Supervised learning-derived tailored risk-stratification in patients with severe secondary mitral regurgitation. (3rd October 2022)

• Record Type: Journal Article
• Authors: Heitzinger, G; Spinka, G; Prausmueller, S; Pavo, N; Dannenberg, V; Dona, C; Kammerlander, A; Nitsche, C; Kastl, S; Strunk, G; Huelsmann, M; Rosenhek, R; Hengstenberg, C; Bartko, P E; Goliasch, G
• Is Part Of: European heart journal. Volume 43(2022)Supplement 2
• Publication Date: 2022-10-03
• Subjects: Cardiology -- Periodicals; Heart -- Diseases -- Periodicals; 616.12005
• DOI: 10.1093/eurheartj/ehac544.1643
• Languages: English
• ISSNs: 0195-668X

Abstract

Abstract: Background: Mitral regurgitation secondary to heart failure (sMR) has considerable impact on quality of life, heart failure (HF) rehospitalizations and mortality. A diverse burden of comorbidities suggests multifaceted aspects of individual risks. This risk-spectrum has never been studied but is essential to understand disease trajectories. Objectives: To provide a comprehensive and structured decision-tree-like approach to risk-stratification in patients with severe sMR. Methods: This large-scale, long-term observational study included 1317 patients with severe sMR from the entire HF spectrum (preserved, mid-range and reduced ejection fraction). Primary endpoint was all-cause mortality and survival tree analysis, a supervised learning technique, was applied to identify patient subgroups with excessive risk of mortality (Figure 1). Results: Eight distinct subgroups that differed significantly in long-term survival were identified (Figure 2). Subgroup 7, characterized by younger age (≤66), higher hemoglobin (>12.7 g/dl) and higher albumin levels (>40.6 g/l) had the best survival. In contrast, subgroup 5 displayed a 20-fold risk of mortality (HR 95% CI: 20.38 ([0.78–38.52]), P<0.001) and presented with older age (>68 years) and low serum albumin (≤40.6 g/l) and higher NT-proBNP levels (≥9750 pg/ml). Results were consistent in internal and temporal validation. Conclusion: Supervised machine learning reveals an unexpected heterogeneity in the sMR risk-spectrum, indicating the clinical challenges tied to severe sMR. A decision-tree-like model can guide through the risk spectrum and provide tailored risk-stratification. This structured approach provides the foundation to generate hypotheses towards improved therapeutic strategies and optimized patient care. Funding Acknowledgement: Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Austrian Science Fund