In our study, semiautomated LGE CMR semiautomated 5-SD gray-scale threshold technique in percentage revealed the best correlation with the visual assessment analysis, and an optimal reproducibility in patients with MVP. We demonstrated for the first time that semi-quantitative the 5-SD gray-scale technique is a reliable and reproducible method to accurately quantify LGE in MVP patients, thus providing a standardized approach to arrhythmic risk stratification. Finally, differentiating MVP patients on the basis of presence or absence of ventricular arrhythmias, in arrhythmic MVP patients, only 5-SD measurements in percentage demonstrated a good correlation with LGE quantification visually assessed.
Quantification of LGE in MVP
CMR has been shown to accurately recognize myocardial LV macroscopic fibrosis with good correlation with pathology, both for ischemic and non-ischemic diseases [
24‐
26]. Due to its excellent spatial resolution and high contrast-to-noise ratio, CMR clearly identifies LV macroscopic fibrosis as an evident bright high signal intensity area, visibly different from the dark normal myocardium, on T1 inversion recovery post-contrast images. In addition, on the basis of the LGE localization, CMR could differentiate ischemic cardiomyopathy, characterized by subendocardial to transmural LGE distribution, from the non-ischemic one, presenting with intramural or subepicardial LGE localization [
25‐
27].
Overall, fibrosis is common in MVP (28–37%), [
6,
7] usually located close to the annulus in the basal left ventricular wall, moreover only LGE within the mitral apparatus (papillary muscles and peri-annular region) has a clear pathophysiological association with arrhythmia. In fact, myocardial LV macroscopic fibrosis has a pivotal role in ventricular arrhythmogenesis, impacting on the prognosis of many cardiomyopathies [
28‐
31].
In previous studies, different semi-automated techniques for LGE have been evaluated in various clinical settings, leading to variable amounts of LGE. Therefore, these semi-automated techniques may not be interchangeable, but should rather be tailored to LGE pattern and distribution and can be used as a validation tool.
Different studies showed demonstrated LGE threshold or FWHM techniques to best correlate to gross examination of LGE in various pathology. Ischemic LGE is characterized by a dense and homogeneous bright infarct core surrounded by a peri-infarct border zone, responsible for a high contrast area compared with the normal myocardium. Bondarenko et al. demonstrated a good correlation between LGE quantification with visual assessment and the 5-SD grey-scale threshold technique in patients with chronic ischemic heart disease [
32]. According to this study, Vermes et al. confirmed that the use of low threshold considerably overestimated the real infarct size when compared to visual assessment [
33].
Another paradigmatic cardiac disease in which LGE quantification is crucial for arrhythmic stratification, as supported by recent Guidelines [
34], is represented by HCM in which has been demonstrated that the 6-SD gray-scale threshold technique presents the closest approximation with visual assessment and the best reproducibility [
9,
16].
Although the pivotal role of LGE in ventricular arrhythmogenesis and prognosis in MVP patients has been widely demonstrated [
35], a standardized method of LGE quantification has yet to be developed [
2,
3]. Moreover, the semi-quantitative methods evaluated in other cardiomyopathies cannot be applied to arrhythmic MVP patients because the LGE pattern distribution and localization in this clinical setting are peculiar. The pathogenesis of myocardial fibrosis in arrhythmic MVP is complex and not really completely clarified. In our population, MVP patients with LGE demonstrated a greater mitral prolapse distance in the left atrium, a larger systo-diastolic variation of the mitral annular diameter, as well as a more severe MAD and curling, in comparison with patients without fibrosis (Table
1). Accordingly, arrhythmic MVP patients presented a mid-wall or subepicardial LGE stria in the basal or mid (in the site of papillary muscles attachment) inferolateral wall [
3]. Our previous study [
2] revealed that fibrosis in MVP patients is due to a replacement-type, but also to an interstitial fibrosis, leading to a different concentration of LGE. As well as for HCM in which LGE isn’t a synonymous of fibrosis, a specific LGE quantification method should be considered for MVP patients. Supporting this hypothesis, in our population the LGE amount, both in percentage and in grams, resulted slighter than in MI, HCM and suspected myocarditis evaluated in previous studies [
10,
16,
17].
In accordance to prior studies, we measured the greatest amount of LGE with the 2-SD method and the lowest one with the 5-SD method, both in percentage and grams [
9,
17] (Table
2; Fig.
3). Therefore, we used visual assessment, representing the human thresholding in bright signal identification, as a comparison method with the different semi-automated techniques (Table
4). In comparison with visual assessment, only the 5-SD quantification method in percentage didn’t significantly differ from the used gold standard (p = 0.543), reporting the least difference with the visual assessment (bias−0.04, Table
5). Intriguingly, the 5-SD quantification in grams showed a not neglectable difference with the visual assessment quantification (p = 0.005; bias−0.11, Table
5). This result could be explained by two phenomena. First, we identified a smaller LGE amount in MVP patients than other cardiomyopathies, that could minimize the differences between the used methods in percentage and grams. Secondly, the enrolled population is highly selected representing the true “malignant MVP” without the bias due to hemodynamic impairment since those with valve regurgitation were excluded. In fact, we excluded from the study all the MVP patients with other possible causes of LV fibrosis (such as moderate-to-severe MR or LV systolic dysfunction), different from the replacement-type, in order to quantify the real amount of LGE in MVP. Similarly, the other CMR semi-automated quantification methods revealed a significant difference with the visual assessment (p < 0.001).
In addition, we noticed that the use of a low gray-scale thresholding was associated with a great difference in the fibrosis amount compared with the visual assessment, responsible for a noteworthy overestimation of LGE (Fig.
4; Table
5), as previously reported in HCM patients [
9,
16].
In comparison to the low gray-scale thresholding methods (the 2-SD and the 3-SD both in percentage and in grams), in our study FWHM demonstrated the least difference with the visual assessment (in comparison with visual assessment, respectively bias in percentage and in grams, 1.15 and 0.60, Table
5). However, despite it was more accurate than the low gray-scale thresholding methods, FWHM tended to overestimate the LGE amount when compared with the 5-SD (respectively, bias in percentage 1.15 vs. − 0.04 and in grams 0.60 vs. − 0.11, Table
5). This result seems to be in contrast with Flett AS et al. that demonstrated that FWHM technique represented the most accurate and reproducible LGE quantification method, regardless the disease [
10]. In our case, this difference could be probably due to the different LGE pattern observed in MVP patients in comparison to MI and HCM included in the study of Flett AS et al. [
10].
Intra and inter-operator agreement demonstrated a good reproducibility (in all used methods, ICC > 0.8). In particular, the visual assessment presented the best reproducibility, both in percentage and in grams. After the used gold standard, confirming the previous results, the 5-SD revealed an optimal concordance, both in percentage and in grams.
Nevertheless, due to the role of fibrosis in the natural history of arrhythmic MVP, prospective studies are necessary to evaluate the possible LGE remodeling in this setting.
LGE measurements and arrhythmic MVP
Until about ten years ago, the risk of SCD was overt for MVP with severe regurgitation, only subsequently the extensive use of CMR a definite “malignant MVP” phenotype (beyond the valve incompetence), previously seen only by autoptic studies, has been recognized.
Morpho-functional anatomy of mitral valve, in term of severe myxomatous degeneration, MAD, replacement-type fibrosis, appears to play a crucial role in arrhythmic risk. This clinic-instrumental profile recognizes the LGE on CMR as a gatekeeper for arrhythmic risk stratification as highlighted by the recent EHRA Expert Consensus Statement [
34]. Despite the increasing number of papers confirming the presence of LGE in MVP there is no standardization in the postprocessing of methods of delineating LGE, limiting comparison of results among different studies and sites [
2,
3,
5,
6,
8,
35‐
37]. The postprocessing for LGE quantification has been standardized for several and different cardiac diseases [
38] in order to provide not only a standardization for clinical purpose, but also for the research aims.
In the subset of malignant MVP data are limited to small study population, sometimes including also different grades of valve regurgitation, and multiple different methods of delineating LGE extent and defining the presence and extent of MAD further increase data heterogeneity [
8]. Our study represents the application of different quantitative CMR tools for LGE assessment in a specific MVP population. The results indicate that the semiautomated 5-SD gray-scale threshold technique in percentage represents the more suitable methods in this setting, especially in those with ventricular arrhythmias. The question regarding the role of LGE in arrhythmogenesis in MVP, if it matters more the presence/absence or the total amount, remain to be elucidate, after the standardization of postprocessing protocol.