Late-Breaking Science - Innovations in drug treatment
Personalised lifetime prediction of survival and treatment benefit in patients with HFrEF: The LIFE-HF model
27 Aug 2022
The LIFE-HF model
Absolute treatment effects vary among patients with heart failure and reduced ejection fraction (HFrEF) based on individual risk factors and life expectancy, yet reliable ways to individually estimate risk and benefit from contemporary treatments are lacking.
This unmet need led to the development and validation of the LIFEtime-perspective for Heart Failure (LIFE-HF) model as presented at a Late-Breaking Science session yesterday by Doctor Stefan Koudstaal (Green Heart Hospital - Gouda, Netherlands). The LIFE-HF project was developed in close collaboration with various experts from the ESC’s Cardiovascular Risk Collaboration (CRC), which endorses its clinical use. Dr. Koudstaal and a stellar team of HF experts initially derived the model from the PARADIGM-HF and ATMOSPHERE trials (n=15,415), using left-truncated Cox proportional hazards functions. The primary outcome was CV death or first hospitalisation for HF, and the secondary outcome was CV death alone. Complementary functions were developed for non-CV mortality. The predictors included age, sex, NYHA class, prior hospitalisation for HF, diabetes, extracardiac vascular disease, systolic blood pressure, LVEF, NT-proBNP, uric acid and total bilirubin.
Life-tables were constructed to predict lifelong HF hospitalisation-free survival and overall survival, with external validation using data from the SwedeHF registry (n=42,063), the ASIAN-HF registry (n=4,479) and the DAPA-HF trial (n=4,744). The model’s discriminative ability was satisfactory with C statistics of 0.66 to 0.70 for HF hospitalisation-free survival and 0.64 to 0.73 for overall survival, and calibration was adequate in all populations. Individual lifetime benefit from optimising HFrEF pharmacotherapy was predicted using the model. Compared with background therapy consisting of an angiotensin-converting enzyme inhibitor (ACEi) and a beta-blocker, adding a mineralocorticoid receptor antagonist (MRA), a sodium-glucose co-transporter-2 inhibitor (SGLT2i) and switching from an ACEi to an angiotensin receptor-neprilysin inhibitor (ARNI) led to a median increase of 3.7 years gained in HF hospitalisation-free survival, ranging between 0.3 years to 18.1 years for individual patients (interquartile range [IQR] 2.4 to 5.5). For overall life-years gained, optimised pharmacotherapy led to a median increase of 2.5 years, ranging from 0.3 to 11.8 years (IQR 1.7 to 3.7).
The LIFE-HF model was then developed into an interactive calculator to allow individual estimation of a patient’s risk and treatment benefit. For example, for a 60-year-old woman with NYHA class II HF, LVEF of 35%, no diabetes and NT-proBNP of 1000 pg/mL, the estimate for 5-year risk of CV death/HF hospitalisation was 24% on an ACEi and a beta-blocker, but optimising pharmacotherapy was expected to improve this to 11%. At the same time, HF hospitalisation-free life expectancy was estimated to improve by 9 years, from 71 to 80 years.
The authors concluded that the LIFE-HF model has the potential to improve HFrEF management by facilitating personalised medicine and shared decision making. As a next step, the authors aim to develop the model into a web-based calculator in the near future, which will be freely available to all physicians and patients worldwide.