Structural and Functional Phenotyping of the Failing Heart

Abstract

Diagnosis, prognosis, treatment, and development of new therapies for diseases or syndromes depend on a reliable means of identifying phenotypes associated with distinct predictive probabilities for these various objectives. Left ventricular ejection fraction (LVEF) provides the current basis for combined functional and structural phenotyping in heart failure by classifying patients as those with heart failure with reduced ejection fraction (HFrEF) and those with heart failure with preserved ejection fraction (HFpEF). Recently the utility of LVEF as the major phenotypic determinant of heart failure has been challenged based on its load dependency and measurement variability. We review the history of the development and adoption of LVEF as a critical measurement of LV function and structure and demonstrate that, in chronic heart failure, load dependency is not an important practical issue, and we provide hemodynamic and molecular biomarker evidence that LVEF is superior or equal to more unwieldy methods of identifying phenotypes of ventricular remodeling. We conclude that, because it reliably measures both left ventricular function and structure, LVEF remains the best current method of assessing pathologic remodeling in heart failure in both individual clinical and multicenter group settings. Because of the present and future importance of left ventricular phenotyping in heart failure, LVEF should be measured by using the most accurate technology and methodologic refinements available, and improved characterization methods should continue to be sought.

Heart failure remains a major health care problem, affecting 6.5 million adults in the United States (1). Although progress has been made in developing effective drug and device therapies, the pace of new development has slowed (2). The beneficial therapies that have been developed, encompassing 8 drug and 2 device classes, have been based on clinical trials using the major inclusion criterion of a reduced left ventricular ejection fraction (LVEF), typically ≤0.40, which defines the heart failure with reduced ejection fraction (HFrEF) phenotype. In contrast, entry criteria that have included a relatively preserved LVEF, so-called HF with preserved EF (HFpEF), have been uniformly unsuccessful. Thus LVEF is able to successfully identify heart failure therapeutic phenotypes.

Recently Konstam and Abboud (3) argued that the LVEF has “exhausted its usefulness as a presumed marker of contractility and a means of categorizing cardiomyopathies. In fact, the latter practice has stymied advances in pathophysiological understanding and therapeutics.” Their argument centers on the limitations of EF as a measurement of intrinsic ventricular contractile performance, primarily because of its load dependency. Based on this argument, could the LVEF-based method of classifying heart failure be limiting progress in the development of new therapeutic approaches, which ideally would be based on some other phenotypic classification? The answer to this question depends on the validity of the LVEF measurement for estimating fundamental ventricular myocardial pathophysiologic abnormalities underpinning heart failure, on its utility in identifying distinct heart failure phenotypes amenable to specific therapeutic approaches, and on the utility and availability of alternative methods of phenotyping the failing heart. We address these issues by reviewing historical work on LVEF, defining what is specifically measured by LVEF, and providing original data for the relationship between LVEF and other hemodynamic measurements as well as molecular changes in the failing human heart.

HISTORICAL EMERGENCE OF THE EJECTION FRACTION CONCEPT

The concept of ventricular EF was a byproduct of the development of methods to reliably measure stroke volume (SV), which if done angiographically requires accurate measurements of ventricular volumes (4). In their classic paper measuring left ventricular volumes of normally functioning mammalian hearts with 67-fold variation in heart weight (dogs to horses), Holt et al (4) noted that cardiac output, ventricular and SVs increased with body and heart weights, but the residual and ejected fractions of ventricular end diastolic volume remained constant from the smallest to the largest animals investigated. In other words, in the absence of pathophysiologic perturbation, the fraction of end diastolic volume ejected in systole is tightly regulated across mammalian species. The term ejection fraction was first used by Kennedy et al. (5) to describe the ejected component of ventricular volume, measured as: [SV/end-diastolic volume]. Sonnenblick (6) was the first to relate EF to sarcomere shortening, the basis for LVEF as a measurement of contractile function.

LVEF MEASURES THE 2 MAJOR CHARACTERISTICS OF PATHOLOGIC ECCENTRIC REMODELING

The most common ventricular myocardial disease process causing heart failure in patients <75 years age (7) is eccentric pathologic hypertrophy characterized by increased LV volumes and mass with no or little increase in wall thickness (8–11). Although heart failure–associated eccentric remodeling is usually described in anatomical terms (8–11), it also includes progressive contractile dysfunction (12–14) and gene expression changes associated with both hypertrophy and decreased contractility (15,16). Eccentric hypertrophy is associated with increased ventricular end diastolic and systolic volumes but also includes chamber geometric changes, with the LV transitioning from a prolate ellipse to a more spherical shape (15). These remodeling changes (Figure 1) can occur in both the right and left ventricles.

FIGURE 1. Remodeling of a Normal Heart Into an HFrEF Phenotype.

Remodeling of a normal heart into a HFrEF phenotype (LV or RV) characterized by contractile dysfunction and eccentric hypertrophy as detected by a reduction in EF. ARVC = arrhythmogenic right ventricular cardiomyopathy; COPD = chronic obstructive pulmonary disease; CRT = cardiac resynchronization therapy; EDV = end diastolic volume; ESV = end systolic volume; HFrEF = heart failure with reduced ejection fraction; HTN = hypertension; IDC = idiopathic dilated cardiomyopathy; LV = left ventricle; LVF = left ventricle failure; MI = myocardial infarction; PAH = pulmonary arterial hypertension; RV = right ventricle; VHD = valvular heart disease.

In the formulaic definition EF = SV/end diastolic volume (EDV), the numerator (SV) is a measurement of contractile function, whereas the denominator (EDV) estimates the degree of chamber dilation due to eccentric hypertrophy. Stroke volume is the result of [EDV – end-systolic volume (ESV)], which means that SV is related to the degree that ESV, an excellent measurement of intrinsic contractile function (17), is altered relative to EDV. Accordingly, the LVEF ratio combines elements of systolic function and eccentric hypertrophic remodeling in a single measurement.

DOES LVEF RELIABLY MEASURE CONTRACTILE FUNCTION?

In order to provide contemporary hemodynamic and molecular data for an examination of the hypothesis that LVEF reliably measures contractile function and eccentric hypertrophy, we report results from 2 longitudinal clinical studies in which hemodynamics and septal ventricular myocardial gene expression were measured before and after left ventricular reverse remodeling produced by β-blockade (Online Methods).

Figure 2A shows left ventricular volume and LVEF measurements in 32 nonischemic dilated cardiomyopathy (DCM) patients treated for 3 or 12 months with β-blocking agents (18,19), which, in two-thirds of the patients was associated with the reverse remodeling changes shown in Figure 1. The same relationships at baseline prior to the administration of β-blocking agents are given in Online Figure 1. As shown in Figure 2A, across a wide range of LVEFs (0.14 to 0.62; median: 0.43 expressed as absolute percentage in Figure 2), LVEF is inversely related to indexed EDV (EDVI) (r = −0.81; p < 0.0001) but is unrelated to SVI (r = 0.11). SVI is weakly (r = 0.51) related to EDVI (Figure 2A) and, as previously reported in coronary artery disease patients without heart failure (20), LVEF is highly inversely related to ESVI (r = −0.87; p < 0.0001) (Figure 2A). Therefore, over a wide range of degrees of pathologic eccentric remodeling: 1) SV remains relatively constant compared with varying values of LVEF; 2) SV increases slightly with increasing EDVI as LV dilation compensates for decreased contractile function (14,21); and 3) LVEF is inversely and closely related to ESVI. Therefore, in eccentric remodeling, LVEF and ESVI but not SVI reflect the decline in contractile function that accompanies HFrEF progression. Both LVEF and ESVI express the degree of eccentric remodeling, but LVEF does so as a unitless ratio that more directly relates to eccentric hypertrophy because EDVI, to which LVEF is closely inversely related, is measured in end diastole. Notably, because of its close relationship to EDVI (r = 0.98; p < 0.0001) (Online Figure 2), ESVI is also a relatively good estimate of eccentric hypertrophy.

FIGURE 2. LVEF, Ventricular Volume (EDVI and ESVI), and SVI Relationships.

(A) LVEF, ventricularvolume (EDVIand ESVI), and SVIrelationships are shownin 32nonischemicDCM patients treated for 3(n = 5)and12(n = 27)months with metoprolol, metoprolol plus doxazosin or carvedilol (18,19). EF and volume measurements were made by radionuclide SPECT imaging (18).

(B) Change from baseline (%) in patient and parameters shown in A. DCM = dilated cardiomyopathies; EDVI = end diastolic volume index; ESVI = end systolic volume index; LVEF = left ventricle ejection fraction; SPECT = single-photon emission computed tomography; SVI = stroke volume index.

The temporal behavior of LVEF and ventricular and SVs in HFrEF patients undergoing reverse eccentric remodeling in response to treatment with β-blocking agents is shown in Figure 2B, which gives changes from baseline for the patients shown in Figure 2A. The reverse remodeling effect of chronic (≥3 months) treatment with b blockade in ischemic or nonischemic causes of dilated cardiomyopathies (DCMs) is known to improve intrinsic systolic function, reduce left ventricular volumes and mass, and restore geometry to a more normal shape (15). The range of LVEF change shown in Figure 2B was from −0.09 to +0.47 (median: +0.16; expressed in Figure 2B as percent of change from baseline). The percent of change in LVEF is inversely related to changes in both EDVI (r = 0.68: p < 0.0001) and ESVI (r = −0.70; p < 0.0001) (Figure 2B), that is, respectively related to improvements in indices of eccentric hypertrophy and intrinsic systolic function. In the data shown in Figure 2B, SV index (SVI) is now statistically significant (p = 0.006) related to LVEF change, unlike in the cross-sectional data shown in Figure 2A, but is unrelated to EDVI (p = 0.33). These findings underscore the role of EDV/ventricular dilation to maintain SV in a relatively constant range (mean: SVI 33 ± 10 ml/m²) (Figure 2B), and demonstrate that SVI as well as LVEF are sensitive to the increase in intrinsic systolic function associated with β-blocker treatment.

The well-known major theoretical drawback of EF, as reiterated by Konstam and Abboud (1), is its load dependency. The fact that LVEF is sensitive to changes in preload and afterload has been known for more than 50 years and, as originally described, was based on changes in LVEF in response to large (25% to 33%) acute changes in loading conditions (22–24). However, in patients with LV systolic dysfunction and chronic heart failure, LVEF is typically changed little by the smaller, more chronic variations in loading conditions observed clinically (13,23,25,26). In patients with chronic heart failure and eccentric remodeling, afterload typically varies within a narrow range dictated by pump dysfunction and neurohormonal inhibitor treatment and is not typically appreciably changed during therapy. Preload, which has an attenuated effect in the chronically volume overloaded failing LV (23), is typically prevented from increasing substantially or becoming too low through adjustments of diuretic therapy and inhibitors of the renin-angiotensin system.

Although medium-sized (12,13) and large (14) cross-sectional studies of hemodynamic and LVEF measurements in heart failure patients support the lack of meaningful effects of changes in afterload and preload in stable patients with chronic heart failure, serial studies superimposed on changes in ventricular phenotype can offer additional evidence. Two previously published studies (18,27) conducted in nonischemic DCM patients treated with β-blockers are presented (Online Table S1). These studies indicate that during chronic therapy of HFrEF that includes β-blockade there is little or no change in loading conditions despite relatively large changes in LVEF.

In addition, when LVEF has been compared to more direct and/or load-independent contractility measurements (12), including end-systolic elastance (28), LVEF is as sensitive as these isovolumic measurements for the detection of pump dysfunction in patients with a history of heart failure or myocardial disease. Last, experienced clinicians know that LVEF needs to be interpreted “in the context of the pathology present” (29), with consideration of any major loading condition abnormalities or changes such as in heart failure from or complicated by valvular heart disease.

LVEF BY ITS EDV COMPONENT IS AN EXCELLENT MEASUREMENT OF ECCENTRIC HYPERTROPHY.

The eccentrically hypertrophied failing ventricle dilates as a compensatory means to maintain SV in the face of declining pump function, as the increase in chamber dimension allows for preservation of SV with less absolute myocyte shortening (21). Direct support for LV dilation being compensatory for decreased pump function can be observed from the reverse remodeling that results from cardiac resynchronization therapy (30). Biventricular or single chamber LV pacing immediately reverses depressed LV function (31), and chronic biventricular pacing subsequently reduces ESV, EDV, and LV mass and increases LVEF (32). For β-blocker–associated reverse remodeling, the therapy-induced increase in intrinsic systolic function is delayed for several weeks (15), and its relationship to reverse remodeling is not as obvious.

In advanced HFrEF, the eccentrically hypertrophied LV consists of cardiac myocytes that exhibit increased length relative to transverse diameter due to in-series sarcomere generation (11). In the failing, eccentrically hypertrophied ventricle sarcomere length is at close to maximum (11,32), and although the Frank-Starling mechanism is at least partially intact in the failing heart (33), it does not appear to have an important compensatory role in HFrEF (21). Rather, the basis for the compensatory effect of eccentric hypertrophy on LV systolic function is the volume displacement advantage of dilated ventricles; compared to normally sized ventricles a much smaller amount of circumferential shortening is associated with a larger SV albeit at expense of increased wall stress during myocyte shortening (21,34). Despite the compensatory effect of chronic LV dilation on SV, from a natural history standpoint eccentric hypertrophy in the failing heart has maladaptive features mediated through the increase in wall stress, neurohormonal activation and proarrhythmia mechanisms (15,16,21). Because these can be mitigated by medical therapy, it is critically important to have a reliable and easily accessible clinical measurement of eccentric remodeling.

The position of EDV in the denominator of the EF equation means that EDV and EF will be inversely related, and previously published radionuclide single-photon emission computed tomography (SPECT) imaging data (35) as well as the data presented in Figure 2A indicate that this relationship is linear. EDV measured by 3D echocardiography correlates highly (correlation coefficient: >0.90) with myocardial mass (36), indicating that EDV is an appropriate surrogate for eccentric hypertrophy. EDV is the dominant component of the LVEF calculation, appearing as a single value in the denominator and in the numerator as the value from which ESV is subtracted to yield SV. This gives rise to the close inverse relationship between LVEF and EDVI shown in Figures 2A and 2B, and is the basis for why LVEF is an excellent estimate of the degree of eccentric hypertrophy.

EXPERIENCE WITH LVEF AS A REMODELING INDEX

Because of the pathophysiologic importance of its defining components, widely available noninvasive imaging measurement methods and methodologic standardization, LVEF has been extensively used as a remodeling index. Based on these results, LVEF has been incorporated into the prevailing model of HFrEF progression (Central Illustration) (14) and into consensus documents on cardiac remodeling (Online Ref. 1). As shown conceptually (Central Illustration), HFrEF begins with an initial index event that damages or eliminates functioning myocardium, followed by activation of compensatory mechanisms that include eccentric hypertrophy and attempts to stabilize contractility and cardiac output through neurohormonal activation, which lead to further development of contractile dysfunction (15,16). These events are followed by eventual progression of and decline in LVEF as the compensatory mechanisms produce secondary damage and the LV undergoes progressive remodeling (Central Illustration) (15,16; Online Ref. 2).

CENTRAL ILLUSTRATION. Natural History of HFrEF Phenotype.

Natural history of a HFrEF phenotype as assessed by serial LVEF measurements. Following initial damage to or loss of sarcomere, compensatory mechanisms are activated to support myocardial function, but these compensatory mechanisms eventually lead to progression of remodeling and heart failure natural history (16). C.O. = cardiac output; ECM = extracellular matrix; EDV = end diastolic volume; HFrEF = heart failure with reduced ejection fraction; LVEF = left ventricle ejection fraction; RAAS = renin-angiotensin-aldosterone systems; SV = stroke volume.

REDUCED OR CHANGED LVEF REFLECTS FUNDAMENTAL ALTERATIONS IN THE BIOLOGY OF THE HUMAN VENTRICLE.

Consistent with its contractile dysfunction and chamber enlargement features, HFrEF ventricular myocardium is characterized by cognate directional changes in the expression of genes that regulate contractility and hypertrophy (Table 1, Online Table 2) (18,19,27; Online Refs. 3,4). Moreover, with reverse remodeling, changes in gene expression more closely align with LVEF than they do with changes in SVI, EDVI, ESVI, cardiac index, or PWP (Table 2). ESVI gives the closest agreement with LVEF for changes in gene expression that have been previously shown to be associated with eccentric remodeling (18,27). For ESVI, 9 of 20 genes were significantly correlated with gene expression changes, compared with 10 of 20 with LVEF. The number of gene expression changes correlating with EDVI, SVI, cardiac index, and PWP changes were 7, 1, 0, and 3 of 20, respectively. Not surprisingly, all 9 of the ESVI significantly correlated genes also correlated with LVEF changes. These data indicate that LVEF and ESVI are the parameters most directly related to a subset of gene expression changes involved in eccentric remodeling.

TABLE 1.

Septal Ventricular Myocardial mRNA Expression of Contractility or Hypertrophy-Regulating or Associated Candidate Genes

Abundance of Gene mRNA	Nonfailing (n = 4)	DCM/Failing (n = 47)	p Value*
Contractility
MYH6	0.73 ± 0.33	0.46 ± 0.65	0.037
MYH6/MYH7†	0.024 ± 0.015	0.014 ± 0.014	0.057
ATP2A2	0.69 ± 0.18	0.49 ± 0.34	0.037
PLN†	1.87 ± 0.22	1.60 ± 0.79	0.20
RYR2†	0.61 ± 0.20	0.66 ± 0.37	0.86
ASQ2‡	0.74 ± 0.20	1.12 ± 0.57	0.14
ADRB1	0.0092 ± 0.0030	0.0049 ± 0.0037	0.014
ADRB2†	0.0092 ± 0.0043	0.0097 ± 0.0052	0.86
ADCY5‡	0.17 ± 0.05	0.13 ± 0.07	0.058
ADRA1A†	0.022 ± 0.011	0.016 ± 0.010	0.13
#p <0.05/# total	3 of 10
Hypertrophy
ACTC1	7.21 ± 2.38	4.62 ± 1.92	0.020
ACTC1/A1	1.76 ± 1.01	0.77 ± 0.63	0.024
MYL2	11.1 ± 2.49	22.0 ± 15.4	0.047
MYL3†	5.33 ± 1.75	6.20 ± 6.50	0.48
MYL4‡	0.072 ± 0.033	0.12 ± 0.15	0.89
TNNT2‡	8.92 ± 4.54	7.82 ± 4.99	0.33
TNNI3	2.11 ± 0.87	4.19 ± 2.92	0.033
BNP	0.048 ± 0.087	0.74 ± 0.97	0.013
ANP	0.083 ± 0.076	3.51 ± 5.00	0.002
CTF1	0.0009 ± 0.0002	0.0017 ± 0.0008	0.029
#p < 0.05, Hty	7 of 10
Contractility + Hty	10 of 20

Septal ventricular myocardial mRNA expression of contractility or hypertrophy-regulating or -associated candidate genes (n = 50 in total dataset) expression (mRNA abundance is expressed as 2^−dCT referenced to GAPDH in Nonfailing vs. mean ± SD DCM/Failing human myocardium) (18). Genes listed are those with a p value of <0.05 between Nonfailing and DCM/Failing on baseline measurements prior to receiving β-blocking agents (18), those that increased or decreased after 3 (n = 7) or 12 (n = 40) months of treatment (18), or control (n = 2 each) contractility or hypertrophy-regulating genes not different at baseline or unchanged with β-blocker treatment. Hemodynamic data are given in Online Table 1.

^*Wilcoxon rank-sum test.

^†Increased or decreased in β-blocker responders vs. in nonresponders but not different in Nonfailing vs. IDC at baseline (18).

^‡Unchanged with β-blocker treatment, not different Nonfailing vs. IDC at baseline.

^#p value, Nonfailing vs. DCM/Failing.

DCM = dilated cardiomyopathies; GAPDH = glyceraldehyde 3-phosphate dehydrogenase; Hty = hypertrophy; IDC = idiopathic dilated cardiomyopathy.

TABLE 2.

Correlation of Changes in mRNA Abundance/GAPDH Abundance Vs. Indices of Remodeling and Myocardial Performance for Contractility and Hypertrophy-Regulating Genes

Δ in Abundance of Gene mRNA	Spearman Correlation Coefficient (rho) and p Values of Cardiac Parameters Shown:*
	LVEF		LV SVI (ml)		LV ESVI (ml/M2)		LV EDVI (ml/m2)		C.I.(l/min/m2)		PWP (mm Hg)
	rho	p Value	rho	p Value	rho	p Value	rho	p Value	rho	p Value	rho	p Value
Contractility
MYH6	0.55	0.003	0.35	0.061	-0.48	0.008	-0.41	0.028	0.30	0.11	−0.23	0.23
MYH6/MYH7	0.61	0.0006	0.17	0.38	−0.53	0.004	−0.49	0.009	0.32	0.094	−0.20	0.29
ATP2A2	0.54	0.003	0.45	0.016	−0.47	0.012	−0.30	0.11	0.47	0.014	−0.18	0.35
PLN	0.22	0.25	0.15	0.44	−0.17	0.39	−0.09	0.66	0.17	0.39	−0.04	0.82
RYR2	0.33	0.08	0.27	0.16	−0.27	0.17	−0.20	0.32	0.25	0.18	−0.17	0.36
CASQ2	−0.46	0.015	0.10	0.61	0.40	0.034	0.42	0.026	−0.13	0.50	−0.07	0.70
ADRB1	0.34	0.081	0.27	0.16	−0.21	0.28	−0.13	0.51	0.27	0.16	−0.21	0.27
ADCY5	−0.13	0.51	0.14	0.49	0.11	0.57	0.19	0.34	−0.04	0.84	0.06	0.76
ADRA1A	0.38	0.043	0.25	0.19	−0.31	0.10	−0.26	0.17	0.05	0.78	−0.17	0.36
†Contractility genes p < 0.05/total contractility genes	5/10		1/10		4/10		3/10		1/10		0/10
Hypertrophy
ACTC1	−0.24	0.22	−0.02	0.90	0.06	0.75	0.27	0.16	−0.11	0.57	0.30	0.11
ACTC1/A1	0.56	0.002	0.07	0.72	−0.57	0.002	−0.61	0.0006	−0.04	0.85	0.37	0.046
MYL2	−0.61	0.0008	0.08	0.67	0.61	0.0007	0.66	0.0002	−0.02	0.93	0.50	0.006
MYL3	0.31	0.12	−0.02	0.93	−0.36	0.069	−0.34	0.081	0.01	0.97	−0.05	0.78
MYL4	−0.01	0.95	−0.05	0.80	−0.03	0.90	0.02	0.92	0.16	0.42	−0.07	0.73
TNNT2	0.13	0.52	0.26	0.18	−0.17	0.38	−0.10	0.61	0.04	0.82	0.14	0.46
TNNI3	−0.44	0.021	0.04	0.84	0.44	0.021	0.49	0.010	−0.11	0.57	0.35	0.064
NPPB	−0.49	0.008	−0.13	0.50	0.45	0.015	0.45	0.016	−0.05	0.78	0.27	0.14
NPPA	−0.52	0.005	−0.48	0.009	0.46	0.014	0.30	0.12	−0.34	0.074	0.42	0.022
CTF1	−0.05	0.81	0.10	0.60	−0.19	0.35	−0.16	0.42	0.20	0.30	−0.05	0.78
‡Hty genes p < 0.05/total Hty genes	5/10		1/10		5/10		4/10		0/10		3/0
‡Hty genes p < 0.05/total Hty genes	10/20		2/20		9/20		7/20		1/20		3/0

Data show correlation (by Wilcoxon rank sum test) of changes in mRNA abundance/GAPDH abundance (fold change using the 2^−ddCT method) vs. indices of remodeling and myocardial performance for contractility and hypertrophy-regulating genes in DCM patients who had baseline (n = 33) and follow-up (n = 32) studies at 3 (n = 32) or 12 (n = 27) months that included LV volume measurements by radionuclide SPECT imaging after receiving beta-blocking agents (Online Ref. 18). Listed genes were either p < 0.05 IDC vs. nonfailing controls (Table 1) or p <0.05 in reverse remodeling responders vs. nonresponders (Online Ref. 18) or were control mRNAs (CASQ2, ADCY5, MYL4, TNNT2) that were neither.

^*Values in bold indicate p < 0.05.

C.I. = cardiac index; DCM = dilated cardiomyopathy; EDVI = end diastolic volume index; ESVI = end systolic volume index; Hty = hypertrophy; IDC = idopathic dilated cardiomyopathy; LV = left ventricle; LVEF = left ventricle ejection fraction; PWP = pulmonary wedge pressure; SPECT = single-photon emission computed tomography; SVI = stroke volume index.

LVEF IS STRONGLY INVERSELY ASSOCIATED WITH MORTALITY ACROSS ALL CAUSES OF PATHOLOGIC ECCENTRIC REMODELING.

Not only is LVEF related to the molecular changes that occur in failing human ventricular myocardium, but more importantly, it is associated with mortality and other adverse outcomes in multiple types of heart disease or syndromes including heart failure (Online Refs. 5,6), coronary artery disease (Online Refs. 7,8), valvular heart disease (Online Ref. 9), and ventricular arrhythmias (Online Ref. 10). There is no other single, widely available cardiovascular test that can make this claim.

LVEF AS A PREDICTOR OF CLINICAL OUTCOMES IN HF REGARDLESS OF TREATMENT.

LVEF as a single measurement has long been known to predict mortality and other major clinical outcomes in heart failure. Evidence that lower LVEF values are associated with increased mortality first emerged in the 1970s (Online Refs. 5,11), and subsequent investigations (Online Refs. 6,12–14) have continued to report the relationship between reduced LVEF and mortality as well as other major adverse outcomes. In addition, serial change in LVEF is one of the most statistically powerful predictors of outcomes in HFrEF (Online Refs. 12,14). For example, in the Beta-Blocker Evaluation of Survival Trial (BEST), in a comparison of data measured at baseline and during follow-up, an absolute 5% change in LVEF was the most significant dynamic predictor of all-cause mortality or heart failure hospitalization (Online Ref. 14).

LVEF is also a strong predictor of outcomes in HFpEF, as in the Irbesartan in Heart Failure with Preserved Systolic Function (I-PRESERVE) trial, a HFpEF risk calculator identified lower LVEFs above the inclusion criterion of ≥0.45 as one of the strongest independent predictors of major outcomes including mortality (Online Ref. 15). In the Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist Trial (TOPCAT), which had the same LVEF entry criterion, lower LVEFs predicted major outcomes as well as a favorable treatment effect of spironolactone (Online Ref. 16). In addition, in a longitudinal population study from Olmsted County, Minnesota, survival was predicted by serial change in LVEF in both HFrEF and HFpEF patients (Online Ref. 17).

LVEF AS A PREDICTOR OF RESPONSE TO MEDICAL THERAPY.

The use of LVEF to identify patients with HFrEF has resulted in the development of numerous drugs and devices to improve clinical outcomes. An extensive portfolio of effective heart failure therapies has been developed, approved, and established in clinical practice based on the entry criterion of a reduced LVEF (≤0.40). In comparison these same therapies have exhibited no efficacy in heart failure patients with HFpEF, typically defined as LVEFs ≥0.45. The indisputable success of LVEF as a reliable method for identifying heart failure patients for clinical trials has led to the universal adoption of the original proposal of dividing heart failure into the two major categories HFrEF and HFpEF (Online Refs. 18–20). When a major discipline of medicine uses a clinical test to phenotypically divide its patient population into major subgroups that predict a wide range of therapeutic response, this should be viewed as an unambiguous validation of that test.

Based on this success an attempt is underway to further delineate heart failure phenotypes by LVEF, through the creation of a third, intermediate/borderline category (Online Ref. 19), HF with mid-range EF (HFmrEF) (Online Ref. 20) or, alternatively, “mildly reduced” LVEF (defined by the range of 0.41 to 0.49). Preliminary data suggest that the validity of this modification of the LVEF-based heart failure classification is supported by gene expression differences between HFrEF and HFmrEF (Online Ref. 21). However, in order to further characterize the HFmrEF patient population and determine whether it is associated with a distinct therapeutic phenotype, precise LVEF measurements will be required in this relatively narrow range, using the best available technologies and methodologies.

ADVANTAGES OF LVEF AS AN IMAGING MEASUREMENT

Because of its value in diagnosis, prognosis and prediction of response to medical therapy, the cardiac imaging market has leveraged the development of multiple highly reliable methods of LVEF measurement that are available in medical facilities worldwide. In contrast, absolute measurements of ventricular volume are highly variable and affected by endocardial characteristics, body size and cardiac disease states. Thus, serial changes of ventricular volumes in the same subject are useful, but group comparisons are limited by these phenotypic differences. Indexing to body size is helpful, but even with this correction there is significant variability between subjects. As a result, the use of ESVI as an index of eccentric remodeling in multicenter trials must be carefully managed by an experienced core laboratory (Online Ref. 22). Ejection fraction measurement avoids most of these pitfalls, because the assessment is a unitless ratio that does not depend on the precision of tedious measurements of volume and mass that are based on geometric assumptions. In addition, LVEF is constant across all sizes of humans and large and small animals (4), so it does not have to be indexed to body size. Given these issues, it is not surprising that, across multiple modalities of cardiac imaging, LVEF is a strong predictor of cardiovascular outcomes (Online Refs. 6,12–17) and that it has endured many assaults advocating for more technically advanced endpoints, from adrenergic innervation (Online Ref. 23) to myocardial metabolism assessments (Online Ref. 24).

METHODS FOR OPTIMIZING MEASUREMENTS OF LVEF, AND FUTURE DIRECTIONS FOR VENTRICULAR CHAMBER PHENOTYPING IN HEART FAILURE

Determination of LVEF by gated radionuclide ventriculography (RNV) made a major contribution and received clinical acceptance because EF is count-based and does not require LV volume determination (Online Ref. 25). The measurement of LVEF using radionuclide counts is noninvasive and reproducible and became a powerful means to stratify risk and survival after myocardial infarction (Online Ref. 8) and other cardiac diseases (Online Refs. 9,14,26). Accordingly, LVEF became a principal selection criterion for pharmacological and device therapy in heart failure clinical trials. Although RNV LVEF measurements are less often used in contemporary clinical settings, advances in computer computational speed has made calculation of LVEF universally available using planar multigated acquisition, SPECT, perfusion imaging, and tomographic RNV methods. This has led to improved accuracy of ventricular volume measurements including the ability to separate LV and RV measurements, and 3D measurements of LVEF by SPECT imaging are highly correlated with those obtained by cardiac magnetic resonance (CMR) (Online Ref. 27).

Determination of LVEF by CMR is currently widely considered the clinical gold standard, particularly in heart failure patients, in whom geometry may be altered (Online Ref. 28). The superiority of image resolution and the resultant definition of myocardial structures with CMR improve accuracy and reproducibility of LV volume determination by this technique. Image processing still requires operator interaction for selecting the blood cavity–myocardial wall interface, but the field is moving toward 3D acquisition techniques and truly automated segmentation of cardiac volumes. However, cost and incompatibility with cardiac devices, although the latter is being addressed through the development of “magnetic resonance imaging [MRI]-conditional” pacemaker leads and other hardware, remain issues with CMR.

In a clinical practical sense, LVEF by echocardiographic biplane imaging continues to be the most widely used imaging modality, due to availability, lower cost, shorter scan times, lack of concern about interference from intracardiac hardware, and easier adoption in multicenter trials (Online Ref. 29). Contrast injection can greatly improve LVEF determination in the subset of patients with suboptimal echocardiographic windows. Increasing use of 3D echocardiographic imaging is an advance over biplane imaging and has been shown to improve temporal reproducibility (Online Ref. 30), and with adequate image quality LVEF results are comparable to those of CMR (Online Ref. 31). The challenge in 3D ultrasonographic acquisition is in subjects with large cardiac volumes in which inclusion of the entire LV is difficult with current 3D probes. This, coupled with subjects with limited echocardiographic windows, results in 10% to 15% of studies that are not able to be analyzed. Continued technological advances in 3D echocardiographic imaging, such as smaller probes and more high-density phase array transducers, adds promise to this approach that may emerge as the most likely practical clinical methodology for LVEF determination.

The major unmet need in heart failure ventricular phenotyping is in the categories of HFpEF and HFmrEF. It is clear that an LVEF within normal limits (≥0.50) does not preclude systolic dysfunction (Online Ref. 32) that is likely associated with adverse outcomes (Online Refs. 15,16) or lack of treatment effects (Online Ref. 16). For HFmrEF, in where one of the issues that needs resolution is whether the phenotype is distinct enough to require its own precision therapy, additional phenotyping such as strain measurements (Online Ref. 32) may be helpful. Thus, LV phenotyping in HFpEF or HFmrEF is likely fertile ground for the development of new imaging approaches based on current and emerging technologies.

CONCLUSIONS

We have presented evidence against the contention that LVEF is “misunderstood and overrated” (3), and offer a counterargument in favor of LVEF as the most important ventricular phenotypic measurement yet devised. It remains to be seen if, under disease conditions affecting ventricular myocardium, LVEF can be surpassed for its practical usefulness in establishing risk, predicting outcomes and measuring effects of medical treatment on remodeling. However, improved phenotype characterization methods are needed and should be vigorously sought, particularly within the LVEF heart failure categories of HFpEF and HFmrREF.

ABBREVIATIONS AND ACRONYMS

CMR

cardiac magnetic resonance

EDVI

end diastolic volume index

ESVI

end systolic volume index

HFmrEF

heart failure with mid-range left ventricular ejection fraction

HFpEF

heart failure with preserved left ventricular ejection fraction

HFrEF

heart failure with reduced left ventricular ejection fraction

LVEF

left ventricular ejection fraction

PWP

pulmonary wedge pressure (mean)

RNV

radionuclide ventriculography

SVI

stroke volume index (in reference to body surface area)