21 Heart failure imaged by cardiac magnetic resonance imaging

Heart failure (HF) is a growing clinical condition resulting from a variety of primary or systemic disorders that impair the ability of the heart to meet systemic demands. Coronary artery disease (CAD), hypertension, and dilated cardiomyopathy (DCM) represent the most common aetiologies in the Western world, with a genetic background found in up to 30% of DCM.¹

Although HF is largely a clinical diagnosis, imaging has become an essential part of patients’ work-up complementing more invasive testing (e.g. coronary angiography, endomyocardial biopsy) and genetic testing. It is key to characterize myocardial and valvular structure and function; identify an underlying treatable substrate; risk-stratify patients; and guide decision-making for medical, surgical, or device therapies such as internal cardioverter-defibrillators (ICD) and cardiac resynchronization therapy (CRT). In addition, it provides measurements for assessing the effect of treatment, including percutaneous or surgical procedures.

As the prognosis of HF remains poor, particular emphasis has been placed on detection of early disease in patients at risk and in those with asymptomatic evidence of left ventricular damage as well as screening of relatives.

Although echocardiography remains the first imaging step,¹ cardiac magnetic resonance imaging (CMR) is now widely available and appears as an ideal complementary technique with the potential to address in a single 45–60 min scan an exhaustive evaluation of three-dimensional cardiac anatomy, function, tissue characterization, coronary and microvascular perfusion, valve disease, and coronary angiography.

The diagnostic and prognostic strengths of CMR as an integral part of the clinical workup of a HF patient are reviewed in a stepwise approach.

Through a wide range of dedicated sequences, CMR can image in any selected plane, without interference from bones or lungs, regardless of patient’s build, without ionizing radiation, and using relatively safe gadolinium contrast agents.

A standard protocol in a new-onset HF patient (Fig. 21.1) includes:

Despite the use of a standardized approach to optimize performance and interpretation of CMR scans, available sequences are clinical instruments to achieve a diagnosis and are dynamically selected throughout the scan on the basis of the evolving picture.

Thus, CMR velocity mapping sequences can be added to the standard cardiomyopathy protocol for suspected valve disease or for quantification of intraventricular gradients with reproducible results.²

Limitations encountered in HF patients are mainly related to previous devices (mostly ICD and pacemakers). Imaging difficulties related to the patient’s difficulty in performing breath-holds, or to atrial fibrillation, were previously considered to be major scanning limitations but can now be overcome by real-time imaging and fast single-shot LGE imaging acquiring one slice per heartbeat.

In HF, treatment and survival are directly related to the cause. While patients are generally rapidly classified as having normal (heart failure with normal ejection fraction, HeFNEF) or impaired left ventricular systolic function (heart failure with reduced ejection fraction, HeFREF) following a standard echocardiography, the next important diagnostic and prognostic step is to define whether patients have ischaemic or nonischaemic cardiomyopathy (NICM). This is usually based on the presence of epicardial CAD as imaged by invasive coronary angiography. However, this approach does not account for patients with NICM who also have concomitant coronary disease which may be incidental or partly contributing to myocardial dysfunction.

CMR offers the unique potential to assist the differential diagnosis step by step from delineation of EF to risk stratification (see Fig. 21.1).

Although many conditions causing HeFNEF in the early stages can progress to HeFREF, CMR diagnostic and prognostic features of the most frequent pathologies responsible for HF are discussed here according to their commonest presentation with a reduced or normal EF.

Aside from searching for a reversible condition responsible for HeFREF, CMR contributes to risk stratification and family screening. It can guide invasive therapy (e.g. percutaneous or surgical revascularization of viable myocardium, left ventricular lead positioning away from infarcted myocardium in CRT) and procedures such as endomyocardial biopsy and provide reproducible data for adequate follow-up under treatment.

CMR accurately assesses global and regional ventricular function, proximal coronary artery stenoses, and their repercussions. Following gadolinium contrast injection, first-pass perfusion can be quantified at rest and during stress. Microvascular obstruction and thrombus are visualized during early imaging, and infarction and residual viability are assessed during the late phase (Fig. 21.2).

Ischaemic cardiomyopathy (ICM) is characterized by a highly specific pattern of LGE validated by histopathology, typically affecting the subendocardium extending up to the epicardium in a pattern consistent with the wavefront phenomenon of ischaemic cell death (Fig. 21.2 C–D).³ Recognition of this characteristic pattern coupled with assessment of perfusion has shed a new light on the diagnosis of ICM.

In presence of left ventricular dysfunction, LGE-CMR is more sensitive than coronary angiography at detecting CAD.⁴ Interestingly, McCrohon et al. showed that among HF patients undergoing CMR, those with known CAD all had subendocardial or transmural enhancement, while DCM patients with unobstructed coronaries displayed three distinct patterns of LGE. The majority had no LGE (59%), but 13% exhibited an ischaemic pattern as a result of transient coronary occlusion and 28% had mid-wall LGE similar to the fibrosis found at autopsy. Similarly, LGE-CMR had a sensitivity of 86% and specificity of 93% for detection of CAD in new-onset HF patients with unobstructed coronaries.⁵

Transmurality of LGE predicts lack of postrevascularization functional improvement in ICM but also outperforms traditional markers of risk such as left ventricular ejection fraction (LVEF) by predicting the risk of inducible ventricular tachycardia, which increases the risk of death.⁶ Similarly, silent infarction detected by LGE, even if small in size, carries a sevenfold increased risk of major cardiovascular events.⁷

To determine whether infarct size correlates with arrhythmias and risk of sudden death (SCD) in HF patients with left ventricular dysfunction, the large DETERMINE trial is still ongoing.

CMR contributes to predicting the success of CRT and to guiding patient selection. For example, the presence and transmurality of posterolateral scar has been linked to lack of response to CRT with a higher mortality when pacing occurred over the scarred area. CMR-derived dyssynchrony indices have also been used as predictors of CRT response.⁸

To ease left ventricular lead positioning, CMR has been used to image coronary venous anatomy prior to device implantation.⁹

Dilated cardiomyopathy (DCM) is the third commonest cause of HF and the most frequent indication for heart transplantation. The aetiology of up to 50% of DCM remains unexplained following exclusion of significant CAD, active myocarditis, and a primary or secondary myocardial disease by coronary angiogram, echocardiography, and rarely endomyocardial biopsy.¹⁰

In addition to providing precise and reproducible quantification of LVEF and biventricular volumes, CMR offers the unique potential of searching noninvasively for an underlying aetiology by its ability to detect fibrosis, scarring, and infiltration.

Up to 28% of patients with systolic dysfunction of unknown aetiology have mid-wall LGE fibrosis, similar to autopsy findings (Fig. 21.3).⁴ While the exact pathophysiology underlying this pattern of fibrosis remains uncertain, it is of strong prognostic significance. Assomull et al.,¹¹ who specifically studied the impact of mid-wall LGE in symptomatic DCM patients, showed that it was associated with increased mortality and cardiovascular events and was the best predictor of sudden cardiac death (SCD) and ventricular tachycardia. Similarly, Wu et al.¹² reported an eightfold increased prevalence of ventricular arrhythmias among ICD-eligible NICM patients with LGE, regardless of the segmental pattern of LGE (midwall, transmural or patchy) and persisting after adjustment for left ventricular volume index and functional class.

Of interest, a proportion of HF patients with unobstructed coronary arteries and systolic dysfunction classified as HF of unknown aetiology display an ischaemic pattern of LGE, highlighting the limitations of standard testing.^4,5 Valle et al.⁵ reported a higher mortality rate and HF admissions among known ischaemic heart disease patients compared to DCM. In unrecognized ICM (normal angiogram but ischaemic pattern of LGE), the risk is similar to that of ischaemic patients, highlighting the role of LGE as a strong predictor of cardiac events beyond EF.

Myocarditis, largely underdiagnosed clinically, is an important underlying cause of several myocardial diseases such as DCM and arrhythmogenic right ventricular cardiomyopathy (ARVC). Although endomyocardial biopsy has been the gold standard diagnostic tool, it is limited by insensitivity and its invasive nature. Among all imaging modalities, CMR appears to be the most powerful in diagnosing myocarditis, by detecting myocardial oedema in the early stages and irreversible fibrotic changes later in the disease process, distinguishing it from infarction and stress-induced cardiomyopathy.

Several aspects of the disease can be imaged by CMR:

As a consequence of myocardial necrosis, several LGE patterns can be seen:

Myocardial iron overload as a result of transfusion-dependant anaemia leads to diastolic and systolic HF, the leading cause of death in these patients despite iron-chelating therapy.

Intensive chelation therapy in the early stages of the disease appears essential to increase life expectancy. However, serum ferritin levels or liver iron are poorly correlated with myocardial iron load. CMR has made possible the noninvasive quantification of myocardial and liver iron as well as cardiac function in the same scan.

Anderson et al.¹⁶ first described the utility of T₂^* relaxation time to measure iron levels from signal intensity decay. A single short-axis midventricular slice is acquired at nine different echo times to derive the T₂^* value arising from field inhomogeneities. The typical epicardial deposition of iron can be visualized in vivo (Fig. 21.5). Aside from monitoring accurately chelation therapy, T₂^* is also a predictive marker of HF and arrhythmias.¹⁶ In a cross-sectional study, 89% of thalassaemia patients with new-onset HF had a T₂^* of less than 10 ms, defining severe cardiac iron overload.

Modell et al.¹⁷ studied the impact of T₂^* measurement in changing outcome among United Kingdom thalassaemia patients and noticed a dramatic reduction in mortality as a result of identification of severe myocardial iron overload and subsequent intensification of iron chelation.

Many myocardial and nonmyocardial conditions can cause HeFNEF. Abnormal diastolic function is the main cause, easily diagnosed by Doppler echocardiography from evidence of abnormal relaxation, decreased compliance and increased filling pressures while left ventricular dimensions and LVEF are normal.¹⁸ Alteration in left ventricular distensibility results from hypertension, CAD, restrictive, obstructive, and infiltrative cardiomyopathies.

While HeFNEF is diagnosed by two-dimensional and Doppler echocardiography, CMR is an ideal complementary tool to identify its underlying aetiology to target further management better.

By providing precise and reproducible quantification of left ventricular volumes, LVEF, left ventricular mass, and wall thickness in any segment, CMR allows a precise detection of serial changes in an individual patient after initiation of treatment.¹⁹

In addition, secondary causes of hypertension, such as renal artery stenosis, coarctation, or adrenal adenomas can be sought during the same scan. The consequences of long-standing hypertension on aortic dimensions can also be assessed precisely.

LGE-CMR provides unique information to assist the differential diagnosis of hypertrophic cardiomyopathy or infiltrative diseases.

Interestingly, although up to 50% of patients may demonstrate patchy LGE, visibly enhanced myocardial regions are usually absent using LGE-CMR even in the presence of diffuse interstitial fibrosis, prompting the use of specific T₁-mapping techniques to quantify amount of collagen.²⁰

Severity of diastolic dysfunction has been correlated to amount of CMR-detected fibrosis.²¹

Hypertrophic cardiomyopathy (HCM) is the most common cause of SCD in the young, including trained athletes, and an important substrate for HF disability at any age.²²

Symptoms can be caused by a variety of mechanisms including left ventricular outflow tract obstruction, arrhythmias, impaired filling due to diastolic dysfunction or impaired systolic function.

Although ECG abnormalities can be the initial and sole clinical clue to HCM, the diagnosis is generally suspected by two-dimensional echocardiographic identification of an asymmetrically hypertrophied, nondilated left ventricle in the absence of another systemic or cardiac disease that is capable of producing the magnitude of wall thickening evident.

However, the distribution and magnitude of left ventricular hypertrophy (LVH) is highly heterogeneous among individuals harbouring the same HCM-causing mutant gene and a proportion of HCM patients are seemingly free from LVH during at least part of their clinical course.²³

In addition, the detection of mild and localized increase in left ventricular wall thickness in trained athletes or long-term hypertensive patients adds to the diagnostic challenge, raising the differential diagnosis between HCM and physiological or secondary hypertrophy. This is further complicated by the presence of disease mimicking HCM such as amyloidosis or Fabry’s disease and by the coexistence of hypertension and HCM in a proportion of patients.

By providing nonoblique images of high spatial resolution, with uniform contrast at the endocardial borders, encompassing all regions of the left ventricle, CMR has the potential to detect segmental wall thickening in any area of the left ventricle (Fig. 21.6A).

Importantly, with the use of CMR, Maron et al.²⁴ identified a spectrum of distribution and pattern of left ventricular wall thickening in HCM patients. About 50% had regional hypertrophy, affecting mainly the basal anterior wall, with a normal left ventricular mass. A minority of HCM patients undetected with conventional imaging were characterized by areas of hypertrophy confined to the anterolateral wall, inferoseptum, and apex, more difficult to image by transthoracic echocardiography. CMR better delineated left ventricular apical aneurysms, affecting 2% of HCM patients, conferring increased risk of SCD, ventricular arrhythmias, thromboembolic strokes, and progressive HF.²⁵

Microvascular dysfunction induced by coronary arteriole dysplasia or mismatch between increased left ventricular mass and coronary flow appears as circumferential stress perfusion defects on CMR and may contribute to the risk attributable to HCM.²⁶

In vivo identification and quantification of fibrosis by LGE-CMR contributes to the diagnostic features found in up to 80% of HCM patients. Commonly, those segments with the greatest hypertrophy are those displaying more LGE, probably as the consequence of long-standing microvascular ischaemia, myocyte death, and fibrosis. Typical patterns include transseptal or RV septal fibrosis, confluent septal or multifocal LGE (Fig. 21.6B). The extent of LGE correlated with increased risk of SCD and HF and was predictive of nonsustained VT and atrial fibrillation.^27,28

While HCM patients without LGE have an excellent prognosis (100% event-free survival at 6-year follow-up), LGE involving 5% or more of left ventricular mass, septal thickness 30 mm or more, and AF are independent predictors of death and ICD discharges.²⁸

Among other prognostic markers, the extent of LVH with marked CMR-calculated left ventricular mass correlate both with the presence of a left ventricular outflow tract gradient and worse HF functional class.²⁴

This infiltrative disease is characterized by the deposition of fibrillary amyloid proteins leading to thickened myocardial walls and diastolic dysfunction resulting in restrictive cardiomyopathy. The most common form, systemic AL amyloidosis, is derived from immunoglobulin light chains. Familial and age-related forms are also described. Cardiac involvement is frequent in the AL form and is associated with a poor prognosis, representing the main cause of death in 50%.

Although echocardiography can raise the suspicion of the diagnosis, endomyocardial biopsy provides the definitive diagnosis. Detection of early stage disease, which may respond to therapy, and exclusion of other disease mimicking amyloidosis, is crucial for patient management.

CMR offers unique diagnostic and prognostic information, helping the early detection of the disease. Typical findings include a small left ventricle with concentric left ventricular (and, inconsistently, right ventricular) wall thickening. Asymmetrical septal thickening mimicking HCM is found in up to 50% of patients. Other features include impaired long-axis function, thickened atrial walls and valve leaflets, dilated atria, and pericardial and pleural effusion. CMR excels by its unique ability to diagnose the macroscopic changes of myocardial tissue composition induced by amyloidosis, by the typical LGE-CMR pattern not seen in any other hypertrophic disease.²⁹

Accumulation of amyloid in the myocardial interstitium results in peculiar gadolinium kinetics with faster washout of gadolinium from blood and myocardium. This often leads to a challenging LGE acquisition with inability to null the myocardium. Typically a predominant diffuse, global, and subendocardial LGE distribution (up to 69%) is found matching the distribution of amyloid on histology (Fig. 21.7).^29,30 The left ventricular midwall is often spared giving rise to a characteristic ‘zebra striped’ pattern of enhancement (Fig. 21.7B). LGE identified cardiac involvement in patients with AL amyloidosis with a sensitivity and specificity of 86% and correlated with severity of HF.³¹

Recently, Syed et al.³² observed that LGE-CMR may detect early cardiac abnormalities in patients with amyloidosis and normal left ventricular wall thickness. Interestingly, while LGE helped the diagnosis, gadolinium kinetics, measured as intramyocardial T₁ gradient, reflecting cardiac amyloid burden, predicted survival. Thus, abnormal T₁ mapping may identify patients in whom early use of more intensive chemotherapy might be justified.³³

Arrhythmogenic right ventricular cardiomyopathy (ARVC) is an under-recognized clinical entity characterized by a fibrofatty replacement of the right ventricular myocardium, involving the left ventricle in up to 75% of cases.

Dilated right ventricle of unknown aetiology, arrhythmias originating from the right ventricular outflow tract, or SCD is often the initial presentation. Despite the fact that right ventricular enlargement and dysfunction are essential features of ARVC, signs of right ventricular failure were seen only in 6% of patients.³⁴ In spite of recent advances in genotyping, the clinical diagnosis remains challenging and relies on the ARVC Task Force criteria, poorly sensitive for detection of gene-carriers with limited disease expression and those with left-sided disease features.³⁵

CMR, is the gold standard for assessing the right ventricle, and offers a complete morphological assessment of the right ventricle without restriction by acoustic windows, detecting global or regional right ventricular dysfunction, wall motion abnormalities, areas of thinning, or aneurysm formation (Fig. 21.8A, B). High temporal resolution transaxial ciné, done in dedicated centres by experienced observers, improves this assessment, reaching a sensitivity of 96% and specificity of 76%.³⁶

T₁-weighted spin echo images were used initially with wide enthusiasm to detect fatty infiltration of the right ventricle free wall but proved of limited value due to difficulties in imaging the thin right ventricular wall and because some healthy individuals have right ventricular adipose infiltration.³⁷

Initial reports focused on detection of right ventricular LGE as a marker of the disease. However, subsequent studies supported left ventricular LGE as the most discriminating diagnostic variable,³⁶ highlighting difficulties in distinguishing right ventricular LGE from myocardial fat, requiring substantially different inversion times compared to the left ventricle.

Left ventricular LGE commonly affects the subepicardium or the midwall (Fig. 21.8C, D) and predicts inducibility of sustained ventricular tachycardia (VT), fibrosis on endomyocardial biopsy, and right ventricular impairment even if its prognostic role still remains unclear.

While no single variable yet allows the detection of ARVC, CMR is an integral component of the diagnostic process in addition to ECG, echocardiography, and other standardized criteria.

Left ventricular noncompaction (LVNC) is an uncommon cardiac abnormality characterized by excessive and prominent trabeculations in the left ventricle (Fig. 21.9) associated with deep recesses. It results from failure of the trabecular regression which occurs during normal embryological development.

LVNC can be an isolated feature or associated with other congenital disorders or genetic syndromes, and may lead to progressive HF, ventricular arrhythmias, and thromboembolic manifestations.³⁸ It usually affects the left ventricle alone, but in fewer than 50% of cases, it can also involve the right ventricle. Features can overlap with DCM, HCM, and restrictive cardiomyopathy.

Although the diagnosis is generally made by echocardiography, CMR has the advantage of higher spatial resolution at the apex and the lateral wall. Hypertrabeculation, with a diastolic ratio of noncompacted over compacted myocardium of 2.3, distinguishes pathological from nonpathological conditions with a sensitivity of 86% and specificity of 99%.³⁹ Dursun et al.⁴⁰ reported three morphological findings: extensive spongy myocardium, prominent trabeculations with deep recesses, and thinned dysplastic myocardium with excessive trabeculations. The absence of well-formed papillary muscles also represents a clue to the diagnosis.

Interestingly, recent reports showed that trabecular LGE subendocardially, in the midwall or transmurally was a common finding, probably as a consequence of microvascular dysfunction.^40,41 The extent of LVNC and the amount and degree of trabecular LGE correlate with LVEF.⁴¹

This multisystem granulomatous disease of unknown aetiology affects the myocardium in 50% of cases of fatal sarcoidosis, with cardiac dysfunction and SCD occurring in up to 67% of patients with evidence of cardiac sarcoidosis found at autopsy. Interestingly, only 23% of patients with cardiac involvement die from HF.⁴² Clinical manifestations depend on the location and extent of granulomatous inflammation and vary from conduction defects and ventricular arrhythmias to diastolic and systolic HF.⁴³ As myocardial involvement alters the prognosis, an early diagnosis is crucial as current therapy may prevent death from cardiac failure and ventricular arrhythmias.

Endomyocardial biopsy and echocardiography are insensitive in identifying myocardial involvement: due to the patchy nature of the disease, relatively advanced stages of the disease tend to be detected. CMR detected myocardial involvement in patients with clinically diagnosed cardiac disease with a sensitivity of 100% compared with 50% for thallium SPECT and 20% for gallium SPECT.⁴⁴ Using T₂-weighted and STIR imaging, CMR identified areas of active inflammation which are reversible with treatment from areas of irreversible myocardial scarring detected by LGE.

In patients with biopsy-proven extracardiac sarcoidosis, LGE is twice as sensitive for cardiac involvement as the widely used clinical Japanese Ministry of Health and Welfare criteria.⁴⁵ Typical CMR findings include areas of focal signal hyperintensity on STIR imaging, corresponding to localized myocardial inflammation. Regional wall motion abnormalities and thinning inconsistent with anatomical coronary artery distribution can be identified on ciné imaging.

LGE as a consequence of granulomatous sarcoid infiltration includes midwall, subepicardial, or patchy patterns (Fig. 21.10).⁴⁶ In addition to providing diagnostic clues to the disease, LGE-CMR may be associated with adverse events, as even small regions of myocardial damage provide a substrate for ventricular arrhythmias and conduction disturbances.

Patel et al.⁴⁵ reported a 9-fold higher rate of adverse events and an 11-fold higher rate of cardiac death in sarcoidosis patients with myocardial damage detected by LGE-CMR. However, the negative predictive value of a negative CMR still remains unknown. Of interest, LGE can be used also to guide localization for endomyocardial biopsy and to monitor the efficacy of steroid therapy.⁴⁷

Through a wide range of sequences, CMR is an ideal complementary tool, providing priceless information in the work-up of a cardiomyopathy patient. The ability of CMR to provide in vivo tissue characterization assists the diagnostic process and provides new measures for risk stratification. In addition, the absence of ionizing radiation offers the opportunity of family screening and regular follow-up.