The influence of constitutive law choice used to characterise atherosclerotic tissue material properties on computing stress values in human carotid plaques

Zhongzhao Teng, Jianmin Yuan, Jiaxuan Feng, Yongxue Zhang, Adam Brown, Shuo Wang, Qingsheng Lu, Jonathan H. GillardJournal of Biomechanics 2015 · DOI

TL;DR

This study investigates how different mathematical models (constitutive laws) for describing the mechanical properties of atherosclerotic plaque tissue affect the calculated stress levels within these plaques. The authors found that while most models fit experimental data well, only a subset (Demiray and modified Mooney-Rivlin) proved stable and suitable for complex 3D simulations, suggesting these are better choices for predicting plaque rupture risk.

What Problem Does This Paper Address?

Atherosclerotic plaques in carotid arteries are a major risk factor for stroke. While the degree of artery narrowing (stenosis) is a known indicator, it's not always reliable, especially in mild to moderate cases. Researchers are increasingly looking at the mechanical stresses within plaques as a better predictor of vulnerability and rupture. However, accurately calculating these stresses depends heavily on how the plaque's material properties are modeled. Different mathematical "constitutive laws" or "strain energy density functions" (SEDFs) exist, and it's unclear which ones best represent plaque tissue and lead to reliable stress predictions. This paper aims to bridge this gap by evaluating several common SEDFs for their ability to fit experimental data, their computational stability, and their impact on stress calculations in realistic plaque models.

Methods

The researchers used a multi-pronged approach:

  • Experimental Data Fitting: They collected stress-stretch data from uniaxial extension tests on human carotid plaque tissues (media, fibrous cap, lipid, and intraplaque hemorrhage/thrombus). Seven different SEDFs (neo-Hookean, one-term Ogden, two-term Ogden, Yeoh, 5-parameter Mooney-Rivlin, Demiray, and modified Mooney-Rivlin) were used to fit this experimental data.
  • Semi-Analytic Solution: They used a simplified model of a thick-walled circular tube under internal pressure to assess how SEDFs influence stress distribution in a basic arterial geometry. This provides a baseline comparison.
  • 2D Structure-Only Analysis: They created an idealized 2D model of a stenotic carotid plaque and performed simulations using each SEDF to calculate stress concentrations, particularly in the vulnerable shoulder region.
  • 3D Fluid-Structure Interaction (FSI) Analysis: This is the most complex simulation, modeling the plaque's interaction with blood flow. They used an idealized 3D plaque geometry and performed FSI simulations. This method is computationally intensive and requires stable material models.
  • Material Stability Analysis: They assessed the "material stability" of each SEDF by examining the convexity of their energy contours and analyzing stability curves under different deformation conditions. This is crucial for ensuring that simulations don't produce unphysical results.

Datasets

The study utilized experimental data from 21 human carotid plaque specimens. This data consisted of stress-stretch curves obtained from uniaxial extension tests performed on excised tissue samples of media, fibrous cap, lipid, and intraplaque hemorrhage/thrombus. No new datasets were introduced; rather, existing experimental data was used to evaluate different constitutive models.

Key Insights and Ideas

  • Not all models are created equal: While most SEDFs could reasonably fit the experimental stress-stretch data, this doesn't guarantee their suitability for complex simulations.
  • Computational stability is key: The ability to perform robust 3D FSI simulations was a major differentiator. Only neo-Hookean, Demiray, and modified Mooney-Rivlin models allowed these complex simulations to complete successfully.
  • Material stability matters: SEDFs with non-convex energy contours or negative stability indicators (like one-term Ogden, two-term Ogden, and 5-parameter Mooney-Rivlin) led to simulation failures in the 3D FSI analysis. This suggests that even if a model fits data, it might be mathematically unstable under certain conditions.
  • Demiray and modified Mooney-Rivlin are strong contenders: These two models not only fit experimental data well but also demonstrated good computational stability and produced nearly identical stress results in the simulations.
  • Neo-Hookean is problematic: Despite being computationally stable, the neo-Hookean model showed significant deviations in fitting experimental data and predicted much lower stress concentrations, making it less suitable for accurately assessing plaque vulnerability.

Key Results

  • Data Fitting: All SEDFs except neo-Hookean provided a good fit to the experimental stress-stretch data.
  • Semi-Analytic Solution: Stress distributions were similar across most SEDFs, with neo-Hookean showing some variation.
  • 2D Structure-Only Analysis: All seven SEDFs could be used in 2D simulations, but stress predictions varied significantly, with neo-Hookean yielding the lowest values.
  • 3D FSI Analysis: Only neo-Hookean, Demiray, and modified Mooney-Rivlin models successfully completed the 3D FSI simulations.
  • Material Stability: Energy contours for one-term Ogden, two-term Ogden, and 5-parameter Mooney-Rivlin were not strictly convex, and their stability curves showed negative eigenvalues at certain stretch levels, indicating potential instability.
  • Demiray vs. Modified Mooney-Rivlin: Stresses calculated using these two models in 3D FSI were nearly identical.

Limitations and Future Work

  • Anisotropy: The study did not account for the fiber-oriented nature of atherosclerotic tissues, which could lead to an underestimation of stress levels.
  • Residual Stresses: Residual stresses, which can influence stress distribution, were not modeled, potentially leading to an overestimation of stress.
  • Calcium: The presence of calcium, a common component in plaques, was not included in the idealized models.
  • FSI Convergence: The convergence of FSI simulations is complex and depends on numerous factors (mesh, time steps, etc.), making it difficult to definitively attribute all simulation failures solely to material instability.
  • Numerical Packages: The findings are based on a specific simulation software (Adina), and results might differ with other packages.

The authors suggest that future work should consider anisotropic material properties and residual stresses. They also highlight the ongoing debate about the optimal stress measure (e.g., Stress-P₁ vs. von Mises stress) for characterizing plaque failure.

Summary updated

Abstract

Calculating high stress concentration within carotid atherosclerotic plaques has been shown to be complementary to anatomical features in assessing vulnerability. Reliability of stress calculation may depend on the constitutive laws/strain energy density functions (SEDFs) used to characterize tissue material properties. Different SEDFs, including neo-Hookean, one-/two-term Ogden, Yeoh, 5-parameter Mooney-Rivlin, Demiray and modified Mooney-Rivlin, have been used to describe atherosclerotic tissue behavior. However, the capacity of SEDFs to fit experimental data and the difference in the stress calculation remains unexplored. In this study, seven SEDFs were used to fit the stress-stretch data points of media, fibrous cap, lipid and intraplaque hemorrhage/thrombus obtained from 21 human carotid plaques. Semi-analytic solution, 2D structure-only and 3D fully coupled fluid-structure interaction (FSI) analyses were used to quantify stress using different SEDFs and the related material stability examined. Results show that, except for neo-Hookean, all other six SEDFs fitted the experimental points well, with vessel stress distribution in the circumferential and radial directions being similar. 2D structural-only analysis was successful for all seven SEDFs, but 3D FSI were only possible with neo-Hookean, Demiray and modified Mooney-Rivlin models. Stresses calculated using Demiray and modified Mooney-Rivlin models were nearly identical. Further analyses indicated that the energy contours of one-/two-term Ogden and 5-parameter Mooney-Rivlin models were not strictly convex and the material stability indictors under homogeneous deformations were not always positive. In conclusion, considering the capacity in characterizing material properties and stabilities, Demiray and modified Mooney-Rivlin SEDF appear practical choices for mechanical analyses to predict the critical mechanical conditions within carotid atherosclerotic plaques.
BibTeX
@article{teng2015influence,
  author = {Zhongzhao Teng and Jianmin Yuan and Jiaxuan Feng and Yongxue Zhang and Adam Brown and Shuo Wang and Qingsheng Lu and Jonathan H. Gillard},
  title = {{The influence of constitutive law choice used to characterise atherosclerotic tissue material properties on computing stress values in human carotid plaques}},
  journal = {Journal of Biomechanics},
  year = {2015},
  doi = {10.1016/j.jbiomech.2015.09.023},
}
Figure