrotic ILD groups showed a numerically higher BMI than controls (though typically in the mild obesity range) (E-Table 1, see related supplementary materials at http://www.rcjournal.com), and increased mass displacement is known to worsen dyspnea; thus, additional studies are required to investigate whether dyspnea further increases when moderate-to-severe obesity and D_LCO < 40% coexist. Although the relative contribution of lung mechanical versus gas exchange mechanisms of breathlessness vary as fibrotic ILD etiologies progress,¹⁰ this seems a less relevant shortcoming in a transversal study. Cycling is associated with less hypoxemia compared to walking;¹¹ consequently, we may have underestimated the severity of O₂ desaturation in daily life. Nevertheless, less exercise-induced hypoxemia (Table 1) gave us the opportunity to investigate the integrated physiological responses to exercise and their sensory consequences without the overruling influence of a markedly increased hypoxic drive.¹¹ Also from a positive perspective, only a minority of our subjects were using long-term O₂ therapy (n = 11), thereby reducing the potential confounding effects of acute O₂ supplementation withdrawal (for CPET performance) on exertional dyspnea. We restrained our analysis to dyspnea intensity, and it remains unknown whether dyspnea descriptors^14,60 would change in those with a severely reduced D_LCO. Since severe hypoxemia may compromise cerebral oxygenation on exertion in fibrotic ILD,²⁶ it remains unclear whether central fatigue⁶¹ may contribute to poorer exercise tolerance in the D_LCO < 40% group.

Conclusions

A severely reduced single-breath D_LCO (< 40%) marked poor exercise tolerance in subjects with fibrotic ILD due to multiple interconnected mechanisms (disturbed hemodynamics, a precocious shift to skeletal muscle anaerobic metabolism, excess exertional ventilation secondary to heightened afferent stimuli including hypoxemia, and critically high inspiratory constraints) that ultimately lead to intolerable respiratory (dyspnea) and peripheral (leg discomfort) symptoms. Given the close association between impaired physical performance and symptom burden with morbidity and poor health-related quality of life in fibrotic ILD,^3-5 this subset of patients might derive particular benefit from interventions to lessen the ventilatory requirements (walking aids, exertional O₂ supplementation,⁶ treatment of cardiovascular comorbidities known to increase afferent stimuli,³⁴ exercise reconditioning⁴); volume constraints (anti-fibrotic therapy as indicated,⁶² weight loss in obesity); and, in selected patients, dyspnea perception (low dose opiates²⁸).