Cardiovascular magnetic resonance demonstration of the spectrum of morphological phenotypes and patterns of myocardial scarring in Anderson-Fabry disease

X-linked mutations in the α-galactosidase gene cause Anderson Fabry disease (AFD), a lysosomal storage disorder [1]. Although genetic testing is used to diagnose AFD, cardiovascular magnetic resonance (CMR) is often performed for accurate volumetric and functional analysis in this disease and to characterize the myocardium. It is recognized from the echocardiography literature that AFD may mimic the morphological characteristics of the various subtypes of hypertrophic cardiomyopathy. In addition, a small proportion of AFD patients have no extracardiac manifestations and therefore, they may be misdiagnosed as having hypertrophic cardiomyopathy (HCM) or hypertensive heart disease [2‐8]. Although it is known that AFD can mimic the morphologic manifestations of HCM on echocardiography [5‐7, 9‐13], CMR literature on this subject is lacking. There is limited information in the published literature on the distribution of myocardial late gadolinium enhancement (LGE) in patients with AFD, with enhancement reported principally in the basal inferolateral midwall [14‐16]. We aimed to use the superior myocardial characterization and spatial resolution of CMR to catalogue the full spectrum of LGE patterns and distribution of left ventricular wall thickening seen in this rare disease.

Thirty-nine patients met inclusion criteria (41 patients with confirmed AFD were identified from clinic database, 1 patient had CMR but no LGE sequences due to renal impairment and 1 patient was excluded due to an apical myocardial infarct. Patient characteristics are given in Table 1). Twenty-two (56 %) had wall thickening (Fig. 1). Patients with wall thickening were more likely to be male (68 % vs. 29 %, p = 0.025) and older (49.9 years [44.9–57.0] vs. 38.9 years [30.7–46.5], p = 0.008). The septal to lateral wall ratio ranged from 0.76 to 1.61 (median 1.00 [0.92–1.08], Fig. 2). Two (5 %) had an apical predominant pattern of hypertrophy (Figs. 2 and 3), 3 had ASH morphology (8 %, Fig. 4) and the remainder had concentric wall thickening (n = 17, 44 %, Fig. 5). For statistical analysis, the asymmetric wall thickening group and apical thickening group were collapsed into a single ‘non-concentric wall thickening’ group (n = 5), Fig. 1. Patients with non-concentric wall thickening had significantly greater EDWTmax than those with concentric wall thickening and there were trends towards greater LVMI (p = 0.066) (Figs. 3 and 4, Table 2).

Previous CMR literature on AFD has focused on scar patterns in AFD. Although CMR, with its high spatial resolution, is ideally suited to assess patterns of LV wall thickening [24], to our knowledge, this is the first paper in the CMR literature that has set out to document the spectrum of wall thickening patterns in AFD and by implication, the potential overlap with HCM with which it may be confused. The echocardiographic literature on this subject (summarized in Additional file 2) confirms that AFD can manifest as ASH and apical hypertrophic patterns of wall thickening with variable proportions. In our cohort, we were able to identify subgroups of patients with concentric wall thickening (19/24, 79 %) and non-concentric wall thickening (5/24, 21 %). The non-concentric wall thickening subgroup included patients with asymmetric septal and apical hypertrophic patterns and we found that these patients had greater wall thicknesses and LVMI than patients with concentric wall thickening. In one previously reported cohort, over two thirds of previously undiagnosed AFD patients (ranging in age between 47 and 79 years) had an ASH pattern of wall thickening [7], which is at odds with our own cohort in whom only one quarter demonstrated asymmetric thickening.

Published CMR findings to date have emphasized the typical pattern of myocardial scar seen in AFD [14, 16, 25‐27] (Additional file 1). However, it was our objective to obtain a clear picture of the spectrum of disease expression in this rare condition and we found substantial variability in scar burden and distribution. We found that four fifths (79 %) of patients with pathologic LGE had typical inferolateral scar but that almost two thirds (60 %) of those with inferolateral scar had other LV scar as well. Therefore, in our cohort, a fifth (21 %) of patients with pathological LGE had atypical patterns of scar. Furthermore, patients with non-concentric thickening had more LV scar and more mid and apical LV scar than those with concentric thickening. Previous echocardiography-based research into a small cohort of asymmetric septal hypertrophy patterns in AFD revealed more short term adverse events, which was statistically linked to increased asymmetry and ‘thinning’ of the posterior wall, but there was no myocardial characterization [7]. We found a higher percentage of patients (24 %) with atypical scar distribution than most of the reported cohorts (Additional file 1), emphasizing the importance of including AFD in the differential diagnosis of hypertrophic myocardial disease. Although a recent study has described LGE in myocardial segments outside the typical locations, further detail or correlation with other imaging or clinical parameters was not provided [15]. Our data suggest that patients with atypical patterns of wall thickening have more total scar as well as more apical and mid ventricular scar, whereas most AFD patients have basal inferolateral LGE. Recent studies have identified non-contrast T1 mapping as a highly sensitive and specific early marker of cardiac involvement in AFD [28‐30]. At present, however, this sequence remains in the research arena and was not utilized in evaluation of our retrospective cohort. The presence of LGE remains an important biomarker of disease since it is one of the indicators that permits funded treatment with enzyme replacement therapy.

In addition, we found 5 patients with midwall scar at fibrous-muscular interfaces including valve annuli/LV wall and chordae tendinae/papillary muscle heads respectively, which to our knowledge has not been described in the imaging literature to date. The presence of papillary scar is unsurprising given the prior reports of enlarged papillary muscles and disproportionate contribution of trabeculae and papillary muscles to total LV mass in AFD [15, 31]. We hypothesize these interfaces between the fibrous skeleton of the heart and the LV midwall or mesocardial layer are sites of increased stress. Viewing the mitral annulus as a bucket handle and considering the basal anteroseptum as the least mobile basal segment similar to the hinge of the handle, the basal inferolateral segment (diagonally opposite on the clock face) is the most mobile of the basal segments and likely faces the most junctional stresses transmitted from the fibrous skeleton into the mesocardial layer. Images G-H from Fig. 7 demonstrate continuity of the scar from the mitral annulus into the mesocardial layer/midwall of the basal inferolateral segment. This hypothesis may explain why midwall basal inferolateral scar is the most common pattern of myocardial scar in AFD.

Our analysis was limited by the small sample size, but this is to be expected given of the rarity of the disease, estimated at between 1 in 7000–8000 live births [32]. For the first time in the literature, our findings have linked the CMR phenotype to arrhythmia. Intriguingly, our data raise the possibility of a link between elevated LV mass and ventricular arrhythmia, as is also seen in HCM in which significant hypertrophy predicts adverse events [33]. We found that patients with elevated LV mass had more overall arrhythmia and that greater degrees of LV hypertrophy, were associated with ventricular arrhythmia and sustained ventricular tachycardia. Increased trabecular and papillary muscle volume above the normal range was associated with overall arrhythmia, atrial fibrillation, ventricular arrhythmia and sustained ventricular tachycardia. Given the disproportionate contribution that LV papillary muscle and trabeculations make to LV mass [15], this finding may be a surrogate marker for disease severity in Fabry disease rather than the actual cause of any arrhythmia.

While we found that patients with both ventricular arrhythmia of any duration had more LV scar, our sample size was too small to prove a definite association. Lastly, multivariable logistic regression was not possible due to the small sample size. As non-concentric patterns of wall thickening were associated with increased LV mass and scar, larger multicenter long term outcome studies focusing on AFD patients with these patterns of hypertrophy and apical predominant scar may be helpful in teasing out other risk factors for ventricular arrhythmia.

Our cohort demonstrates that Anderson-Fabry disease has a number of different phenotypic expressions both in extent and location of hypertrophy and pattern of scar. There is consequently a direct overlap between the AFD phenotype and HCM phenotype. It is therefore unsafe to rely on imaging appearances alone when trying to exclude or make the diagnosis of AFD, and genetic testing is still indicated where there is still a reasonable degree of clinical suspicion. Some centers, including ours, offer AFD screening to all potential HCM patients because of the existence of AFD variants with only cardiac manifestations. Concentric thickening and inferolateral mid-myocardial scar are the most common manifestations of AFD, but the spectrum includes cases morphologically identical to apical and ASH subtypes of HCM and these have more apical and mid-ventricular LV scar than cases with concentric thickening. Further research into differences in clinical outcomes between concentric and non-concentric morphologic subtypes is warranted in larger cohorts. Patients with elevated indexed left ventricular mass in our cohort had a greater incidence of ventricular arrhythmia. Further research collaborations should focus on the potential link between LV mass, LV scar burden and distribution, and ventricular arrhythmia in Anderson Fabry Disease.