Assessment of sex differences in ventricular-vascular coupling of left ventricular and aortic flow derived from 4D Flow MRI in healthy, young adults

Abstract

To gain further insight into male-female differences in cardiovascular conditions it is important to understand sex differences in healthy populations. A previous study from our group of 39 healthy young volunteers (20–35 years) paradoxically found that men had greater left ventricular (LV) kinetic energy (KE) but women had greater LV vorticity. We reanalyzed cardiac four-dimensional flow MRI data from 20 of the original subjects (10 male and 10 female) to quantify aortic flow in addition to LV flow. The combination of LV and aortic flow parameters were then used to calculate ventricular vascular coupling of KE and vorticity. The sex difference found in LV flow were not found in aortic flow and the ventricular-vascular coupling of LV-to-aortic flow was similar between men and women. Dimensional analysis to account for differences in cardiac output and ventricular volume explained the differences found in LV flow. The analysis methods and results of this study may be of further use in understanding ventricular vascular coupling of transported flow variables in healthy sex differences, healthy aging, and various cardiovascular conditions.

INTRODUCTION

Prior studies have investigated sex differences in the cardiovascular function of healthy individuals (Andre et al., 2015; Kolar and Ostadal, 2013) as a method to gain insights to the clinical question of why men and women responded differently to cardiovascular diseases (Aggarwal et al., 2018; Humphries et al., 2017). Cardiovascular sex differences in healthy populations have previously been identified in resting heart rate, arterial blood pressure, and left ventricular strain (Andre et al., 2015; Kolar and Ostadal, 2013). Our group has previously published a 4-dimensional flow sensitive (4D Flow) MRI study assessing sex differences in the left ventricle (LV) flow parameters kinetic energy and vorticity (Rutkowski et al., 2020). The results of this study were counterintuitive however, as male subjects had greater kinetic energy while female subjects had greater vorticity, both of which would indicate more inefficient flow patterns in the LV.

4D Flow MRI is an increasingly used tool to study cardiac function by non-invasively measuring time resolved velocity fields in the heart and large vessels (Francois et al., 2011; Hirtler et al., 2016; Stoll et al., 2019). Metrics such as kinetic energy and vorticity are commonly calculated to both assess ventricular flow efficiency and detect abnormal flow patterns that could lead to adverse ventricular remodeling. These 4D Flow MRI parameters have been found to correlate with more traditional metrics such as ventricular strain (Rutkowski et al., 2020), serum markers of ventricular remodeling, myocardial energetics, and exercise capacity (Stoll et al., 2019).

However, to date the 4D Flow MRI literature, including our prior studies, has neglected that cardiovascular function is a combination of both the status of the heart and the status of the vasculature. The conceptual framework of the interactions of the heart and vasculature was first formulated by Sagawa et al., as ventricular vascular coupling (VVC) (Sagawa, 1981). VVC has traditionally been measured by invasive catheterization to quantify the ratio of ventricular and arterial elastance however more recently non-invasive approximations have been developed (Sanz et al., 2012). In this work, we apply the concept of VVC to define a novel metric 4D Flow MRI VVC based on the hypothesis that elevated kinetic energy in the LV is not necessarily inefficient if LV kinetic energy is successfully transferred to aortic KE. In addition, we apply dimensional analysis to our results to determine if the previous paradoxical finding of elevated KE in men and elevated vorticity in women is due to differences in cardiac output and the size of cardiac chambers.

METHODS

Human Subjects

This study was reviewed and approved by the University of Wisconsin Health Sciences institutional review board. The study cohort overlaps with 39 healthy volunteer (19 retrospective and 20 prospective) from a prior study focused on sex differences in ventricular flow (Rutkowski et al., 2020). Only the 20 prospective volunteers (10 male and 10 female) were included in the present study given that for the 19 retrospective subjects the field of view of the 4D Flow MRI acquisition did not include the totality of the aortic arch.

MRI Acquisition

The study methods are the same as previously described (Rutkowski et al., 2020). Briefly, MRI was performed on a 3.0T imaging system (MR750 and Signa Premier, GE Healthcare, Waukesha, WI) without contrast agent. 4D Flow MRI was acquired with retrospective cardiac gating using the phase-contrast vastly undersampled isotropic projection (PC-VIPR) sequence (Johnson et al., 2008). 4D Flow MRI parameters were 1.25×1.25×1.25mm spatial resolution, 150cm/sec velocity encoding, and a 6.2ms repetition time. Cardiac anatomy was assessed from balanced steady state free precession (bSSFP) MRI for which acquisition parameters included 1.4×1.4×7mm spatial resolution, 1.18ms echo time, 3.26ms repetition time, and a 45° flip angle.

bSSFP MRI Analysis

Ventricular volumes for calculating end diastolic volume (EDV), end systolic volume (ESV), stroke volume (SV), cardiac output (CO), and ejection fraction (EF) were calculate from short axis cine bSSFP images using the software Segment as previously described (Rutkowski et al., 2020).

4D Flow MRI Analysis

The left ventricle was segmented time-averaged 4D Flow MRI magnitude and the aorta was segmented from time-averaged 4D Flow MRI complex-difference images using the software Mimics (Materialise, Leuven, Belgium). The thoracic aorta was segmented from just distal to the aortic root to the diaphragm. Ascending aorta cross sectional area was calculated just distal to the aortic root using the software Mimics. Flow data was reconstructed into 14 time frames per cardiac cycle. Flow parameters were analyzed and visualized using a combination of MATLAB (Mathworks, Natic, MA) and Ensight (Ansys, Canonsburg, PA). MATLAB was used to calculate kinetic energy (KE), vorticity, viscous energy loss (EL), and wall shear stress (WSS). Ensight was used to calculate flow waveforms and also to visualize streamlines and velocity vectors to check data quality.

LV and aortic KE were calculated

$$KE\left(t\right)=\int \frac{\rho }{{ }_{V}2}{u}_i{(t)}^{2}dV$$ #(1)

where ρ is density (1060 kg/m³), u_i is the velocity vector, and V is the segmented LV or aortic volume. LV KE was calculated at peak systole and peak diastole and aorta KE is reported at peak systole. Peak diastole and peak systole are defined as the cardiac time frames where maximum KE was observed in diastole and systole. KE was reported as a raw value, normalized by stroke volume, normalized by the mask volume, and in dimensionless form

$$K{E}_{LV}^{\ast }=\frac{K{E}_{LV}}{\rho \hspace{0.17em}C{O}^2{V}^{-\frac{1}{3}}}orK{E}_{Ao}^{\ast }=\frac{K{E}_{Ao}}{\rho \hspace{0.17em}{\left(\frac{CO}{A}\right)}^{2}V}$$ #(2)

for the LV and aorta respectively where * represents the dimensionless variable. Q is the cardiac output, V is volume and A is the ascending aorta cross-sectional area.

LV and aorta average vorticity (ω) were calculated

$$\omega =\frac{1}{V{T}_{HB}}\int {}_{T_{HB}}\int {}_V\nabla \times u_{i}dVdt$$ #(3)

Where ∇ × is the curl operator and T_HB is the heart beat duration. Vorticity was reported as a raw value, normalized by stroke volume, normalized by the mask volume, and in dimensionless form

$${\mathit{\omega }}_{\mathit{L}\mathit{V}}^{\mathbf{*}}=\frac{{\mathit{\omega }}_{\mathit{L}\mathit{V}}}{CO\hspace{0.17em}{V}^{-1}}or\hspace{0.17em}{\mathit{\omega }}_{\mathit{A}\mathit{o}}^{\mathbf{*}}=\frac{{\mathit{\omega }}_{\mathit{A}\mathit{o}}}{CO\hspace{0.17em}{A}^{-\frac{3}{2}}}$$ #(4)

for the LV and aorta respectively.

LV and aorta EL were calculated

$$EL=\int {}_{T_{HB}}\int {}_V\mathrm{\text{μ}}{\mathbf{\text{S}}}_{\mathbf{\text{ij}}}^{2}dVdt$$ #(3)

Where S_ij is the strain rate tensor and T_HB is the heart beat duration. Vorticity was reported as a raw value, normalized by stroke volume, normalized by the mask volume, and in dimensionless form

$$K{L}_{LV}^{\ast }=\frac{K{E}_{LV}}{\mu \hspace{0.17em}C{O}^2{V}^{-1}{T}_{HB}}or\hspace{0.17em}K{L}_{Ao}^{\ast }=\frac{K{L}_{Ao}}{\mu \hspace{0.17em}C{O}^{2}V\hspace{0.17em}{T}_{HB}{A}^{-3}}$$ #(4)

for the LV and aorta respectively.

Ventricular-vascular coupling (VVC) for both KE and vorticity were calculated by taking the ratio of the LV-to-aorta value.

Aorta wall shear stress (WSS) was calculated from triangulated surface meshes by taking the cross product of the velocity vector and the surface normal vector and then averaged over the aorta surface,

$${\tau }_w=\frac{1}{A}{\int }_A\mu \frac{\partial u_i}{\partial n_i}dA$$ #(5)

where μ is viscosity (4 centipoise) and n_i is the surface normal vector. WSS was calculated both as a time-averaged value and as the peak systolic value. WSS was reported as a raw measurement and in dimensionless form.

$${\tau }_w^{\ast }=\frac{{\tau }_w}{\mu \hspace{0.17em}CO{A}^{-\frac{3}{2}}}$$ #(6)

Statistics

Male and female groups were compared using the non-parametric Mann-Whitney U test which is appropriate for small groups. P-values less than 0.05 were considered statistically significant. Results are presented as median and the interquartile range.

RESULTS

General Cardiac Anatomy and Function

HR, LV EF, and ascending aortic area were similar between male and female participants (Table 1). CO, LV EDV, LV ESV, and LV SV were all greater in male participants than female participants.

Table 1:

General Cardiac Anatomy and Function

	Male	Female	p-value
HR (BPM)	58 [55–74]	62 [54–75]	0.8
CO (L/min)	6.0 [5.0–7.0]	3.8 [3.7–5.8]	0.04*
LV EDV (mL)	161 [152–197]	130 [118–143]	0.02*
LV ESV (mL)	72 [59–85]	53 [50–60]	0.02*
LV SV (mL)	93 [78–100]	70 [60–79]	0.04*
LV EF (%)	56 [54–61]	56 [55–60]	0.7
AAo Area (mm2)	111 [91–120]	96 [80–105]	0.39

^*p<0.05, HR – heart rate, BPM - beats per minute, CO – cardiac output, LV – left ventricle, EDV – end diastolic volume, ESV – end systolic volume, SV – stroke volume, EF – ejection fraction, AAo – ascending aorta

Left Ventricular Flow

Results of LV flow analysis are shown in Table 2 and Figure 2. Raw systolic and diastolic KE trended towards being greater in male participants than female participants (systolic: 3.6 [2.4–4.6] mJ vs 2.3 [2.2–3.1] mJ p=0.13, diastolic: 6.2 [4.2–7.6] mJ vs 4.1 [3.7–5.3] mJ p=0.13). Systolic KE normalized by stroke volume was greater in male participants than female participants (354 [207–474] J/L vs 178 [113–262] J/L p=0.04) and diastolic KE normalized by stroke volume trended towards being greater in male participants (588 [312–733] J/L vs 301 [250–361] J/L p=0.07). KE normalized by EDV was similar for both groups. Average vorticity trended towards being greater in female versus male participants (190 [147–265] 1/s vs 279 [205–363] 1/s p=0.11) and significantly larger when normalized by SV (p=0.02) and EDV (p=0.04). EL trended towards being greater in female participants when normalized by EDV (p=0.13).Dimensionless LV flow paremeters were similar between male and female participants (systolic p=0.81, diastolic p=0.42).

Table 2:

Left ventricular flow analysis

	Male	Female	p-value
Systolic KE (mJ)	3.65 [ 2.48 – 4.57]	2.30 [2.22 – 3.08]	0.13
Systolic KE / SV (mJ/L)	38.8 [27.5 – 49.1]	37.2 [27.8 – 41.7]	0.04*
Systolic KE / EDV (mJ/L)	23.5 [ 13.8 – 27.4]	20.7 [17.6 – 22.5]	0.74
Diastolic KE (mJ)	6.16 [ 4.19 – 7.63]	4.07 [3.69 – 5.32]	0.13
Diastolic KE / SV (mJ/L)	63.0 [ 56.6 – 68.2]	55.0 [49.2 – 73.4]	0.07
Diastolic KE / EDV (mJ/L)	35.9 [ 30.7 – 40.5]	30.2 [30.1 – 37.3]	0.74
Vorticity (1/s)	19.0 [ 14.7 – 26.5]	27.9 [20.5 – 36.3]	0.11
Vorticity / SV (1/s/L)	207 [ 154 – 280]	384 [282 – 491]	0.02*
Vorticity / EDV (1/s/L)	95 [ 89 – 167]	228 [145 – 251]	0.04*
EL (mJ)	0.83 [ 0.50 – 1.11]	0.89 [0.70 – 1.25]	0.80
EL / SV (mJ/L)	7.4 [ 5.7 – 11.2]	14.5 [11.3 – 20.3]	0.24
EL / EDV(mJ/L)	4.6 [ 3.3 – 6.5]	7.3 [5.9 – 8.8]	0.13

^*p<0.05, KE – kinetic energy, mJ - millijoule, SV – stroke volume, EDV – end diastolic volume, L – liter,s – second, EL – energy loss

Figure 2:

Left ventricle flow analysis: A. dimensionless systolic KE, B. dimensionless diastolic KE, C.dimensionless average vorticity, D. dimensionless EL.

Aortic Flow

Results of aortic flow analysis are shown in Table 3 and Figure 3. For aortic flow, peak systolic KE, EL, and WSS metrics were similar between both groups. When normalized by either SV or aorta volume, vorticity trended towards being greater in females (p=0.08 and p=0.12 respectively). Dimensionless aorta flow parameters were similar between both groups.

Table 3:

Aorta flow analysis

	Male	Female	p-value
Systolic KE (mJ)	34.2 [ 27.3 – 49.0]	31.1 [30.4 – 34.2]	0.44
Systolic KE / SV (mJ/L)	353 [282 – 513]	440 [419 – 494]	0.28
Systolic KE / VAo (mJ/L)	343 [279 – 432]	351 [301 – 383]	0.91
Vorticity (1/s)	65.6 [ 59.5 – 79.3]	73.5 [68.4 – 83.4]	0.39
Vorticity / SV (1/s/L)	713 [ 580 – 913]	1,000 [892 – 1,261]	0.08
Vorticity / VAo (1/s/L)	586 [ 558 – 703]	788 [605 – 908]	0.12
EL (mJ)	5.3 [ 2.7 – 10.8]	6.7 [3.6 – 8.3]	0.91
EL / SV (mJ/L)	50.8 [36.5 – 114]	86.1 [58.9 – 121]	0.32
EL / VAo (mJ/L)	536 [ 372 – 858]	1,059 [630 – 1,164]	0.39
Avg. WSS (N/m2)	0.52 [ 0.51 – 0.67]	0.51 [0.36 – 0.61]	0.48
Max. WSS (N/m2)	1.84 [ 1.72 – 2.40]	1.91 [1.21 – 2.08]	0.44

^*p<0.05, KE – kinetic energy, mJ - millijoule, SV – stroke volume, V_Ao – aorta volume, L – liter, s – second, EL – energy loss, WSS – wall shear stress, N – Newton, m – meter

Figure 3:

Aorta flow analysis: Adimensionless systolic KE, B. dimensionless average vorticity, C. dimensionless EL, D. dimensionless time-averaged WSS, E. dimensionless peak systolic WSS.

Ventricular Vascular Coupling

Results of ventricular vascular coupling (VVC) are shown in Figure 4. VVC was calculated by dividing the LV KE and vorticity by the corresponding aorta values. Differences in KE VVC and vorticity VVC were not found between male and female participants. KE and vorticity were consistently greater in the aorta giving KE VVC values less than 0.25, and vorticity VVC values less than 1.0.

Figure 4:

A. Kinetic energy ventricular vascular coupling (VVC), B. vorticity ventricular vascular coupling.

DISCUSSION

This study applied the concept of ventricular vascular coupling to 4D Flow MRI analysis of healthy young men and women. The major findings of this study include that while males have greater kinetic energy and females have greater vorticity in the left ventricle, aortic flow was similar between the two groups. No differences in ventricular vascular coupling were found between male and female participants suggesting that both flow patterns equally transfer transported flow variables from LV flow to aortic flow. Lastly, dimensional analysis suggests that previously identified sex differences in LV flow (Rutkowski et al., 2020) were largely due to men having greater cardiac output and ventricular volumes than women.

Previous studies have investigated sex differences in cardiovascular function in young healthy populations (Daimon et al., 2011; Natori et al., 2006; Okura et al., 2009). Women have been found to have elevated ejection fraction and greater ventricular strain which would indicate superior ventricular function compared to men. Our group’s previous study of sex differences in LV flow however gave confounding results (Rutkowski et al., 2020). Men had greater LV KE which in the ventricular flow literature is associated with inefficient flow. Women were found to have elevated vorticity which would be associated with greater strain rates and energy dissipation in ventricular flow. The present study applied the concept of ventricular vascular coupling to 4D Flow MRI analysis and found that men and women equally transfer kinetic energy and vorticity from ventricular flow to aortic flow, which would indicate similar efficiencies. In the original formulation of VVC, it was hypothesized that in adaptive cardiovascular remodeling, changes in arterial elastance of the vasculature would be coupled to changes in ventricular contractility (Sagawa, 1988). Our concept of 4D Flow MRI based VVC is not attempting to non-invasively calculate the traditional VVC like other approaches based on ventricular volumes (Sanz et al., 2012). Instead, we applied a similar conceptual framework to develop a novel metric of the coupling of aorta flow and LV flow.

Differences in myocardium contraction and relaxation due to pathological processes are known to alter the physics of ventricular diastolic vortex formation and helicoidal flow patterns that ultimately negatively impact cardiovascular function. We ound that dimensionless LV flow parameters were similar between men and women. These results indicate that the fundamental physics and functionality of ventricular diastolic vortex formation and helicoidal flow patterns are similar in both men and women and that quantitative differences measured between men and women should be expected based on differences in cardiac output and ventricular volumes. In this way, dimensional analysis could be a useful tool in comparing flow data between healthy and diseased populations as dimensional analysis successfully accounts for the anatomic differences between healthy men and women. The standard normalizations of flow parameters to SV or EDV have the advantage of a more direct physical interpretation as an energy efficiency or energy density in addition to having an established literature to compare results with. The advantages of presenting results in dimensionless form as shown in this study include that the impacts of SV, EDV, and HR are accounted for in addition to being the proper way to normalize variables from a fluid dynamics perspective.

While the differences in ventricular flow are expected on a dimensional basis, women do still have higher LV vorticity which would indicate greater shear stresses at the endocardial surface. WSS is known to be a potent biomechanical stimulus for the vascular endothelium (Ballermann et al., 1998; Lu and Kassab, 2011) and is believed to be similarly important for myocardial remodeling (Pasipoularides, 2012). Correlations found between LV flow parameters and myocardial energetics in heart failure patients have been hypothesized to be the results of mechanobiological interactions between forces at the endocardial surface and myocyte metabolism (Stoll et al., 2019).

The primary limitation of the present study was the small sample sizes. The current study may not be sufficiently powered to detect differences in aortic flow and several comparisons that were previously found to be statistically significant (p<0.05) now only trended towards significance (p=0.07–0.13). In addition, other relevant factors, such as if participants exercise, were not controlled for. Another limitation of this study was that LV masks for 4D Flow MRI analysis were segmented from time-averaged magnitude images. Our group has previously compared a time-averaged mask with a dynamic mask and found good agreement, with the time-averaged mask overestimating systolic KE by ~10% due to inclusion of some myocardium in the mask (Hussaini et al., 2017). Segmenting LV volumes from 4D Flow MRI magnitude images can also underestimate the total LV volume although it does eliminate errors associated with co-registration from another imaging sequence. As the fastest flow is in the center of the LV, underestimation of LV flow metrics should be less than the underestimation of LV volume. We have previously shown good inter-observer agreement using these methods (Hussaini et al., 2017), so even if LV flow metrics are being systematically underestimated, we are confident in the repeatability of our analysis to identify inter-group differences. Lastly, there could also be observer variability in the volume and area measurements that we used to calculate dimensionless variables. Future studies should investigate the observer variability of dimensionless flow variables.

In conclusion, male-female sex difference in left ventricular flow were not found in aortic flow and the coupling of LV-aortic flow was similar between men and women. Dimensional analysis to account for differences in cardiac output and ventricular volume explained the differences found in LV flow. The analysis methods and results of this study may be of further use in understanding ventricular vascular coupling of transported flow variables in healthy sex differences, healthy aging, and various cardiovascular conditions.

Figure 1:

A. Maximum intensity projection of LV and aorta velocity overlaid on the complex-difference angiogram, and B. contours of aorta wall shear stress.