Abstract
The thoracic aorta stores strain energy during systole and returns it during diastole to sustain steady organ perfusion in a phenomenon known as the Windkessel function. Elastic fibers, assembled during a narrow developmental window and never replaced, are the structural basis of this function. When elastic fiber integrity is compromised by genetic mutation, aging, or an acquired insult, the extracellular matrix (ECM) reorganizes: load shifts from elastin to collagen and ground substance, glycosaminoglycans (GAGs) and proteoglycans accumulate, and the wall stiffens. That same reorganization simultaneously alters how fluids and solutes move through the wall. Yet the mechanical and transport consequences of elastic fiber deficiency have been studied largely in parallel, with no framework connecting them through their shared dependence on ECM composition. Current clinical assessment relies on aortic diameter which does not reliably predict adverse outcomes. This dissertation bridges the mechanical and transport consequences of elastic fiber loss and develops experimental and computational tools to extract functional signatures of wall remodeling that structural assessment alone cannot provide. Chapter 1 establishes the biological and physical foundations: elastic fiber assembly and its disruption in heritable and acquired vascular disease, the mouse models that isolate distinct failure modes, the constitutive modeling frameworks used to characterize arterial mechanics, and the biphasic and multiphasic mixture theories that govern transmural transport. It identifies three gaps: whether the conventional first-invariant (I1) ground substance model captures the mechanical consequences of elastic fiber loss, whether intrinsic transport properties can be separated from the confounding effects of wall geometry and deformation, and whether the mechanical and transport signatures can be unified through their shared ECM basis. Chapter 2 addresses the first gap through constitutive modeling of biaxial mechanical data. Section 2.1 applies the unified fiber distribution (UFD) model to ascending aortas and carotid arteries from Eln+/- and wild-type mice at 6 and 24 months of age in both sexes. The results reveal that fiber coupling mode is determined by artery type rather than genotype or disease state, that the ascending aorta preserves near-homeostatic strain energy despite ~40% less elastin while the carotid does not, and that sex modulates ECM-material constant relationships. A central finding is that the Neo-Hookean ground substance parameter c is unchanged in the Eln+/- ascending aorta despite reduced elastin-raising the question of whether I1 dependence adequately captures the non-collagenous matrix response. Section 2.2 addresses that question directly with a data-driven discovery pipeline, Beam Search Variable Projection Sparse Identification of Nonlinear Constitutive Models (BSVP-SINC), applied to 387 vessels spanning Marfan (Fbn1mgR/mgR), elastin-haploinsufficient (Eln+/-), and neonatal knockout cohorts. The algorithm selects the constitutive form from a 24-term library of invariant-based candidates without prior constraint. All five top-ranked three-term equations contain the second strain invariant I2. The selected equation is an exponential I1 term, a linear I2 term, and an exponential anisotropic I5 term with six free parameters-achieves higher median R2 (0.971) than the four-fiber Holzapfel-Gasser-Ogden model (0.969, 8 parameters) and the two-fiber HGO model (0.953, 4 parameters) across 328 aortic vessels. The I2 coefficient xi2 increases with disease severity and matrix accumulation, while the I1 coefficient xi1 decreases with loss of elastic fiber integrity, producing a quantitative xi1/xi2 inversion that serves as a mechanical signature of ECM remodeling invisible to I1-only models. Chapter 3 addresses the second gap through transmural transport characterization. Section 3.1 develops biphasic and multiphasic finite element models in FEBio to isolate intrinsic hydraulic permeability (k0) and effective diffusivity (deff) from geometric and mechanical confounders across a graded-severity spectrum of thoracic aortic aneurysm in the Fbln4E57K/E57K mouse. Parameter sensitivity analysis confirms that k0 and deff are the dominant determinants of fluid and solute flux, respectively. Fitted k0 decreases with aneurysm severity, while deff shows a non-monotonic pattern: elevated in non-aneurysmal mutants and returning to wild-type levels in aneurysmal tissue. This pattern reflects competing ECM remodeling processes-elastic fiber fragmentation enlarges pore spaces while proteoglycan accumulation fills them-and maps onto the two-phase temporal model of Marfan aortopathy. Section 3.2 extends the transport characterization in two directions. New experimental measurements in the Fbn1mgR/mgR mouse demonstrate that the aortic wall is intrinsically size-selective (solute flux decreases with molecular weight while fluid flux is unchanged), that Fbn1mgR/mgR aortas show elevated solute transport relative to controls despite little change in fluid flux, and that fluid flux correlates negatively with aortic diameter (R2 = 0.89, p = 0.0005) while solute flux does not-reproducing the transport phenotype identified in the Fbln4 system in a second genetic model. A tandem machine-learning architecture comprising a forward DeepONet surrogate, a virtual field sensor, and a regime-aware inverse operator recovers coupled transport parameters (k0, deff, solid volume fraction phi0, and partition coefficient kappa) simultaneously from boundary-accessible measurements on held-out synthetic data, with test-set R2 values of 0.82-0.98 across parameters and molecular weights. Chapter 4 synthesizes the mechanical and transport findings through their shared dependence on aortic remodeling. The I2-dependent isotropic term discovered in Chapter 2 becomes more prominent as elastic fiber integrity is lost and the non-collagenous matrix is transformed and this same transformation alters hydraulic permeability and solute diffusivity in Chapter 3. The dissertation argues that elastic fiber loss triggers a broader ECM reorganization whose functional consequences appear in both how the wall stores strain energy and how it admits, retains, and excludes molecules. These functional signatures are not captured by diameter or histopathology alone. The tools developed here-data-driven constitutive discovery for mechanics and tandem neural-operator inversion for transport-provide a computational foundation for extracting those signatures from experimental and, ultimately, clinical data.
Committee Chair
Jessica Wagenseil
Committee Members
Carmen Halabi; Guy Genin; Lori Setton; Matthew Bersi; Mohamed Zayed
Degree
Doctor of Philosophy (PhD)
Author's Department
Biomedical Engineering
Document Type
Dissertation
Date of Award
4-29-2026
Language
English (en)
DOI
https://doi.org/10.7936/ay8t-pj74
Recommended Citation
Kailash, Keshav, "Mechanics and Transport Across the Thoracic Aortic Wall: Experimental and Computational Characterization of Elastic Fiber-Deficient Arteries" (2026). McKelvey School of Engineering Graduate Student Theses & Dissertations. 1376.
The definitive version is available at https://doi.org/10.7936/ay8t-pj74