Importance of Cardiopulmonary Exercise Testing amongst Subjects Recovering from COVID-19

The cardiopulmonary exercise test (CPET) provides an objective assessment of ventilatory limitation, related to the exercise minute ventilation (VE) coupled to carbon dioxide output (VCO2) (VE/VCO2); high values of VE/VCO2 slope define an exercise ventilatory inefficiency (EVin). In subjects recovered from hospitalised COVID-19, we explored the methodology of CPET in order to evaluate the presence of cardiopulmonary alterations. Our prospective study (RESPICOVID) has been proposed to evaluate pulmonary damage’s clinical impact in post-COVID subjects. In a subgroup of subjects (RESPICOVID2) without baseline confounders, we performed the CPET. According to the VE/VCO2 slope, subjects were divided into having EVin and exercise ventilatory efficiency (EVef). Data concerning general variables, hospitalisation, lung function, and gas-analysis were also collected. The RESPICOVID2 enrolled 28 subjects, of whom 8 (29%) had EVin. As compared to subjects with EVef, subjects with EVin showed a reduction in heart rate (HR) recovery. VE/VCO2 slope was inversely correlated with HR recovery; this correlation was confirmed in a subgroup of older, non-smoking male subjects, regardless of the presence of arterial hypertension. More than one-fourth of subjects recovered from hospitalised COVID-19 have EVin. The relationship between EVin and HR recovery may represent a novel hallmark of post-COVID cardiopulmonary alterations.


Introduction
Shortly after discharge, survivors of COVID-19 present lung function alterations with reduction in diffusion capacity for carbon monoxide (DL CO ) [1] and severe impairments in physical function during activities of daily living [2]. Few data are available about a comprehensive evaluation of COVID-19 clinical alterations during a more extended period.
The cardiopulmonary exercise test (CPET) provides an objective assessment of exercise capacity, adding physiological aspects that limit the individual's performance [3]. In particular, the exercise minute ventilation (V E ) relative to carbon dioxide output (V CO2 ) (V E /V CO2 ) shows complementary information about ventilatory limitation and ventilatory control [4,5]. During incremental exercise, the relationship between V E and V CO2 may be plotted on a y-axis (V E ) and x-axis (V CO2 ); the slope of this regression line (V E /V CO2 slope ) may be considered an indicator of ventilatory efficiency [4,5]. Lower and upper limits of normal range of V E /V CO2 slope are reported from approximately 21 to 31 [4,6,7]. High values of V E /V CO2 slope define an exercise ventilatory inefficiency (EVin) [4,5]; this pathophysiological feature may explain the out-of-proportion breathlessness of patients with chronic obstructive pulmonary disease (COPD) [5]. Smokers with normal spirometry but with low values of DL CO may have EVin [8].
In our pilot study, we explored the methodology of CPET to post-COVID subjects in order to evaluate the presence of cardiopulmonary alterations.

Materials and Methods
A dedicated outpatient clinic has been organised at our tertiary hospital enrolling all adult subjects previously hospitalised for interstitial pneumonia due to COVID-19, with or without respiratory failure. The prospective RESPICOVID study has been designed to evaluate the prevalence and the clinical impact of pulmonary damage in subjects recovered from COVID-19. In a subgroup of subjects (RESPICOVID2) a CPET has been performed. The study protocol was approved by the local Ethics Committee (no. 2785CESC), according to the Good Clinical Practice recommendations and the requirements of the Declaration of Helsinki. Written informed consent was obtained from all participants.

Inclusion Criteria
All consecutive patients discharged were considered.

Exclusion Criteria
The study has not considered subjects with the following criteria: (a) age > 65 years; (b) all concomitant previous respiratory or non-respiratory diseases; (c) chronic respiratory failure or need for oxygen-therapy under exertion; (d) moderate obesity defined by a body mass index (BMI) ≥ 35 kg/m 2 ; (e) inability to perform functional tests; (f) inability to perform a CPET with a peak respiratory exchange ratio (RER) < 1.05 (to exclude poor motivation). Among chronic diseases, only stable arterial hypertension was accepted.

Measurements
All measures were collected prospectively beginning on 17 July 2020, after more than five months from subjects' discharge (mean time 169 days, standard deviation (SD) 28 days). We recorded demographic and anthropometric variables, data concerning the hospitalisation, clinical symptoms, and gas-analysis.
Lung function and CPET procedures were performed according to international recommendations [3,9]. A flow-sensing spirometer connected to a computer for data analysis (Jaeger MasterScreen PFT System) was used to measure lung function. Forced vital capacity (FVC), forced expiratory volume in the first second (FEV 1 ), total lung capacity (TLC), and inspiratory capacity (IC) were recorded. FEV 1 /FVC ratio and IC/TLC ratio were taken as the index of airflow obstruction and resting hyperinflation, respectively. Diffusion capacity for carbon monoxide (DL CO ) was measured by the single breath method. FEV 1 , TLC, and DL CO were expressed as percentages of the predicted values [10,11]. For the CPET, according to the ATS/ACCP Statement [3], we used a cycle ergometer (Cosmed, Milan, Italy) with a ramp protocol of 10 to 25 watts increment every minute and based on the predicted peak power output, in order to achieve an exercise time between 8 and 12 min. Subjects were asked to avoid caffeine, alcohol, cigarettes, and strenuous exercise 24 h before the day of testing; to eat a light breakfast; and to avoid eating for the 2 h before the test. Subjects suspended β-blockers before testing, but they could take their current antihypertensive therapies. During the test, subjects were asked to maintain a pedal frequency of 65 per minute and were continuously monitored [3]. Patients were continuously monitored with a 12-lead electrocardiogram (ECG) and a pulse oximeter; blood pressure was measured every two minutes. Stopping criteria consisted of symptoms, such as unsustainable dyspnoea, leg fatigue or chest pain, a significant ST-segment depression at ECG, or a drop in systolic blood pressure or oxygen saturation ≤84% [3]. Oxygen uptake (V O2 ) at the peak was expressed in mL/kg/min. The ventilatory response during exercise was expressed as a linear regression function by plotting minute ventilation (V E ) against carbon dioxide production (V CO2 ) obtained every 10 s, excluding data above the ventilatory compensation point, and the slope (V E /V CO2 slope ) and Y-intercept (V E /V CO2 intercept ) values were obtained from the regression line. We used the regression equation of V E /V CO2 slope for healthy subjects, according to Sun et al. [6], considering three standard deviations as the upper limit. Then, we considered subjects having a normal range of V E /V CO2 slope (exercise ventilatory efficiency-EVef ) and subjects with over the upper limit of V E /V CO2 slope (EVin). The cardiovascular response to exercise was expressed by the oxygen pulse (O 2 pulse), the double product (DP) reserve, and the heart rate (HR) recovery, considering the value of heart rate measured after 1 min of exercise stops. At the end of the exercise, dyspnoea and leg fatigue were measured by a Borg 6-20 perceived exertion rate (RPE) scale [12]. Reasons for considering a maximal test were (a) a plateau of the V O2 more than 20 s; (b) a RER > 1.15; (c) a rate of perceived exertion >18 on the Borg RPE scale [3].
As a measure of physical tolerance, walking capacity was assessed by the 6 min walking distance (6MWD) and performed according to the recommended guidelines [13]; the better of two consecutive tests was considered for the analysis. The reference equation for healthy adults was also used [14].
The Italian version of the International Physical Activity Questionnaire (IPAQ) was administered to measure the physical activity of the subjects in the last seven days, deriving three levels of metabolic equivalent of task (METs): inactive, minimally active, and healthenhancing physical activity (HEPA) active [15].
A preliminary Shapiro-Wilk test was performed. Data are reported as percentages for categorical variables, as mean (SD) or median [first quartile; third quartile] for continuous variables with a normal or non-normal distribution. Categorical variables were compared by the χ2 test or the Fisher exact test, while continuous variables were assessed by the independent t-test or the non-parametric Mann-Whitney test. Pearson (r) and Spearman (ρ) correlations have been carried out between parametric variables. The area under a receiver operating characteristic curve (AUC) measured the diagnostic discrimination property of significant predicting ventilatory inefficiency. All analyses were performed using IBM SPSS, version 17.0 (IBM Corp., Armonk, NY, USA), with p-values of <0.05 considered statistically significant.

Results
The RESPICOVID study enrolled 130 subjects, but according to the selective criteria for the RESPICOVID2, defined to avoid baseline bias influencing the ventilatory response to exercise, we performed the CPET in 28 subjects. All subjects performed a maximal exercise test, and 8 out of 28 (29%) had EVin. As compared to subjects with EVef, subjects with EVin showed a reduction in HR recovery and V E /V CO2 intercept , with an increase by definition of the V E /V CO2 slope and vigorous METs (Table 1). V E /V CO2 slope was inversely correlated with HR recovery (r −0.537, p = 0.003) ( Figure 1); this correlation was confirmed in a subgroup of older subjects (age > 55 y, N = 14, ρ −0.611, p = 0.020), males (N = 22, r −0.543, p = 0.009), non-smokers (N = 19, r −0.611, p = 0.005), regardless of the presence of arterial hypertension (yes, N = 9, r −0.669, p = 0.049; no, N = 19, r −0.487, p = 0.034).  Figure 1. Scatterplot between V E /V CO2 slope and HR recovery. Lines represent the regression with the 95% confidence intervals. Abbreviations: V E /V CO2 slope represents the slope of minute ventilation-V E to carbon dioxide output-V CO2 ratio; HR, heart rate.

Discussion
Our pilot study is the first evaluating, in survivors of COVID-19 pneumonia, of CPET variables during an extended follow-up after hospital discharge. Altho considered subjects had a normal lung function and a preserved maximal exercis ity, surprisingly, more than one-fourth had an EVin, which is a determinant of H ery, especially older male non-smokers, regardless of the presence of arterial h sion.
In smokers with normal spirometry but low values of DLCO, EVin may be pr as well as [8] an impaired peripheral endothelial function [16]. In the context of a capillary membrane damage, decrements in DLCO may be more likely related to nary microvascular abnormalities than impaired gas distribution [8]. We may hyp a similar mechanism in our post-COVID subjects, in which we observe five mon discharge a selective lung function impairment in DLCO reduction.
Ventilatory inefficiency in the healthy population is not a common occurre normal upper limit of VE/VCO2 slope is 31 [6,7]. Variables related to age and sex [6,1 as chronic pulmonary and cardiovascular conditions, may influence the exercise tory efficiency [3,4]; however, there is no concrete evidence that the fitness leve impact on exercise ventilation [18,19]. In addition, regular endurance training m prove exercise ventilatory efficiency (potentially with a reduction in VE/VCO2 slop ducing peripheral chemoreceptor sensitivity [18]. In our sample, subjects having E a coexisting presence of higher values of 6MWD, workload, and VO2 at peak, si higher aerobic capacity. The levels of vigorous weekly METs were higher comp subjects with EVef; however, this was not correlated to VE/VCO2 slope or HR recove not shown). Interestingly, ventilatory efficiency is not related to residual lung limitations and specific treatment during the COVID-19 hospitalisation. Althoug lative, our findings of exercise ventilatory alterations in post-hospitalised subject ated without baseline bias, seems to be specific and COVID-related.

Discussion
Our pilot study is the first evaluating, in survivors of COVID-19 pneumonia, the role of CPET variables during an extended follow-up after hospital discharge. Although our considered subjects had a normal lung function and a preserved maximal exercise capacity, surprisingly, more than one-fourth had an EVin, which is a determinant of HR recovery, especially older male non-smokers, regardless of the presence of arterial hypertension.
In smokers with normal spirometry but low values of DL CO , EVin may be present [8] as well as [8] an impaired peripheral endothelial function [16]. In the context of alveolarcapillary membrane damage, decrements in DL CO may be more likely related to pulmonary microvascular abnormalities than impaired gas distribution [8]. We may hypothesise a similar mechanism in our post-COVID subjects, in which we observe five months from discharge a selective lung function impairment in DL CO reduction.
Ventilatory inefficiency in the healthy population is not a common occurrence. The normal upper limit of V E /V CO2 slope is 31 [6,7]. Variables related to age and sex [6,17], such as chronic pulmonary and cardiovascular conditions, may influence the exercise ventilatory efficiency [3,4]; however, there is no concrete evidence that the fitness level has an impact on exercise ventilation [18,19]. In addition, regular endurance training may improve exercise ventilatory efficiency (potentially with a reduction in V E /V CO2 slope ) by reducing peripheral chemoreceptor sensitivity [18]. In our sample, subjects having EVin had a coexisting presence of higher values of 6MWD, workload, and V O2 at peak, signs of a higher aerobic capacity. The levels of vigorous weekly METs were higher compared to subjects with EVef ; however, this was not correlated to V E /V CO2 slope or HR recovery (data not shown). Interestingly, ventilatory efficiency is not related to residual lung function limitations and specific treatment during the COVID-19 hospitalisation. Although speculative, our findings of exercise ventilatory alterations in post-hospitalised subjects, evaluated without baseline bias, seems to be specific and COVID-related.
HR recovery represents a marker of cardiac autonomic dysfunction and a predictor of mortality in adults without heart disease history [20]. In COPD patients, HR recovery is associated with endothelial dysfunction, representing peripheral impairment [21]. Moreover, EVin in COPD is a predictor of the delay of HR recovery [22]. Although our post-COVID survivors do not have an airways obstruction as in COPD, recent reports highlighted frequent extrapulmonary manifestations, especially involving the cardiovascular system (myocardial dysfunction, arrhythmia, and acute coronary syndromes), attributed to virus-mediated endothelial-cell damage [23]. Our findings on EVin and HR recovery could therefore represent a novel hallmark of post-COVID cardiopulmonary alterations. In these subjects, also with low cardiovascular risk (non-smokers without arterial hypertension), an in-depth assessment of exercise-induced ventilatory and cardiovascular parameters by CPET allows the identification (or monitoring) of early specific alterations. Further studies are needed to evaluate the progressive persistence and the prognostic role of these alterations.
As a limitation, we acknowledge the small number of subjects included, related to selective criteria considering younger subjects without previous diseases. Moreover, we lack data concerning the residual organic pulmonary damage (by lung ultrasound or thorax computed tomography scan) with indirect signs of pulmonary hypertension (by echocardiography).

Conclusions
More than one-fourth of post-COVID subjects present an exercise ventilatory inefficiency related to lower heart rate recovery; this aspect may be a sign of systemic alterations present in these subjects. Further studies in a very large cohort of subjects need to confirm our finding. In the future, it may be interesting to apply the methodology of CPET in elderly patients with or without coexisting diseases to evaluate the impact of COVID-19 on single chronic conditions.
We suggest CPET as a potentially useful tool for identifying ventilatory and cardiovascular alterations in subjects recovered from COVID-19. Moreover, CPET may be useful as a monitoring system for exercise capacity and cardio-ventilatory limitations in subjects admitted to a rehabilitation program.