Heart failure subphenotypes based on repeated biomarker measurements are associated with clinical characteristics and adverse events (Bio-SHiFT study)
International Journal of Cardiology, 2022
de Lange I., Petersen T., de Bakker M., Akkerhuis K., Brugts J., Caliskan K., Manintveld O., Constantinescu A., Germans T., van Ramshorst J., Umans V., Boersma E., Rizopoulos D., Kardys I.
| Disease area | Application area | Sample type | Products |
|---|---|---|---|
CVD | Patient Stratification | Plasma |  Olink Target 96 |
Abstract
Background: This study aimed to identify heart failure (HF) subphenotypes using 92 repeatedly measured circulating proteins in 250 patients with heart failure with reduced ejection fraction, and to investigate their clinical characteristics and prognosis.
Methods: Clinical data and blood samples were collected tri-monthly until the primary endpoint (PEP) or censoring occurred, with a maximum of 11 visits. The Olink Cardiovascular III panel was measured in baseline samples and the last two samples before the PEP (in 66 PEP cases), or the last sample before censoring (in 184 PEP-free patients). The PEP comprised cardiovascular death, heart transplantation, Left Ventricular Assist Device implantation, and hospitalization for HF. Cluster analysis was performed on individual biomarker trajectories to identify subphenotypes. Then biomarker profiles and clinical characteristics were investigated, and survival analysis was conducted.
Results: Clustering revealed three clinically diverse subphenotypes. Cluster 3 was older, with a longer duration of, and more advanced HF, and most comorbidities. Cluster 2 showed increasing levels over time of most biomarkers. In cluster 3, there were elevated baseline levels and increasing levels over time of 16 remaining biomarkers. Median follow-up was 2.2 (1.4-2.5) years. Cluster 3 had a significantly poorer prognosis compared to cluster 1 (adjusted event-free survival time ratio 0.25 (95%CI:0.12-0.50), p < 0.001). Repeated measurements clusters showed incremental prognostic value compared to clusters using single measurements, or clinical characteristics only. Conclusions: Clustering based on repeated biomarker measurements revealed three clinically diverse subphenotypes, of which one has a significantly worse prognosis, therefore contributing to improved (individualized) prognostication.