Barcelona Bio-HF Calculator Version 2.0: incorporation of angiotensin II receptor blocker neprilysin inhibitor (ARNI) and risk for heart failure hospitalization.

Abstract

Estimating risk for an individual with heart failure (HF) is routine for the practising physician. This may sometimes be done using experience and clinical acumen, or by using a risk model. A number of prediction models with broad variation in terms of validation and output have been developed, but only a few are freely available as online calculators.1 The Barcelona (BCN) Bio-HF Calculator (www.bcnbiohfcalculator.org) (Figure 1),2 developed 3 years ago and discussed in the 2016 European Society of Cardiology HF guidelines,3 incorporates three biomarkers that reflect different facets of HF pathophysiology: N-terminal pro-B-type natriuretic peptide (NT-proBNP), a marker of myocardial stretch; high-sensitivity cardiac troponin T (hs-cTnT), a marker of myocyte injury, and high-sensitivity soluble ST2, which reflects myocardial fibrosis and remodelling. The calculator estimates the risk for all-cause death,2 has been externally validated,4 and was highlighted by Levy and Anand5 as a reference for the appropriate methodology for adding single or multiple variables to a risk model. The combination of clinical and treatment data plus routine laboratory data and biomarkers is also valuable for predicting HF-related hospitalization. Furthermore, the incorporation of novel drugs and devices into the HF armamentarium, notably sacubitril–valsartan, which have strong impacts on death and HF hospitalization,6 prompted an update of the BCN Bio-HF Calculator. The BCN Bio-HF Calculator Version 2.0 was derived from a cohort of 864 consecutive treated HF outpatients [72% men; mean age 68.2±12 years; New York Heart Association (NYHA) class I-II/III-IV 73%/27%, left ventricular ejection fraction (LVEF) 36%, ischaemic aetiology 52.2%].3 During follow-up of up to 5 years, 363 deaths and 210 first HF-related hospitalizations were recorded; 430 patients suffered at least one event of the composite endpoint. In the update, three new clinical variables (duration of HF in months, number of HF-related hospitalizations in the preceding year, and diabetes mellitus) and four new treatments [mineralocorticoid receptor antagonists, angiotensin II receptor blocker neprilysin inhibitors (ARNI), cardiac resynchronization therapy (CRT) and implantable cardioverter defibrillator (ICD)] were added to the original variables (age, sex, NYHA functional class, LVEF, serum sodium, estimated glomerular filtration rate, haemoglobin, loop diuretic dose, beta-blocker, angiotensin-converting enzyme inhibitor/angiotensin-II receptor blocker and statin treatments, and hs-cTnT, ST2 and NT-proBNP levels). Beta values for ARNI treatment were derived from the benefit observed in the PARADIGM-HF trial, which involved the largest and best characterized cohort of patients treated with ARNIs.6 HF-related hospitalization was estimated taking into account competing risk for death. Model performance was evaluated using discrimination, calibration and reclassification tools. The C-statistics [area under the curve (AUC)] at 2 years for the model with biomarkers using logistic regression were 0.83 for all-cause death, 0.79 for HF-related hospitalization, and 0.80 for the composite endpoint. Discrimination was significantly better than that obtained in a model without biomarkers for risk for death (P = 0.001), risk for HF hospitalization (P < 0.05) and the composite endpoint (P = 0.001) (supplementary material online, Tables S1–S3). Calibration improved in the model with biomarkers, and reclassification with this model using continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI) was also highly significant (P < 0.001). Using NRI, the BCN Bio-HF Calculator Version 2.0 model with biomarkers reclassified in the correct direction 39% of patients for risk for death, and 42% for risk for the composite endpoint relative to the clinical model (supplementary material online, Tables S4 and S5). Validation for up to 2 years was possible in a subgroup of 1934 patients from the PARADIGM-HF study cohort6 for whom the three biomarkers were available. The C-statistics were 0.70 for both risk for death and risk for HF-related hospitalization at 2 years. Some variables and endpoints differed between the Barcelona derivation cohort and the PARADIGM-HF validation cohort. Indeed, risk prediction for the composite endpoint could not be validated because the composite endpoint in PARADIGM was cardiovascular death or HF-related hospitalization, rather than all-cause death. In a manner similar to the present efforts in a cohort of chronic ambulatory HF patients, the BIOSTAT-CHF study recently developed and validated three risk models to predict all-cause mortality, HF-related hospitalization and the composite endpoint in a cohort of worsening HF patients.7 These researchers obtained C-statistic values of 0.73, 0.69 and 0.71 for the three outcomes, respectively. Both methods are pioneers in their use of HF biomarkers and are appropriate in two different clinical scenarios. In conclusion, the updated version of the BCN Bio-HF Calculator incorporates new clinical variables and allows better individual prediction of all-cause death, HF-related hospitalization and the composite endpoint for up to 5 years. To the best of the present authors' knowledge, this is the first online calculator to incorporate treatment with an ARNI in the prediction of risk in HF patients. Risk prediction is a cornerstone of HF management. Accurate prediction of risk for death and/or HF hospitalization may identify high-risk patients and candidates for intensified monitoring and treatment, such as drug dose increases, switches to ARNI, CRT and ICD implantation, and surveillance of ischaemia or valvular dysfunction among other factors, and should help those caring for these patients to make better management decisions. Supplementary materials are available for beta-values (supplementary material online, Tables S6–S8). Conflict of interest: J.L. and A.B.G. declare relationships with Critical Care Diagnostics and have received lecture honoraria from Roche Diagnostics. J.J.V.M. reports the receipt of other forms of remuneration from Novartis, Cardiorentis, Amgen, Oxford University/Bayer, GlaxoSmithKline, Theracos, Abbvie, DalCor, Pfizer, Merck, AstraZeneca, Bristol Myers Squibb (BMS), and Kidney Research UK (KRUK)/Kings College Hospital, London/Vifor-Fresenius Pharma outside the submitted work. The other authors have nothing to disclose. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.