Left ventricular global function index by magnetic resonance imaging--a novel marker for assessment of cardiac performance for the prediction of cardiovascular events: the multi-ethnic study of atherosclerosis

Abstract

Left ventricular (LV) function is generally assessed independent of structural remodeling and vice versa. The purpose of this study was to evaluate a novel LV global function index (LVGFI) that integrates LV structure with global function and to assess its predictive value for cardiovascular (CV) events throughout adult life in a multiethnic population of men and women without history of CV diseases at baseline. A total of 5004 participants in the Multi-Ethnic Study of Atherosclerosis underwent a cardiac magnetic resonance study and were followed up for a median of 7.2 years. The LVGFI by cardiac magnetic resonance was defined by the ratio of stroke volume divided by LV total volume defined as the sum of mean LV cavity and myocardial volumes. Cox proportional hazard models were constructed to predict the end points of heart failure, hard CV events, and a combined end point of all CV events after adjustment for established risk factors, calcium score, and biomarkers. A total of 579 (11.6%) CV events were observed during the follow-up period. In adjusted models, the end points of heart failure, hard CV events, and all events were all significantly associated with LVGFI (heart failure, hazard ratio=0.64, P<0.0001; hard CV events, hazard ratio=0.79, P=0.007; all events, hazard ratio=0.79, P<0.0001). LVGFI had a significant independent predictive value in the multivariable models for all CV event categories. The LVGFI was a powerful predictor of incident HF, hard CV events, and a composite end point, including all events in this multiethnic cohort.

Figure 1. Schematic comparison of LVEF and LVGFI variations in different sub-clinical pathophysiological settings

As presented in this figure, LVEF is not sensitive to LV mass variations that are present in concentric hypertrophy (such as in the first stages of hypertensive cardiomyopathy) whereas LVGFI is significantly decreased. In the case of LV eccentric hypertrophy (advanced hypertensive cardiomyopathy, early stages of dilated cardiomyopathy), both functional indices are significantly decreased, but LVGFI is still more decreased than LVEF. Of note: LVGFI= [SV/(LVEDV+LVESV/2 +LVmass/1.05)]×100. Concentric hypertrophy is defined by increased LV mass and increased relative wall thickness (RWT). Eccentric hypertrophy is defined by increased LV mass with normal RWT. RWT is calculated as two times the posterior wall thickness divided by the LV end-diastolic diameter and is increased when this ratio is > 0.42. Increased LV mass is defined as an LV mass >115 g/m² in men and >95 g/m² in women.

Figure 2. Relationship between the Left Ventricular Stroke Volume, Mean LV Cavity Size and absolute LV Mass in the whole MESA population

As discussed in the introduction, SV can be predicted from LV mean cavity size as well as LV absolute mass. LV= left ventricle.

Figure 3. Correlation plots between LVEF and LVGFI in the whole MESA population

Linear (black) and non-linear (red) regression lines are shown. LVGFI=left ventricular global functional index; LVEF=left ventricular ejection fraction.

Figure 4. Nelson-Aalen Analyses for the Different Categories of Clinical Endpoints

Nelson-Aalen plots illustrating cumulative hazard by quartile of left ventricular functional index (LVGFI) at baseline for (A) heart failure events, (B) hard cardiovascular events, (C) all clinical events. The cumulative hazard was systematically significantly greater in the 1^st quartile compared to the other quartiles for each end-point (log-rank for difference, p<0.001). The quartile limits were determined in a subset of healthy MESA participants without any cardiovascular risk factors and then applied to the entire MESA population (1^st quartile<37 %; 2^nd quartile from 37 to 42%; 3^rd quartile from 42 to 47% and 4^th quartile >47%).

Figure 5. Receiver Operating Characteristic (ROC) curves of the predictive power of LVGFI, LVEF, LVMVR and LV mass index determined with a non-adjusted Cox-regression model on the presence/absence of the different clinical outcome categories

There were significant differences between the different areas under the curve (AUC) obtained with each LV parameter. The AUCs obtained with LVGFI and LV mass index on heart failure were significantly greater than those obtained with LVEF and LVMVR (p<0.0001). For hard cardiovascular events and combined all clinical events, the AUCs obtained with LVGFI, LVMVR and LV mass index were significantly greater than those obtained with LVEF (p<0.0001 respectively). LVGFI=left ventricular global functional index; LVEF=left ventricular ejection fraction; normalized LV mass=LV mass normalized by body surface area; LVMVR=left ventricular mass to end-diastolic volume ratio; CV=cardiovascular.

Figure 6. Receiver Operating Characteristic (ROC) curves of the predictive power of LVGFI, LVEF, LVMVR and LV mass index determined with a multivariate Cox-regression model on the presence/absence of the different clinical outcome categories

There were no significant differences between the different areas under the curve (AUC) obtained with each LV parameter (p=NS). The ROC curves were not computed if the predictive value of the LV parameter didn't reach significance level<0.05 in the full multivariable regression model. LVGFI=left ventricular global functional index; LVEF=left ventricular ejection fraction; normalized LV mass=LV mass normalized by body surface area; LVMVR=left ventricular mass to end-diastolic volume ratio.