Cardiopulmonary Exercise Testing and Pulmonary Function Testing for Predicting Aggravation of CTEPH

Cardiopulmonary exercise testing (CPET) and pulmonary function testing (PFT) are noninvasive methods to evaluate the respiratory and circulatory systems. This research aimed to evaluate and monitor chronic thromboembolic pulmonary hypertension (CTEPH) noninvasively and effectively. At the same time assess the predictive value of CPET and PFT parameters for the aggravation of CTEPH. We used data from 86 CTEPH patients (55 for test set, and 31 for validation set) at the Shanghai Pulmonary Hospital Aliated to University. The clinical, PFT and CPET parameters of mild, moderate and severe CPET patients classied according to PAP (mm Hg) were compared. Logistic regression analysis was performed to appraise the predictive value of each potential predictor for severe CTEPH. The performance of PFT and CPET parameters for predicting severe CTEPH was determined by receiver operating characteristic (ROC) curves and calibration curves. suggests Load @ Peak was effective and ecient in discriminating patients with severe CTEPH. Some of the limitations of this study are its patient sample size, non-randomized nature and single-center design and potential selection bias. U or chi-square Health Organization; peptide;

loaded pahse in which the workload increased 10-25W/min, and the recovery phase of 5 minutes. The patients should pedal at 55-60 revolutions/min in the unloaded and loaded phase, and when they reached their maximum tolerance they could enter the recovery phase. Patients could stop at any time when they have fatigue, dyspnea, chest tightness and any other discomfort during the process.
Measurements included Load, minute ventilation (VE), carbon dioxide output (VCO 2 ), oxygen uptake (VO 2 ), oxygen pulse (VO 2 /HR), end-tidal partial pressure for carbon dioxide (PETCO 2 ), end-tidal partial pressure for oxygen (PETO 2 ), heart rate (HR), breathing reserve (BR), respiratory exchange ratio (RER) and breathing frequency (BF). Anaerobic threshold (AT) which represents the beginning of anaerobic metabolism was determined by the V-slope method. VE/VCO 2 slope was obtained by linear regression analysis of the relation between VE and VCO 2 . Oxygen uptake e ciency slope (OUES) was computed by a linear squares regression from the oxygen uptake on the logarithm of the minute ventilation according to the following equation: VO 2 =a*lgVE+b. Constant "a" is called the OUES. Oxygen uptake e ciency plateau (OUEP) was at 90 seconds of the highest consecutive values for VO 2 (mL/min)/VE (L/min).

Statistical analysis
Data were analyzed by using SPSS 22.0 and GraphPad Prism 6. The data are presented as mean ± SD, median (interquartile range), or n. One-way ANOVA test, Kruskal-Wallis test, Tukey's multiple comparisons test, Dunn's multiple comparisons test, Unpaired t test, Mann-Whitney U test, chi-square test, univariate logistic regression analysis and multivariate logistic regression analysis were used according to the corresponding situation. A two-tailed P<0.05 was considered signi cant.
Predictive value of the CPET and PFT parameters for the severe CTEPH Furthermore, 55 patients with CTEPH in the test set were re-grouped into "Mild-Moderate" and "Severe" group to analyze predictors for severe CTEPH. All the parameters measured were analyzed with the univariate analysis for the severe CTEPH, and 21 parameters were found to have a P<0.05. They are listed as followed: VE @ AT (L/min), VO 2 /VE @ AT, BR @ AT (%), LOWEST VE/VCO 2 , VE/VCO 2 @ AT, PETO 2 @ AT, PETCO 2 @ Peak, FEV1/FVC (%), PETCO 2 @ Rest, Load @ Peak (W), VE/VCO 2 @ Peak, BR @ Rest (%),VO 2 @ Peak (mL/Kg/min), VCO 2 @ Peak, VE/VCO 2 slope, FEV1 (% Pred), VE/VCO 2 @ Rest, BMI (kg/m 2 ), NT-proBNP (pg/mL), VO 2 /VE @ Peak and RER @ Peak. Considering the sample size, 11 parameters with the minimum P value were tted into the multivariate Combining these three parameters, we got the prediction equation for the severe CTEPH which was Logit(P)=7.397+0.205*VE@AT-0.116*FEV1/FVC-0.059* Load @ Peak. To evaluate the ability of the VE @ AT (L/min), FEV1/FVC (%), Load @ Peak (W) and the prediction equation to differentiate severe from mildmoderate CTEPH, ROC curves and calibration curves were performed. Details are listed in Table 5 and Figure 3. It must be noted that the AUC of the prediction equation was better than that for each parameter, indicating that the equation based on three parameters yielded the highest AUC value and could signi cantly improve the prediction performance for severe CTEPH.
Additionally, we analyzed the prediction equation in the validation datasets to validate the results. Only 4 of 31 patients with CTEPH of the validation set were ambiguous, and the accuracy was 87.10%. Details are listed in Table 6.

Discussion
We retrospectively analyzed the clinical, hematological, PFT and CPET data of 86 patients with CTEPH. Part of the patients were randomly classi ed into the validation set (a total of 31), and the remaining 55 patients were classi ed into the test set. Patients with CTEPH were divided into "Mild", "Moderate" and "Severe" groups according to PAP (mm Hg) detected by RHC. This was used for nding the parameters likely predicting severe CTEPH [12]. Through comparison, we can nd that there are differences in the CPET performance of patients with different severity of CTEPH. For example, Load @ Peak (W) and VO 2 @ Peak (mL/Kg/min) decreased with the aggravation of CTEPH. VE (L/min) and BR (%) of patients with different severity of CTEPH differed only at the AT phase. VE/VCO 2 , VO 2 /VE, PETCO 2 (mm Hg) and PETO 2 (mm Hg) was different between "Mild" and "Severe" groups. After that, patients with CTEPH were regrouped into "Mild-Moderate" and "Severe" groups. By univariate analysis for analyzing risk factors for severe CTEPH, 21 parameters were found to have a P<0.05. While by multivariate analysis, only Load @ Peak (W), FEV1/FVC (%) and VE @ AT (L/min) were found to have a P<0.05. Combining these three parameters, we got a prediction equation with AUC of 0.897 (Sensitivity: 80.0%, Speci city: 85.0%). ROC curves and calibration curves proved the prediction equation was good in discrimination and calibration. Additionally, its application in the validation set further con rmed the e ciency.
CPET could be used for the detection of CTEPH in patients with suspected PH but normal echocardiography . The number of patients with undiagnosed CTEPH may be signi cantly higher [13]. Estimates suggest that up to a quarter of patients with CTEPH may have no history of previous pulmonary embolism [14]. Compared with healthy cohorts, patients with CTEPH had higher value of VO 2 /VE @ AT, VE/VCO 2 @ AT, P(c-ET)CO 2 while lower value of PETCO 2 @ AT. P(c-ET)CO 2 was a diagnostic parameter of CTEPH with the highest sensitivity (85.7%) and speci city (88.2%) [13]. Ventilatory e ciency parameters including P(c-ET)CO 2 , PETCO 2 decrease patterns, VD/VT @ Peak, VE/VCO 2 slope, VE/VCO 2 @ AT, OUEP and OUE @ AT can help to distinguish CTEPH from idiopathic pulmonary arterial hypertension (IPAH) [15][16][17][18]. The lowest VE/VCO2 ratio best predicts CTEPH from the chronic PE patients [19]. Godinas et al reported that in distal CTEPH patients, higher VD/VT was associated with lower peak oxygen consumption and worse survival [20]. Qi Jin et al reported that after BPA, patients with inoperable CTEPH had signi cant improvement in CPET and PFT parameters including Load @ Peak, VO 2 @ Peak, OUES, FVC, FEV1 and MVV[21]. As we have seen, CPET can be used for the diagnosis/differential diagnosis, prognostic evaluation and treatment evaluation of CTEPH.
For the rst time, we have evaluated the value of CPET in the process of CTEPH aggravation. Despite current evidence not supporting routine screening for CTEPH after pulmonary embolism, it's in the best interest of the up to a quarter of CTEPH patients that have no history of PE at least have a noninvasive test like CPET to better understand disease progression. We hope that CPET can be used for routine monitoring of CTEPH patients in the future, and then the application of this formula can provide value for guiding the further examination and treatment of patients. Our study shows that patients with severe CTEPH, the Load @ Peak (W) and FEV1/FVC (%) value was lower while VE @ AT (L/min) was higher than in patients with mild-moderate CTEPH. The prediction equation Logit(P)=7.397+0.205*VE @ AT-0.116*FEV1/FVC-0.059* Load @ Peak was effective and e cient in discriminating patients with severe CTEPH.
Some of the limitations of this study are its patient sample size, non-randomized nature and single-center design and potential selection bias.

Declarations
Ethics approval and consent to participate All patients in this study were informed at admission that their medical records were likely to be used for clinical studies. Ethical approval by the medical ethics committee of Shanghai pulmonary hospital was obtained.

Consent for publication
Not applicable.

Availability of data and materials
All the related data are presented in the manuscript.