Abstract 15241: Prediction of Incident Adverse Cardiovascular Disease Events Using Left Ventricular Mass: Impact of Adding Trabecular and Papillary-Muscle Mass to Conventional (Compacted) Mass
Background: Left ventricular mass (LVM) by volumetric cardiovascular magnetic resonance (CMR) is an independent predictor of incident adverse cardiovascular disease (CVD) events. Prior studies have used compacted LVM (cLVM), ignoring trabecular and papillary-muscle mass (TPM). We assessed whether use of total LVM (TotLVM), the sum of cLVM and TPM, augments prediction of incident adverse CVD events over a cLVM-based model.
Methods: A total of 1494 Framingham Offspring cohort members (65±9y, 46.8% men) underwent ECG-gated, contiguous multislice LV short-axis bSSFP CMR during 2002-2006 and had cLVM and TotLVM determined. cLVM was calculated from manually traced endo- and epicardial borders. TPM was quantified by custom software using thresholds based on signal intensities of cLVM and the LV bloodpool. Both cLVM and TotLVM were indexed (i) to body surface area. Adverse CVD events comprised CVD death, myocardial infarction, unstable angina, ischemic stroke, and first admission for heart failure. Seventy Offspring with prevalent (prior to CMR) CVD were excluded. We compared two Cox proportional hazards models using Harrell’s C statistic: both models adjusted for CVD risk factors (age, sex, systolic blood pressure, total and HDL cholesterol, diabetes, smoking, hypertension treatment), but incorporated either cLVMi or TotLVMi.
Results: Overall, men had greater cLVMi than women (62.3±11.1 vs. 48.6±8.0 g/m2, p<0.0001) and greater TotLVMi (79.3±12.9 vs. 63.0±9.3 g/m2, p<0.0001). There were 71 (of 1424, 5.0%) new incident adverse CVD events over median 8.4-y follow up. Those experiencing an incident event had greater cLVMi and TotLVMi than those free of an event. The hazard ratio (HR) for an incident adverse CVD event was HR=1.37 [95%CI = 1.11 - 1.68], per 10 g/m2 of cLVMi, p=0.003; and HR=1.30 [1.08 - 1.55] per 10g/m2 TotLVMi, p=0.005. The C statistics were similar for the model with cLVMi, C=0.757 [CI 0.697 - 0.816] and the TotLVMi model, C=0.753 [0.694 - 0.813].
Conclusions: In a community-dwelling cohort of middle-aged and older adults, both cLVMi and TotLVMi augmented prediction of incident adverse CVD events over standard CVD risk factors. However, the performance of TotLVMi did not differ from that of conventional cLVMi in multivariable-adjusted Cox models.
Author Disclosures: S. Qazi: None. P.N. Gona: None. C.J. Salton: None. M.G. Larson: None. C.J. O’Donnell: None. W.J. Manning: None. M.L. Chuang: None.
- © 2016 by American Heart Association, Inc.