The Potential of Clinical Phenotyping of Heart Failure With Imaging Biomarkers for Guiding Therapies

Abstract

The need for noninvasive assessment of cardiac volumes and ejection fraction (EF) ushered in the use of cardiac imaging techniques in heart failure (HF) trials that investigated the roles of pharmacological and device-based therapies. However, in contrast to HF with reduced EF (HFrEF), modern HF pharmacotherapy has not improved outcomes in HF with preserved EF (HFpEF), largely attributed to patient heterogeneity and incomplete understanding of pathophysiological insights underlying the clinical presentations of HFpEF. Modern cardiac imaging methods offer insights into many sets of changes in cardiac tissue structure and function that can precisely link cause with cardiac remodeling at organ and tissue levels to clinical presentations in HF. This has inspired investigators to seek a more comprehensive understanding of HF presentations using imaging techniques. This article summarizes the available evidence regarding the role of cardiac imaging in HF. Furthermore, we discuss the value of cardiac imaging techniques in identifying HF patient subtypes who share similar causes and mechanistic pathways that can be targeted using specific HF therapies.

Cardiac imaging in heart failure (HF) plays 3 specific roles: imaging identifies HF phenotype, assesses severity of systolic and diastolic dysfunction, and monitors responses to interventions (1–5). To facilitate early recognition of HF, the American College of Cardiology/American Heart Association (ACC/AHA) practice guidelines for HF have divided the disorder into 4 stages, 2 of which (stages A and B) are in asymptomatic individuals (4). Cardiac imaging techniques play a pivotal role in defining abnormal myocardial structure and function that appear at and beyond stage B. A phenotypic characterization of HF into heart failure with reduced ejection fraction (HFrEF) or preserved ejection fraction (HFpEF) has also become commonplace. However, debate exists as to whether these 2 are really 2 separate entities or merely represent 2 different stages in the overall continuum of HF.

To address the limitations in existing clinical HF classifications, investigators have recently used cardiac imaging techniques for emphasizing integration of imaging biomarkers, using novel informatics platforms to define unique HF patient clusters (6,7). Recognition of such cardiac imaging-related HF phenotype, underpinned by omics strategies, might finally have utility in identifying patient subtypes who may benefit from individualized HF therapies. This focused update reviews newer developments in cardiac imaging strategies, imaging biomarkers and their integration with traditional metrics of cardiac structure and function assessment for understanding the phenotypic presentation, establishing the specific causes, and guiding the HF therapy. As a focused update, we specifically address new steps in integrating the HF phenotypic presentation with ACC/ AHA HF stages, assessment of causes of infiltrative disorders with specific examples like cardiac sarcoidosis and amyloidosis, drug and therapeutic interventions, and the novel use of miniaturized pocket ultrasound devices for guiding clinical HF therapies.

CARDIAC CHAMBER QUANTIFICATION

Cardiac imaging has played a pivotal role in defining ventricular remodeling for patient selection and for therapeutic endpoints in HF. Ventricular remodeling has been measured traditionally as a change in ventricular volume, left ventricular (LV) mass and EF and has been correlated with pathological changes such as myocyte hypertrophy, myocyte apoptosis, myofibroblast proliferation, and interstitial fibrosis (8,9). All 4 cardiac imaging modalities used for tracking this process, echocardiography, radionuclide ventriculography, computed tomography, and cardiac magnetic resonance (CMR), have been well standardized. Although CMR is considered a reference standard, echocardiography remains the predominant imaging technique used in clinical trials, based on broader availability and cost efficacy. However, due to the quantitative and reproducible nature of their EF results, equilibrium radionuclide angiography, and CMR techniques have been used in large clinical trials to identify systolic dysfunction. The simultaneous assessment of LV systolic function as well as stress and rest perfusion by gated single-photon emission computed tomography (SPECT) imaging can provide a range of information relevant to the care and clinical decision making for patients with HF, including the state of LV function, the probability of coronary artery disease (CAD) as the cause of HF, and the presence and extent of viability and ischemia. Similarly, the use of contrast-enhanced CMR for the assessment of late-gadolinium enhancement (LGE) has emerged as an established marker of myocardial scarring associated with cardiac remodeling.

Multiparametric approaches by combining the changes in LV volumes, mass, and EF can be helpful in characterizing different patterns of concentric and eccentric LV remodeling and have been used as a prognostic marker for predicting cardiovascular death, myocardial infarction, HF, stroke, or cardiac arrest. Moreover, studies have shown favorable effects of drugs on these variables that improve clinical outcomes in patients with decreased LVEF. Although the changes in LV volumes, mass, and EF can be measured accurately, changes in myocardial contractile behavior allows superior characterization of patterns of remodeling, leading to alternative and an intuitive approach to HF phenotyping. This has led to enthusiasm in using myocardial deformation parameters, such as strain, for understanding the mechanistic diversities underlying the development of myocardial dysfunction at the subclinical level and its subsequent progression to clinical HF (10–12). The Central Illustration shows a classification of HF stages where the observations from cardiac imaging have been combined with ACC/AHA classification of HF. In this sense, LV systolic dysfunction, a prerequisite for HFrEF, and LV hypertrophy, LV myocardial fibrosis, and abnormalities of LV diastolic dysfunction occurring as a part of the pathophysiology of HFpEF, can exist subclinically (stages A and B). Further progression of either systolic and/or diastolic dysfunction and the additive effect of several other cardiac and extracardiac comorbidities can lead to overt symptoms, with development of HFpEF or HFrEF corresponding to stages C or D in the ACC/AHA classification of HF later in the course of the disease. Integration of observations from cardiac imaging techniques with clinical classification schemes would allow future clinical trials to have precision in phenotypic characterization of HF patients. Figures 1 and 2 list the echocardiographic and CMR indices that have been used in clinical trials for characterizing the extent of structural and functional changes associated with HF. Table 1 lists nuclear imaging approaches and potential molecular targets in HF.

CENTRAL ILLUSTRATION. Cardiac Imaging for Selecting Patients With Subclinical or Overt Heart Failure.

ACC = American College of Cardiology; AHA = American Heart Association; DM = diabetes mellitus; EF = ejection fraction; GCS = global circumferential strain; GLS = global longitudinal strain; GRS = global radial strain; HFpEF = heart failure with preserved ejection fraction; HFrEF = heart failure with reduced ejection fraction; HTN = hypertension; LV = left ventricular.

FIGURE 1. Echocardiographic Biomarkers for Assessing Heart Failure.

EF = ejection fraction; LA = left atrial; LV = left ventricular; MR = mitral regurgitation; PASP = pulmonary artery systolic pressure. Reproduced with permission from Omar et al. (11).

FIGURE 2. CMR Biomarkers for Assessing Heart Failure.

CMR = cardiac magnetic resonance; LGE = late gadolinium enhancement; other abbreviations in Figure 1.

TABLE 1.

Nuclear Imaging Approaches and Potential Molecular Targets in HF

Myocardial Functional and Structural Target	Imaging Techniques and Agents
I. Myocyte compartment
a. Perfusion	PET: [15O]water, [13N]ammonia, 82rubidium, [18F]flurpiridaz
	SPECT: 201thallium, 99mTc-sestamibi, 99mTc-tetrofosmin
b. Metabolism
i. Substrate use
1. Glucose	2-Deoxy-2-[18F]fluoro-D-glucose
2. Fatty acid Metabolism	[11C]palmitate
3. Ischemic memory	[123I]-β-methyl[p-123I]iodophenyl-pentadecanoic acid
ii. Oxidative metabolism	[11C]acetate
c. Function	Gated SPECT or PET
d. Cell death
i. Necrosis	99mTc Pyrophosphate
ii. Apoptosis	99mTc-annexin V
II. Non-myocyte compartment
a. Renin angiotensin system
i. Angiotensin-converting enzyme	18F-labeled captopril, [18F]fluorobenzoyl-labeled lisinopril, 99mTc-benzoyl-labeled lisinopril
ii. Angiotensin II type 1 receptor	[11C]KR31173, 99mTc-labeled losartan, 18F-labeled losartan
b. Extracellular
i. Amyloid fibrils	99mTc-pyrophosphate, 99mTc-methylene Diphosphonate, 99mTc 3,3,-diphosphono-1,2-propanodicarboxylic acid
ii. Inflammatory sarcoidosis	2-deoxy-2-[18F]fluoro-D-glucose, 67gallium
c. Cardiac innervation	[123I]meta-iodobenzylguanidine, [11C] hydroxyephedrine
d. Cardiac Microvasculature	PET absolute hyperemic blood flow

Radiotracers in bold are approved by the U.S. Food and Drug Administration and are available commercially.

PET = positron emission tomography; SPECT = single-photon emission tomography.

ESTABLISHING CAUSE: ISCHEMIC VERSUS NONISCHEMIC

Identifying the exact cause of HF is important as this allows clinical trials to link patient outcomes directly to the underlying cause of HF. An important diagnostic consideration in patients with reduced LV systolic dysfunction (HFrEF) involves distinguishing those whose cardiomyopathy may be primarily due to CAD (many of whom have potentially reversible LV dysfunction) from those with idiopathic dilated cardiomyopathy. Among cardiomyopathies with systolic dysfunction (HFrEF), they are generally grouped as ischemic or nonischemic cardiomyopathies. Nonischemic cardiomyopathy can be further classified as dilated, hypertrophic, restrictive, valvular, and arrhythmogenic cardiomyopathies. Cardiomyopathies associated with specific cardiac or systemic disorders, such as hypertensive cardiomyopathy or cardiac sarcoidosis, are categorized as “specific” cardiomyopathies, and all other cardiomyopathies that do not fit into any single group are termed “unclassified.”

CORONARY ARTERY DISEASE

Clinical trials often have used imaging as a screening strategy for identifying coronary artery disease as a cause for HF. Direct visualization of the coronary arteries either with invasive coronary angiography or noninvasively with CT angiography allows early and direct distinction between ischemic and nonischemic cardiomyopathies, including anomalies. Noninvasive imaging techniques, such as echocardiography, nuclear imaging, and CMR provide information about the consequences of CAD on the myocardium (inducible ischemia, such as decreased perfusion or contraction) or myocardial fibrosis/scarring pattern (transmural extent, mural distribution, and pattern) that could lead to determining the correct underlying cause and diagnosis of the cardiomyopathy. CMR with its high resolution has more often been used to detect fibrosis (13), a technique that, in observational studies, has identified ischemic versus nonischemic cardiomyopathy in HF patients and linked fibrosis with clinical outcome (14,15). Given that CAD is the most common cause of HF in developed countries (16), noninvasive assessment of myocardial ischemia and viability in HFrEF would identify the subgroup of patients who have a potentially reversible degree of LV dysfunction and may benefit from revascularization (17). Causes of LV systolic dysfunction in ischemic cardiomyopathy include transmural scar, nontransmural scar, repetitive myocardial stunning, hibernating myocardium, and remodeled myocardium. The LV remodeling process, which is activated by the reninangiotensin system, stimulates toxic catecholamine actions and matrix metalloproteinases (18) resulting in maladaptive cellular and molecular alterations (19) with a final pathway to interstitial fibrosis (20,21). These responses to LV dysfunction and interstitial fibrosis lead to progressive worsening of LV function. Clinical trials attempting to target interventions that improve dysfunctional but viable myocardium may significantly affect global LVEF, LV remodeling, and survival. Identification of CAD in patients with HF also has implications in secondary prevention strategies, as recurrent myocardial infarction is a common mechanism of death in patients with HF.

ASSESSMENT FOR MYOCARDIAL PERFUSION AND VIABILITY

Although many patients with nonischemic cardiomyopathy may have perfusion abnormalities detected in myocardial perfusion imaging studies, a normal result in stress myocardial perfusion study in a patient with HFrEF is highly predictive for the absence of CAD (19). The modest specificity of myocardial perfusion imaging to rule out CAD is explained, in part, by pathologic and CMR studies (13) demonstrating that patients with nonischemic cardiomyopathy may have patchy or larger confluent territories of fibrosis or scarring, manifested as fixed defects on myocardial perfusion studies. For instance, abnormalities in myocardial perfusion in patients with hypertrophic cardiomyopathy (despite normal epicardial coronary arteries) can be attributed to microvascular disease, abnormalities in arterioles, or demand/supply mismatch (22,23). Several positron emission tomography (PET) studies have demonstrated abnormal coronary flow reserve in dilated and hypertrophic cardiomyopathy patients, with the subset of patients exhibiting blunted flow reserve having a more unfavorable natural history (24,25). Such data support the relevance of the perfusion abnormalities rather than simply classifying them as false-positive if epicardial CAD is not present (26). On the other hand, extensive perfusion abnormalities in the setting of LV dysfunction are essentially always associated with CAD rather than with dilated cardiomyopathy, especially when the perfusion defects are segmental.

Requirements for myocellular viability include sufficient myocardial blood flow, sarcolemma membrane integrity, intact mitochondria and ionotropic reserve, and preserved glucose or fatty acid metabolism (17). Because sarcolemma or mitochondria membrane integrity is dependent on preserved intracellular metabolic activity to generate high-energy phosphates, radiotracers that reflect cation flux (e.g., ²⁰¹thallium) (27), electrochemical gradients (technetium 99m [Tc^99m]-labeled sestamibi or tetrofosmin) (28), or metabolic processes with ¹⁸F-labeled fluorodeoxyglucose ([¹⁸F]FDG) (29) provide insight into myocardial viability. There are several different protocols used clinically for the detection of myocardial viability. The first and foremost protocol is used to determine the presence and extent of viable, ischemic myocardium, particularly in dysfunctional vascular territories. In the absence of myocardial ischemia, 2 nuclear imaging protocols are optimized for detection of viable, hibernating myocardium: the first is resting-redistribution thallium SPECT, and the second is resting rubidium-82- or ¹³N-labeled ammonia perfusion and [¹⁸F]FDG metabolism with PET. It has been shown that even at a similar mass of viable myocardial tissue (reflected by resting myocardial perfusion studies), the presence of inducible ischemia (reversible defect) is associated with an increased likelihood of functional recovery (30). For a dysfunctional myocardial region or vascular territory, the probability of functional recovery after revascularization is related to the magnitude of radiotracer uptake, which is a continuous variable; that is, the magnitude of tracer uptake directly reflects the magnitude of preserved tissue viability (31). Conversely, LGE CMR provides the magnitude of interstitial fibrosis and myocardial scar (extent of transmural and/or non-transmural infarction) and the probability of functional recovery after revascularization. Thus, a dysfunctional territory that does not exhibit LGE or has preserved radiotracer uptake has a high likelihood of improved function after revascularization. In contrast, a territory with evidence of LGE or a severe reduction in radiotracer uptake would represent predominant infarction, and the likelihood of improved function after revascularization would be low. The magnitude of potential improvement of global LV function after revascularization is, in turn, determined by the extent of viable dysfunctional myocardial regions or vascular territories. Novel functional strategies using assessment of myocardial deformation with speckle tracking echocardiography can be used readily at the bedside. Several studies have suggested that assessment of peak systolic longitudinal strain after revascularization in patients with acute myocardial infarction can be used for predicting global improvement in LVEF (32).

Guidelines suggest that patients with HF and active angina benefit in terms of natural history from revascularization and thus should be referred directly for angiography (1–5). For patients with HF and no angina, studies suggest that hibernation and stress-induced ischemia may be common in such patients (33). Hence, a search for underlying ischemia and viability would be an appropriate clinical strategy at some point in their evaluation. If substantial ischemia or viability of dysfunctional territories is found in the setting of narrowed coronary arteries that are technically amenable to revascularization, studies suggest a clinical benefit from revascularization (34). In the absence of substantial ischemia or viability, such a benefit is less likely. The imaging data can be used in decision making to help balance the risks and benefits of revascularization in a patient with HFrEF by supplying information on potential benefit of a revascularization strategy.

Clinical studies of the use of stress myocardial perfusion imaging have focused on the utility of ischemia as a marker of downstream improvement in LV function in HFrEF. The CHRISTMAS (Carvedilol Hibernation Reversible ischemia Trial: Marker of Success) trial was a double-blind, randomized trial of carvedilol versus placebo in 305 HFrEF patients, where both nuclear and echocardiographic imaging techniques were used to identify hibernation or myocardial ischemia (33). There was a gradient relationship between the extent of hibernating and/or the ischemic segments and improvement in LV function with carvedilol treatment, with no significant increase in LV function in the absence of viability. In contrast, patients in the placebo arm showed an increase in the number of nonviable segments over time, including segments that were previously deemed to be hibernating.

POTENTIAL BENEFIT OF REVASCULARIZATION

A meta-analysis of the prognostic value of viability testing and the impact of therapeutic choice on survival demonstrated a significant reduction in the risk of cardiac death during long-term follow-up in patients with evidence of preserved myocardial viability that underwent revascularization (34). Revascularization conferred no natural history advantage in patients without substantial myocardial viability. Similar findings were observed with ischemic cardiomyopathy patients who underwent LGE CMR. Patients with evidence of preserved myocardial viability by CMR who underwent revascularization had better outcome than those who were treated with medical therapy alone (35). These data suggest that noninvasive imaging of viability and ischemia can potentially play a role in selecting patients for revascularization, with the expectation of improving symptoms and natural history. However, given the observational and/or retrospective nature of the studies, a certain evaluation bias may apply in such meta-analysis. It is likely that the decision-making process to perform coronary artery bypass grafting (CABG) and/or percutaneous intervention was driven by the results of viability imaging studies.

The STICH (Surgical Treatment for Ischemic Heart Failure) trial enrolled 1,212 ischemic HF patients who were randomly assigned to receive medical therapy alone or medical therapy plus CABG. Although there were no significant differences between medical therapy alone and medical therapy plus CABG with respect to the primary endpoint of death of any cause during the follow-up period (median 56 months), patients assigned to undergo CABG had lower rates of cardiovascular death or hospitalization for cardiovascular causes (36). However, a recent follow-up publication of the same cohort of patients with ischemic cardiomyopathy nearly 10 years later showed that the rates of death from any cause, death from cardiovascular causes, and death from any cause or hospitalization for cardiovascular causes were significantly lower over 10 years among patients who underwent CABG in addition to receiving medical therapy than among those who received medical therapy alone (37).

The viability substudy of STICH, which consisted of 601 patients, examined whether the presence of viable myocardium as determined by the local participating site (myocardial perfusion SPECT, dobutamine echocardiography, or both) and restoration of coronary flow improved patient outcome (38). Notably, [¹⁸F]FDG PET and delayed enhanced CMR viability techniques were not studied in STICH. At first, it appeared that the presence of viable myocardium was associated with a greater likelihood of survival in patients with CAD and LV dysfunction, yet the improved survival did not hold after adjustments were made for other baseline clinical variables. However, there are several limitations to the STICH viability substudy. Study limitations included: 1) nonrandomized nature of the viability substudy; 2) clinicians were not blinded to the results of viability imaging, which might have excluded those with extensive myocardial ischemia or viability/ hibernating myocardium on the basis of imaging (selection bias); 3) patients had fewer comorbid conditions and less prior CABG than the prior observational studies; 4) 3-vessel CAD was present in only about one-third of the patients; 5) outcomes were not analyzed with viability being considered as a continuous variable; and 6) more than >85% of patients enrolled were from outside the United States. This was due to the very difficult and slow patient accrual rate in the United States over the initial 3-year period (the clinical sites were subsequently expanded to international countries). Moreover, there were a substantial number of patients who crossed over from medical therapy to CABG due to instability (angina symptoms, cardiac decompensation) in the first year after enrollment. In a post-hoc subanalysis, it was appreciated that these patients with ischemic cardiomyopathy who crossed over to CABG had a far better outcome than those who received medical therapy alone. Although an ongoing trial, AIMI-HF (Alternative Imaging Modalities in Ischemic Heart Failure), aims to randomize patients with ischemic HF to undergo SPECT, PET, or CMR to assess myocardial ischemia and viability, there is the potential for a significant ethical dilemma by the clinician-investigators of enrolling HFrEF patients with viability into a randomized trial who would otherwise benefit from revascularization (17). This dilemma becomes particularly more difficult when the standard clinical practice and ACC/AHA practice guidelines (Class IIA, Level of Evidence: B) suggest significant symptomatic and/or survival benefit among HF patients with significant myocardial viability who undergo revascularization.

Altered glucose metabolism and myocardial efficiency have also been studied in high-risk patients with HF imaged with PET and have shown that the magnitude of enhanced use of glucose relative to perfusion in PET (termed “mismatch pattern”) correlates with the potential for improved HF symptoms after revascularization (Figure 3) (39–41). The Canadian PARR 2 (PET and Recovery Following Revascularization) trial used a randomized, controlled design in order to evaluate whether the use of [¹⁸F]FDG-PET in the clinical decision making process manifested a better clinical outcome than standard care, where PET was not available (42). This multicenter trial included patients with LVEF of ≤35% and suspected ischemic heart disease, who were being considered for coronary revascularization, transplantation, or HF. Unfortunately, 25% of the study subjects did not adhere to the PET-guided recommendation. Although there was a trend for improved clinical outcome and survival using the PET strategy compared with standard care in the overall group, it did not reach statistical significance. However, post hoc analysis that considered only subjects who had adhered to the PET-guided recommendation showed a significantly improved outcome in the PET-guided group compared to the outcome in the standard care group (43).

FIGURE 3. PET Scan Showing Perfusion-Metabolism Mismatch in Hibernating Heart Tissue as an Example of Preserved Cardiometabolic Reserve.

(A) Rubidium-82-labeled positron emission Tomography (PET) in short-axis view shows markedly decreased perfusion defects in the apical, inferior, inferolateral, and septal regions of the left ventricle at rest, which extends from distal to basal slices. (B) Images acquired under glucose-loaded conditions, labeled with [¹⁸F]fluorodeoxyglucose (¹⁸F-FDG), show perfusion-metabolism mismatch pattern (the scintigraphic hallmark of hibernation) in all abnormally perfused myocardial regions at rest. An exception is the anteroseptal region, which demonstrates matched perfusion-metabolism pattern (compatible with scarred myocardium). Reproduced with permission from Taegtmeyer and Dilsizian (41).

Beyond myocardial perfusion and metabolism studies, recent studies have determined the prognostic value of myocardial sympathetic innervation in HF (44,45). Results of the ADMIRE-HF (AdreView Myocardial Imaging for Risk Evaluation in Heart Failure) study have published the first large, prospective, multicenter data to validate the independent prognostic value of the [¹²³I]meta-iodobenzylguanidine heart-to-mediastinal (H/M) ratio in assessment of patients with LVEF of ≤35% and New York Heart Association functional classes II to III HF (46). Results showed that, for patients with H/M <1.6, 2-year cardiac death and all-cause mortality were 11.2% and 16.1%, respectively, versus 1.8% and 3.0%, respectively, for those with H/M of ≥1.6. There was a progressive decline in both cardiac and all-cause mortality from >20% in patients with H/M <1.10 to 0% for patients with H/M ≥1.8. This was followed by the PAREPET (Prediction of Arrhythmic Events with Positron Emission Tomography) study, a prospective, observational trial that examined the association between myocardial sympathetic innervation and carbon-11-labeled hydroxyephedrine ([¹¹C]HED) PET and sudden cardiac arrest in 204 patients with ischemic cardiomyopathy and a primary prevention indication for implantable cardioverter-defibrillator (ICD) over a median of 4.1 years (47). Myocardial perfusion, viability, and sympathetic denervation were quantified in all patients. Variables associated with sudden cardiac arrest were volume of denervation and volume of viable denervated myocardium, whereas the volumes of hibernating myocardium and infarcted myocardium were not associated with sudden cardiac arrest.

ASSESSMENT FOR SPECIFIC CAUSES

Cardiac imaging plays important roles in identifying causes that predispose or directly cause HF including structural alterations, congenital or acquired, where there is a disorder of the peripheral vessels, coronary circulation, pericardium, myocardium, endocardium, or cardiac valves. Although a comprehensive discussion of all causes is beyond the scope of this review, the following sections illustrate the value of multimodal imaging in a few specific situations where considerably new developments in imaging assessments of HF have been applied in recent years.

CARDIAC SARCOIDOSIS

Cardiac involvement of sarcoidosis can be characterized by the presence of non-necrotizing granulomatous inflammation and/or fibrosis, which may account for nearly 25% of patients with systemic sarcoidosis. The clinical presentation may be nonspecific, but new right bundle branch block or conduction abnormalities, such as atrioventricular block, idiopathic tachyarrhythmia, or bradyarrhythmia, and HF, should alert the clinician for cardiac involvement of sarcoidosis. Given the focal nature of inflammatory cells, granulomas or scarring in the heart, endomyocardial biopsy may underestimate cardiac sarcoidosis. Noninvasive imaging provides the opportunity to evaluate the presence and extent of the inflammatory phase of cardiac sarcoidosis along with non-necrotizing granulomas and fibrosis, which has implications for treatment choice, monitoring response to treatment, and patient outcome (48).

In the absence of coronary artery disease or history of myocardial infarction, myocardial perfusion defects at rest, with either SPECT or PET, may identify areas of scarring or microvascular disturbances by granulomatous inflammation (49). Focal fibromuscular dysplasia found in the small coronary arteries may provide an explanation for focal ischemic injuries and scar defects described in myocardial perfusion studies. Such left ventricular myocardial perfusion defects have been associated with atrioventricular block and HF, and right ventricular myocardial perfusion defects have been associated with ventricular tachycardia of right ventricular origin (48).

Gallium-67 is a nonspecific indicator of inflammation with a high specificity but low sensitivity (<50%) for diagnosing cardiac sarcoidosis (50). In the absence of anatomical colocalization of the gallium-67 signal, it may be rather challenging to distinguish cardiac gallium-67 uptake from pulmonary and mediastinal uptake. However, a combined myocardial perfusion assessment with gallium-67 inflammation may improve the sensitivity of detecting cardiac sarcoidosis (51,52) and identify patients with cardiac fibrosis without inflammation (matched pattern of severe perfusion defect and lack of gallium-67 uptake) who are less likely to respond to immunosuppressive treatment (52).

[¹⁸F]FDG PET imaging (with or without anatomic colocalization with CT or CMR) combined with perfusion imaging has recently gained interest because of its ability to detect inflammation and fibrogranulomatous replacement of the myocardium and potential follow-up of cardiac sarcoidosis (53). [¹⁸F]FDG PET imaging has been included in the most recent American Society of Nuclear Cardiology/Society of Nuclear Medicine and Molecular Imaging (ASNC/SNMMI) PET guidelines (48). Because inflammatory cells, such as macrophages, contain increased membrane glucose transporters and significantly high hexose monophosphate shunt pathway activity, [¹⁸F]FDG can accumulate within areas of granulomatous inflammation and identify macrophage-dense regions in the heart (54). As a granuloma matures, the number of macrophages and inflammatory cells decrease, with subsequent fibrous replacement that can be detected best with delayed gadolinium enhancement in regions of myocyte replacement fibrosis with CMR (55). Thus, [¹⁸F] FDG PET imaging in concert with either CMR or CT may be optimal for monitoring the efficacy of corticosteroid or other immunosuppressive therapies directed at the active inflammation in cardiac sarcoidosis, transition to fibrosis, and for detection of recurrence (55,56). Moreover, whole-body PET/CT images can provide information on extrathoracic sarcoidosis (beyond the heart, lungs, and hilar/mediastinal nodes) such as the eyes, liver, spleen, skin, peripheral lymph nodes, and bone marrow involvement.

CMR can readily identify characteristic features of cardiac sarcoidosis including biventricular dilation and dysfunction. There is a variable pattern of LGE seen with cardiac sarcoid with a classic pattern of midwall or epicardial LGE, but it can also present with subendocardial or transmural enhancement in almost any distribution. Two studies have shown that LGE by CMR identifies a higher prevalence of sarcoidosis than identified by standard Japanese Ministry of Health guidelines (57). In a recently published meta-analysis of 10 studies and 760 patients, the presence of LGE was associated with a 3-fold increase in all-cause mortality and a 10-fold increase in a composite outcomes of mortality plus arrhythmic events. In addition, T2 mapping may have a role in delineating activity of sarcoidosis (58). In a study of 28 patients, regions of LGE demonstrated decreased T2 which may reflect an inactive phase of the disease (59). The value of CMR in predicting arrhythmic events and death, however, remains to be tested in trials that would involve an intervention to prevent these arrhythmias.

CARDIAC AMYLOIDOSIS

Noninvasive imaging provides the opportunity to evaluate cardiac amyloid load quantitatively and to monitor disease progression and response to treatment (60). Echocardiographic criteria have been most commonly used for clinical diagnosis, including the simple observation of increased wall thickness, although typical features of cardiac amyloidosis occur late in the disease. An important observation was made by Phelan et al. (61) using speckle tracking longitudinal strain segmental analysis to demonstrate relative apical sparing to differentiate wall thickening from cardiac amyloidosis from other causes of LV wall thickening, such as hypertensive heart disease, aortic stenosis or rare disorders like Fabry’s disease and Friedrich’s ataxia (62).

At present, nuclear imaging of cardiac amyloidosis is accomplished predominantly with bone-seeking radiotracers, with conflicting results in terms of diagnostic performance, with the exception of ^99mTc-3,3,-diphosphono-1,2-propanodicarboxylic acid and ^99mTc-pyrophosphate, which appear to differentiate light-chain amyloidosis from transthyretin-related cardiac amyloidosis (63). In particular, ^99mTc-pyrophosphate scintigraphy has recently been shown to identify patients with cardiac amyloidosis, demonstrating diffuse uptake throughout the myocardium, and help in differentiating deposition due to light-chain amyloid versus the transthyretin-related (ATTR) variant (64). However, the future may lie in labeling derivatives of thioflavin-T. With the recent success of visualizing deposition of amyloid β in the brain, the U.S. Food and Drug Administration-approved PET imaging agent [¹⁸F]florbetapir may be used to target cardiac amyloidosis next (60).

CMR has an important role in diagnosing and assessing prognosis in amyloidosis. Amyloid protein in the myocardial interstitium is associated with characteristic patterns of LGE due to abnormal gadolinium kinetics in the infiltrated myocardium (65). The circumferential pattern of subendocardial LGE has a sensitivity of 80% and specificity of 94% against endomyocardial biopsy (66). LGE is much more extensive in ATTR amyloidosis as 90% of patients demonstrated transmural LGE compared to 37% in amyloid. An LGE scoring system can differentiate between the 2 types with 87% sensitivity and 96% specificity (67). The extent of LGE is prognostically important. In a cohort of 250 patients followed for 2 years, a transmural pattern of LGE predicted death with a hazard ratio of 5.4 (68). A meta-analysis of 7 studies and 425 patients demonstrated that LGE was present in 73% of patients, and its presence was associated with an odds ratio for mortality of 5.0 (69).

The native T1 of the myocardium is significantly longer in patients with amyloidosis than that in age-matched individuals without amyloidosis (70) or in those with LV hypertrophy due to aortic stenosis (71). Native T1 tends to be higher in AL amyloid disease than in ATTR (72). The utility of native T1 in suggesting the diagnosis of amyloidosis is particularly important in patients who cannot receive gadolinium contrast due to class 4 or 5 chronic kidney disease, which is quite prevalent in this setting.

SELECTION OF DEVICE THERAPY

Cardiac resynchronization therapy (CRT) with biventricular pacing has advanced the therapy of HF beyond standard medical therapy (73). However, 35% to 50% of patients do not benefit from CRT (42,74). One of the goals of recent imaging studies is the ability to predict response to CRT. Guiding placement of the LV lead for individualization of LV pacing is essential due to variability of patient response (75). Potential reasons for lack of response include pacer placement in scar or in regions that are not late activating, both of which are potential targets for imaging.

Identifying dyssynchrony as a marker of potential CRT response has been a goal of several imaging studies. The PROSPECT (Predictors of Response to CRT) trial enrolled 498 patients in 53 centers and tested 12 echocardiographic markers of dyssynchrony using a blinded core laboratory analysis. Due in part to a lack of reproducibility of the large number of markers tested, the area under the receiver operating characteristic curve for clinical or LV volume response to CRT was only ≤0.62. Some of these markers measure mechanical dyssynchrony rather than electrical dyssynchrony, and CRT does not necessarily reverse the former (76). Some markers that are undergoing study but have not been subjected to multicenter study include septal flash, apical rocking, left bundle branch block pattern by longitudinal strain, septal strain patterns, and systolic stretch index (44). Another echo approach is to identify lack of scar using 2-dimensional radial strain imaging by speckle tracking. In the TARGET (Targeted Left Ventricular Lead Placement to Guide Cardiac Resynchronization Therapy) study, 220 patients were randomized to undergo echocardiographically guided CRT or unguided CRT, and the group with echo-guided LV lead placement demonstrated a significantly higher clinical response and lower rates of combined death and HF hospitalization (77).

Prior CMR studies suggest that scar in the lateral wall is predictive of nonresponse due to lack or poor pacer capture in this location (78–80). Strain imaging by CMR can also be used to identify mechanical dyssynchrony and late activating myocardium that is likely to respond well to CRT (81,82). A single-center study of 75 patients demonstrated that the combination of mechanical dyssynchrony, pacer placement in the site of latest activation, and absence of scar could predict CRT response with an area under the curve of 0.95 (81). A U-shaped pattern of contraction identified by feature tracking cine CMR identified those more likely to respond to CRT in a study of 52 patients (83). None of these CMR endpoints have been studied in a multicenter fashion.

Presently, an EF cutoff of ≤35% is used as the criterion for ICD placement in patients with NYHA functional classes II and III HF. More precise markers of indications for ICD are needed. One potential imaging marker is the presence and extent of myocardial scar as identified by CMR. In a study of 137 patients with cardiomyopathy and EF of >30%, scar size >5% of LV mass increased the risk of death or appropriate ICD discharge similarly to those with an EF ≤30% (84). The absence of scar in patients with EF of <30% lowered the risk similarly to those with EF of ≥30%(12). In one study of 124 patients with ischemic and dilated cardiomyopathy referred for ICD therapy, total scar was predictive of appropriate ICD therapy and had a negative predictive value of 86% (85). Retrospective studies in nonischemic cardiomyopathy demonstrate that LGE presence and extent are predictive of cardiovascular death and appropriate ICD therapy (86,87). A meta-analysis of 19 studies involving 2,850 patients with ischemic and nonischemic cardiomyopathy suggested that the presence of LGE was associated with an arrhythmic event in 24% compared to 5% in those patients without LGE (88). Another potential marker of arrhythmic risk in HF is cardiac innervation as identified by [¹²³I]meta-iodobenzylguanidine. A heart-to-mediastinal ratio of <1.6 had a hazard ratio of 3.5 for arrhythmic events in the ADMIRE-HF study (89).

Measurements of mechanical dispersion and global strain using speckle tracking echocardiography after myocardial infarction can aid assessment of risk of arrhythmia (90). Although mechanical dispersion may overcome the limitation of EF, which predicts poor outcome only in those with obviously reduced ventricular function, the value of mechanical dispersion in patients with relatively preserved EF remains incompletely characterized.

USE OF MINIATURIZED IMAGING DEVICES FOR GUIDING HEART FAILURE THERAPIES

The role of miniature, low-cost hand-held ultrasonographic scanners has been recently evaluated in a few randomized clinical trials. These devices can be used as diagnostic aids for determining the clinical signs of congestive HF and delineating cardiac structural abnormalities associated with HF. For example a randomized trial has investigated whether nurses can be trained in outpatient HF clinics to use ultrasonographic examinations of the pleural cavities and inferior vena cava for improving diagnostics and patient care (91). There was a good agreement between the cardiologist and the heart-failure nurses when pocket-sized ultrasonography was used, with regard to the number of comet tail artifacts, a sign of interstitial fluid. Moreover, the nurses found almost the same number of pleural cavities as found by reference echocardiography. Similarly, another randomized study showed that the determination of volume status based on ultrasonography in outpatient clinics predicted the total diuretic dose adjustments in first and follow-up HF visits (92). Because volume assessment can be performed in a few minutes, this strategy may have role for every routine HF outpatient visit.

With regard to hospitalized patients, a recent study suggested remarkable underutilization of echocardiography (93). Moreover, patients who had echocardiograms had lower hospital mortality. In this regard, use of hand-held ultrasonography units may have important roles for influencing hospitalization outcomes like length of stay and number of hospital visits. For example a randomized study of hospitalist care guided by hand-carried echocardiography was shown to reduce length of stay in participants who were referred for HF (94). The role of hand-held ultrasonography units in combination with other health technologies and other newer technologies, like telemonitoring and teleimaging, require testing.

FUTURE STUDIES

Imaging has important roles in identifying, selecting, and monitoring patients with HF. However, more research and clinical studies are required for development and standardization of several imaging algorithms. First, noninvasive indices of right- and left-sided filling pressures have been suggested to be useful in titrating medical therapy in patients with HF. For example, the size of the inferior vena cava and its collapsibility has been used as an indirect surrogate of right atrial pressure. However, the variability and overlap between patients with normal right pressure and those with elevated right atrial pressure can be substantial (95). Moreover, an increase in right atrial pressure beyond a certain level may cause only minimal increases in inferior vena cava diameter and the degree of inferior vena cava collapsibility with inspiration. Other parameters that have been suggested to provide an estimate of right atrial pressure include right atrial size in 2- or 3-dimensions, hepatic veins, tricuspid inflow, and tissue Doppler indices. However, no single parameter can provide the precision, and perhaps in the future, weighted algorithms or scoring system using the various available indices of these parameters would need to be tested for their accuracy in estimating right atrial pressures (95).

Similarly, with regard to estimation of left atrial filling pressures, several echocardiographic variables have been proposed. However, these variables have complex nonlinear distributions and numerous clinical confounders such as age-dependency. Therefore, no single measurement is completely able to define left atrial or LV diastolic filling pressures. Expert consensus-driven guidelines have suggested complex decision trees with multiple echocardiographic variables; however, these decision trees have been imprecise and frequently changed over the years (96). The most recent guidelines have simplified the diagnostic algorithm and included fewer parameters with the intention of increasing the ease of application in busy modern cardiology practice (97). The validity of the algorithm presented in the current recommendations for guiding clinical therapies but remains a matter of ongoing investigations.

More emphasis needs to be placed on linking LV remodeling with the underlying cause. The use of LV volumes and EF as surrogate endpoints for trials of novel medical therapies has been a staple of HF trials in the past and will continue to be. However most previous HF trials have focused on the presence of coronary artery disease established invasively on cardiac catheterization and presence of LV systolic dysfunction, using imaging techniques for selecting therapies for HF patients. Combining noninvasive imaging (e.g., scar imaging by CMR and coronary imaging by CTA) to differentiate ischemic from nonischemic CMR in patients with new-onset HF is an attractive alternative to invasive coronary angiography, which may mistakenly label a case as nonischemic based on extent of obstructive CAD alone rather than in combination with infarct scar.

Novel PET markers of inflammation and/or CMR markers of edema, inflammation, and/or necrosis imaging in the future could be used as a surrogate for myocardial biopsy, for example, in patients with infiltrative disorders and myocarditis and for identification of transplant rejection. Most of the evidence supporting the use of such new imaging algorithms is either observational or retrospective in nature, with the predominant supportive evidence being prognostication rather than identification of therapeutic targets. Imaging for identifying therapeutic targets in clinical trials remains an area of intense investigation. For example, targeted imaging of myocardial renin angiotensin aldosterone system with radiolabeled lisinopril (Figure 4) may identify patients with increased myocardial angiotensin converting enzyme activity, prospectively and in the early stages of HF, before the transition to replacement fibrosis and remodeling occurs (60,98–101).

FIGURE 4. Noninvasive Micro-SPECT/Micro-CT Imaging of ACE-1 Activity.

Micro-SPECT-CT imaging provides simultaneous scintigraphic and morphological localization of Tc-99m-labeled lisinopril uptake 60 min after tracer administration in a control animal (left), ACE-1 overexpression in a transgenic animal (middle), and a transgenic animal after administration of cold lisinopril (right). Intense lung uptake (white arrowheads) and myocardial ACE-1 activity (yellow arrows) are also demonstrated. In the control rat, micro-SPECT-CT images show tracer uptake in the lungs but not in the myocardium, suggesting normal ACE activity in the lungs and a lack of appreciable ACE activity in normal myocardium. Conversely, ACE-1 overexpression in transgenic rats demonstrated substantial tracer uptake within the myocardium and lungs after Tc-99m-labeled lisinopril administration. Pretreatment of the ACE-1 overexpressing transgenic rats with nonlabeled cold lisinopril almost completely abolished the radiotracer uptake within the myocardium and lungs (white arrows), demonstrating targeting specificity of the radiotracer. ACE = angiotensin-converting enzyme; SPECT-CT = single-photon emission tomography-computed tomography. Reproduced with permission from Dilsizian et al. (99).

Imaging will likely have an important role in selecting proper patients for cardiac devices in the future. As more data accumulates in terms of the predictive value of scar for identifying patients at risk for sudden cardiac death, randomized trials of the presence and/or extent of LGE by CMR in ischemic and nonischemic CM and subsequent SCD/appropriate ICD discharge are low-hanging fruit. Echocardiography and CMR will vie for a place at the table in the identification of patients who are likely to respond to CRT. Both of these techniques have potential in identifying mechanical dyssynchrony, site of late activation, and presence of scar in the lateral wall at the pacing site. These markers in concert are likely to improve upon patient selection for CRT. Comparative imaging trials should be able to discern which technique(s) alone, or in combination, perform best.

Finally, it should be recognized that HF is a disease with large phenotypic variations in cause, clinical features, and natural history. This heterogeneity in presentation is perhaps a reason why several mechanistically sound therapeutic interventions in clinical trials have not shown efficacy. Cardiac imaging provides diverse insights, but the ability to integrate large sets information for identifying specific subtypes of patients with greatest likelihood of response remains a challenging proposition. In this regard, the use of novel phenotypic mapping and machine learning approaches for data mining and identifying patient subtypes will play a critical role in future clinical trials (102).

ABBREVIATIONS AND ACRONYMS

CABG

coronary artery bypass surgery

CAD

coronary artery disease

CMR

cardiac magnetic resonance

CT

computed tomography

HF

heart failure

HFpEF

heart failure with preserved ejection fraction

HFrEF

heart failure with reduced ejection fraction

ICD

implantable cardioverter-defibrillator

LGE

late gadolinium enhancement

LV

left ventricle

PET

positron emission tomography

SPECT

single-photon emission computed tomography