Value of Cardiopulmonary Exercise Testing in Prognostic Assessment of Patients with Interstitial Lung Diseases

Background: Interstitial lung disease (ILD) is associated with high rates of comorbidities and non-infectious lung disease mortality. Against this background, we aimed to evaluate the prognostic capacity of lung function and cardiopulmonary exercise testing (CPET) in patients with ILD. Materials and Methods: A total of 183 patients with diverse ILD entities were included in this monocentric analysis. Prediction models were determined using Cox regression models with age, sex, body mass index (BMI), and all parameters from pulmonary function testing and CPET. Kaplan–Meier curves were plotted for selected variables. Results: The median follow-up period was 3.0 ± 2.5 years. Arterial hypertension (57%) and pulmonary hypertension (38%) were the leading comorbidities. The Charlson comorbidity index score was 2 ± 2 points. The 3-year and 5-year survival rates were 68% and 50%, respectively. VO2peak (mL/kg/min or %pred.) was identified as a significant prognostic parameter in patients with ILD. The cut-off value for discriminating mortality was 61%. Conclusion: The present analyses consistently revealed the high prognostic power of VO2peak %pred. and other parameters evaluating breathing efficacy (VÉ/VCO2 @AT und VÉ/VCO2 slope) in ILD patients. VO2peak %pred., in contrast to the established prognostic values FVC %pred., DLCO/KCO %pred., and GAP, showed an even higher prognostic ability in all statistical models.


Introduction
The main entities of interstitial lung diseases (ILDs) are idiopathic pulmonary fibrosis (IPF), idiopathic nonspecific interstitial pneumonia (NSIP), sarcoidosis with fibrosis, and hypersensitivity pneumonitis [1]. Some revisions of ILD entities have been performed recently [1,2]. Due to the diagnostic complexity of ILDs, a dynamic integrated approach using multidisciplinary discussion (MDD) is considered the standard for classification of these diseases [3]. ILDs (sarcoidosis included) occupy the third position among non-infectious pulmonary diseases in terms of the mortality rate [4]. The German INSIGHT-IPF-Registry data indicated a 1-year and 2-year survival rate of 87% versus 46% and 62% versus 21%, respectively, for patients with versus without antifibrotic therapy [5]. In comparison to the general population, patients with ILDs more frequently show several comorbidities, especially cardiac diseases, diabetes mellitus, dyslipidaemia, obstructive sleep apnoea, cancer, and depression [6][7][8][9]. Additionally, these diseases are associated with limited cardiorespiratory fitness (CRF). The CRF, which is measured as 'power of work' [10] or 'maximum oxygen uptake (VO 2 peak)' [11] in addition to other parameters [12,13], has a prognostic influence in ILD subgroups. There are multiple pathophysiologic reasons for

Patients
In this single-centre retrospective study, a total of 215 patients with ILD were included. This centre follows the IPF-guidelines on diagnosis and therapy such as multidisciplinary discussion (MDD) [2,6,42]. No CPET data were available for 32 of the 215 patients. Therefore, the final study population consisted of 183 patients.
For the characterisation of the patients, sex, age, body mass index (BMI), and number of comorbidities (according to the Charlson comorbidity Index [50]) were analysed. The Charlson comorbidity Index predicts 10-year survival in patients focusing on 19 comorbidities with different assessment scores. The severity grading of IPF patients followed the modified GAP index [21].

Cardiopulmonary Exercise Testing
CPET was performed according to the modified JONES-protocol using a bicycle ergometer as a symptom-limited test. Performance and analysis methods have been previously described in detail [54]. Briefly, the test started with a 3-min resting phase and unloaded cycling of 1 min followed by a protocol with a step-increment protocol of 16 W·min −1 .

Echocardiography
Resting echocardiography was performed by experienced physicians according to relevant guidelines [58,59]. TR was classified according to the American College of Cardiology/European Society of Cardiology (ESC) recommendations, and PAPsys was estimated using a simplified Bernoulli equation via TR velocity (v) as RVsys (mmHg) = 4v2, with the addition of 5 mmHg if the inferior vena cava was not dilated and there was visible respiratory variability and 10 mmHg if the inferior vena cava was dilated or without respiratory variability.

Follow-Up Assessments
The patients were contacted by phone and provided written informed consent for data collection. The date of evaluation was 01.03.2020 (mean observation time, 3.0 ± 2.5 years). The study was approved by the ethics committee of the University of Greifswald (Reg.-Nr. BB 057/2017).

Methodological Limitations
The selection of patients was inherently biased, since only patients who underwent CPET were included in the analyses. As a result, despite the retrospective nature of the study, only five patients were lost to follow-up (2.7%). Moreover, due to the retrospective approach, not all clinical and functional data were available. Due to the fact that our institution is a supra-regional centre for PH-diagnosis and therapy, the proportion of PH-patients in this study is high.
In all patients, the modified JONES-protocol was used on a braked cycle ergometer, which has been only evaluated for COPD patients so far and did not significantly influence the comparability of exercise parameters to other protocols [60,61]. Studies comparing different exercise protocols are not known for patients with ILD. Moreover, the current study focused on the prognostic evaluation of CPET parameters. Therefore, data on therapy are not provided. However, the effect of available antifibrotic medication on pulmonary function and cardiopulmonary exercise capacity is undisputable [8,[62][63][64].

Statistical Analyses
Continuous variables, stratified by group status, were reported as the median and interquartile range (IQR, in brackets). Categorical variables were reported as absolute numbers and percentages. Differences among groups were verified using Wilcoxon (continuous data) and χ 2 tests (categorical data). Potential associations of group status and parameters from pulmonary function testing and CPET with mortality were tested using Cox regression models adjusted for age and sex. For group status, the follow-up duration was calculated based on the time of diagnosis; for the other variables, the time of first examination was defined as the starting point. Prediction models were determined using Cox regression models with age, sex, body mass index (BMI), and all parameters from pulmonary function testing and CPET as explanatory variables. For the final model, variables using a backward selection procedure with a cut-off p-value of 0.1 were eliminated. The discrimination of these models was reported using Harrell's C-statistic. Based on logistic regression models with the outcome 'death: yes/no', conducted receiver operating characteristic (ROC) analyses for selected variables were conducted. Kaplan-Meier curves were plotted for selected variables; for continuous variables, cut-off values were defined as the point which maximised the Youden index for the outcome 'death'. The Youden index was defined as sensitivity + specificity − 1. All analyses were carried out using Stata 14.1 (Stata Corporation, College Station, TX, USA).

Patient Characteristics
The median age of the included patients (n = 183, 68% male) was 68.1 ± 10.4 years. The median age of patients with a UIP pattern was 72.8 ± 7.9 years and that of those with CPFE syndrome was 71.1 ± 8.3 years (Table 1). In the latter group, 95% of the patients were men. Patients with EAA (64.3 ± 10.5 years) and sarcoidosis (60.6 ± 12.1 years) were significantly younger than those with IPF (p < 0.001). The percentage of male patients was the smallest in the sarcoidosis group (53%). At the time of study inclusion, the diagnosis had been established for 2.7 ± 6.1 years in ILD patients and 2.0 ± 1.7 years in CPFE syndrome patients.
The average Charlson comorbidity index of the patients was 2 ± 2 points. The mean number of comorbidities was 2.4, and more than three comorbidities were reported in 41% of the patients ( Figure 1). The GAP index in the IPF subgroup was ≥3 in 88% of the cases (Figure 1a).

Echocardiography
On average, 71% of all patients had normal left ventricular function, and the corresponding value in patients with the UIP pattern was 59%. Reduced left ventricular function was documented in 11% of the patients. Right ventricular function was comparable in all patients, with an approximate tricuspid annular plane systolic excursion (TAPSE) of 21 ± 5 mm (Table 1).

Echocardiography
On average, 71% of all patients had normal left ventricular function, and the corresponding value in patients with the UIP pattern was 59%. Reduced left ventricular function was documented in 11% of the patients. Right ventricular function was comparable in all patients, with an approximate tricuspid annular plane systolic excursion (TAPSE) of 21 ± 5 mm (Table 1).

Right Heart Catheterisation
Haemodynamic data were available in a subgroup of 87 patients, and PH was diagnosed in 79% of these patients (68% of IPF patients and up to 100% of patients with sarcoidosis), Table 1. The RHC and non-RHC groups showed significant differences (Supplementary Table S1), especially in relation to the time between diagnosis and inclusion in the present study (1.3 ± 5.4 vs. 3.0 ± 6.8 years, p = 0.008). Patients who underwent invasive diagnostic procedures (RHC, n = 87) experienced diverse comorbidities more often (e.g., arterial hypertension, atrial fibrillation, renal insufficiency, coronary artery disease,

Cardiopulmonary Exercise Testing
The determined maximum power in watts (W) was 67% ± 30% pred., peak oxygen uptake was 62% ± 21% pred., and the VÉ/MVV ratio was 68% ± 21% pred. In 36 (20%) patients of the overall group, this value was >80% and, therefore, demonstrated pulmonary exercise limitation. The respiratory efficacy (measured as the VÉ/VCO 2 slope) was 44 ± 14 in 77% of the patients, with values > 34 considered pathological. The PaetCO 2 max value > 6 mmHg at the end of exercise demonstrated an inhomogeneity of ratio perfusion/ventilation and was pathological in 73% of the overall group. Interestingly, 31% of all ILD patients showed dynamic hyperinflation (defined as EELVmax − EELFrest > 0); in sarcoidosis patients, it was even evident in 65% of the patients (Table 1). Figure 2 depicts the survival rates of the entire group of ILD patients. The 3-year and 5-year survival rates were 68% and 50%, respectively. Figure 3 shows the survival rates of subgroups of ILD patients. The 3-year and 5-year survival rates were the lowest in patients with CPFE. For IPF patients, the rates were 72% and 58%, respectively. Figure 2 depicts the survival rates of the entire group of ILD patients. The 3-year and 5-year survival rates were 68% and 50%, respectively. Figure 3 shows the survival rates of subgroups of ILD patients. The 3-year and 5-year survival rates were the lowest in patients with CPFE. For IPF patients, the rates were 72% and 58%, respectively.

Parameters Relevant to Prognosis
Parameters relevant to prognosis over the years were determined by Cox regression analyses of the data at study entry (adjusted for age, sex, and body mass index, Supplementary
The IPF patient subgroup was too small for these analyses. In two of the three models, VO 2 peak (as mL/kg/min or %pred.) in the entire group of patients with ILD was of significant prognostic relevance. For the entire group, the best cut-off VO 2 peak was 61% pred. for the discrimination of mortality (for VÉ/VCO 2 @AT, 39; for VÉ/VCO 2 slope, 40; and for VO 2 peak/HRpeak, 8.6 mL) (Supplementary Table S3, Figure 4). The IPF patient subgroup was too small for these analyses. In two of the three models, VO2peak (as mL/kg/min or %pred.) in the entire group of patients with ILD was of significant prognostic relevance. For the entire group, the best cut-off VO2peak was 61% pred. for the discrimination of mortality (for VÉ/VCO2@AT, 39; for VÉ/VCO2 slope, 40; and for VO2peak/HRpeak, 8.6 mL) (Supplementary Table S3, Figure 4). Kaplan-Meier curves for IPF patients (VO2peak % pred cut-off: 61%; VO2peak/HRpeak cut-off: 8.6 mL; VE/VCO2@AT cut-off: 39; VE/VCO2 slope cut-off:40). "Low" means beneath cut-off, "normal" means above cut-off.
For the subgroup of patients who underwent RHC, the prognostic relevance of the presence of PH (defined as PAPmean > 20 mmHg) was examined (adjusted for age, sex, and BMI) but did not prove to be of significance for survival (HR, 2.3; 95% CI 0.90-5.94; p = 0.082).

Discussion
Our study with 183 ILD patients clearly demonstrated that the maximum oxygen uptake (VO 2 peak) had a highly significant prognostic influence. The best mortality predictive cut-off value for VO 2 peak in ILD patients was 61% pred., while in IPF patients, it was 69% pred. Only four studies included more than 100 patients. However, these studies lacked detailed information on comorbidities, and the included exercise variables were limited [41].
Dyslipidaemia, as well as the existence of PH (medical records), echocardiographically measured right ventricular function (TAPSE), and PAPsys were of prognostic significance. In addition to male sex and age > 70 years, the degree of dyspnoea; DLCO < 60% pred., 6-MWD < 250 m, and SpO 2 < 88% measured during the 6-MWD; and the existence of PH and cardiovascular comorbidities were considered prognostically relevant baseline data in IPF patients [13,65,66]. Additionally, in the literature, histological data (number of fibroblastic foci), the extent of fibrotic alterations measured with HRCT, and selected biomarkers were prognostically relevant. Some of the prognostically relevant data are also reflected by our data, especially dyslipidaemia and PH. Against this background, it appears strange that comorbidities were not regularly part of investigations of ILD prognosis. With regard to dyslipidaemia, there were no differences between IPF and NSIP patients (20% and 17%) [7]. This is similar to our data. Although the prognostic significance of PH (medical history) is adequately familiar [15,67] it has not yet been described for dyslipidaemia in ILD-patients.
Focusing on lung function, only FEV1 and DLCO (as well as KCO) were reported to be prognostically relevant in our Cox analysis. In contrast to other studies [39,40,68], the prognostic significance for FVC was not observed. However, a good correlation between FVC and FEV1 is known [10]. The small number of included patients with IPF (n = 55/183) might explain the missing prognostic relevance of FVC.
The prognostic significance of diffusion parameters in ILD patients is undisputable [24,39], justifying their inclusion in prognosis scores [22]. DLCO is a sensitive marker for gasexchange pathologies in fibrotic lung diseases. However, it is decisively influenced by the capillary blood volume [69]. Against this background, it is a good parameter for the detection of early pulmonary vascular injury [70]. It is reduced in manifest vascular disturbances and often associated with PH. In our study, 94% of the patients with PH showed pathological diffusion capacity (DLCO < 60% pred.), while the corresponding percentage in evaluations based on KCO pred. was 83%. The optimal prognostic cut-off value was 36% for DLCO pred. and 48% for KCO pred.
The prognostic relevance of CPET parameters for diverse subgroups of patients with ILD is commonly known [41]. In contrast to the straightforward and inexpensive 6-MWD, the most convincing advantage of CPET is the early detection of pathophysiological conditions and cardiopulmonary limitations. The limited lung compliance in ILD patients leads to only a small increase in breathing volume (Vt), which results in an increase in breathing frequency. Predominant dead-space ventilation leads to a pathological breathing efficacy. Hyperventilation is enhanced by the stimulation of mechanoreceptors of the lung. Additionally, the alveolo-arterial oxygen difference (AaDO 2 ) at rest is often elevated in patients with ILD and increases during exercise due to oxygen exploitation in the muscle tissue. This eventually results in acidosis, which aggravates hyperventilation and impairs breathing efficacy. Impaired oxygen exchange in the lung eventually limits oxygen supply to the muscle tissue and exercise performance. The increased breathing effort claims a relevant proportion of oxygen uptake that is no longer available to the peripheral muscles. Depending on the extent of disease, it is comprehensible that ventilation, perfusion, and gas exchange are impaired in ILD patients [14,71]. During exercise, despite lung parenchymal disease, the heart rate is elevated and the stroke volume is reduced in ILD patients. Our data support these statements and prove the prognostic relevance of these parameters (Supplementary Tables S1 and S2).
In summary, the present data revealed that VO 2 peak pred., VÉ/VCO 2 slope, and VÉ/VCO 2 @AT are of prognostic relevance in ILD patients (including those with IPF). This is consistent with other studies [41,78]. Surprisingly, in our patients, dyslipidaemia was prognostically relevant, but has not been described to show prognostic significance in ILD patients so far. None of the established prognostic parameters, including FVC %pred. and DLCO/KCO %pred., other than the VO 2 peak %pred. were shown to be relevant prognostic markers in the entire group of patients with ILD.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcm11061609/s1, Table S1: Patients with/without Right heart catheter; Table S2: Associations of measurements from medical history, echocardiography, right heart catheter, lung function testing and cardiopulmonary exercise testing with mortality; Table S3: Sensitivity, specificity, and Youden index for the perfect cut-offs for mortality (selected data).