What is needed to make low-density lipoprotein transport in human aorta computational models suitable to explore links to atherosclerosis? Impact of initial and inflow boundary conditions

Personalized computational hemodynamics (CH) is a promising tool to clarify/predict the link between low density lipoproteins (LDL) transport in aorta, disturbed shear and atherogenesis. However, CH uses simplifying assumptions that represent sources of uncertainty. In particular, modelling blood-side to wall LDL transfer is challenged by the cumbersomeness of protocols needed to obtain reliable LDL concentration profile estimations. This paucity of data is limiting the establishment of rigorous CH protocols able to balance the trade-offs among the variety of in vivo data to be acquired, and the accuracy required by biological/clinical applications. In this study, we analyze the impact of LDL concentration initialization (initial conditions, ICs) and inflow boundary conditions (BCs) on CH models of LDL blood-to-wall transfer in aorta. Technically, in an image-based model of human aorta, two different inflow BCs are generated imposing subject-specific inflow 3D PC-MRI measured or idealized (flat) velocity profiles. For each simulated BC, four different ICs for LDL concentration are applied, imposing as IC the LDL distribution resulting from steady-state simulations with average conditions, or constant LDL concentration values. Based on CH results, we conclude that: (1) the imposition of realistic 3D velocity profiles as inflow BC reduces the uncertainty affecting the representation of LDL transfer; (2) different LDL concentration ICs lead to markedly different patterns of LDL transfer. Given that it is not possible to verify in vivo the proper LDL concentration initialization to be applied, we suggest to carefully set and unambiguously declare the imposed BCs and LDL concentration IC when modelling LDL transfer in aorta, in order to obtain reproducible and ultimately comparable results among different laboratories.