A simplified and reproducible method to size the mitral annulus: implications for transcatheter mitral valve replacement.

Abstract

Transcatheter mitral valve replacement (TMVR) provides definitive valve replacement through a minimally invasive procedure. In the setting of TMVR, it remains unclear how relevant the differences between different mitral annular (MA) diameters are. We sought to define a simplified and reproducible method to describe the MA size. Using cardiac computed tomography angiography (CTA) studies of 47 patients, 3D MA perimeter (P3D) was annotated. The aorto-mitral continuity was excluded from MA contour either by manual annotation (yielding a saddle-shape model) or by simple truncation at the medial and lateral trigones (yielding a D-shape model). The method of the least squares was used to generate the projected MA area (Aproj) and perimeter (Pproj). Intercommissural (IC) and septolateral (SL) diameters, Dmean = (IC diameter + SL diameter)/2, area-derived diameter (DArea = 2 x √(A/π)) and perimeter-derived diameter (DPerimeter = P/π) were measured. MA eccentricity, height, and calcification (MAC) were assessed. Thirty studies were re-read by the same and by another observer to test intra- and inter-observer reproducibility. Patients (age, 75 ± 12 years, 66% males) had a wide range of mitral regurgitation severity (none-trace in 8%, mild in 55%, moderate-severe in 37%), MA size (area: 5-16 cm2), eccentricity (-8-52%), and height (3-11 mm). MAC was seen in 11 cases, in whom MAC arc occupied 26 ± 20% of the MA circumference. DArea (36.0 ± 4.0 mm) and DPerimeter (37.1 ± 3.8 mm) correlated strongly (R2 = 0.97) and were not significantly different (P = 0.15). The IC (39.3 ± 4.6 mm) and the SL (31.4 ± 4.5 mm) diameters were significantly different from DArea (P < 0.001) while Dmean (35.4 ± 4.0 mm) was not (P = 0.5). The correlation of DArea was stronger with Dmean (R2 = 0.96) than with IC and SL diameters (R2 = 0.69 and 0.76, respectively). The average difference between DArea and Dmean was +0.6 mm and the 95% limits of agreement were 2.1 and -0.9 mm. Similar results were found when the D-shape model was applied. All MA diameters showed good reproducibility with high intraclass correlation coefficient (0.93-0.98), small average bias (0.37-1.1 mm), and low coefficient of variation (3-7%) for intra- and inter-observer comparisons. Reproducibility of DArea was lower in patients with MAC. MA sizing by CTA is readily feasible and reproducible. Dmean is a simple index that can be used to infer the effective MA size.