Mitral Annular Disjunction Assessed Using CMR Imaging

Background Mitral annular disjunction is the atrial displacement of the mural mitral valve leaflet hinge point within the atrioventricular junction. Said to be associated with malignant ventricular arrhythmias and sudden death, its prevalence in the general population is not known. Objectives The purpose of this study was to assess the frequency of occurrence and extent of mitral annular disjunction in a large population cohort. Methods The authors assessed the cardiac magnetic resonance (CMR) images in 2,646 Caucasian subjects enrolled in the UK Biobank imaging study, measuring the length of disjunction at 4 points around the mitral annulus, assessing for presence of prolapse or billowing of the leaflets, and for curling motion of the inferolateral left ventricular wall. Results From 2,607 included participants, the authors found disjunction in 1,990 (76%) cases, most commonly at the anterior and inferior ventricular wall. The authors found inferolateral disjunction, reported as clinically important, in 134 (5%) cases. Prolapse was more frequent in subjects with disjunction (odds ratio [OR]: 2.5; P = 0.02), with positive associations found between systolic curling and disjunction at any site (OR: 3.6; P < 0.01), and systolic curling and prolapse (OR: 71.9; P < 0.01). Conclusions This large-scale study shows that disjunction is a common finding when using CMR. Disjunction at the inferolateral ventricular wall, however, was rare. The authors found associations between disjunction and both prolapse and billowing of the mural mitral valve leaflet. These findings support the notion that only extensive inferolateral disjunction, when found, warrants consideration of further investigation, but disjunction elsewhere in the annulus should be considered a normal finding.

S o-called "mitral annular disjunction" is the separation between the left atrial wall, the hinge point of the mural mitral leaflet, and the base of the left ventricular free wall. 1 First described in 1876, 2 and systematically studied in the 1980s, [3][4][5] the finding went largely unnoticed until recently, despite an early report suggesting it might be related to sudden cardiac death. 6 The recent technical advances in echocardiography, and better accessibility of cardiac magnetic resonance (CMR), have now made it easier to observe this entity. In echocardiographic studies, disjunction has mostly been observed and described only adjacent to the inferolateral ventricular wall because this section is best visualized in the parasternal long axis view. A recent study by Dejgaard et al, 7 however, reported on a detailed analysis using CMR in patients with suspected disjunction on echocardiography. They showed that disjunction was usually spread around a larger part of the annulus, being interspersed with normal hinging, concurring with previous histologic findings. 4 A growing body of evidence has suggested that disjunction might play a role in arrhythmic events in patients with [8][9][10][11][12] and without 7 mitral valvar prolapse.
Most of the published studies, however, have been conducted on preselected populations of patients. 10 Thus far, retrospective studies have mostly been based on images obtained from consecutive patients referred for echocardiography. 13,14 A recent study, nonetheless, observed disjunction with computed tomography (CT) in structurally normal hearts. 15 There is, however, a paucity of data on the prevalence and circumferential extent of disjunction in the general population. Indeed, to our knowledge, there have been no studies on disjunction in subjects without clinical indications for CMR. Our aim, therefore, was to assess the prevalence and extent of disjunction in a large cohort with no clinical indication for CMR.
Additionally, we aimed to seek any association between disjunction and prolapse or incident arrhythmias. Such information is essential if we are better to understand and refine approaches to the diagnosis of this feature, and its risk stratification.

METHODS
STUDY POPULATION. In this observational crosssectional study, we analyzed the CMR images from 2,646 Caucasian subjects enrolled between April 2014, and August 2015, in the UK Biobank imaging study. 16 The selection included 804 subjects without any known cardiovascular disease, other serious illnesses, or risk factors for cardiovascular disease, who have previously been selected for the study, which provided the specific reference ranges for chamber quantification. 17 Of these, 35 were later diagnosed with either cardiovascular disease or other illnesses and removed from the healthy cohort. We then made a random selection of 1,842 scans from the remaining 4,261 scans available in the UK Biobank database, which were obtained within the selected time period (Supplemental Figure 1). Although these participants did not fit the strict criteria used in the aforementioned study, it has been shown that the participants are, in general, healthier, leaner, and with lower rate of all-cause mortality and lower total cancer incidence than the UK population taken as a whole. 18   In this regard, it should be noted that, if considered attitudinally, the segment said to be "anterior" would better be described as being "superior," whereas the "anterolateral" segment is posteriorly located when assessed relative to the bodily coordinates. We have retained, nonetheless, these conventional descriptors when denoting the site of measured disjunction.
Disjunction was defined as present when it measured 1 mm or more, observing the consensus statement for CMR. 20 Where disjunction was observed, it was  Fisher exact tests were used to test for relationships between disjunction status and other features.
Multivariable logistic modelling was used to explore these relationships further, adjusting for age, sex, arterial hypertension, and BMI. Correlation analyses were made using Pearson correlations. Two-sided values of P ˂ 0.05 were considered significant. Intraobserver and interobserver measurement reliability was assessed using intraclass correlation coefficient.

RESULTS
STUDY POPULATION. We analyzed visually a total of 2,646 scans. Of these, 39 were discarded due to insufficient quality of any of the long-axis views, retaining the remaining 2,607 datasets (Supplemental  Table 1). Of the selected  Table 1). Of these, just 6 were ventricular arrhythmias, and 4 people survived cardiac arrest for any cause before imaging.  (Table 1). Significant, albeit weak to moderate, positive correlations of co-occurrence were found between sites, with the strongest correlations found for the rarest co-occurrence, which was  We found a strong association between the disjunction and a general increased prevalence of prolapse ( Table 3), particularly of the mural leaflet ( Table 4). Billowing of the mural leaflet, indicating abundant leaflet tissue without prolapse, was also associated with an increased prevalence of disjunction ( Table 4). Intraobserver reliability was excellent, with intraclass correlation coefficient $0.93 for all sites measured.
Interobserver measurement reliability was good to excellent, with the lowest reliability for the inferolateral site, with intraclass correlation coefficient of 0.72 (Supplemental Table 2, Supplemental Figure 2).

DISCUSSION
Our study reveals new insights into the feature described as mitral annular disjunction, adding to overall understanding of the entity, and its occurrence in a large population. As far as we are aware, ours is the largest study to date, and the first to   Values are n/N (%) unless otherwise indicated. There are 2 logistic models represented, one for the odds of prolapse given any disjunction overall, and another model including disjunction indicators across the 4 sites simultaneously. a Fisher exact test for independence between 2 categorical variables. In each case, this is between prolapse (present/ absent) and the disjunction variable listed (present/absent). b Logistic models have prolapse as the outcome (present/absent), disjunction as the exposure, and are adjusted by age, sex, arterial hypertension, and body mass index.  STUDY LIMITATIONS. Our retrospective study had a cross-sectional design. The average age of our subjects when scanned was 61 years, whereas the median reported age for sudden cardiac death in patients with prolapse is 30 years. 33 Our cohort, therefore, represents subjects at lower risk for sudden death due either to prolapse or disjunction. At the same time, our observed prevalence of disjunction suggests a more benign connotation for asymptomatic disjunction.