Abstract 12810: Data-Driven Clusters of Cardiac Structure and Function and Association With Premature Cardiovascular Disease: The Coronary Artery Disease in Young Adults (CARDIA) Study

Autor: Michael C Wang, Lindsay R Pool, Norrina B Allen, Paul A Reyfman, Yuan Luo, Joao A Lima, Don M Lloyd-jones, Sadiya S Khan
Rok vydání: 2021
Předmět:
Zdroj: Circulation. 144
ISSN: 1524-4539
0009-7322
DOI: 10.1161/circ.144.suppl_1.12810
Popis: Introduction: Abnormal indices of cardiac structure and function likely cluster together. However, the clinical relevance of data-driven patterns of cardiac remodeling across a broad range of echocardiographic measures, particularly in mid-life, is not known. Hypothesis: Latent class analysis (LCA) will identify distinct clusters of cardiac structure and function based on echocardiographic indices in middle-aged adults that differ in risk of CVD. Methods: We included all CARDIA participants without CVD who underwent an echocardiogram. Continuous parameters were indexed to participant height (if appropriate), imputed with the SVDImpute algorithm (if missing), and filtered to remove pairwise correlations >0.6. We applied LCA using a total of 31 echocardiographic indices (standardized to mean 0 and standard deviation 1) to identify distinct groups based on model convergence, the Bayesian Information Criterion, and class membership >5%. We examined the association between groups and risk of premature CVD (coronary artery disease, stroke, or heart failure) with Cox regression adjusted for traditional risk factors. Results: Of 3,361 participants, mean (SD) age was 50.1 (3.6) years, 57% were female, and 46% were non-Hispanic Black. LCA identified 3 groups: (1) optimal cardiac mechanics; (2) impaired systolic function; and (3) impaired diastolic function ( FIGURE ). Over a median 8.9 years of follow-up, risk of CVD was higher in the impaired diastolic function group (unadjusted hazard ratio [HR] 4.08 [95% CI: 2.48, 6.71] and adjusted HR 1.95 [1.12, 3.40]) compared with the optimal group. The impaired systolic function group had a higher unadjusted risk of CVD (1.86 [1.10, 3.15], which attenuated after adjustment (1.36 [0.79, 2.36]). Conclusions: Unbiased, model-based clustering of cardiac structure and function in middle-aged adults identified distinct patterns of cardiac remodeling that were associated with risk of future CVD.
Databáze: OpenAIRE