Linking shape and clinical outcomes: Statistical shape model of Bidirectional Glenn connection in congenital single ventricle patients

In patients with congenital single ventricle physiology, the Bidirectional Glenn procedure is a critical intermediate step toward Fontan circulation. Despite technical advances, morbidity and mortality remain significant. While left pulmonary artery (LPA) stenosis is a recognized complication, the broader impact of 3D anatomical variations on outcomes remains poorly understood. We aimed to characterize morphological variability of the superior cavopulmonary connection (SCPC) following the Bidirectional Glenn and assess its relationship with clinical outcomes. Statistical Shape Modeling (SSM) was applied to cardiovascular magnetic resonance images from 29 single ventricle patients following Bidirectional Glenn, whose SCPC showed no clinically important SCPC or pulmonary arterial stenosis. A population-based 3D anatomical template was generated using the Deformetrica framework. Patient-specific shape deformations were quantified and correlated with clinical metrics—including SCPC pressure, oxygen saturation, intensive care unit (ICU) stay, and total hospital stay—via Partial Least Squares (PLS) regression. Associations were compared to correlations with conventional morphological parameters. SSM revealed 3D shape features significantly associated with SCPC pressure and ICU stay (p < 0.01). Lower SCPC pressure correlated with straighter, wider LPA geometry. In contrast, narrower—but without overt stenosis—LPA morphologies were associated with worse outcomes. Conventional geometric metrics did not show such correlations. No association was observed between SCPC shape and total hospital stay. This study provides a novel 3D shape-based characterization of SCPC anatomy in single ventricle patients post Bidirectional Glenn surgery and its link to outcomes. SSM-derived morphological biomarkers offer potential for improved risk stratification and surgical planning in congenital heart disease.