Supplemental material Cardiopulmonary exercise testing (CPET) protocol

This supplemental material outlines a detailed cardiopulmonary exercise testing (CPET) protocol and working definitions to characterize dysfunctional breathing (DB) phenotypes, particularly in post-COVID-19 patients. The aim is to standardize assessment, objectively determine maximal effort, and identify exercise-limiting factors including hyperventilation syndrome (HVS), erratic breathing/periodic deep sighs (EB/PDS), mixed patterns, respiratory limitation, O2 delivery/utilization impairment, chronotropic insufficiency, and metabolic limitation. The importance lies in distinguishing DB from underlying cardiopulmonary diseases and guiding tailored management.

The protocol references established standards and prediction equations, including the SHIP study for CPET reference values, ERS 2019 statement on standardization of CPET in chronic lung disease, and classic CPET norms (Jones et al., Fairbarn et al.). It also cites literature on ventilatory efficiency (V′E/V′CO2 relationships) in health and disease, DB classification (Boulding et al.), and abnormal CPET response patterns (Neder et al.). These works support the analytical criteria for hyperventilation patterns and ventilatory efficiency assessment.

Incremental CPET was performed seated on a cycle ergometer using a Geratherm Ergostik system with facemask interface. Patients fasted ≥4 hours; usual medications were continued. Protocol phases: resting ≥2 minutes until stable RER <1; 2 minutes unloaded pedaling warm-up; individualized linear ramp targeting 8–12 minutes of exercise; 3–5 minutes recovery. Continuous monitoring: finger pulse oximetry (earlobe oximetry if poor signal), 12-lead ECG, blood pressure (motion-tolerant TANGO M2). Arterial blood gases and lactate were sampled at rest and within 30 seconds post-peak. Borg modified scale assessed dyspnoea and muscular fatigue at peak; reason for test termination recorded. Predicted values used SHIP reference equations. Objective maximal effort determination followed ERS 2019 guidance and included presence of ≥1: V′O2 plateau; V′O2 ≥100% predicted; peak ventilatory reserve <15%; RER >1.05; peak HR ≥100% predicted; lactate ≥8 mmol/L. Analysis used tabulated data, modified Wasserman panels, additional unfiltered ventilation slope graphs (VT and BF over V′E; VT and BF over time), and continuous flow-volume matched with volume-over-time. Steps: (1) Confirm objective maximality; (2) Determine V′O2/W slope as test quality criterion (normal ~8–12 mL·kg−1·min−1·W−1); (3) Graphically determine peak V′O2 and, if ≤84% predicted with maximal effort, identify the limiting factor; (4) Determine anaerobic threshold (AT) via V-slope and secondary criteria; (5) Determine respiratory compensation point (RCP) if applicable; (6) Define dominant pattern via numeric, graphical, and arterial blood gas integration. Working definitions: Hyperventilation (HVS) identified by persistent, episodic, or end-exercise disproportionate hyperventilation based on V′E/V′CO2 slope and nadir, intercept, V′E/V′CO2 at 40 W, VD/VT ratio, PaCO2 (rest/end-exercise), V′E/V′CO2 nadir, V′E/V′O2, PETCO2 and PETO2 kinetics over load; elevation attributed to hyperventilation only after excluding cardiorespiratory disease. Hyperventilation throughout test: high V′E/V′CO2 slope and nadir (>30) with normal VD/VT and no significant cardiorespiratory disease, indicating low PaCO2 set point. End-exercise hyperventilation: disproportionate ventilatory response not explained by lactic acidosis (end-exercise pH >7.35 or >7.45 with low lactates). Episodic HVS: frequent inappropriate acute hyperventilation with decoupling of V′E and V′CO2 seen on time-resolved ventilatory indices and gas exchange. Particularity in DB: loss of linear V′E–V′CO2 relationship can make slope analysis and noninvasive AT determination difficult; V′E/V′CO2 nadir is especially informative then. EB/PDS definitions: EB is erratic, inappropriate ventilation with high VT and BF variability; PDS is abnormally frequent sighs (at rest >1/min or ~15 in 15 minutes). Sigh during exercise defined qualitatively as abnormally high VT not explained by other phenomena. Detection leveraged unfiltered ventilation slope graphs and continuous flow-volume/volume-over-time, with clinician presence to avoid misinterpretation from speaking/coughing or voluntary patterns. Classification after CPET: (1) EB/PDS without HVS; (2) HVS without EB/PDS; (3) mixed HVS and EB/PDS. O2 delivery/utilization impairment defined by low peak V′O2 (≤84% predicted) due to low O2 pulse in a maximal test or early AT (≤50% predicted peak V′O2). Other definitions: Chronotropic insufficiency = peak HR ≤80% predicted in an objectively maximal test. Metabolic limitation (subjective, due to lack of consensus): in BMI ≥30 kg/m2, high cost of unloaded pedaling and dissociation between predicted maximal load and predicted peak V′O2. Respiratory limitation defined by hypoxemia at end-exercise (PaO2 <10 kPa at ~450 m a.s.l.) or ventilatory reserve exhaustion (≤15%) in a maximal test; ventilatory reserve = (1 − [peak ventilation/max voluntary ventilation]) × 100; MVV estimated as FEV1 × 40.

Case examples illustrated DB phenotypes and CPET interpretation: (1) Normal patient with maximal CPET, normal peak V′O2 (99% predicted), normal V′O2/W slope (8.8), and no DB or respiratory limitation. (2) Patient 1 (52M): mixed EB/PDS with HVS; V′O2/W slope 13.2; massive sighs (~5 L) detected on continuous graphs; non-linear V′E/V′CO2; end-exercise respiratory alkalosis; DB caused termination. (3) Patient 2 (31F smoker): EB with possible HVS; non-maximal CPET; elevated but non-linear V′E/V′CO2; wide dispersion in VT and BF; DB limited exercise with dyspnoea, dizziness, palpitations. (4) Patient 3 (48F, BMI 38): HVS without EB/PDS; maximal by RER 1.28; severely reduced peak V′O2 (41% predicted); early AT (31%); respiratory alkalosis with minimal lactate; concomitant O2 delivery/utilization impairment likely deconditioning; metabolic limitation due to obesity. (5) Patient 4 (45F, asthma history): EB/PDS without HVS; maximal by RER 1.14; peak V′O2 88% predicted; AT 51%; DB was limiting factor. (6) Patient 5 (58F): HVS without EB/PDS; maximal by HR 104% predicted and RER 1.06; peak V′O2 89% predicted; tachypnoea with regular pattern; VD/VT normal at end-exercise; no EID or pulmonary vascular features; DB was limiting factor. (7) Patient 6 (17F): EB/PDS without HVS; maximal by RER 1.4; peak V′O2 55% predicted; early AT 38%; frequent sighs confirmed; O2 delivery/utilization impairment likely deconditioning. Subgroup data (supplemental tables): Subgroups included Periodic Deep Sighing (PDS, N=22), Hyperventilation (HVS, N=10), and Mixed (N=16). Significant differences included: history of asthma more common in PDS (31.8%) vs HVS 0% and Mixed 0% (p=0.008); mMRC dyspnoea scale distribution differed across groups (p=0.037); PCFS distribution differed (p=0.008); loss of weight at first consultation differed (p=0.009). Nijmegen total scores: medians PDS 26 [IQR 10], HVS 33.5 [14.3], Mixed 28 [22]; proportion ≥23: 63.2%, 80%, 68.8% (ns). HADS anxiety/depression risk distributions showed high proportions in severe category without significant between-group differences; SF-36 physical functioning showed a trend toward lower scores in HVS/Mixed (p=0.052). Missing data (selected): peak pH/PaO2/PCO2/lactates missing in 7 (14.6%); resting ABG components missing in 3 (6.3%); V′E/V′CO2 intercept/nadir missing in 2 (4.2%); peak VT missing in 1 (2.1%); NT-proBNP missing in 2 (4.2%). Overall, CPET frequently identified DB as the main exercise-limiting factor and helped exclude other causes (e.g., pulmonary vascular disease) in specific cases.

The protocolized CPET with extended graphical analyses reliably identified dysfunctional breathing phenotypes and delineated their contribution to exercise intolerance in post-COVID patients. The use of both filtered (modified Wasserman panels) and unfiltered ventilation slope graphs, coupled with continuous flow-volume/volume-over-time traces and arterial blood gas sampling, allowed differentiation between hyperventilation-driven ventilatory inefficiency and primary cardiopulmonary pathology. In several cases, DB was the predominant limiting factor despite normal routine investigations, while CPET also uncovered concomitant contributors (deconditioning, metabolic limitation in obesity). Recognizing loss of linearity in the V′E–V′CO2 relationship as a hallmark in some DB patients guided reliance on V′E/V′CO2 nadir and ABG patterns when slope analysis and noninvasive AT determination were unreliable. The approach informs clinical decision-making by distinguishing DB from pulmonary vascular disease and by supporting targeted rehabilitation and breathing retraining.

A structured CPET protocol with comprehensive analysis, including unfiltered ventilatory pattern graphs and arterial blood gases, enables robust classification of dysfunctional breathing into EB/PDS, HVS, and mixed phenotypes and identification of exercise-limiting mechanisms in post-COVID patients. This methodology supports differential diagnosis against cardiopulmonary disease, highlights roles for deconditioning and obesity-related metabolic limitation, and can guide individualized management. Future work should establish consensus definitions for sighs during exercise and metabolic limitation, validate these criteria in larger controlled cohorts, and standardize objective measures of DB severity and treatment response.

Interpretation of ventilatory patterns requires clinician presence to avoid misclassification due to artifacts (speaking/coughing/voluntary patterns). Loss of linear V′E–V′CO2 coupling in DB can preclude reliable slope-based analysis and noninvasive AT determination. Definitions for sighs during exercise and metabolic limitation are not yet standardized, relying on qualitative or subjective criteria. Missing data were present for several variables (e.g., peak ABG and lactate in 14.6%), potentially affecting subgroup comparisons. The supplemental material pertains to a prospective case series design rather than a controlled cohort, limiting generalizability.