Discussion
This study provides a quantitative description of morphological changes in CA in a large, consecutive population referred for CMR with the clinical suspicion of myocardial amyloid infiltration. This has been previously evaluated using echocardiography, which has limitations in ventricular volume and mass quantification. The scarce available data specifically for CMR was reported in a small case-control study that found no differences in LV wall thickness between CA (n=16) and concentric hypertrophic cardiomyopathy.
As expected, patients with CA often had increased wall thickness and LV mass. The presence of concentric LVH is considered one of the hallmarks in CA and, accordingly, we found this to be the most common type of remodelling pattern. However, one of the main findings is that 42% of the patients showed an alternative pattern: 18% eccentric hypertrophy, 16% concentric remodelling, and 8% normal geometry. Thus, the absence of concentric LVH on imaging does not reliably rule out myocardial amyloid infiltration.
In a detailed analysis of segmental wall thickness distribution, the maximal thickness was located in the basal anteroseptal segment in 99% of individuals without CA, while it was found in other locations, most commonly the mid-inferoseptal segment, in 33% of patients with a positive diagnosis. Interestingly, this pattern mimics, to some extent, the counter-clockwise spiral trajectory of increased wall thickness recently described in asymmetric hypertrophic cardiomyopathy. An older investigation in a large series of patients with cardiac AL amyloidosis reported a prevalence of 8% of asymmetric septal hypertrophy on echocardiography. Another important finding in our study is that the presence of asymmetry was much higher (69%). Interestingly, a recent histological study of CA described a segmental form with predominance in the septum. As mentioned above and demonstrated in figure 4, this pattern resembles that seen in asymmetric hypertrophic cardiomyopathy, so it is imperative for imagers and clinicians to be aware that asymmetric LVH is still consistent with CA, particularly in the appropriate clinical scenario. Moreover, asymmetric LVH was independently associated with the diagnosis of CA in our patient population.
We believe that the discrepancy between our findings and the classical notion that CA leads to diffusely increased wall thickness may be due to an earlier disease stage in our patients. There are two indirect pieces of evidence to support this hypothesis. First, asymmetric LVH was associated with CA specifically in those patients with normal LV mass. This fact also explains the lack of association of asymmetric LVH with the diagnosis in the EMB subgroup, who presented a higher LV mass, together with the limited number of such patients. In this regard, it is important to note that we defined asymmetric LVH based on conventional imaging criteria (presence of increased wall thickness and asymmetry), not on LV mass criteria. Thus, the coexistence of asymmetric LVH with other patterns of remodelling is possible (see online supplementary table 3 http://heart.bmj.com/content/100/21/1688/suppl/DC1). Secondly, asymmetric LVH was less frequently associated with low voltage. Overall, these findings suggest that amyloid deposition might occur preferentially in the interventricular septum in early disease stages leading to asymmetric LV wall thickening with possibly normal LV mass, and as disease progresses concentric LVH develops (more frequently associated with low voltage). Serial CMR studies for early diagnosis and follow-up could be helpful to clarify this issue.
Asymmetric LVH was independently associated with the diagnosis of CA in the present study, and also a model containing asymmetric LVH, RWTand LV mass index showed good diagnostic performance. Even though LGE is a more accurate approach, kidney involvement in AL amyloidosis is frequent (up to 74%), but it is rare in transthyretin-related amyloidosis, so it may constitute a limitation for the use of gadolinium-based contrast agents in the former group. Assessment of the patterns of LV remodelling with CMR could be of particular relevance in this patient population. It is important, however, to note that these findings are useful in the differentiation of present versus absent CA in patients with high clinical suspicion. These criteria and the model we tested are not meant to differentiate CA from other disorders that may share a similar phenotype such hypertrophic cardiomyopathy. In such cases, the presence of distinct patterns of LGE is needed to reach a diagnosis with CMR. Additionally, new amyloid-specific scintigraphic techniques may be useful for diagnosis and characterisation. 99mTc-3,3-diphosphono-1 ,2-propanodicarboxylic acid (99mTc-DPD) scintigraphy allows for early diagnosis of transthyretin-related amyloidosis with high accuracy (100% sensitivity and 83% specificity against echocardiography,) as well as differentiation from other CA forms. Thus, multimodality imaging may play an important role in the evaluation of amyloidosis.
When traditional (echocardiography and ECG) diagnostic criteria were evaluated in our series, they showed high specificity but low sensitivity, and therefore, poor negative predictive value. Differences in performance when compared with prior reports could again be explained by more advanced cardiac involvement in their patients. Additionally, these studies used echocardiography, that relies on several anatomical assumptions to calculate LV mass, whereas, CMR allows for more precise and reproducible quantification. Consequently, our results suggest that traditional criteria have limited use for the detection of CA, probably more so in its early stages, and overreliance on them might delay diagnosis.
Study Limitations
EMB was not always performed, and in the cases without pathology we used the presence of a typical LGE pattern for diagnosis. Although the true sensitivity of LGE for detecting CA is uncertain, several studies have demonstrated excellent specificity when compared to EMB, so we believe that the possibility of a false positive diagnosis is minute. Conversely, some patients with CA and atypical patterns of LGE may have been included in the non-CA group as false negatives, but this would have reduced the differences between the two groups and would be unlikely to explain our findings. We performed a comparison of LV remodelling patterns according to amyloid subtype (Table 3) which confirmed higher LV wall thickness and lower ejection fraction in transthyretin CA. The absence of significant differences in remodelling patterns should not be interpreted as equivalence among disease subtypes, given the limited power of these analyses. This is a retrospective study, and thus subject to the potential limitations of such design. Finally, our study was performed in a tertiary referral centre, and our results may not be extrapolated to populations with lower clinical suspicion or prevalence of disease.