Development and validation of an integrated biomarker-based congestion score: the BIOTOOL-CHF DISCO study
C Delacroix, G Coutance, D C H Ceelen, J M Ter Maaten, I Diemberger, G Savarese, M Metra, J Nunez, M Masetti, A A Voors, L PotenaAbstract
Background
Congestion is the dominant driver of symptoms, hospitalisations, and adverse outcomes in heart failure (HF), yet its bedside evaluation relies on semi-quantitative signs prone to inter-observer variability, leaving subclinical congestion frequently undetected. An integrative multi-biomarker approach capturing complementary pathophysiological pathways may better reflect the heterogeneous phenotypes of congestion than any single analyte.
Purpose
To develop and validate a standardised multi-biomarker congestion score and evaluate its association with HF outcomes.
Methods
In the BIOTOOL-CHF project, 10 prespecified biomarkers reflecting distinct congestion mechanisms were tested against a prospectively assessed Clinical Congestion Score (CCS, 0–5, integrating a standardised assessment of jugular venous pressure, orthopnea, and peripheral oedema) in the derivation (n=2,516) and validation (n=1,738) cohorts of the BIOSTAT-CHF study. Multivariable ordinal models with complementary statistical selection strategies identified a parsimonious biomarker panel. Discrimination and calibration were assessed using bootstrap internal validation (500 resamples) and independent external validation. A standardised Biomarker-based Congestion Score (BCS, 0–100) was derived for clinical interpretability.
Results
Four biomarkers capturing complementary pathways, namely bio-ADM (vascular permeability), CA-125 (serosal stress), sST2 (myocardial strain), and NT-proBNP (cardiac wall stress) were independently associated with CCS. The model achieved excellent discrimination in development (optimism-corrected weighted mean AUC 0.81, 95% CI 0.78–0.83; ordinal C-index 0.74) and external validation (AUC 0.78, 95% CI 0.75–0.81; calibration slope 0.88). Over 9 months, HF hospitalisation occurred in 16.4% and 14.7%, and cardiovascular death in 9.3% and 8.2% in development and validation cohorts, respectively. BCS was strongly and independently associated with both endpoints in both cohorts (p<0.001 for all) and identified high-risk patients even with none or minimal clinical congestion.
Conclusions
A standardised multi-biomarker congestion score integrating four complementary pathophysiological pathways was derived and externally validated. It accurately quantified congestion on a continuous scale and predicted HF events independently of bedside assessment. These findings support prospective randomised trials testing BCS-assisted diuretic titration within the BIOTOOL-CHF programme.For image description, please refer to the figure legend and surrounding text.