Clinical outcomes among distinct groups of patients with severe tricuspid valve regurgitation. (3rd October 2022)

• Record Type: Journal Article
• Authors: Rao, V; Giczewska, A; Chiswell, K; Felker, G M; Wang, A; Parikh, K; Vemulapalli, S
• 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/ehac544.1656
• Languages: English
• ISSNs: 0195-668X

Abstract

Abstract: Introduction: Severe tricuspid valve regurgitation (TR) is associated with increased 1-year morbidity and mortality. Characterization by valve etiology (primary, secondary, and lead-associated), a classification borrowed from mitral valve disease, has not been universally shown to correlate with outcomes. Purpose: Among a large, racially diverse cohort with newly identified severe TR, we aimed to 1) characterize outcomes of severe TR by etiology, and 2) assess whether unsupervised phenoclustering or supervised outcome-driven prediction trees were more effective in establishing subgroups of TR with differential clinical risk profiles. Methods: We retrospectively analyzed outcomes of 5-year all-cause death and a composite of death or heart failure hospitalization (HFH) among adult patients with new severe TR identified by echocardiography between 2007 to 2018 at a large academic tertiary referral center in the United States. Patients were initially categorized by etiology of TR, including primary, secondary, and lead-associated. Second, we separately applied unsupervised hierarchical clustering to identify distinct clusters using demographics, clinical, and echo data at the time of diagnosis. Third, we applied a supervised recursive partitioning algorithm (survival trees) by each outcome to identify distinct TR subgroups. We estimated the cumulative incidence of death and composite death or HFH over 5 years by 1) etiology of TR, 2) distinct clusters, and 3) groups identified by supervised learning (prediction trees). Results: Among 2, 379 consecutive patients with newly identified severe TR, the median age was 70 years, 61% were female, and 40% were Black. Event rates (95% CI) were 30.9 (29.0 to 32.8) events/100 PY for death and 49.0 (45.9 to 52.2) events/100 PY for composite death or HFH over median follow-up of 1.6 years. Event rates were similarly high across TR etiology groups for both death and composite death or HFH (Figure 1). Multiple methods of unsupervised clustering did not yield distinct clusters by patient demographic and imaging characteristics. After applying supervised survival tree modeling, four phenoclusters with distinct clinical prognoses were separately identified for death and composite death or HFH (Figure 2). Variables identified to partition the cohort to discriminate both death and composite death or HFH were age, albumin, blood urea nitrogen, right ventricular contractility, and right ventricular systolic pressure (all p<0.05). Conclusions: Five-year cumulative incidence of adverse events among patients with newly diagnosed severe TR was 69% for death and 80% for composite death or HFH. TR etiology did not stratify prognosis, while supervised survival tree models identified phenoclusters with distinct clinical risk. The identified subgroups of severe TR with differential outcomes offer insights towards enrichment in clinical trials of TR and risk/benefit analysis in patients undergoing TR therapies. Funding Acknowledgement: Type of funding sources: Private company. Main funding source(s): Abbott