P2433Local blood viscosity and local Reynolds number are associated with coronary plaque calcium and lipid

Abstract

Abstract Background Despite being a shear-thinning non-Newtonian fluid, most computational fluid dynamic (CFD) simulations assume blood to be a Newtonian fluid with constant viscosity. The use of more realistic assumptions may deepen mechanistic understanding of the relationship between blood flow disturbances and atherosclerosis, and improve the diagnostic accuracy of CFD simulations. Purpose To compare associations between plaque composition and local hemodynamics at a single time point using Newtonian versus non-Newtonian rheological models in patient-specific coronary arteries. To investigate whether viscosity-based local haemodynamic indices correlate with plaque composition. Methods Sixteen patient-specific coronary arteries containing non-culprit plaques were reconstructed from optical coherence tomography imaging. CFD simulations using Newtonian and non-Newtonian models were performed to calculate endothelial shear stress (ESS). Local blood viscosity (LBV) and local Reynolds number (ReL) were calculated from non-Newtonian simulation data. Each haemodynamic index was distributed into quintiles, mapped in 5-degree sectors, and compared to plaque composition using logistic regression. Results In total, 69120 sectors from 960 OCT frames were analysed. The lowest ESS quintiles were associated with underlying lipid (ESS<0.8Pa: odds ratio [OR] 1.26, p<0.001, 95% CI 1.15–1.38; ESS 0.8–1.1Pa: OR 1.71, p<0.001, 95% CI 1.58–1.85), while the highest quintile of ESS (>2.2Pa) had lower odds of underlying lipid (OR 0.89, p=0.015, 95% CI 0.82–0.98) compared to the median ESS quintile. However, in the non-Newtonian results, only the second lowest quintile of ESS (1.1–1.5Pa) was associated with lipid (OR 1.54, p<0.001, 95% CI 1.42–1.67). Low ReL was associated with lipid (ReL<28: OR 1.71, p<0.001, 95% CI 1.55–1.89; ReL 28–38: OR 1.47, p<0.001, 95% CI 1.35–1.58). Conversely, the highest quintile of ReL had decreased odds of lipid (ReL>68: OR 0.69, p<0.001, 95% CI 0.62–0.76) (FIGURE). In both the Newtonian and non-Newtonian results, lower ESS was associated with increased odds of underlying calcium. Whereas the lowest quintile of LBV had a lower odds of calcium (LBV<1.4: OR 0.60, p<0.001, 95% CI 0.52–0.71), the highest quintile had significantly higher odds of underlying calcium (LBV>1.5: OR 1.38, p<0.001, 95% CI 1.18–1.63) Conclusions Using the standard Newtonian assumption, low ESS is associated with underlying lipid. However, using a more realistic non-Newtonian rheological model, there is no strong or consistent relationship between ESS and underlying lipid, highlighting the importance of methodological assumptions and lingering questions in arterial CFD simulation. Non-Newtonian indices LBV and ReL are independently associated with calcium and lipid, respectively, suggesting possible mechanistic effects of local blood viscosity in atherosclerosis and implying their use as novel haemodynamic markers of atherosclerosis.