Cardiopulmonary Exercise Test, mMRC and GAP score Can Predict 1-year Mortality in Patients with Idiopathic Pulmonary Fibrosis: A Prospective Study

Idiopathic pulmonary fibrosis (IPF) is a rare lung disease with high mortality. Finding an effective predictor of survival is therefore important for both clinicians and patients. In this prospective observational study, we evaluated the prognostic value of the parameters of cardiopulmonary exercise test (CPET), pulmonary function test (PFT), 6-min walk test (6MWT), and certain questionnaires on mortality in Asian patients with IPF. A total of 34 patients diagnosed with IPF were enrolled and followed up for 12 months, during which 6 patients died. The non-survivors had significantly higher minute ventilation to carbon dioxide output (VE/VCO2) slope, more oxygen desaturation during CPET and 6MWT, less heart rate recovery one minute after CPET, higher dead space, higher Gender-Age-Physiology (GAP) index and Modified Medical Research Council (mMRC) score. GAP index, mMRC score, VE/VCO2 slope, and end-tidal partial pressure of carbon dioxide (PETCO2) at maximal exercise demonstrated an area under curve (AUC) of >0.7, and the corresponding cut-off values were 4, 2, 35.1 and 3.6 kPa. Therefore, GAP index, mMRC scores, VE/VCO2 slope, and PETCO2 at maximal exercise could be important predictors for mortality in Asian patients with IPF. CI, confidence interval; CPET, cardiopulmonary exercise test; DLCO, diffusing capacity of the lungs for carbon monoxide; FAI, functional aerobic impairment; FVC; forced vital capacity; FEV1; forced expiratory volume in s; GAP index, gender, age, and the pulmonary physiology index; HR, hazard ratio; IPF, idiopathic pulmonary fibrosis; mMRC, Modified Medical Research Council; PETCO2, end-tidal partial pressure of carbon dioxide; pulmonary function test; SF-36, Form-36;


Introduction
Idiopathic pulmonary fibrosis (IPF) is a progressive interstitial lung disease characterized by increased alveolar interstitial inflammation and cell proliferation, culminating in irreversible fibrotic changes. 1,2 Compared with common chronic pulmonary diseases, the incidence of IPF is lower in East Asia and South America. 3 However, its incidence and mortality rate are increasing steadily worldwide. 4 Despite the disease course being highly variable, the median survival period is approximately 2-3 years from the time of diagnosis. 5 Poor prognosis and the relatively short survival time have made it imperative to identify the appropriate predictors of survival for this patient population.
Many tools were developed for monitoring and evaluating the mortality risk of patients with IPF include pulmonary function test (PFT), 6,7 cardiopulmonary exercise test (CPET), 8-14 and 6-min walk test (6MWT) [15][16][17][18] . In IPF, a decrease in forced vital capacity (FVC) and diffusing capacity of the lungs for carbon monoxide (DLCO) are associated with increased mortality. 19 Parameters in CPET test included maximum oxygen consumption, reduced ventilatory efficiency and exercise-induced hypoxaemia were all reported to have prognostic value in IPF. 12 In 6MWT, the distance walked in 6 min, 20 decrease in the distance walked at 24 weeks, 16 recovery of heart rate after 1 min of the test, 17 and oxygen desaturation 21 are associated with increased mortality in patients with IPF. Gender, age, and the two physiologic variables in PFT "FVC" and "DLCO" have been recently integrated into the gender, age and pulmonary physiology (GAP) index, 22 which can also precisely predict the survival rate in patients with IPF. 23 For monitoring IPF, the commonly used questionnaires that evaluate the overall health impact and the degree of breathlessness include Saint George's Respiratory Questionnaire (SGRQ) 24 29 Many of the mentioned tests predict the mortality of patients with IPF, but only some of them have been validated in Asians. 7, 23,26,29 The aim of the current study is to investigate the prognostic value of CPET, PFT, GAP index, SGRQ score, mMRC score, and SF-36 score for 1-year mortality in patients with IPF.

2-1. VE/VCP2 slope in CPET, mMRC and GAP score were higher in non-survival group in patients with IPF
In total, 34 patients with IPF were enrolled into to this study, and 6 patients died during the follow-up period within 1 year. Patients' data including age, gender, body mass index (BMI), CPET, PFT, 6MWT, GAP index, mMRC, SGRQ, and SF-36 scores were collected and compared between the survivors and non-survivors (Table   1). We found no significant differences in age, gender and BMI between the groups.
In the data of CPET, higher VE/VCO2 slope, lower heart rate recovery one minute after exercise, higher minimal and maximal Vd/Vt, and more reduction in oxygen saturation was noted in the non-survivors. The more prominent oxygen desaturation was also observed in 6MWT of the non-survivors. In the data of PFT, no significant differences were noted between the two groups. Furthermore, lower GAP index and mMRC scores were also noted in the survivors.
2-2. VE/VCO2 slope>35.1, PETCO2≤3.6 kPa, mMRC ≥2 and GAP score ≥4 were associated with 1-year mortality in patients with IPF In order to find the predictors of mortality in our study participants, univariable cox regression analysis was done, and the data were shown in Table 2. Significant higher hazard ratio of mortality was observed in patients with lower PETCO2, higher VE/VCO2 slope, higher dead space, more desaturation in 6MWT, higher GAP index and higher mMRC score. However, further multivariable cox regression model failed to identify any parameters with significant p value. ROC curve analysis was then performed to further identify the predictors excluding the effect of time to mortality. Table 3, we discovered that mMRC score, GAP index, PETCO2 and VE/VCO2 slope had an AUC > 0.7, and the cutoff value for them was 2, 4, 3.6 kPa, and 35.1, respectively. The curve of ROC analysis for the four parameters was displayed in Figure 1. The result of the De Long test was shown in Table 4, and no significant differences were established between the AUC of each two of the four parameters.

Discussion
In this preliminary study, we established the predictive value of CPET, GAP index and mMRC dyspnea score for mortality in an Asian cohort of patients with IPF.
Although statistical significance could not be achieved in the multivariable Cox regression analysis, the four important parameters GAP index, mMRC dyspnea score, PETCO2 and VE/VCO2 slope in CPET, with a significantly large AUC in the ROC curve analysis were identified. These parameters can provide a reference to enable healthcare workers to manage Asian patients with IPF more effectively. To the best of our knowledge, this is the first study to confirm the mortality predictive role of CPET for Asian patients with IPF. VE/VCO2 slope and PETCO2 have been considered to be two important prognostic factors in patients with interstitial lung disease and COPD, 30 and both the two parameters were shown to be significant predictors of mortality in both the univariable cox regression analysis and ROC curve analysis of our study. In the influential consensus statement by European Association for Cardiovascular Prevention & Rehabilitation and American Heart Association, the likelihood of secondary pulmonary hypertension increases if VE/VCO2 slope > 45 and resting PETCO2 < 33mmHg with less than 3mmHg increase during CPET. 30 Our study further added crucial supplement to the above statement, revealing the likelihood of mortality increases with VE/VCO2 slope > 35.1 and exercise PETCO2 < 3.6kPa (27mmHg). PETCO2 is the partial pressure of carbon dioxide detected in the expired air during end expiration, reflecting the volume of alveolar dead space calculated by Bohr equation. 31 Higher alveolar dead space, signifying worse ventilation/perfusion mismatch, is evidenced to contribute to mortality in critically ill patients. 32,33 As shown in Table 1, our study also revealed higher resting and exercise dead space in the non-survivors. As pulmonary hypertension is a common consequence of IPF, the disease severity of pulmonary hypertension is evidenced to be proportional to PETCO2 at rest and exercise. 34 Our study echoes the above literatures, and emphasizes the importance of mortality prediction in Asian IPF patients.
The indicator of ventilation efficiency, VE/VCO2 slope, can never be overemphasized in the prediction of mortality in IPF patients. Past literatures had used the value of VE/VCO2 in the mortality prediction, with the cut-off value being 34 9 and 45 35 at anaerobic threshold. A past study also showed that VE/VCO2 slope and VO2/kg had the strongest correlation with the outcome of mortality. 8 In our study, although no predictive role was established in VO2/kg, higher VE/VCO2 slope still demonstrated higher hazard ratio of mortality in cox regression analysis (p=0.029) and a large AUC> 0.7 in ROC analysis despite the p value was not significant (p=0.108). The meaning of VE/VCO2 slope is straightforward, representing the amount of ventilation volume needed per unit of carbon dioxide expelled. Therefore, higher value of VE/VCO2 slope signifies ventilatory inefficiency, indicating higher disease severity and higher likelihood of progression into pulmonary hypertension, 36 and thus, higher likelihood of mortality.
Since 2012, the GAP index, which is derived from the patient's gender, age, FVC, and DLCO, has been used for survival prediction. 22 In addition, the index has been successfully validated in Korean patients. 37 However, the AUC for 1-year mortality was only 0.619 for GAP stage in a recnelty nationwide cohort study in Korea. 38 In our study, although the AUCs of FVC and DLCO were not large enough to predict mortality, the GAP index, with an AUC as high as 0.86, is a potential predictor of survival. In contrast to the original version of GAP staging, which has two cut-off values (i.e., ≤3 and ≥6) to define three stages of IPF, our result demonstrated only a single cut-off value of ≥4 between the two original values. Another simple questionnaire, the mMRC, demonstrated the highest AUC of 0.87 in our study, which is in line with the study by Nishiyama et al., who validated the predictive ability of mMRC in patients with IPF. 29 In our study, a cut-off value of ≥2 had predictive mortality for patients with IPF, which happen to be the same for patients with COPD in an earlier study. 39 Our study has several limitations. First, the follow-up period was short; thus, only a limited number of mortality cases were assessed in the one-year follow-up. If any of the data of the nonsurvivors were statistical outliers, the results of our study could be significantly affected. Second, the sample size was small. This was because of the relatively low prevalence of IPF in Taiwan (2.0-4.9 cases/100,000 persons). 40 If more patients were to be included in our study, the results of the multivariable Cox regression and Kaplan-Meier curve analysis would have been more convincing.
However, our study was a prospective single-center observational study of patients with newly diagnosed IPF. These preliminary results still hold value, especially for healthcare workers managing Asian patients with IPF.

Study design
This prospective observational cohort study was conducted in a tertiary medical centre in central Taiwan. Patients diagnosed as having IPF and who agreed to provide informed consent were enrolled from December 2018 to December 2020. The study protocol was reviewed and approved by the Institutional Review Board of Taichung Veterans General Hospital (IRB number, CE18325B; date of approval, Dec 18, 2018).

Study population and setting
In this study, a subgroup analysis of patients with a definite diagnosis of IPF according to the criteria of the American Thoracic Society, the European Respiratory Society, the Japanese Respiratory Society, and the Latin American Thoracic Association was conducted. 41 The parameters included patient demographic data such as age, sex, height, and body weight. The index day was defined as the day on which the patient signed the inform consent form. The participants completed SGRQ, SF-36, and mMRC dyspnea questionnaires on the index day. Within 1 week of enrolment, the participants were administered PFT, 6MWT, and CPET. The patients were then followed up for 12 months to trace the incidence of mortality.

Pulmonary function test
In this study, PFT included two components. First, spirometry was conducted, which involves a period of quiet breathing, followed by expiration into the sensor as quickly and as long as possible after taking the deepest inspiration. FVC and FEV1 data were obtained from the results of the spirometry. Second, the difference in the partial pressure between the inspired and expired carbon monoxide was measured.
After a deep inspiration, the participants had to hold their breath for 8-12 s. The

Cardiopulmonary exercise test
We performed CPET by using an electromagnetically braked cycle ergometer.
All test procedures were performed in accordance with the guidelines of the American Heart Association and comprised 3-min rest, 3-min of unloaded pedalling, pedalling with the brake gradually applied in a ramp manner up to the maximal level tolerated by the patient, and unloaded pedalling for 3 min. 43 The ramp protocol to increase the workload during testing, in contrast to the conventional stepwise protocol, is characterized by the continuously gradual escalation of the work rate with even distribution within every moment of the exercise phase. CPET provided the following data: oxygen consumption in relation to body weight (VO2/kg), functional aerobic impairment (FAI; ratio of the difference between the observed and the predicted peak oxygen consumption to the predicted peak oxygen consumption), percentage of peak oxygen consumption in relation to the heart rate (O2 pulse), end-tidal partial pressure of carbon dioxide (PETCO2) at maximal exercise, minute ventilation to carbon dioxide output (VE/VCO2) slope, heart rate recovery at 1 min after peak exercise (HR recovery), O2 desaturation, ratio of dead space to tidal volume at rest (maximal Vd/Vt), and maximal exercise (minimal Vd/Vt) and change in Vd/Vt during the test. 44 Furthermore, anaerobic threshold was determined by the V-slope method, 45 which is the time when the rate the carbon dioxide production increases out of proportion to the rate of oxygen consumption.

Six-Minute Walk Test
6MWT was performed in accordance with the guidelines of the American Thoracic Society. 46 The patients were instructed to walk as far as possible for 6 min in a corridor between the two orange traffic cones placed 30-m apart. Data on oxygen saturation and the distance walked in 6 min were obtained.

Statistical analysis
All the data were analysed using IBM SPSS Statistics 23.0 (Armonk, New York, USA) software. The Shapiro-Wilk test was used to examine variables with a normal distribution. For non-normally distributed data, the Mann-Whitney U and Fisher exact tests were used to determine the presence of any significant difference between survivors and nonsurvivors. The chi-square test was used to analyze discrete variables.
Cox proportional regression analysis and receiver operating characteristic (ROC) curve analysis were performed for all the parameters measured. The parameters with areas under curve (AUCs) of >0.7 were identified, and the cut-off points were decided to maximize the sum of sensitivity and specificity values of the respective ROC curves. Furthermore, the De Long test was used to determine the differences between the AUCs of each two parameters. A p value of <0.05 was considered significant.