Risk Stratification of Asian Patients with Heart Failure and Reduced Ejection Fraction: The Effectiveness of the Echo Heart Failure Score

Risk stratification in heart failure (HF) is a cornerstone in clinical management.1 Several prognostic indices of HF have been identified;2-4 however, it is increasingly recognized that risk stratification based on the use of a single parameter is less effective than that based on a combination of several parameters.1 Transthoracic echocardiography (TTE) provides a variety of measurements of potential prognostic significance in HF. Although HF guidelines recommend the use of TTE in the management of HF patients, usually only left ventricular (LV) ejection fraction (EF) is incorporated in clinical scoring systems.1 Recently, the Echo Heart Failure Score (EHFS)—an echocardiographic multimarker risk stratification tool—has been developed in a large Italian cohort to predict mortality risk in systolic HF.4 The EHFS incorporates measures of left and right ventricular function and left atrial volume, easily assessable during routine TTE. The EHFS is a simple but powerful stratification score, and appeared to be strongly predictive of mortality risk in patients with chronic systolic HF, independently of, and with incremental predictive power over, known clinical predictors such as age, New York Heart Association (NYHA) class, natriuretic peptides, kidney dysfunction, and anaemia. However, whether EHFS results are generalizable to HF populations of other ethnicities and regions has not been studied. Ethnicity-based differences in echocardiographic parameters are known to exist,5 and carry important implications for the detection of cardiac disease and prognostication. Ethnicity may also influence the performance characteristics of echocardiographic indices, depending on the prevalence of disease and confounding co-morbidities. Comparisons of HF prediction scores in Asian vs. Western settings are lacking. In a recent study, N-terminal pro-brain natriuretic peptide (NT-proBNP) showed superior discriminative capacity in an Asian compared to a New Zealand HF cohort.6 We therefore aimed to evaluate the effectiveness of EHFS in risk stratification in an Asian cohort of systolic HF patients. Asian patients were from the Singapore Heart Failure Outcomes and Phenotypes (SHOP) study, a multicentre prospective study of Southeast Asian patients with a validated history of HF uniformly recruited in a stable state.7 All patients had a validated diagnosis of HF and underwent comprehensive standardized TTE at study entry, with oversight by a central echocardiographic core lab. In a previous study from the SHOP cohort, 438 HF patients underwent comprehensive right-sided echocardiographic measurements to study the impact of right ventricular dysfunction in HF.8 In the current study, a subset of these patients with LVEF <50% were included, in whom measurements of left atrial and ventricular volumes, EF, tricuspid annular plane systolic excursion (TAPSE), pulmonary artery systolic pressure (PASP), diastolic function, and severity of mitral regurgitation were performed as described previously8 and according to guideline recommendations.9 Left atrial and LV volumes were indexed by body surface area. The EHFS was calculated as previously described.4 Briefly, the score is based on five different parameters or ‘predictors’, i.e. left atrial volume index (LAVI) ≥45 mL/m2, LV end-systolic volume index (LVESVI) ≥84 mL/m2, mitral E wave deceleration time (DT) ≤140 ms, TAPSE <16 mm, and PASP ≥45 mmHg. A value of 1 was assigned when a predictor was present. The sum of the five predictors resulted in EHFS ranging from 0 to 5 points. In the previous study, EHFS ≥3 identified a higher-risk group with significantly reduced long-term survival. Follow-up assessment was scheduled periodically and included clinical visits and/or telephone interviews. The primary endpoint was death from any cause. Continuous variables are expressed as mean (± standard deviation), categorical variables as percentages. Univariable and multivariable analyses by Cox proportional hazards models were performed to assess the association between EHFS and outcome, adjusting for other prognostic factors. The Kaplan–Meier method was used to estimate survival probabilities according to EHFS <3 or ≥3. Differences between survival curves were analysed using the log-rank test. Integrated discrimination improvement (IDI), and net reclassification improvement (NRI) with the continuous NRI method, were computed to assess the incremental discriminative value of EHFS when added to significant predictors of mortality. To check the calibration of the model, we compared the observed event probabilities with the event probabilities predicted by the model including the EHFS. We used post-estimation features following Cox model fitting, as suggested.10 We defined three prognostic groups from the 25th and 75th centiles of the predicted survival probabilities from the Cox model, and compared them with the non-parametric Kaplan–Meier estimates. The Hosmer–Lemeshow test was also computed. Informed consent was obtained from each patient, and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee. A total of 175 patients were studied, whose clinical, demographic, and echocardiographic characteristics are described in Table 1. Participants were predominantly male, with a mean age of 64.3 ± 11.3 years. Ischaemic cardiomyopathy was the underlying aetiology in 57.1%. Ethnicity was 67% Chinese, 9% Indian, and 23% Malay, reflecting the Southeast Asian composition of the population of Singapore. Overall mean EHFS was 2.19 ± 1.35, with 40.6% of patients having an EHFS ≥3. During a median follow-up of 730 days, the endpoint all-cause death occurred in 29 patients. As shown in Figure 1A the incidence rate of death (per 100 patients/year) increased as EHFS increased. Figure 1B clearly shows that high-risk patients (EHFS ≥3) had lower survival compared with the low-risk group (log-rank 9.31, P = 0.002). Compared with the low-risk group, high-risk patients had a three-fold increased risk of mortality [hazard ratio (HR) 3.09, 95% confidence interval (CI) 1.44–6.66, P = 0.004]. After adjusting for age, NYHA class, diastolic blood pressure, history of hypertension, LVEF, NT-proBNP, and QRS duration, by multivariable Cox regression model, EHFS ≥3 remained a significant predictor of death (HR 2.28, 95% CI 1.01–5.20, P < 0.05), with the ability to reclassify high-risk patients as demonstrated by analysis of IDI (3.9%, P < 0.05) and continuous NRI (24.4%, P < 0.05) when added to a base clinical model of predictors. As shown in Figure 2, the calibration of the model was satisfactory. Actual survival fitted the predicted in each prognostic group, although CIs tended to be wide reflecting small subgroup sample sizes. The Hosmer–Lemeshow test yielded a χ2 of 4.71 (P = 0.318). These are the first data showing that EHFS might be a useful tool for risk stratification in Asian patients with HF. In addition to its clinical relevance, this observation suggests that echocardiographic risk markers in HF are similar across geographies and ethnicities, regardless of differences in clinical risk factors in different patient populations. Clinical HF risk factors differ greatly between Asian and Western cohorts.11 This may limit the application of a universal clinical risk score. On the other hand, this may not be the case for echocardiographic risk factors, since cardiac remodelling may be expected to be a universal mechanism of adverse prognostic impact. Such an assumption needs to be tested, since ethnic differences in echocardiographic reference values are known to exist.5 Recently, a systematic review of HF risk prediction scores concluded that the majority of studies are conducted in the Western world only, with only 36% of prediction models validated in an independent cohort, and none derived from or validated within Southeast Asian ethnicities.1 To our knowledge, Western-derived HF risk scores have only been validated in Asia in a Japanese cohort (Get With The Guideline–Heart Failure risk score and the Seattle Heart Failure Model, for prediction in the acute setting and long-term survival, respectively).12, 13 The limited availability of global and geographically representative data and lack of validation of risk scores in different Southeast Asian ethnicities illustrates the need for the current study. In a Western Europe cohort, EHFS more accurately identified HF patients at high risk of mortality.4 Our results suggest that the EHFS may offer promising applications for clinical use also in Asian patients with HF. The advantages of the EHFS are that (i) it is based upon variables that can be easily measured at any outpatient clinic or echocardiography laboratory; (ii) it can be used for comparison with European cohorts; and (iii) it is independent of, and incremental to, clinical risk factors. Limitations include a limited sample size and variability in echocardiographic methods, although in both the Italian derivation study and the current Asian validation study, standard techniques recommended by international guidelines were applied and the results were robust. In conclusion, these initial data suggest that the EHFS allows risk stratification in Asian patients with HF. This suggests that echocardiographic risk markers of HF are similar across geographies and ethnicities. The authors would like to thank the clinicians, clinical research coordinators, and sonographers in their institutions who assisted with the conduct of this study. This work was supported by a Centre Grant from the National Medical Research Council of Singapore (grant number R-172-003-2190511) and by the Alexandre Suerman MD/PhD programme from the UMC Utrecht. Conflict of interest: C.S.P.L. is supported by a Clinician Scientist Award from the National Medical Research Council of Singapore; has received research support from Boston Scientific, Bayer, Thermofisher, Medtronic, and Vifor Pharma; and has consulted for Bayer, Novartis, Takeda, Merck, Astra Zeneca, Janssen Research & Development, LLC, Menarini, Boehringer Ingelheim, and Abbott Diagnostics. A.M.R. is supported by a Senior Translational Research Award from the National Medical Research Council of Singapore and is Principal Investigator for the Centre Grant made to the National University Heart Centre, Singapore, which funded this study. He has consulted for, and received research support from, Bayer, Novartis, Merck, Medtronic, Pfizer, Boehringer Ingelheim, Astra Zeneca, Roche Diagnostics, Thermo Fisher, Alere, Critical Diagnostics, and Abbott Diagnostics.