Comparison Of Cardiopulmonary Exercise Test Variables To Predict Adverse Events In Heart Failure

PURPOSE: Stratification of heart failure (HF) patients with an increased risk for adverse events is critical given the rising HF burden. This study aims to compare the predictive value of various maximal and submaximal cardiopulmonary exercise test (CPET) variables for adverse clinical outcomes in HF patients. METHODS: 237 HF patients (66 [58-73] years, 30% women, 70% HFrEF) completed a CPET and had five years of follow-up. Patient and disease characteristics and clinical outcomes (all-cause mortality, major adverse cardiovascular events and unplanned cardiovascular-related hospitalization) were extracted from electronic patient files. Receiver operating characteristics curves for maximal (e.g. peak VO2) and submaximal CPET variables (e.g. VE/VCO2 slope, cardiorespiratory optimal point (COP), VO2 at anaerobic threshold) were compared using the Akaike Information Criterion (AIC) method, whereas their calibration was assessed. RESULTS: 103 participants (43%) reached the composite endpoint of which 55 participants (23%) died. Area under the curve of the different CPET variables are depicted in Figure 1. Percent predicted peak VO2 was the best predictor for adverse outcomes (AIC: 302.6) followed by COP (AIC: 304.3), and relative peak VO2 (mL/(kg·min), AIC: 304.4). Relative peak VO2 (AIC: 217.1) and COP (AIC: 224.4) were also among the three best predictors for all-cause mortality, together with absolute peak VO2 (in ml/min, AIC: 220.5). A good calibration between the observed and predicted event rate was observed for these variables. CONCLUSION: Percent predicated and relative peak VO2 had the best predictive accuracy for adverse events and mortality, but the submaximal COP had a non-inferior predictive accuracy for adverse events in patients with HF. These findings highlight the potential of submaximal exercise testing in HF patients.This project is supported by an Eurostars Grant (E!114585). EAB has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No [101064851]. The funding sources were not involved.