Embracing the unknown: risk stratification in heart failure with preserved ejection fraction with the EPYC score

This editorial refers to ‘Predicting the transition to and progression of heart failure with preserved ejection fraction: a weighted risk score using bio-humoural, cardiopulmonary, and echocardiographic stress testing’ by N.R. Pugliese et al. pp. 1650–1661. Heart failure with preserved ejection fraction (HFpEF) accounts for a significant proportion of patients with heart failure (HF), ranging between 22% and 76%, depending on the characteristics of the observed population.1–3 It is current understanding that HFpEF results from different aetiologies and pathophysiological substrates and is associated with various comorbidities, which may have an impact on adverse outcomes.4 In-hospital and long-term mortality in HFpEF patients are comparable to those with HFrEF; hence, HFpEF has been increasingly recognized as an emerging clinical concern.5,6 Therefore, predicting outcomes in HFpEF is an unmet clinical need that may have implications for the management. In the current issue of the European Journal of Preventive Cardiology, Pugliese et al. present a novel risk score, Exercise-induced PulmonarY Congestion (EPYC) based on the prospective assessment of 274 patients with HFpEF, Stages A, B, and C, compared with 30 control subjects.7,8 The multimodal patient evaluation included biomarker assessment, conventional and speckle tracking echocardiography, cardiopulmonary exercise testing (CPET), as well as lung ultrasound.8 The median follow-up was 18.5 months, and the composite endpoint of HF hospitalization and cardiovascular (CV) death was observed in 81 cases. The EPYC score is based on five independent predictors, including N-terminal pro-B type natriuretic peptide (NT-proBNP), peak-exercise oxygen consumption (VO2), the ventilatory efficiency slope, defined as minute ventilation to carbon dioxide production (VE/VCO2) slope, peak-exercise systolic pulmonary artery pressure (PAP), and the change in stress-to-rest count in B-lines on lung ultrasound. Based on the EPYC score, three risk groups are defined, as follows: (i) low risk (<3 points), (ii) intermediate risk (3–6 points), and (iii) high-risk (>6 points), with the estimated probability of survival free of CV death or HF hospitalization being 93%, 52%, and 20%, respectively. Importantly, the score was predictive not only of the composite endpoint but also of each of the components, CV death, and HF hospitalization. The worst prognosis was expectedly noted in the patients with the EPYC score >6 points. Each score point increased the hazard ratio for the composite endpoint by 46%.8 Thus far, several risk scores have been developed to predict outcomes in HFpEF. The recent additions include three scores (Modified Glasgow Prognostic Score, CONUT, 3A3B score) based on baseline characteristics, comorbidities, laboratory findings, and natriuretic peptides.9–14 The H2FPEF score, additionally includes echocardiographically derived left ventricular (LV) filling pressure, and PAP.15 This score has been shown to predict all-cause mortality and HF hospitalization in patients with acute HF,10 as well as the risk of 1-year rehospitalization and other adverse CV events in HFpEF.14,16 The highest clinical value is attributed to risk scores that have been externally validated e.g. MAGGIC and CHARM.13,17 The MAGGIC score, composed of 13 variables, has been validated in an HFpEF cohort to predict both morbidity and mortality during 4-year follow-up.17 The CHARM score, composed of 21 variables, demonstrated that age >60 years, diabetes, and LV ejection fraction <45% were the most accurate predictors of all-cause mortality and the composite of HF hospitalization and CV death.13 The EPYC score represents a comprehensive prognostic tool that combines five components selected to provide high overall prognostic value across the stages of HFpEF severity. NT-proBNP value >300 pg/mL was used as an inclusion criterion, while the cut-off value for scoring 1 point in the EPYC score was ≥900 pg/mL. The difficulty of providing an appropriate cut-off for NT-proBNP in HFpEF lies in a diversity of factors and comorbidities that modify NT-proBNP levels.18–20 Accordingly, values as low as 110 pg/mL have a significant predictive value to diagnose isolated diastolic dysfunction in patients with preserved LV ejection fraction.21 Obese patients tend to have lower values of NT-proBNP in HF, unlike patients with atrial fibrillation (AF) in whom higher values are required for the diagnosis. Hence, the clinical applicability of the score may be limited in select population, although the authors have shown that the EPYC score ≥3 can predict adverse outcomes in a small subset of AF patients. The EPYC score also utilizes exercise-derived parameters reflecting exercise-induced changes in haemodynamics. This has been recognized in the recent HFA-PEFF algorithm, developed to facilitate HFpEF diagnosis.18 Likewise, combining stress-echocardiographic testing and CPET has been proposed for improving diagnostic capacity for HFpEF in asymptomatic patients.22,23 Moreover, exercise-induced increase in LV filling pressure and PAP have been shown to predict major CV events and all-cause mortality in HFpEF.24 The score developed by Pugliese et al. includes two indices derived from CPET, namely peak exercise VO2, and VE/VCO2. As previously demonstrated by Guazzi et al. abnormalities in these indices are associated with cardiac dysfunction in patients with HFpEF.25 Importantly, CPET derived decrease in VO2 and oscillatory ventilation pattern have been linked to adverse outcomes and increased all-cause mortality in HFpEF.26,27 Increased PAP is indicative of pulmonary hypertension and is an independent predictor of all-cause mortality, HF hospitalizations, and/or cardiac death in HFpEF.28,29 An increase in PAP during exercise testing in HFpEF correlates with invasive haemodynamic monitoring,30 suggesting that non-invasive PAP assessment is reliable. The detection of B-lines on lung ultrasound is an established method to assess pulmonary congestion.31 B-lines may be absent at rest and become apparent during exercise due to an increase in LV filling pressure in HF.31 An increase in the number of B-lines at exercise is predictive of a greater risk of all-cause mortality and non-fatal myocardial infarction in subjects referred to stress echocardiography.32 The study by Pugliese et al. confirmed that assessing peak-exercise B-lines is feasible and prognostically significant, providing a rationale for including this parameter in the EPYC score. The EPYC score has certainly gained in prognostic accuracy from combining multimodality evaluation that includes exercise-derived parameters. The relevance is most apparent in HFpEF patients that have normal LV diastolic function and filling pressure at rest but develop haemodynamic changes and symptoms during effort.18 However, the clinical utility of the EPYC score may suffer due to its complexity. Indeed, exercise-derived components, and in particular CPET, may be time consuming and require equipment and personnel proficient in expert measurements. In addition, the study is limited by a small sample size, single-centre design, male predominance (uncharacteristic for HFpEF), and the absence of external validation. The use of the EPYC score is also conditioned by the capability of subjects to perform stress-testing, specifically in the elderly that constitute the majority of HFpEF population. Furthermore, only CV mortality was assessed, while the predictive validity for all-cause mortality (largely attributed to non-cardiac aetiologies) remains unknown.33,34 Taking into account the growing burden and considerable morbidity and mortality of HFpEF, the establishment and validation of risk scores is highly important. The implementation of advanced methods such as artificial intelligence for the detection of HFpEF based on electrocardiography, might be also investigated in future research on risk scores.35 Although there are no therapies that can improve outcomes in HFpEF, the development of the more sophisticated prognostic scores is relevant, as they enable monitoring of high-risk patients may translate into improved survival, pending advances in treatment options.36 Conflict of interest: none declared. The opinions expressed in this article are not necessarily those of the Editors of the European Heart Journal-Cardiovascular Pharmacotherapy or of the European Society of Cardiology.