We explore the use of functional and topological data analysis for the unsupervised stratification of patients with Marfan syndrome, a genetic disorder that can lead to severe cardiovascular complications, including aortic dissection. Our analysis focuses on 3D reconstructions of the aorta derived from CT scans, extracting key features that capture essential geometric properties of the aortic arch, such as centerline curvature and local radius. To stratify Marfan patients, we employ two clustering approaches: one utilizing functional principal component analysis and another based on persistence diagrams. Both methods identify two distinct patient subgroups: one with a smooth, rounded aortic arch and the other with a straighter and tighter profile. Our results align with existing clinical risk assessment practices for Marfan syndrome and may contribute to the development of personalized treatment strategies.