Left ventricular shape predicts arrhythmic risk in fibrotic dilated cardiomyopathy

Figures

• 

  Figure 1. Methodology to capture the LV shape. (A) Top: CMR image with segmented myocardium and RV landmark to break the LV symmetry. Bottom: 3D volumetric representation of short-axis contours. (B) An idealized ellipsoid (transparent orange) tailored to fit the length and diameter of the contours (red). (C) Result of fitting process with points of the contours colour coded according to the distance to the nearest epicardial contour (rainbow scale, from blue—0 cm to red—1 cm). 3D, three-dimensional; CMR, cardiac magnetic resonance; LV, left ventricular.

• 

  Figure 2. Modes of LV shape variation. The average patient LV geometry is shown in the middle column, and is shifted by 3 std along Modes 1 and 2, and the arrhythmic mode. The red dot points towards the manually selected RV landmark. LV, left ventricular.

• 

  Figure 3. Spearman correlations between Modes 1 and 2 the arrhythmic mode and baseline variables. Correlations with absolute values ≥0.17 are statistically significant in a two-tailed t-test P < 0.05. M1 is significantly correlated with body surface area, moderate alcohol excess, aldosterone antagonist use, loop diuretic use, CMR LV ejection fraction, CMR LV ejection fraction index, CMR LV mass index, CMR RV ejection fraction, 3D LV model sphericity, and 3D LV model length. M2 is significantly correlated with CMR LV ejection fraction, CMR LV ejection fraction index, CMR LGE volume, 3D model sphericity, and 3D model LV length. The arrhythmic mode (LVAS) is not significantly correlated with any of the baseline variables. 3D, three-dimension; CMR, cardiac magnetic resonance; LGE, late gadolinium enhancement; LV, left ventricular; LVAS, left ventricular arrhythmic shape.

• 

  Figure 4. Detailed anatomical changes associated with the arrhythmic mode of variation. (A) Geometries representing plus (red) or minus (green) 3 std variation along the arrhythmic shape mode. The red sphere marks the direction towards the RV landmark, and the arrows highlight the deformation from the low- to high-risk LV morphology at the base (yellow arrows) and at the septum (red arrows). (B) Bullseye plot representation of the change in wall thickness when moving from the low- to the high-risk LV morphology, illustrating an irregular pattern with wall thinning in the basal lateral and the mid antero-lateral regions, and wall thickening in the septal apex. LV, left ventricular; RV, right ventricular.

• 

  Figure 5. Arrhythmic shape analysis applied to two example patients. (A) LV arrhythmic shape analysis from two example patients, one of whom suffered an arrhythmic event during follow up. MRI images are a selection of each patients short-axis image stack. (B) Key characteristics of the two patients in comparison with each other and the entire cohort. The patient who met the arrhythmic end-point had a higher LVAS score, a greater LV ejection fraction, a lower NYHA class (1 vs. 3), and was younger than the patient who was event-free. LV, left ventricular; LVAS, left ventricular arrhythmic shape; MRI, magnetic resonance imaging; NYHA, New York Heart Association.

• 

  Figure 6. Association of baseline variables with earlier time to arrhythmic event in univariate cox regression. aPer quartile. bPropensity score includes age, sex, diabetes, moderate alcohol excess, NYHA III/IV, beta-blocker use, LV mass index, and LGE volume. cCategorical variables, referent for smoker and ex-smoker is non-smoker, referent for NYHA III/IV is NYHA I/II. ARB, angiotensin receptor blocker; LGE, late gadolinium enhancement; LV, left ventricular; LVAS, left ventricular arrhythmic shape score; NYHA, New York Heart Association. Red = P < 0.05.

• 

  Figure 7. Association of LVAS with arrhythmic events after multivariate adjustment. aPer quartile. LVAS, left ventricular arrhythmic shape score, Red = P < 0.05. ICD and CRT are time-varying covariates. CRT, cardiac resynchronization therapy; ICDs, implantable cardioverter-defibrillators.