Supplemental oxygen users with pulmonary fibrosis perceive greater dyspnea than oxygen non-users

Background Exertional dyspnea is a hallmark symptom of fibrosing interstitial lung disease (fILD), and oxygen (O2) desaturation is common among patients with fILD. Supplemental O2 is prescribed to maintain normoxia and alleviate dyspnea. We sought to better understand the associations between O2 and dyspnea in fILD during the 6-min walk test (6MWT). Methods 1326 fILD patients compose the sample group. Borg dyspnea and other 6MWT variables were compared between subjects who performed the test without (non-users) versus with O2 (users). Results There were 812 users and 514 non-users; users were older, more likely to have smoked, had greater body mass index, and had more severe fILD. Despite a similar 6-min SpO2, users perceived greater dyspnea than non-users (Borg 3.9 ± 2.0 vs 2.9 ± 1.7, p < 0.0001). Whether subjects became hypoxemic (6-min SpO2 < 89 %) or not during the walk, the results were the same: users perceived greater dyspnea than non-users (hypoxemic: users 3.5 ± 2.1 vs non-users 2.7 ± 1.8, p < 0.0001; non-hypoxemic: users 3.4 ± 1.9 vs non-users 2.4 ± 1.6, p < 0.0001). Among subjects who did not desaturate (SpO2 drop < 4 %), users walked a shorter distance (944.9 ± 367.0 vs 1385.3 ± 322.4 feet, p < 0.0001) but perceived greater dyspnea than non-users (3.3 ± 1.6 vs 2.3 ± 1.7, p = 0.005). No combination of potentially influential predictor variables entered in multivariate models explained more than 11 % of the variance in dyspnea ratings. Conclusion Dyspnea is a complex perception, and in patients with fILD, O2 may lessen, but does not resolve, it. Further research is needed to clarify why fILD patients who use O2 perceive greater levels of dyspnea with activity than O2 non-users.


Background
The interstitial lung diseases (ILD) comprise several diffuse parenchymal lung diseases whose causes are unknown or include exposures (e.g., dust, drug, aerosolized organic antigen) or underlying connective tissue disease (CTD). Regardless of cause, fibrotic ILD (fILD) is typically progressive and incurable. Exertional dyspnea, the hallmark symptom of fILD, impairs physical functioning and quality of life (QOL) and is often associated with peripheral oxygen desaturation (SpO 2 ).
The six-minute walk test (6MWT) is commonly used as a measure of submaximal exercise capacity in patients with fILD. Along with distance walked (6MWD), SpO 2 , heart rate and dyspnea ratings are often collected as part of the 6MWT and used to assess disease status. Dyspnea-the perception of "breathing discomfort"-is due to a number of complex physical, psychological, social, environmental and interwoven physiological factors [1]. In fILD, dyspnea is due to reduced lung compliance, inability to expand tidal volume in response to respiratory drive, as well as the elevated work and oxygen cost of breathing [2]. Although dyspnea is a personalized perception, it is experienced and described similarly among patients with the same respiratory disease. For example, during symptom-limited incremental cycle exercise tests, 'unsatisfied inspiratory effort' and 'rapid breathing' are used to describe dyspnea by patients with fILD-but not by healthy controls [3]. Investigators have observed that although patients with fILD desaturated to a greater degree than patients with chronic obstructive pulmonary disease (COPD), patients with COPD perceived greater dyspnea. In that study, SpO 2 was an independent predictor of dyspnea severity in patients with fILD but not in those with COPD. Among patients with fILD, SpO 2 explained only a quarter of the variance in dyspnea ratings [4].
Although supplemental oxygen (O 2 ) is commonly prescribed to patients with fILD to maintain normoxia, in hopes of relieving dyspnea (and by extension, improving physical functioning and QOL), few studies have aimed to decipher the beneficial effects of O2 in these patients [5,6]. Through this study, we sought to examine how dyspnea ratings from patients who use O 2 compare with those from patients who do not use O 2 .

Study subjects
The study group was composed of 1326 patients with fILD evaluated at National Jewish Health (NJH) from January 1, 2008 to December 30, 2014. We formed the cohort by querying the NJH research database for patients with fILD who completed at least one 6MWT. Patients with underlying connective tissue disease (CTD) were excluded; thus, the overwhelming majority of subjects had idiopathic pulmonary fibrosis (IPF), idiopathic nonspecific interstitial pneumonia (iNSIP) or chronic hypersensitivity pneumonia (cHP), with diagnoses made in accordance with accepted criteria [7][8][9][10]. The study was approved by the NJH Institutional Review Board (IRB; study #2868) which waived the requirement for written, informed consent.

6MWT
The 6MWT was conducted similarly in all patients (whether users or non-users), by trained technicians at NJH, according to published guidelines with slight modification [11]. Per standard operating procedure at NJH, the 6MWT is terminated if SpO 2 drops below 80 %. To maintain reliability in the 6MWT outcome of most interest (distance), we tried to hold constant as many other variables as possible. Thus, a patient performed all 6MWT on the same O 2 l flow, unless or until he was unable to walk for a full six minutes without SpO 2 falling below 80 %. We included in our analyses data only from patients who walked for a full six minutes. For patients who completed multiple 6MWT, we selected the first test. Ratings for dyspnea and exertion were assessed immediately after completion of the test by the technician and using the CR10 Borg scale (range 0-10, with higher scores connoting greater dyspnea or exertion as appropriate) [12]. The minimal clinically important difference for the Borg scale is reported to be one point [13].

Statistical analysis
Summary statistics were generated for baseline data with the sample stratified on whether O 2 was used (users) or not (non-users) during the 6MWT. Student's t-tests were used for between-groups comparisons of continuous variables. Cochran-Mantel-Haenszel, Chi square or Fisher's exact tests were used as appropriate for between-groups comparisons of categorical variables. We used Pearson correlation coefficients to express associations between dyspnea ratings and other variables. We used linear regression to examine associations between dyspnea ratings and other variables while controlling for potentially influential predictors. We considered p < 0.05 to represent statistical significance. Analyses were performed using SAS version 9.3 statistical software (SAS, Inc.; Cary, NC).

Results
The study group comprised 812 users and 514 nonusers. On average, users were older, had greater impairments in pulmonary physiology and had shorter 6MWD than non-users. Despite a similar SpO 2 at six minutes (88.1 % vs. 88.7 %), users perceived significantly greater dyspnea than non-users (Table 1).
In both users and non-users, dyspnea was correlated with certain other variables; however, all correlations were weak (Table 2). In both subgroups, dyspnea was inversely correlated with 6MWD. Among the 791 subjects whose SpO 2 fell below 89 %, there were nearly twice as many users as non-users (Table 3). Although the SpO 2 at six minutes was similar (85.3 % vs. 86.0 %), dyspnea ratings among users were significantly higher than in non-users. The same was true for subjects whose SpO 2 remained 89 % or greater for the duration of the 6MWT: despite identical mean SpO 2 values at six minutes (91.4 % vs. 91.4 %), dyspnea ratings were significantly higher among users than in non-users (Table 4).
Results were similar for the 883 subjects (572 users and 311 non-users) with a history of smoking: the SpO 2 values at six minutes were similar (87.9 % vs. 88.4 %), and dyspnea ratings were higher among users than in non-users (3.8 ± 2.0 vs. 2.9 ± 1.7, p < 0.0001). For the 473 subjects (303 users and 170 non-users) with IPF, the SpO 2 values at six minutes were the same (86.7 % vs. 86.9 %), and dyspnea ratings were higher among users than in nonusers (3.8 ± 2.1 vs. 2.9 ± 1.6, p < 0.0001). Among the 118 subjects whose SpO 2 never dropped by more than three points from baseline (rest), although minute-six SpO 2 was higher in users than in non-users, dyspnea ratings among users were significantly higher than in non-users (Table 5).
Results from the linear regression analysis are presented in Table 6. While controlling for various combinations of predictors, O 2 use remained a significant predictor of dyspnea rating. As revealed by the R-squared values, none of the combinations of variables explained more than minimal variance in dyspnea ratings.

Discussion
We examined patients with fILD and found that those who used O 2 during a 6MWT consistently experienced more severe dyspnea than those who did not use O 2 . Data on the effects of O 2 in patients with fILD are surprisingly limited, and much of the information on the potential benefits of O 2 that is used in clinical decisionmaking with fILD patients, is based solely on scientific rationale or borrowed from the COPD literature. In two Letters to the Editor, investigators described the results of retrospective studies in which they examined the within-subject beneficial effects of O 2 on various outcome measures collected around 6MWTs [5,6]. In one study, investigators observed that, in 52 patients with fILD, during a second 6MWT for which O 2 was administered according to a semi-quantitative algorithm aimed at maintaining SpO 2 at (closer-to) acceptable levels, distance walked, nadir SpO 2 and Borg Values are mean and standard deviation or count (percent); IPF idiopathic pulmonary fibrosis; BMI body mass index; FVC% percent predicted forced vital capacity; DLCO% percent predicted diffusing capacity of the lung for carbon monoxide; 6MWD distance walked during six-minute walk test (6MWT); SpO2 peripheral oxygen saturation; HR heart rate; O2 completed 6MWT using supplemental oxygen; No O2 completed 6MWT without using supplemental oxygen; *N = 715 for O2 users and 483 for non-users; **N = 446 for O2 users and 251 for non-users; ***N = 211 for O2 users and 123 for non-users; ****N = 672 for O2 users and 415 for non-users  Dyspnea is a complex perception that depends on the integration of multiple inputs from several sources. Blood oxygen level is but one of those sources, and the weak correlation we observed between dyspnea rating and minute-six SpO 2 affirms it is far from the main contributor. Other contributors include neural inputs arising from receptors in the airways and lung parenchyma, peripheral locomotor and respiratory muscles, and central and peripheral chemoreceptors, along with corollary neuronal discharge arising from the brainstem and cortical motor centers [14].
In our study, patients who used O 2 consistently started with a higher SpO 2 than non-users, and although SpO 2 declined to a greater degree during the walk in users than non-users (9 % vs. 6 %), minute-six SpO 2 was the same in both groups (88 %). Perhaps has dyspnea more to do with SpO 2 decline from baseline than the absolute SpO 2 at the time of dyspnea rating? Our results suggest this is not the case: the correlation between dyspnea and SpO 2 decline was the same (weak) as the correlation between dyspnea and minute-six SpO 2 . However, in the subgroup of patients who did not desaturate at all during the walk (SpO 2 decline < 4 points), despite O 2 users having a higher minute-six SpO 2 than non-users (95 % vs. 93 %), O 2 users perceived greater dyspnea (mean Borg scores 3.4 vs. 2.4).
In our statistical models controlling for either minutesix (data not shown) or decline-from-baseline in SpO 2 (i.e., SpO 2 drop), each of these SpO 2 measures was a significant predictor of dyspnea, and in each model, O 2 use remained an independent predictor of dyspnea. However, each model explained minimal variance in dyspnea scores-again, confirming that dyspnea relies on more inputs than simply blood oxygen.
We suspect users in our study perceived greater dyspnea intensity than non-users because of a complex interaction of elements, including those related to conduct of the 6MWT and perhaps certain neurophysiological factors. At our center, in an attempt to maintain reliability in the 6MWT outcome of most interest (distance), we try to hold constant as many other variables as possible. Thus, a patient performs all 6MWT on the same O 2 l flow, unless or until he is unable to walk for a full six minutes (in which case flow is reset for subsequent 6MWTs). Because of this practice, on certain occasions, patients at our center may perform their 6MWT on O 2 l flows below what they use with exertion at home. In our study, this "intentional under-dosing" of O 2 flow-to maintain reliability-could have driven dyspnea ratings up in O 2 users. Unfortunately, with this data set, we are unable to determine when this underdosing might have occurred. Regardless, this "intentional under-dosing" explanation would seem not to apply to subjects whose O 2 was dosed adequately enough to maintain an acceptable SpO 2 throughout the test, including the over 600-patient subgroup whose SpO 2 remained above 88 %, or the greater than 100-patient subgroup  Values are mean and standard deviation; 6MWD distance walked during six-minute walk test (6MWT), O2 completed 6MWT using supplemental oxygen, No O2 completed 6MWT without using supplemental oxygen, HR heart rate, SpO2 peripheral oxygen saturation; *N = 30 for O2 users and 67 for non-users whose saturations did not decline at all. In both these subgroups, O 2 users rated their dyspnea as more intense than non-users. Also at our center, patients who use O 2 at home either carry or pull their O 2 delivery device while completing their 6MWT. Having to move this excess weight over distance-or altered chest wall mechanics resulting from carrying or pulling the device-could add to dyspnea. To our knowledge, this has yet to be examined, but we believe it deserves investigation. If carrying or pulling the delivery device is found to add significantly to dyspnea intensity, this could be a target for therapeutic intervention.
Another alternative explanation is that dyspnea truly depends greatly on arterial oxygen but SpO 2 was an especially inaccurate reflection of it in this cohort; we doubt this was the case, but if it were, we would expect the inaccuracies to affect both users and non-users equally. We excluded patients with underlying CTD in the hopes of limiting the influence of vascular abnormalities like Raynaud's or pulmonary hypertension.
Various physical factors unrelated to the lungs, SpO 2 or other aspects of oxygen delivery also must be considered as potential explanations for our results. While exerting (and into recovery), subnormal lung compliance in patients with ILD induces rapid-shallow breathing. The physical sensation-and mental/emotional impact-of this breathing pattern, which occurs to some degree when patients with ILD exert to any degree, is expected to influence dyspnea ratings. How subjects internally considered, weighed and integrated each component (physical or mental/emotional) as they made their ratings for "breathlessness" is unknown. Compared with nonusers, O 2 users had lower FVC% and, by deduction, lower lung compliance than non-users-a factor expected to hasten and heighten rapid-shallow breathing. We do not measure respiratory rate during the 6MWT at our center, so we are unable to determine whether O 2 users had higher respiratory rates than non-users. Additional studies aimed at discerning whether or how much respiratory rate (and other physical or emotional components) contributes to exertional dyspnea ratings are needed. Although we are unable to comment on directly-observed respiratory rate, FVC% as a marker of lung compliance could be considered a reasonable surrogate for respiratory rate. In statistical models that included O 2 use, and controlled for FVC%, O 2 use remained a significant predictor of dyspnea. In a comprehensive appraisal of dyspnea in patients with chronic interstitial lung disease, Faisal and colleagues observed that dyspnea intensity during exertion climbed as inspiratory neural drive increased and tidal volume became constrained (thus blunting the mechanical respiratory response during exercise) [15].
Another consideration is whether the greater disease severity in O 2 users might have contributed to physical inactivity and deconditioning. Given the practical challenges of using O 2 and the possibility that it prohibits patients from living a more active, carefree lifestyle, we suspect that, on average, fILD patients who require O 2 are less physically active (and thus less well-conditioned) than patients who do not require O 2 . De-conditioned skeletal muscles are less efficient and fatigue-resistant Values are coefficients and standard error (top) and p value (bottom); 6MWD distance walked during six-minute walk test (6MWT), O2 completed 6MWT using supplemental oxygen, No O2 completed 6MWT without using supplemental oxygen, HR heart rate, SpO2 peripheral oxygen saturation than conditioned muscles, and peripheral locomotor [14] muscle fatigue contributes to dyspnea. Compared with non-users, O 2 users walked shorter distances during the 6MWT; this was true even for the subgroup that did not desaturate to < 89 %. Whether the shorter distance walked was due to deconditioning or some other factor(s) is unknown; however, deconditioning could well explain the interesting and perhaps somewhat paradoxical finding that dyspnea severity was inversely correlated with 6MWD: subjects who walked further, were better conditioned and thus perceived less dyspnea.

Conclusion
Dyspnea is a complex perception that impacts the lives of patients with fILD. It is important for a patient with fILD to know what to expect when being prescribed O 2 : it will likely decrease dyspnea (compared with not using O 2 ), but because dyspnea is driven by so many inputs (with SpO 2 being just one), O 2 will not resolve dyspnea. Further research is needed to better understand the mechanisms driving dyspnea in patients with fILD and to devise strategies to lessen it.