1350459 - Defining the “Usual” Right Ventricle in Hypoplastic Left Heart Syndrome Patients with Statistical Shape Modeling

Background: Hypoplastic left heart syndrome (HLHS) is a congenital heart disease characterized by incomplete development of the left heart and requires surgical palliation which results in the right ventricle (RV) becoming the systemic ventricle. Beyond postoperative outcomes of single ventricle palliation, there is great interest in determining the functional status of the RV in HLHS as RV dysfunction has been identified as an important risk factor for adverse outcomes in HLHS. Current diagnostic imaging such as cardiac magnetic resonance imaging (cMRI) can risk-stratify HLHS patients, however conventional cMRI measurements (including RV size and function) do not consider the heterogenous morphology of the RV.

Few studies in the literature thus far have utilized shape analysis and statistical shape modeling in this population (1,2). This study utilized statistical shape modeling to determine a mean shape template of the “usual” RV in HLHS, identifying shape modes that may be relevant for guiding therapy.

Methods: This was a retrospective cohort study of HLHS patients utilizing cMRI datsets. 3D end-diastolic models of the RV were created from cine, MR angiography and 3D SSFP datasets using Mimics (Materialise, Leuven, Netherlands). The 3D models were then uniformly re-meshed and rigidly aligned using the iterative closest point technique. The mean shape template was computed using an iterative technique based on the Large Deformation Diffeomorphic Metric Mapping (LDDMM) implemented into the open-source software Deformetrica. Shape modes (deviations from the mean shape template) were computed using the Principal Component Analysis (PCA) technique implemented into in-house MATLAB software.

Results:

Our cohort of 38 HLHS studies (body surface area 1.50±0.44, age 17.0±7.7) were included. The cohort had RV ejection fraction 45±4% and indexed RV end-diastolic volume 115±29 mL/m2. From PCA analysis, 20 independent principal component modes were extracted, with 95% of shape variance achieved by Shape mode 8. Shape mode 4, which was associated with longitudinal elongation of the RV inflow (Fig 1), correlated with indexed RV end-diastolic volume (r=0.361, p=0.026) and RV Cardiac Index (r=0.348, p=0.032) shown in Figure 3. Shape mode 9, which was associated with effacement of the RV inflow and RV apical bulging (Fig 2), correlated with indexed RV end-diastolic volume (r=0.426, p=0.007) and RV Cardiac Index (r=0.361, p=0.025) also shown in Figure 3.

Conclusion: Statistical shape modeling can be applied to HLHS and other forms of systemic RVs. Deriving a mean shape template of the “usual” RV” and shape mode variations may serve as advanced biomarkers that can be clinically relevant and help determine RV dysfunction in HLHS beyond traditional 2D or 3D-based metrics.