The Role of Multimodality Imaging in Cardiomyopathy

CMR ventricular measurements are made from a short-axis stack that covers the entire ventricle. The LV and RV ejection fraction and volume are calculated using the Simpson’s summation of discs method. The myocardium and cavity of each short axis slice is contoured to create a stack of disks that accurately reflect the myocardial structure. This differs from 2D echocardiography, which uses the Simpson’s biplane method with the 2-chamber and 4-chamber long-axis views to make assumptions about the shape of the disks and therefore an estimate of the volume. As a result, the LVEF can often vary between the modalities. CMR offers high reproducibility for both intra- and inter-observer comparisons. In lower LVEF ranges, 2-dimensional (2D) echocardiography often overestimates LVEF compared to CMR [19, 20•, 21]. In a study evaluating the impact of CMR on implantable cardioverter-defibrillators (ICD), CMR reclassified 41% of patients with LVEF between 25–40% by echocardiography [19]. Additionally, LVEF by CMR has been shown to be a better predictor of mortality in patient referred for primary prevention ICD when compared to echocardiography [20•].

Cine CMR can acquire cardiac views with high spatial and temporal resolution, which can characterize morphologies that may be unique or of particular importance to specific cardiomyopathies. Hypertrophic cardiomyopathy (HCM) is the most common genetic cardiomyopathy occurring in 1 in 200 to 500 individuals [22]. Sarcomere gene mutations lead to hypertrophy and replacement fibrosis, presenting with heterogeneous phenotypes and variable expression. HCM is diagnosed in adults based on a maximum LV thickness ≥ 15 mm, with a lower cut-off of 13-14 mm in those with family history of HCM or positive genetic testing [23]. The risk of sudden cardiac death (SCD) is correlated with the maximum wall thickness, with 20-year cumulative risk of nearly zero for patients with wall thickness ≤ 19 mm and 40% for those with ≥ 30 mm [24]. There are multiple phenotypic variants of HCM based on location of hypertrophy, septal curvature, and the level of blood flow obstruction if present. Apical HCM is an important variant that is a more common cause of HCM in Asian populations and results in a “spade-like” appearance in the LV cavity (Fig. 1). Apical aneurysms can develop presumably from repetitive wall stress and impaired perfusion leading to subsequent scarring [25]. The aneurysm has a particularly high risk of thromboembolic event, arrhythmias including ventricular tachycardia (VT) and HF. HCM related deaths are threefold greater in patients with apical aneurysms when compared to those without aneurysms [26]. However, around 40% of apical aneurysms are missed by echocardiography but may be identified on CMR [25]. Other morphologic abnormalities such as myocardial crypts, systolic motion of the mitral valve, elongated mitral leaflets, apically displaced papillary muscles, and right ventricular hypertrophy can also be seen with HCM. These findings can help differentiate HCM from other mimics such as glycogen/lysosomal storage diseases, cardiac amyloidosis, or hypertensive heart disease.

Left ventricular non-compaction cardiomyopathy (LVNC) is also easily identified by CMR based on the prominent trabeculae and deep recesses in non-compacted myocardium. LVNC is due to failure of the embryological spongy myocardium to compact after coronary vasculature development [27]. LVNC is a rare cause of HF in adults, but is present in 9.2% of children with a primary cardiomyopathy, placing itself as the third most common cardiomyopathy in the pediatric population after HCM and dilated cardiomyopathy [28]. Diagnosis can be challenging as prominent myocardial trabeculations can be seen in other cardiomyopathies as well as healthy patients. Patient who carry the diagnosis of LVNC tend to have more segmental trabeculations in the inferior, lateral, and apical areas [29]. The most commonly accepted criteria for diagnosis is a ratio > 2.3 for the thickness of noncompacted to compacted myocardium measured at end-diastole by CMR, which has a reported sensitivity and specificity of 86% and 99% [30]. Another proposed method is to measure total non-compacted myocardial mass index and percentage, which has a higher diagnostic accuracy [31]. However, limitations of all the diagnostic criteria for LVNC are well documented [32]. The most common complications of LVNC are HF, ventricular arrhythmias, and thromboembolic events, with the latter being due to stasis of blood in the recesses between the trabeculae. In a study following 106 patients with LVNC, 26% patients died or underwent heart transplant over a 2.9 year follow-up period [33].

Arrhythmogenic cardiomyopathy (ACM) is another diagnosis for which CMR is the modality of choice. ACM was originally known as arrhythmogenic right ventricular cardiomyopathy given that fibro-fatty replacement was primarily occurring in the RV [34]. This contributes to RV dysfunction and ventricular arrhythmias. However, there are other variants with either left ventricular or biventricular involvement [35]. ACM accounts for 10% of unexpected sudden cardiac death based on autopsies, with one third of cases occurring in the fourth decade of life [36]. ACM is diagnosed based on the 2010 Task Force criteria, which incorporates imaging, histology, EKG abnormalities, arrhythmias, and family history [37]. The imaging component requires evidence of RV regional wall motion abnormalities or dyssynchrony and either reduced RV ejection fraction or dilated end-diastolic cavity size [37]. However, these findings can be subtle and difficult to acquire with echocardiography due to complex structure of the RV. As a result, CMR is heavily used to obtain accurate volumetric measurements and evaluate the RV free wall and outflow tract. The proposed 2020 international criteria also include findings for LV phenotypes [38]. CMR is additionally very helpful to differentiate ACM from mimics such as the cardiac sarcoidosis, congenital heart disease, and normal variants such as the “butterfly apex.” This is a normal anatomic variation in which the LV and RV have separate apices in the shape of a butterfly. However, the separate RV apex is often misdiagnosed as an aneurysmal or dyskinetic segment [39].

One of the strengths of contrast CMR is its ability to identify scar in the myocardium. LGE imaging specifically enhances areas of retained gadolinium contrast in the myocardium, which is due to expanded extracellular space from myocyte loss and replacement fibrosis [40]. There is unequivocal evidence that the presence of scar detected using LGE is associated with an increased risk of all-cause mortality, HF hospitalizations, and SCD in patients with NICM [41]. A higher LGE burden is also associated with worse outcomes. In patients with HCM, LGE that is greater than 15% of the LV mass had a threefold increase in SCD and ICD discharge [42]. Although there is no consensus of the best quantification method for LGE in HCM, it is still considered by experts to be a powerful risk stratification tool for SCD. Given the prevalence of scar in akinetic and aneurysmal segments, LGE is also useful for both detecting LV thrombus and identifying patients at risk of subsequent embolic events [43].

The pattern for LGE can be used to distinguish between types of cardiomyopathies (Fig. 2) [44]. For example, the location of LGE within the myocardial wall can easily differentiate between ischemic and non-ischemic cardiomyopathy. In the former, ischemia occurs first in the subendocardium due to reduced perfusion pressure from epicardial coronary disease. As a result, LGE is seen in the subendocardium of a diseased coronary territory and can extend to the epicardium during myocardial infarction. Transmural involvement of the coronary territory suggests poor viability and decreased likelihood of recovery of function after revascularization [45]. In dilated cardiomyopathy, a little less than a third of patients will have mid-wall stripe LGE in the interventricular septum [46], for which the presence and extent is an independent predictor SCD, HF hospitalization, transplant, and death [47]. Sarcoidosis is an inflammatory disorder with multiorgan involvement of noncaseating granulomas. LGE can be seen in commonly involved segments such as the LV basal septum and lateral wall in the epicardium and mid-myocardium. The presence of LGE is associated with a 3.5-fold increase in annualized mortality rate [48]. Cardiac amyloidosis is an infiltrative disorder in which proteins such as immunoglobulin light chains and transthyretin, are deposited in the myocardium and result in expansion of the extracellular space. The 2 most common types are light chain (AL) amyloidosis from plasma cell-dyscrasias or transthyretin amyloidosis (ATTR) from misfolded albumin produced by the liver. LGE is commonly seen as diffuse subendocardial involvement of the base and middle of the left ventricle [49]. In late-stage cardiac amyloidosis, the degree of LGE can be so diffuse that the images are difficult to interpret due to alterations in the inversion time, which is considered pathognomonic the cardiac amyloidosis and a strong predictor of mortality [50]. In chronic Chagas disease, LGE is seen in the apex and inferolateral wall that can be either focal, transmural, or diffuse in extent [51]. RV insertion point LGE is a common but nonspecific finding in patients in patients with HCM and pulmonary hypertension [52, 53].

CMR offers a quantitative method to characterize myocardial tissue using parametric mapping. By measuring the magnetic relaxation times in the longitudinal (T1) and transverse (T2) directions to the static magnetic fields from the scanner, color-encoded maps can be constructed for which each pixel represents a T1 or T2 value (Fig. 3) [54]. This allows the reader to visualize and measure the global and regional T1 and T2 values [55]. T2 elevation is generally considered to be more specific to myocardial edema. In canine models that underwent myocardial infarction, T2 values were shown to strongly correlate with percent water content in the infarct territories and therefore to be an excellent marker of edema [56]. Native T1 values are elevated in any disease process that alters intracellular and extracellular content such as edema, fibrosis, and necrosis [57, 58]. After the administration of gadolinium contrast, T1 relaxation times will shorten predominantly based on volume of distribution of contrast in the extracellular space [59]. Extracellular volume (ECV) maps can be calculated based on the pre- and post-T1 maps of the myocardium and blood pool and then adjusted for the hematocrit [60]. Conceptually, ECV mapping represents the change in tissue T1 compared to plasma T1 with gadolinium contrast and is therefore an indirect measurement interstitial volume fraction of the myocardium. Elevated ECV can reflect expansion of the extracellular space due to diffuse fibrosis [61] or infiltrative processes [62]. A synthetic ECV can be calculated without blood sampling based the T1 of the blood pool, which has been validated in multiple cohorts [63, 64] but can be less accurate in extremes of hematocrits [65]. Although ECV can vary depending on field strength, T1 mapping sequence, and MRI vendor, the normal range is generally more consistent and reproducible than that of T1 and T2 mapping. Additionally, ECV values are reported as fractions, as opposed to millisecond units used for T1 and T2 values.

Parametric mapping sequences are crucial for evaluating inflammation of the myocardium [66]. The Lake Louise Criteria II requires both T1- and T2-based imaging to diagnose acute myocarditis [67]. In the MyoRacer-Trial, patients underwent endomyocardial biopsy and CMR imaging with T1 and T2 mapping for acute and chronic symptoms from suspected myocarditis. T1 and T2 mapping had a diagnostic accuracy of 81% and 80% for acute myocarditis. Interestingly, only T2 mapping was able to diagnose chronic myocarditis with an accuracy of 73%. COVID-19 related myocarditis has also been increasingly recognized by CMR. In a study by Puntmann et al. [68], a total of 100 patients who recovered from mild to moderate COVID-19 underwent CMR 2 to 3 month after their initial COVID test. Surprisingly, 78 of the patients had abnormal CMR findings, which included reduced biventricular function and elevated T1 and T2 times; however, the study has been criticized for the lack of control subjects who did not have COVID-19 [69]. Immune checkpoint inhibitor (ICI) myocarditis is a well-known but rare complication of the immunotherapy, with a reported fatality rate of up to 40% [70]. A large multicenter registry of 136 patients with biopsy proven ICI myocarditis found abnormal T1 and T2 values in 78% and 43% respectively. Furthermore, only native T1 was independently associated with MACE. These findings suggest that there is more myocardial injury than the extent of edema detected using T2 imaging that occurs in ICI myocarditis.

Parametric mapping with CMR is also useful for diagnosing infiltrative diseases. Fabry disease (FD) is rare X-linked lysosomal storage disease that results in the accumulation of glycosphingolipids throughout the body. Its accumulation in the myocardium results in asymmetric septal hypertrophy and eventual systolic dysfunction with inferolateral wall thinning [71]. In early stages of Fabry disease, reduction of native T1 due to glycosphingolipids storage has a high specificity of up to 99% [72]. As the disease progresses, the glycosphingolipids accumulation lead to replacement fibrosis and pseudonormalization of the native T1 [72]. Similar to LGE and focal fibrosis, ECV and diffuse fibrosis provide important diagnostic and prognostic information. In a prospective study by Cadour et al. [73••], 225 patients with non-ischemic dilated cardiomyopathy underwent CMR and were followed for 2 years. They showed that ECV was independent predictor of HF and arrhythmia related events [73••]. In cardiac amyloidosis, pooled analyses have shown that ECV has both a higher diagnostic odds ratio and mortality hazard ratio than LGE [74]. Because the condition is entirely driven by the deposition of amyloid proteins in the extracellular space, cardiac amyloidosis often has the highest ECV of all NICM, nearing levels seen in infarcted myocardium [75]. CMR can also detect iron overload in cardiac siderosis, which occurs in transfusion-dependent anemias or hemochromatosis [76]. T2* relaxation is the decay of transverse magnetization (T2) in the presence of magnetic field inhomogeneity, which can be induced by iron deposition. T2* relaxation time decreases linearly with increasing iron load and predict the development of ventricular dysfunction [77].

With scintigraphy or SPECT, bone tracers can be used to differentiate ATTR and AL cardiac amyloidosis, which have vastly different treatment strategies and survival rates. In the United States, ^99mTc-pyrophoscate (PYP) and ^99mTc-hydroxy-methylene diphosphonate (HMDP) are used off-label to diagnose ATTR cardiac amyloidosis, while ^99mTC-3,3-diphsphono-1,2-pro-panedicarboxylic acid (DPD) is only available in Europe [78, 79]. The mechanisms of the radiotracers remain unclear but is potentially related to the binding of microcalcifications seen in ATTR deposits [80]. The myocardial uptake of radiotracers can be visually scored by comparing it to the uptake in the ribs (grade 0-absent uptake, grade 3-uptake greater than bone) or quantitatively measured as the ratio of the heart to contralateral chest uptake for PYP (Fig. 4). This ratio is not recommended for HMDP due to significantly more background noise that confound the results [79]. For PYP scintigraphy, both methods are comparable with the quantitative ratio having a sensitivity of 97% and specificity of 100% based on a ratio ≥ 1.5 [81]. In a recent study by Delbarre et al. [82], they trained a deep learning model with routine whole-body bone scintigraphy planar images to identify positive studies (visual grade ≥ 2) and achieved an accuracy of 99% on internal and external validation. This offers a potential means for screening patients for ATTR cardiac amyloidosis when undergoing bone scintigraphy for unrelated oncologic or musculoskeletal indications. More recently, it has been recognized that the use of SPECT imaging, especially when combined with CT to localize the location of PYP uptake can significantly improve the diagnostic accuracy of PYP imaging for the detection of ATTR cardiomyopathy [83].

Active cardiac sarcoidosis refers to the episodes of granulomatous inflammation that lead to myocardial injury and eventual fibrosis. ¹⁸F-fluorodeoxyglucose (FDG) PET can detect the active phase, thus allowing for the diagnosis of cardiac sarcoidosis and treatment with immunosuppression. FDG is a glucose analog that targets activated macrophages in the granulomas, which have elevated metabolic rates. Patients must undergo either a fasting or dietary modification protocol in order to suppress consumption of glucose by healthy myocytes and promote the use free fatty acids for energy. The diet consists of the consumption of a high-fat and low-carbohydrate meals [84]. About one fourth of patients fail to successfully achieve FDG myocardial suppression due to incomplete dietary preparation [85], resulting in diffuse homogenous uptake in the myocardium. This technical challenge has encouraged the investigation of new tracers that do not accumulate in myocytes [86]. The current cardiac sarcoid PET protocols also incorporate perfusion imaging to identify microvascular dysfunction or scar in the absences of any known coronary disease. Early cardiac sarcoidosis will show focal uptake of FDG with normal perfusion, suggesting the presence of only inflammation without any scarring. A “mismatch” pattern refers to the focal areas of both FDG uptake and reduced perfusion, which represents active inflammation with myocardial injury (Fig. 5). In late stages of cardiac sarcoidosis, there may be only reduced perfusion without FDG uptake due to burned-out granulomatous tissue and scar formation [87]. A meta-analysis with 33 studies comparing the diagnostic performance of FDG PET and CMR demonstrated sensitivities of 84% and 95% and specificities of 82% and 85% respectively [88]. A few recent studies suggest using hybrid imaging with CMR and FDG PET to better classify active inflammation vs extensive scar [89, 90].

FDG PET also plays an important role in guiding the use of immunosuppression and monitoring response. A study from the Granulomatous Myocarditis Registry found that quantitative FDG uptake and LVEF > 40% was a predictor of complete response to immunosuppression with area under the curve (AUC) of 0.85 [91]. Complete response was defined as improvement in New York Heart Association functional class, freedom from ventricular arrhythmia and HF admission, and improvement of LVEF of ≥ 10%. Interestingly, none of the patients who demonstrated complete response to immunosuppression had residual FDG uptake on subsequent PET scans. However, residual FDG uptake was present in 58% of partial responders and 91% of non-responders, suggesting that PET can assess disease progression and treatment response [91].

FDG PET can also be used as alternative assessment for inflammatory cardiomyopathies including myocarditis. Although FDG PET is not routinely used for myocarditis, it can be advantageous in patients with irregular heart rates or ICDs, which contribute to imaging artifacts on CMR. In fact, parametric mapping values are significantly less reliable due cardiac implantable electronic devices [92]. When compared to endomyocardial biopsy, FDG PET had a modest sensitivity and specificity of 75% and 67% respectively [93]. There have also been a few cases reports demonstrating the use FDG PET combined with either CMR or CCT to further characterize myocarditis [94].

Since the integration of CCT with electroanatomic mapping software, cross-sectional imaging has become a necessity for many electrophysiology procedures. In newly diagnosed atrial fibrillation (AF), HF is the most common complication and cause of death in patients globally [98]. Pulmonary vein isolation (PVI) is a treatment option for AF when anti-arrhythmic medication therapy fails. Prior to PVI, CCT can be used to define pulmonary vein anatomy and identify variants such as common ostium or accessory veins. CCT guided pre-procedural planning allows for optimal selection of catheter sizes and ablation approaches [99]. Additionally, CCT can assess for post-operative complications such as pulmonary vein stenosis and atrio-esophageal fistula. CCT with delayed imaging is an effective modality for excluding left atrial appendage (LAA) thrombus prior to ablation [100, 101], with a sensitivity and negative predictive value of 100% (Fig. 6). This is an attractive option for patients undergoing cardioversion or evaluation for cryptogenic stroke but are not suitable candidates for transesophageal echocardiogram (TEE) [101]. Expert consensus also recommends either preprocedural CCT or TEE to assess the LAA morphology and exclude thrombus prior to implantation of occlusion devices including the Amplatzer Amulet and the Watchman Device [102]. In a subanalysis of the SWISS APERO trial, operators that were unblinded to the pre-procedural CCT had lower radiation exposure, contrast doses, major procedure-related complications, and residual peri-device leaks [103]. CCT has also been used to identify scar with delayed enhancement imaging for ventricular tachycardia ablation [104] and define coronary sinus anatomy for cardiac resynchronization therapy [105].

With increasing life expectancy following heart transplant or left ventricular assist device (LVAD) implantation, CCT has seen an expanded role in evaluating for complications. Coronary computed tomography angiography (CCTA) offers a reliable non-invasive alternative to coronary angiography for early detection of cardiac allograft vasculopathy (CAV). The prevalence of CAV is 20% at 3 years following transplant with a 10% mortality rate after diagnosis [106]. CAV is characterized as the intimal hyperplasia and diffuse concentric luminal narrowing that leads to eventual graft failure. Patients with CAV often have minimal or non-specific symptoms due to denervation of the transplanted heart, eventually presenting with overt HF [107]. According to meta-analysis by Wever-Pinzon et al. [108], CCTA can detect CAV with stenosis ≥ 50% by invasive angiography with high sensitivity and specificity of 94% and 92% respectively. Additionally, 64-slice coronary CTA could detect intimal thickening > 0.5 mm by intravascular ultrasound (IVUS) with sensitivity and specificity of 81% and 75% respectively [108].

With the limited availability of donor hearts, there is growing use of LVAD as destination therapy for end-stage HF. Pump thrombosis and graft obstruction are common mechanical complications of LVAD. Low attenuation filling defects on CCT can represent thrombus in the pump, inflow cannula, or outflow graft (Fig. 7) [109]. In a study with 24 patients with suspected LVAD thrombosis [110], CCT demonstrated a high specificity of 100% for surgically confirmed thrombosis. However, the sensitivity was low given that most thrombi are found in the pump motor, which is not well visualized on any modality [110]. In addition, multiphase CCT can identify dynamic motion of cardiac structures that obstruct the inflow cannula and twisting of outflow graft that results in kinking of the lumen. When CCT is added to echocardiography, the diagnostic accuracy for cardio-mechanical complications of LVAD increases from 41 to 73% [111]. CCT, however, can be limited by metal artifact generated by implanted device, which can obstruct the view of the structures in question. Metal artifact reduction techniques are being developed to reduce beam hardening [112, 113]. FDG PET with CCT is another effective multimodal test that combines metabolic and anatomic findings in patients with suspected device infections including LVADs, prosthetic valves, and cardiac implantable electronic devices [114].