Myocardial Fibrosis in Congenital Heart Disease and the Role of MRI

Abstract

Progress in the field of congenital heart surgery over the last century can only be described as revolutionary. Recent improvements in patient outcomes have been achieved through refinements in perioperative care. In the current and future eras, the preservation and restoration of myocardial health, beginning with the monitoring of tissue remodeling, will be central to improving cardiac outcomes. Visualization and quantification of fibrotic myocardial remodeling is one of the greatest assets that cardiac MRI brings to the field of cardiology, and its clinical use within the field of congenital heart disease (CHD) has been an area of particular interest in the last few decades. This review summarizes the physical underpinnings of myocardial tissue characterization in CHD, with an emphasis on T1 parametric mapping and late gadolinium enhancement. It describes methods and suggestions for obtaining images, extracting quantitative and qualitative data, and interpreting the results for children and adults with CHD. The tissue characterization observed in different lesions is used to examine the causes and pathomechanisms of fibrotic remodeling in this population. Similarly, the clinical consequences of elevated imaging biomarkers of fibrosis on patient health and outcomes are explored.

Keywords: Pediatrics, MR Imaging, Cardiac, Heart, Congenital, Tissue Characterization, Congenital Heart Disease, Cardiac MRI, Parametric Mapping, Fibrosis, Late Gadolinium Enhancement

© RSNA, 2023

Summary

Cardiac MRI is an important tool to identify myocardial fibrosis, which is prevalent in congenital heart disease and may have important clinical implications for patient management and outcomes.

Essentials

• ■ The cause of fibrotic remodeling in congenital heart disease (CHD) is multifactorial and complex.

• ■ Myocardial fibrosis is prevalent in CHD and has important implications for patient outcomes and management.

• ■ Parametric mapping has potential strength in longitudinal monitoring and guidance of antifibrotic therapies.

Introduction

In the early years of congenital heart disease (CHD) surgery, the focus was on the operation itself as the major determinant of outcomes. With the development of cardiopulmonary bypass and improvements in surgical technique and perioperative care, establishing a viable circulation is no longer the rate-limiting step for many patients, and the majority now survive to adulthood. As a consequence, myocardial changes have more time to develop, and long-term preservation of myocardial health has taken on greater importance in the field. Starting in the 1960s, the "myocardial factor" was recognized as a driver of patient outcomes (1). Although the underlying mechanisms were incompletely understood, the occurrence of arrhythmias, ventricular dysfunction, and sudden death despite an "excellent surgical result" made it evident that additional myocardial drivers were at play (1,2). Jones and Ferrans (3) posited that the "muscle cells that degenerate do not contribute to the mechanical function of the heart, and are ultimately replaced by noncontracting, noncompliant interstitial tissue components," describing the essence of fibrotic myocardial remodeling. Today, that is what we call the myocardial factor: the rearrangement in myocardial structure and fibrosis. The promotors of these changes are multifactorial and include potential combinations of surgical, physiologic, and genetic factors that vary based on individual patient characteristics. The pursuit to characterize fibrosis at a tissue level started with invasive biopsy samples and electron microscopy, which demonstrated interstitial fibrosis and cellular hypertrophy (3,4). More recently, noninvasive techniques, foremost cardiac MRI, have been aiding in the evaluation and diagnosis of these degenerative changes. In this narrative review, we summarize the cardiac MRI approaches to investigate myocardial fibrosis, its potential for evaluation and management of fibrosis, and the origins and role of myocardial fibrosis in CHD.

Use of Cardiac MRI to Assess Fibrosis in CHD

Cardiac MRI provides a noninvasive method for quantification and anatomic evaluation and is therefore most commonly performed in a subset of CHD where this information is particularly valuable. The most common forms of CHD in which routine cardiac MRI is performed for evaluation of ventricular function, hemodynamics, and anatomy include tetralogy of Fallot (TOF), transposition of the great arteries, single ventricle physiology, aortic valve and arch disease, and shunt lesions. Fibrosis imaging has been incorporated in guidelines and position statements within the last decade for some of these lesions, including TOF and transposition of the great arteries (5,6). As a result, these forms of CHD are highlighted most prominently in this review.

Cardiac MRI Methods for Detecting and Quantifying Fibrosis

The T1-shortening properties of gadolinium-based contrast agents combined with contrast agent retention within fibrotic tissue underlies our ability to assess fibrosis by using cardiac MRI. Following intravenous administration, gadolinium enters the myocardial microvascular network and extravasates into the interstitial fluid. Over a period of time, the contrast agent washes out into the venous system. This process is delayed in injured or scarred tissue, resulting in contrast enhancement in regions of scar. Late gadolinium enhancement (LGE) and T1 parametric mapping are the two most commonly applied methods for the detection and quantification of fibrosis.

LGE Imaging

LGE imaging exploits the T1-shortening effect of gadolinium. A T1-weighted inversion recovery image acquisition several minutes after gadolinium administration is timed so that signal from healthy myocardium is nulled. At that point, myocardium that contains gadolinium appears bright, or enhanced (Fig 1). Developed in the late 1980s, LGE has been extensively studied and validated as a method to assess for reversible and irreversible injury in various disease states. The presence of LGE has been linked to adverse clinical outcomes and is a routine component of cardiac MRI for myocardial tissue characterization (7–9). While assessment of LGE is predominantly qualitative in nature, quantitative and semiquantitative (ie, scoring based on location and extent) assessment has been introduced in some forms of congenital heart disease, including TOF and transposition of the great arteries (10–14).

Figure 1:

Late gadolinium enhancement at cardiac MRI. Mid short-axis delayed enhancement image in a 4-year-old boy with bicuspid aortic valve status after Ross operation and resultant coronary obstruction shows areas of subendocardial and midmyocardial enhancement in the inferoseptal and inferior segments, signifying areas of myocardial scarring (arrows). Note the thin-walled right ventricle obscuring identification of scar.

Despite its unique ability to demonstrate myocardial damage, LGE has numerous limitations. First, it requires the use of a gadolinium-based contrast agent. Administration of contrast media is contraindicated in patients who are pregnant or who have history of gadolinium-based contrast media allergies, and contrast media should be used cautiously in patients with end-stage renal disease after careful risk-benefit assessment. Furthermore, the long-term effects, including the potential for brain deposition and associated consequences, of contrast agent administration in the pediatric population is unknown; therefore, it is important to avoid the use of gadolinium-based contrast agent in children unless deemed necessary. Second, LGE relies on the contrast between diseased and neighboring myocardium. When the heart is uniformly, or diffusely, fibrosed, LGE may not demonstrate fibrosis and the myocardium may appear normal. Third, LGE has limited ability to help identify the specific mechanism of injury, as it may represent either chronic scar (eg, ischemic heart disease) or acute necrosis (eg, myocarditis). Inflammatory and infiltrative diseases also result in myocardial enhancement. For example, cardiac sarcoidosis may result in LGE because of the presence of disrupted cell membrane integrity and noncaseating granulomas, and cardiac amyloidosis may result in LGE from deposition of amyloid fibrils. These systemic diseases are more common in adulthood, so LGE presence must be cautiously interpreted particularly in older adult patients with CHD. Often, the clinical context and imaging findings allow for the distinction between fibrosis or scarring and infiltrative and inflammatory conditions (15,16). Fourth, given the finite resolution of LGE images, the scar or injury has to be of a minimum size to be detected. Last, in the right ventricular (RV) free wall and outflow tract, LGE imaging is challenged by the thin myocardial layer, sandwiched between T1 signal-intense blood pool and epicardial fat. Artifacts from sternal wires and the proximity to the surface coil pose additional challenges to signal and field homogeneity (17).

Parametric (T1) Mapping

T1 mapping is a quantitative method to evaluate diffuse myocardial fibrosis. The longitudinal relaxation time constant (T1) tracks with the degree of fibrosis, and an equation containing T1 times before and after administration of gadolinium-based contrast media provides a quantitative estimate of the extracellular volume (ECV) expansion associated with diffuse fibrosis, as described below (18,19).

The modified Look-Locker inversion recovery (MOLLI) pulse sequence is the most commonly applied T1 mapping technique. It uses an electrocardiographically gated single-shot approach to acquire several diastolic images, each with slightly different inversion times. These images allow for the creation of a T1 map (20). For precontrast native T1 mapping, the most common method of acquisition today includes an inversion pulse followed by acquisitions over five heartbeats, a pause (P), and, finally, a second inversion pulse with images acquired over three beats (5[P]3 method). The pause is measured either in beats or in seconds and has to be long enough to allow for full relaxation in between the two inversion experiments. A "shortened" MOLLI sequence, known as shMOLLI, can reduce acquisition time to nine heartbeats using a 5(1)1(1)1 acquisition (21). Saturation-based (saturation recovery single-shot acquisition, or SASHA) and combined saturation and inversion recovery (saturation pulse prepared heart-rate-independent inversion recovery, or SAPPHIRE) approaches reduce acquisition times and are less heart rate dependent than their inversion recovery counterparts (22). Images are typically acquired in the basal, midventricular, and apical short-axis levels and sometimes the four-chamber view, depending on the lesion and protocol used (20).

T1 times are traditionally derived by drawing corresponding regions of interest on each of the source images and using the average signal intensities of each to fit the curve. Inline generated T1 maps are now available on most platforms, providing a less time-consuming method for T1 analysis by drawing a region of interest on the automatically generated map (Fig 2). The mapping algorithm measures signal intensities for corresponding pixels on each of the maps and uses them to fit the T1 recovery curve. While time efficient, careful attention is necessary to recognize potential artifacts or coregistration errors. This may be particularly important in children who are unable to lie completely still or hold their breath for the acquisition. Both methods have good-to-excellent inter- and intrarater reliability, with inter- and intraclass correlation coefficients generally in the 0.8–0.9 range (23). However, given the systematic bias between the two methods, the same approach should be employed consistently at any one center (24). Furthermore, special attention should be given to regions of LGE; T1 and ECV values are elevated in areas of patchy scarring, and most experts exclude areas of LGE when assessing T1 mapping.

Figure 2:

T1 mapping in patients with single ventricle physiology. Native T1 maps from a mid short-axis plane in patients with (A) hypoplastic left heart syndrome and (B) a hypoplastic right heart. * indicates the hypoplastic chamber with evidence of hypertrophy and dilatation of the dominant ventricles. Free wall regions of interest (violet) are drawn on the free wall of the dominant ventricles, and blood pool regions of interest (blue) are drawn within the blood pool of the dominant ventricle.

ECV Calculation

While native T1 times can provide valuable information about myocardial relaxation properties, the use of gadolinium-based contrast media allows for the integration of pre- and postcontrast T1 values to generate ECV using the following formula:

where postT1myo and preT1myo represent post- and precontrast myocardial T1 values, respectively, and postT1blood and preT1blood represent post- and precontrast blood pool T1 values, respectively.

ECV can be derived via two different approaches. First, an ECV map can be generated after coregistration of pre- and postcontrast T1 maps and adjusting for the patient’s hematocrit level. Following ECV map generation, regions of interest can be drawn. Alternatively, T1 times derived independently from pre- and postcontrast images and corresponding regions can be entered into the ECV formula. This allows for greater control of the imaging information underlying the values, but is more time-consuming.

Native T1 and ECV have been shown to correlate with collagen volume fraction and extracellular matrix expansion, as measured by histologic samples in adults and children (19,25,26). Clinically, they correlate with all-cause mortality and severity of diastolic dysfunction among adult patients with hypertrophic cardiomyopathy and heart failure with preserved ejection fraction (27,28). Whether ECV offers a diagnostic or prognostic advantage over native T1 in chronic, noninflammatory conditions such as CHD is unclear at this time.

T1 Mapping Challenges

Similar to LGE, other, nonfibrotic causes may result in abnormal T1 and ECV values. For example, the presence of infiltrative diseases such as amyloidosis or iron overload will result in abnormal T1 mapping values (increased native T1 and ECV values in amyloidosis and decreased T1 in iron overload). However, amyloidosis can normally be differentiated from fibrotic causes because of the extreme elevations in values (ie, native T1 values > 1200 msec and ECV values of 50%–60%) (20,29).

Native T1 values are age, sex, vendor, field strength, pulse sequence, and temperature dependent. This variability with patient, hardware, and software factors inhibits the clinical use of T1 mapping because of the nongeneralizability of normal values. It also limits the ability to perform effective multicenter research by using native T1 values as an outcome. The Society of Cardiovascular Magnetic Resonance recommends establishing normal reference ranges with individual scanners within an institution, using 15–50 healthy controls to establish normal values (19,20). Therefore, published reference values serve as an approximate guide only, with the upper limit of normal native T1 values in adults varying from 950 to 1050 msec with 1.5-T magnets and as high as 1050–1200 msec with 3.0-T platforms (30,31). In the pediatric population, larger data sets or meta-analyses are lacking. Smaller studies have reported normal values of 1008 msec ± 31 (SD) and 1018 msec ± 25 in children. Many experts regard values more than 2 SDs above the mean as abnormal (23,32). Furthermore, at this time there is no consensus statement on the best T1 mapping analysis technique. Guideline statements do not provide any specific recommendations about software and methods for analysis (inline vs source images, region of interest locations, etc).

In comparison to native T1, ECV is less strongly impacted by patient- and scanner-specific factors because the use of pre- and postcontrast myocardial imaging relative to blood results has some degree of standardization across different platforms and field strengths (30,31). However, while ECV reference ranges from large adult cohorts have been published, with typical values between 24% and 28%, reference data in children are limited to smaller, single-center studies (23,33–36). Larger, multicenter initiatives to determine normal values in pediatric patients are currently underway.

Given the finite spatial resolution, T1 quantification is challenging in thin myocardium because of partial voluming with luminal blood and epicardial fat, such as in the RV. As a result, the assessment of RV T1 and ECV values is controversial. A history of prior cardiac surgery further complicates assessment because of the presence of sternal wires and other metallic implants that distort the magnetic field (37). These issues are somewhat mitigated in conditions with significant RV hypertrophy, such as pulmonary hypertension, TOF, systemic RVs, or single ventricle physiology. Newer approaches that are not currently widely available include diaphragm-navigated, cardiac-gated imaging with compressed sensing and/or a segmented readout to improve resolution and image quality (38).

The need for intravenous access to inject gadolinium-based media and draw blood for hematocrit assessment are barriers to ECV quantification. Of note, the hematocrit values may vary with body position, and best practice is to draw the blood at the end of the examination while the patient is still supine; however, this is not always practical, especially in children in whom the most reliable way to obtain sufficient blood is at the time of intravenous line insertion (39). The use of point-of-care hematocrit analysis as an alternative to a laboratory blood draw may be acceptable, though this method may be less reliable and presents more variability than laboratory hematocrit assessment (40). As a result, the method and device used for hematocrit measurement should be standardized at each institution.

Similar to LGE, sternal wires and other metallic implants, such as stents, coils, or mechanical valves, may degrade the quality and accuracy of T1 mapping. Artifacts are typically more problematic at higher field strengths.

Considerations for Pediatric and CHD Populations

While LGE and T1 mapping can generally be performed in patients with CHD using similar parameters as those used in patients without CHD, certain considerations apply specifically to the pediatric population. For example, modifications to standard imaging protocols are necessary to account for higher baseline heart rates in younger children. While traditional precontrast MOLLI T1 mapping sequences wait for three heartbeats between the two inversion experiments, the pause needs to be extended to five beats with higher heart rates to allow for full T1 relaxation (23). The shorter RR interval at higher heart rates may render it impossible to fit in the acquisitions at longer inversion times, so it may become necessary to use two beats per acquisition. The same can be accomplished, if feasible with the platform, by manually entering a heart rate that is half the patient’s heart rate (at the expense of a longer acquisition time). To avoid motion artifact, decreasing acquisition time per beat may also be necessary, especially considering the shorter length of diastole per beat at higher heart rates (41).

In pediatric patients and those in heart failure, ability to maintain a breath hold for an entire acquisition may not be possible. The use of single-shot imaging acquisitions, rather than a segmented approach, for LGE provides a high-quality free-breathing alternative and often yields an image quality that is superior to that achieved by multiple averages (42). Automated motion correction for T1 mapping has been developed to negate some effects of cardiac motion, but its performance for respiratory motion is more variable; as a result, breath holding is still preferred for T1 mapping. At this time, a widely available free-breathing T1-mapping alternative is not available.

Mechanisms of Fibrosis in CHD

Surgical Intervention

Surgical intervention in many forms of CHD requires incisions in the ventricular myocardium. Myocardial scarring occurs as a result of ventriculotomy (43). Beyond direct surgical manipulation, there is evidence that cardiopulmonary bypass–mediated inflammation also plays a role in fibrotic remodeling. Cardiopulmonary bypass results in a systemic inflammatory response and oxidative stress, which culminate in cellular injury (44). The length of cardiopulmonary bypass and aortic cross-clamp correlates with native T1 times in TOF (45). Additionally, surgical patch material routinely enhances late after gadolinium-based media administration (12,46,47).

Ventricular Loading

Ventricular adaptation that occurs in response to abnormal loading conditions, including ventricular dilatation (in the setting of volume overload) and hypertrophy (in the setting of pressure overload), may be a driver of ventricular fibrosis in CHD. The presence of LGE and elevated ECV values, T1 times, and circulating biomarkers of fibrosis have been correlated with increased ventricular afterload and volume overload in various forms of CHD, including TOF, aortic stenosis, systemic RV, single ventricle physiology, and simple shunting lesions such as atrial and ventricular septal defects (48–52). For pressure-overload conditions, the resultant ventricular hypertrophy, increased oxygen demand, elevated wall stress, and end-diastolic pressure, as well as capillary paucity, may lead to relative microvascular ischemia and reactive fibrosis (53). Conversely, volume overload may precipitate myocardial fibrosis through increased noncollagen extracellular matrix production as opposed to collagen synthesis (54).

Neurohormonal Activation and Genetically Mediated Fibrosis

Neurohormonal activation appears to be involved in the development of fibrosis in ventricular adaptation. Interestingly, fibrosis in one ventricle is commonly associated with similar changes in the contralateral ventricle, even if that ventricle is not burdened by abnormal loading conditions (55). This observation has been made in numerous forms of CHD, including TOF and systemic RV, thus supporting the notion that humoral factors are at play (45–47,56,57). The renin-angiotensin-aldosterone system controls fluid volume, sodium-potassium balance, and blood pressure. Its activation results in an upregulation of numerous profibrotic and proinflammatory genes, including those encoding for matrix metalloproteins, plasminogen activator inhibitor 1, transforming growth factor B, and connective tissue growth factor, ultimately leading to increased extracellular matrix protein deposition (58). Genetic predisposition, such as mutations in myocyte enhancer factor 2, have been implicated in the development of fibrosis through various neurohormonal pathways, including in CHD (59–61). Specific examples include increased production of hypoxia-inducible factor 1-α (HIF1A), which upregulates transforming growth factor B and matrix metalloproteins, in TOF (62,63), renin in patients with a single ventricle (64), and procollagen type III peptide in various forms of CHD, including single ventricle disease, systemic RV, and TOF (50,65,66). Histopathologic findings and cardiac MRI markers demonstrated the association between presence of certain genetic polymorphisms and higher functioning HIF1A alleles and increased RV fibrosis in patients with TOF (67,68). Additionally, patients with single ventricle disease with polymorphisms for renin-angiotensin-aldosterone system pathway genes showed maladaptive ventricular remodeling following the second stage of single ventricle palliation (69). These associations bear opportunities for potential therapeutic targets and will be discussed in more detail later.

There are also potential sex-specific differences in gene expression that may result in differential fibrosis risk among men and women with CHD. Female patients with TOF have been shown to have higher left ventricular (LV) and RV native T1 and ECV values (45,70). Furthermore, higher collagen and matrix metalloproteinase gene expression has been demonstrated in men as compared with women undergoing aortic valve replacement (71).

Hypoxemia

Hypoxemia likely contributes to the development of fibrosis. In patients with Ebstein anomaly, ECV is strongly associated with decreasing oxygen saturations (72,73). Associations between hypoxia, upregulation of HIF1A, and fibrosis in TOF also support the role of hypoxemia in fibrosis development (67).

Clinical Correlates of Myocardial Fibrosis in CHD

Tetralogy of Fallot

TOF is the most common form of cyanotic CHD and is defined by the following features: (a) an anterior malalignment ventricular septal defect, (b) RV outflow tract obstruction with or without pulmonary stenosis or atresia, (c) an overriding aorta, and (d) RV hypertrophy. Timing of surgery for TOF is dependent on degree of obstruction and ability to maintain adequate saturations. Currently, surgical timing most commonly varies from the neonatal period (in cyanotic TOF or pulmonary atresia) to 3–6 months of age for infants with adequate oxygen saturations (ie, "pink" TOF). Surgical repair normally consists of ventricular septal defect closure and relief of RV outflow tract obstruction or pulmonary stenosis, often with a transannular patch, resulting in an unguarded pulmonary valve with severe regurgitation. Long-term regurgitation results in RV volume overload, which over time can result in RV and, occasionally, LV dysfunction. Cardiac MRI is therefore a routine part of monitoring for progressive RV dilatation and dysfunction, as well as evaluating fibrosis and anatomy (5).

Myocardial scarring and fibrosis in patients with TOF can develop early. Histopathologic fibrosis is present in up to 65% of patients at the time of repair (74) and is associated with poor ventricular function in the postoperative period (75). LGE has been extensively reported long term after TOF repair, identified in about two-thirds of patients. It is frequently found at the inferior RV insertion points and areas of surgical intervention, as discussed above (Fig 3) (12,14,46,66,76). The presence of LGE is associated with adverse ventricular mechanics, including worse biventricular ejection fraction, increased RV volumes, and restrictive physiology (12,46,76,77). It is linked to elevations in atrial and brain natriuretic peptide, as well as adverse clinical markers, including all-cause mortality, worse New York Heart Association functional class, poor exercise tolerance, and increased arrhythmia burden and QRS duration (11–13,46,78).

Figure 3:

Late gadolinium enhancement (LGE) patterns at MRI in patients with tetralogy of Fallot (TOF). (A) Short-axis delayed enhancement image shows LGE in the area of the transannular patch (arrow). (B) Left ventricular outflow tract delayed enhancement image shows LGE at the ventricular septal defect patch (arrow) in an 18-year-old woman with history of TOF with pulmonary atresia after repair.

Biventricular elevations in T1 values, in addition to patchy LGE, are also common in repaired TOF and, in patients from the current surgical era, may be the predominant mechanism of adverse myocardial remodeling. Multiple groups have described elevated native RV and LV T1 and ECV values in these patients compared with controls, although others have demonstrated normal values (70,79).

Elevated ECV has been associated with adverse events and outcomes, including increased arrhythmia burden, worse New York Heart Association functional class and other heart failure metrics, exercise intolerance, hospitalization, and death (80,81). T1 and ECV have also been linked to abnormal myocardial mechanics, including mechanical dyssynchrony and worse longitudinal, circumferential, and radial strain at cardiac MRI, as well as RV dilatation and hypertrophy, further highlighting the potential role of chamber remodeling in the development of fibrosis and vice versa (34,35,46,47,70,79,81).

Systemic RV in a Biventricular Circulation

Subsystemic RVs occur in two categories: (a) d-transposition of the great arteries following the Senning and Mustard operations, prior to the advent of the arterial switch operation (51); and (b) l-transposition (or congenital corrected transposition), where atrioventricular (ie, right atrium to LV and left atrium to RV) and ventriculoarterial (ie, LV to pulmonary arteries and RV to aorta) discordance result in a systemic RV (6,82).

Presence of LGE and fibrosis at histologic analysis has been reported frequently in this population (prevalence up to 60%), particularly in the RV free wall, along the basal septum, and at the inferior and superior insertion points (Fig 4) (10,51,52,83,84). In these patients, LGE is associated with arrhythmias and impaired exercise performance and is an independent predictor of adverse outcome, including decompensated heart failure, transplantation, or death (10,51,83).

Figure 4:

Late gadolinium enhancement (LGE) at MRI in a patient with a systemic right ventricle. Short-axis image shows superior insertion point LGE with midmyocardial anteroseptal extension (arrows) in a 27-year-old male patient with congenitally corrected transposition of the great arteries after mechanical tricuspid valve replacement.

Native T1 and ECV values are elevated in patients with systemic RV, as compared with the LV of healthy controls (37,57), and tend to be higher than in other forms of CHD (34,85).

Single Ventricle Physiology

In a subset of patients with CHD, neonatal anatomic considerations (ie, a hypoplastic LV or RV, abnormal or hypoplastic atrioventricular valves, or unusual orientation of the great arteries) make the establishment of a biventricular circulation impossible. In these patients, a series of staged surgeries can be performed that ultimately result in the establishment of a passive mechanism for pulmonary blood flow, starting with a superior vena cava–to–pulmonary arterial anastomosis (superior cavopulmonary connection, also known as bidirectional Glenn or hemi-Fontan) and culminating with a conduit from the inferior vena cava and hepatic veins to the pulmonary arteries (total cavopulmonary connection, also known as Fontan palliation) (86).

Histologic data on myocardial fibrosis in patients with single ventricle physiology has been mixed. One study found elevated fibrosis levels only in patients who had undergone placement of an RV-to–pulmonary artery conduit as compared with the Blalock-Taussig-Thomas shunt during the first stage of palliation (87). Approximately one-third of all patients with Fontan circulation have LGE, which tends to occur most commonly in the ventricular free walls or in the region of a prior ventriculotomy (Fig 5) (88,89). The presence of LGE has not been associated with hospitalization, Fontan failure, or all-cause mortality, although it has been associated with larger ventricular volumes and mass and lower ejection fraction, as well as a higher incidence of ventricular tachycardia, independent of ventricular dominance (88,90).

Figure 5:

Late gadolinium enhancement (LGE) patterns at MRI in patients with single ventricle physiology. (A) Short-axis image shows inferior insertion point LGE (arrow) in a 15-year-old adolescent boy with unbalanced atrioventricular canal after Fontan operation. (B) Short-axis image shows LGE (arrow) at the region of ventriculotomy for right ventricle–to–pulmonary artery conduit (Sano shunt) insertion along the right ventricular anterior wall in a 13-year-old adolescent girl with hypoplastic left heart syndrome after Fontan operation.

Compared with controls, ECV and native T1 values are elevated in patients with Fontan circulation and associated with worse myocardial deformation at cardiac MRI. They are also associated with a composite outcome of hospital admission, cardiac reintervention, Fontan failure, or clinically significant arrhythmias (91,92). The association between parametric mapping values and ventricular dominance has been more variable, with one study demonstrating higher ECV and T1 values in patients with single RV compared with patients with single LV and controls (91).

Aortic Stenosis

Histologic fibrosis has been demonstrated in patients with severe aortic stenosis and has been associated with worse New York Heart Association functional class and the need for earlier aortic valve replacement (93,94). Typical patterns of LGE are basal, subendocardial fibrosis, believed to be secondary to decreased myocardial perfusion combined with myocardial wall sheer stress (94). Reported native T1 and ECV values are consistently higher in adults with aortic stenosis and correlate with stenosis severity (95,96).

Similar findings have been demonstrated in patients with bicuspid aortic valve physiology. Young patients with congenital aortic stenosis have demonstrated more extensive LGE compared with healthy controls, which correlates with surrogates of myocardial systolic and diastolic dysfunction, such as strain and left atrial dilatation, respectively (36,97). Presence of LGE in bicuspid aortic valve has also been associated with a higher degree of stenosis, as well as a need for earlier aortic valve replacement (93).

Ebstein Anomaly

Ebstein anomaly is a rare form of CHD that involves delamination and apical displacement of the tricuspid valve leaflets and atrialization of the RV cavity, with a variable degree of tricuspid regurgitation and RV dysfunction. The spectrum of clinical presentation ranges from mild and asymptomatic to severe neonatal presentation of cyanosis and heart failure, which is often dependent on the degree of apical displacement, tricuspid regurgitation, and RV dysfunction.

Literature on myocardial fibrosis in Ebstein anomaly is limited to relatively few publications. In adults with Ebstein anomaly, LGE has been demonstrated in around 20% of patients and is predominantly seen in the RV free wall and along the RV portion of the septum, with sparing of the LV myocardium (73). Presence of LGE is associated with larger RV volumes, lower oxygen saturations, and worse New York Heart Association functional class (73).

Whereas LV LGE has not been demonstrated, patients with Ebstein anomaly have significantly higher LV native T1 and ECV as compared with controls. Presence of higher native T1 and ECV has been directly associated with disease severity, RV volumes, and LV ejection fraction and strain. Furthermore, higher T1 and ECV values are associated with clinical status, including higher New York Heart Association functional class, lower oxygen saturations, and impaired exercise performance (72,73).

Patterns of LGE and T1 mapping trends, as well as respective clinical associations in various forms of CHD, are summarized in the Table.

LGE and T1 Mapping in Congenital Heart Disease

Implications for Clinical Care and Future Directions

Despite the expanding volume of literature on fibrosis, most studies to date are cross-sectional in nature. The lack of well-defined clinical implications of fibrosis and cardiac MRI biomarkers in the CHD population, as well as the substantial overlap of T1 and ECV between health and disease, make it challenging to derive individual treatment decisions. While lesion-specific guidelines and position statements have incorporated an assessment for myocardial scarring into the standard cardiac MRI protocol, specific recommendations on the management of LGE and/or elevated T1 mapping values, if found, are lacking to date (5,82). Limited data in the TOF population suggest that LGE, when present, is not progressive in the intermediate-term follow-up, though this may not be generalizable to other forms of CHD (14). Furthermore, recommendations on the frequency of LGE and T1 mapping are lacking from available guidelines. Considering the need for contrast media administration for LGE imaging, a reasonable approach would be to perform LGE and T1 imaging at the first routine scan and at subsequent examinations if there is clinical concern (ie, regional wall motion abnormalities or abrupt change in ventricular function), a need for gadolinium-based contrast media for other indications (ie, anatomic assessment), or every few years. As a result, these markers must always be viewed in the patient’s larger clinical context. Future research focused on longitudinal changes within the same patient is necessary and holds particular promise toward clinical decision-making and prognostication (47).

Preventing or slowing down adverse myocardial remodeling is an important therapeutic aim in many types of heart disease, including congenital conditions. LGE, and particularly T1 mapping, provide noninvasive metrics that are increasingly used as end points in clinical trials. Considering not only the potential for cardiomyocyte regeneration in younger individuals, but also the observation that fibrotic remodeling begins in childhood, finding preventative antifibrotic strategies is especially attractive in children and young adults (98).

Previous studies point toward a potential role for medications targeting the renin-angiotensin-aldosterone pathway in remodeling and fibrosis reduction, supported by evolving knowledge about its role in fibrogenesis. Potential therapeutic options include aldosterone inhibitors (eg, spironolactone and epleronone), angiotensin-converting enzyme inhibitors (eg, ramipril and enalapril), and angiotensin receptor blockers (eg, losartan and valsartan). These medications have been shown to improve morbidity and mortality in heart failure and following myocardial infarction (99–101); furthermore, they have a potential role in slowing the progression of myocardial fibrosis, as has been suggested in cardiomyopathies such as Duchenne muscular dystrophy (102).

In the realm of CHD, epleronone has been studied in a small group of patients with systemic RVs, with a trend toward reduction in circulating biomarkers of collagen synthesis; however, this reduction did not meet statistical significance (49). The reported success of angiotensin receptor blockers in CHD has been mixed. In patients with systemic RVs, one study demonstrated an improvement in RV ejection fraction and exercise tolerance (103), while other studies showed no effect on neurohormone levels, including N-terminal pro-B-type natriuretic peptide and angiotensin II, or exercise performance (104,105). Similarly, a randomized controlled trial of losartan in patients with TOF failed to show an effect on ventricular function, exercise capacity, and N-terminal pro-B-type natriuretic peptide (106). Finally, ramipril and enalapril failed to result in improvement in cardiac MRI and clinical markers in TOF and systemic RVs (107,108) or improvement in survival and ventricular function in interstage patients with single ventricle physiology (109). Of note, studies evaluating the direct antifibrotic effects of medications targeting the renin-angiotensin-aldosterone system pathway are rare, with only one study assessing T1 and ECV (110). Randomized controlled trials in adults with acquired heart disease have demonstrated that careful patient selection is key when attempting to favorably affect myocardial remodeling. In a large study on the effects of spironolactone for patients with heart failure with reduced ejection fraction, only patients with an activated collagen metabolism benefited from the drug (111). Cardiac MRI markers, especially from parametric mapping, may be able to help select patients with CHD who are most likely to benefit from treatments. These markers may also be able to help with the type, timing, and monitoring of targeted antifibrotic therapy. Ultimately, precise tissue characterization of the myocardium harbors the promise of a personalized approach to antifibrotic therapy and prevention.

Advances in Imaging

Despite substantial advances in parametric mapping, there are still numerous limitations, some of which have already been discussed. T1 mapping sequences, although offered by all the major vendors, are still not available or routinely employed at many institutions. Published normal values in children remain limited to single-center series, and normal baseline values specific to patients with CHD are missing. Evaluation of LGE and parametric mapping in small children, the nonhypertrophied RV, and thin-walled atria is technically limited because of limits of spatial resolution. Development and widespread availability of sequences with improved spatial resolution are needed to further expand the applicability of fibrosis imaging. Finally, image acquisition, analysis, and postprocessing are time-consuming. More recently, a technique known as MR fingerprinting has been developed, which acquires both T1 and T2 maps in addition to tissue perfusion, diffusion, fat signal fraction, T2*, and ECV value in a single sequence without the need for exogenous contrast agents (112–115). Machine learning algorithms that segment and analyze imaging data have also been developed to reduce analysis time (116). These advances in reducing acquisition and processing times have the potential for more widespread adoption of advanced tissue characterization techniques. Currently, these techniques are predominantly research-based, and further validation among large data sets is necessary before they can be used clinically.

Conclusion

With substantial surgical advances in the management of patients with CHD, patient myocardial health has emerged as a major determinant of outcomes. Myocardial tissue characterization using cardiac MRI has become a valuable tool for the detection and quantification of fibrosis, with the potential for earlier, as well as more targeted, interventions and treatment monitoring. Long-term longitudinal studies in various disease states are needed to define risk factors for the development of fibrosis and to understand the prognostic implications of these imaging markers on long-term outcomes.