Effect of blood rheology and hematocrit on the estimation of hemodynamic parameters in hypertrophic cardiomyopathy.

Background: Hemodynamic parameters estimated from 4DFlow MR images such Wall Shear Stresses (WSS), Oscillatory Shear Indexes (OSI), and viscous Energy Loss (ELoss) have been thoroughly evaluated in large vessels^1-4. These parameters depend on the viscosity μ, which is assumed to be constant (range of 0.03 to 0.045 poise). However, blood viscosity has an intrinsic shear thinning nonlinear behavior, which depends mainly on the hematocrit percentage (Hct) and can be effectively described using the power-law model^5,6. We aim to investigate the influence of blood rheology and hematocrit (Hct) percentage on the estimation of WSS, ELoss and OSI, in patients with obstructive (HOCM) and non-obstructive (HNCM) hypertrophic cardiomyopathy, using the power-law model.

Methods: 43 patients (28 HOCM and 15 HNCM) with Gd enhanced triple VENC 4DFlow MR images were retrospective acquired (20 males, mean age 65 ± 16 years and mean weight 63 ± 12 Kg) using a 3.0T Achieva scanner (Philips Healthcare, Best, The Netherlands) ⁷. The imaging parameters included voxel size of 1.7 x 1.7 x 2.0 mm, VENCs of 50, 150, and 450 cm/s, temporal resolution of 40 ms, and 15 to 21 frames depending on heart rate. Each patient had a blood test with the Hct measurement (40.5 ± 4.08). The acquisition was conducted at Nippon Medical School Hospital, with the study approved by the Institutional Review Board. All subjects provided written informed consent.

The hemodynamics parameters quantifications (WSS_cv, OSI_cv, ELoss_cv for constant viscosity and WSS_vv, OSI_vv and ELoss_vv for variable viscosity using power-law^5,6) were estimated using the 4D Flow Matlab Toolbox⁸, which employs a previously validated finite element scheme^9,1 (Figure 1).

Two observers (JS and CV) conducted an intra- and inter-observer analysis. Normality of the data was evaluated using one-sample Kolmogorov-Smirnov test. Significant differences were calculated using Wilcoxon signed-rank test for dependent samples with a significance limit of 0.05.

Results: Intra and inter-observer variability for velocity and each hemodynamic parameter was shown in Figure 2. Most of the significant differences are in the interobserver analysis (observer 1.1 and observer 2.1) for WSS. In Figure 3 (A) shown the differences between the use of constant viscosity and variable viscosity, for WSS, OSI and ELoss, in each section of the aorta at peak systolic cardiac phase. The WSS was less sensitive to variable viscosity (VV) and similar patterns can be found in figure 3 (B). In figure 3 (C) we show a 3D map of each hemodynamics parameter for constant viscosity (CV) and VV, using one representative HOCM and HNCM patient.

Conclusion: The present study evaluated the effect of non-linear blood viscosity models on key hemodynamic parameters derived from 4DFlow MRI in patients with HCM. Our findings indicate that the use of a VV model yields significant differences in estimated hemodynamic parameters when compared to the CV assumption traditionally employed.

Figure 1. Steps of the proposed quantification process. A) From the 4DFlow MRI, we generate a semiautomatic segmentation of the vessel of interest, then a tetrahedral finite element mesh was created from the vessel segmentation, and finally the velocity vector field was transferred from the 4D flow MRI to the mesh using a linear interpolation method. B) The WSS, OSI and ELoss, were calculated for constant and variable viscosity. C) the statistical analysis was performed in three different sections: Section 1 = ascending aorta (AAo), Section 2 = aortic arch (AArch) and Section 3 = proximal descending aorta (pDAo).
Figure 2. Intra and inter observer box chart analysis. In this experiment we evaluate the velocity and each hemodynamic parameter (WSS, OSI and ELoss) calculated for both observers (JS is the observer 1.1-1.2 and CV is the observer 2.1), in each section (AAo, AArch, and pDAo). Marks (*) indicate significant differences.
Figure 3. Peak systolic analysis. In A) we compare each hemodynamics parameter using constant viscosity (CV) and variable viscosity (VV), in each analyzed section. In B) we compare each hemodynamics parameter using CV and VV between groups of obstructive and nonobstructive patients, in each analyzed section. In C) we show a 3D map of each hemodynamics parameter analyzed in this study, for CV and VV, using one representative obstructive and nonobstructive patient. Marks (*) in A) and B) indicate significant differences.