Most heart failure diagnoses arrive too late for meaningful prevention — by the time symptoms emerge, structural damage is often irreversible. A large multicenter study now suggests that the texture and composition of fat surrounding the heart, detectable from scans patients are already receiving, may carry a legible biological warning signal years before clinical heart failure develops. This reframes a routine diagnostic scan as a potential early-warning instrument.
The investigation enrolled 72,751 adults across nine UK centers who had undergone coronary computed tomographic angiography (CCTA) without prior heart failure or myocardial infarction. A fully automated pipeline extracted 1,655 radiomic features — encompassing volume, shape, and higher-order texture characteristics — from epicardial adipose tissue (EAT). A survival autoencoder model synthesized these into a single risk signature termed the fat radiomic profile for HF (FRPHF). Developed in nearly 59,000 individuals and externally validated in a geographically distinct cohort of over 13,400, the signature demonstrated predictive power independent of age, sex, coronary artery disease severity, and EAT volume alone. Roughly 2.7–2.9% of participants developed heart failure over median follow-up periods of four to five years.
The scientific weight of this work lies in its scale and external validation — two features that are conspicuously absent from most radiomic research, which tends to be small, single-center, and retrospective. Epicardial fat is not merely structural padding; it is metabolically active tissue that exchanges paracrine signals with adjacent myocardium, and prior research has linked EAT inflammation to atrial fibrillation and coronary plaque vulnerability. The FRPHF model extends that biology into heart failure prediction by capturing subtle textural changes that simple volume measurements miss. Key limitations include an observational design preventing causal inference, a predominantly UK population that may limit generalizability across diverse ethnic groups, and the relatively short follow-up window. Whether integrating FRPHF into clinical triage would alter outcomes — rather than simply identify risk — remains untested. Still, as a potentially paradigm-shifting use of existing imaging infrastructure, this finding warrants serious attention from cardiologists and preventive medicine specialists.