A 1D pulse wave propagation model of the hemodynamics of calf muscle pump function.

The calf muscle pump is a mechanism which increases venous return and thereby compensates for the fluid shift towards the lower body during standing. During a muscle contraction, the embedded deep veins collapse and venous return increases. In the subsequent relaxation phase, muscle perfusion increases due to increased perfusion pressure, as the proximal venous valves temporarily reduce the distal venous pressure (shielding). The superficial and deep veins are connected via perforators, which contain valves allowing flow in the superficial-to-deep direction. The aim of this study is to investigate and quantify the physiological mechanisms of the calf muscle pump, including the effect of venous valves, hydrostatic pressure, and the superficial venous system. Using a one-dimensional pulse wave propagation model, a muscle contraction is simulated by increasing the extravascular pressure in the deep venous segments. The hemodynamics are studied in three different configurations: a single artery-vein configuration with and without valves and a more detailed configuration including a superficial vein. Proximal venous valves increase effective venous return by 53% by preventing reflux. Furthermore, the proximal valves shielding function increases perfusion following contraction. Finally, the superficial system aids in maintaining the perfusion during the contraction phase and reduces the refilling time by 37%.

1 Introduction

Orthostatic intolerance is observed in astronauts after spaceflight 1, in patients with spinal cord injury 2, in de-conditioned subjects, for example, after bed rest 3, and is strongly associated with aging 4. In the upright position, orthostatic intolerance may result in critical events such as syncope. Because of their high compliance and high storage capacity, veins are believed to play a major role in blood volume shifts from the upper towards the lower body, which is associated with orthostatic intolerance 5. In healthy subjects, this fluid shift is compensated for by among others maintaining venous return, whereas venous return is impaired in patients with orthostatic intolerance. Venous return can be restored by muscle contraction in the lower limb, which promotes emptying of the venous system. Therefore, improved understanding of the muscle pump effect on lower limb hemodynamics is important to better comprehend the sources of orthostatic intolerance.

The muscle pump is a mechanism that promotes venous return via muscle contraction (Figure 1). During muscle contraction, the extravascular pressure increases and exceeds the intravascular pressure, resulting in the collapse and emptying of the embedded deep veins 6–8. Unidirectional flow is guaranteed by the presence of distal venous valves. During muscle relaxation, the pressure in the deep veins drops and the deep veins refill from the micro-circulation, resulting in distal valve opening and proximal valve closure, preventing reflux 6. In the upright position, the refilling time is expected to decrease due to increased perfusion pressure, as closure of the proximal venous valves reduces the distal venous pressure (shielding) 9. It is hypothesized that superficial veins, connected to the deep veins by perforating veins, might further promote the efficiency of the muscle pump. Additional valves in the perforating veins allow flow only in superficial-to-deep direction 10 and, as a consequence, promote venous return during contraction and venous filling during the relaxation phase. The collapsibility of the veins and the dynamics of the venous valves, that direct the blood towards the heart and shield hydrostatic pressure, are believed to be the main physiological factors in the muscle pump effect 7.

Figure 1

Schematic representation of the muscle pump effect during the contraction (A) and relaxation (B) phases. During contraction, the deep vein collapses due to the extravascular pressure pex exceeding the intravascular pressure. Venous return is increased, whereas backflow and flow to the superficial system is blocked by the distal and perforator valves. During relaxation, the deep vein is refilled from both the artery and the superficial vein, while the perforator valves open and the proximal valve is closed to prevent backflow.

In this study, venous hemodynamics during muscle contraction and relaxation are investigated using a mathematical modeling approach that enables simultaneous analysis of venous collapse and valve dynamics. Furthermore, it allows study of the influence of independent parameter variation. A one-dimensional pulse wave propagation model is used, which has been well-established to examine and predict arterial hemodynamics 11–14 but has only been applied in few studies to model venous hemodynamics 15,16. Modeling venous hemodynamics is challenging because, in contrast to arteries, vessel collapsibility and venous valves are important for many applications. Furthermore, more anatomical variation is present in the venous system, for example, the deep and superficial veins. Most models of venous hemodynamics account for the venous collapsed state under negative transmural pressure by interpreting the collapsible tube law as a non-linear cross-sectional area-pressure A, p-relation 17–19. In addition to collapsibility, changes in hydrostatic pressure between supine and upright positions are also important when considering fluid volume shifts and therefore cannot be ignored as is common in arterial models. Gravity has been included in lumped-parameter models using an additional pressure source 20 and can straightforwardly be included in the momentum equation of 1D pulse wave propagation models 21, resulting in a continuous gravity field. The inclusion of venous valves within such models allows the transient shielding of hydrostatic pressure to be simulated 22 and their action to prevent backflow. Valves are often represented by a perfect diode in circulatory models 23, whereas more detailed models include gradual opening and closing behaviors based on the transvalvular pressure drop 24 or force balance on the valve leaflets 25.

Numerical studies of the muscle pump have been reported by Marchandise et al. 21 and Fullana and Zaleski 26. Marchandise et al. 21 simulated the contraction phase of the muscle pump considering changes in cross-sectional area and wave speed of a collapsing tube, which were validated against in vitro measurements. A normal and pathological case were simulated by inclusion and exclusion of a venous valve in the distal boundary conditions. Fullana and Zaleski 26 performed a computational study of the muscle pump, including a more extensive anatomy of the venous system, and although they did include valves, they did not examine resulting valve dynamics. Although these studies have been able to model specific venous physiology, an analysis of venous valve dynamics combined with an increased perfusion pressure and the communication of deep and superficial venous systems during a muscle contraction has, to our knowledge, not yet been undertaken.

Therefore, the aim of this study is to investigate the physiological mechanisms of the calf muscle pump and its effect on venous return in the upright position using a modeling approach. The 1D pulse wave propagation model of Kroon et al. 27 is extended to include venous properties: a collapsible tube law is implemented to account for the non-linear A,p-relation, gravity is included in the momentum equation to account for the hydrostatic pressure, and venous valves are included to prevent backflow. We will show how the model can be used to explain the different physiological mechanisms of muscle pump function, including the effect of (1) venous valves, (2) hydrostatic pressure, and (3) the deep and superficial venous systems. In three configurations with different levels of detail, muscle contraction and relaxation will be simulated by applying changes in extravascular pressure to study the valve dynamics and the hemodynamics.

2 Methods

To study calf muscle pump function, a numerical model of the arteries and veins in the calf was built. The calf muscle is perfused mainly by three arteries: peroneal, anterior tibial, and posterior tibial, which are each associated with two veins 10. Apart from these vessels, all embedded in the calf muscle, a superficial system, of which the great and small saphenous veins are the largest vessels, supplies the skin with blood 10. The superficial veins are connected to the deep venous system by a varying number of perforating veins, which contain venous valves that open to allow flow towards the deep venous system 10. The model used in this study contains all of these types of vessels. However, to reduce the complexity of the system, but retain all essential elements, a generic approach is chosen and one vessel of each type was included. Also, for the inclusion of venous valves, a generic approach is chosen, by including a single proximal and distal valve for each deep and superficial vein and one valve in each perforating vein.

Three anatomical configurations are considered in this study (Figure 2); from the simplest valveless single artery–vein configuration, via the inclusion of valves, to the most complex arrangement which includes the superficial and perforating veins. By comparing the different configurations, the importance of venous valves, hydrostatic pressure, and the superficial system is studied. Simulation of hemodynamics was undertaken in both the supine and upright positions, where the gravity vector is aligned with the vessel axis in the upright position. The model formulation includes both 1D and 0D components for which the governing equations are provided in Sections 2.1–2.7, and the numerical formulation is provided in Section 2.8. The geometrical parameters used for each region of the model were taken from the literature 16 and are provided in Table 1, with the numbering of each anatomical segment given in Figure 2. The constant parameters used for the governing equations presented in Sections 2.1–2.7 are given in Table 2.

Figure 2

Three different model configurations: (A) reduced configuration without valves, (B) reduced configuration including valves, and (C) deep-superficial configuration. All three configurations consist of different elements: 1D arteries (AR), and deep (DV) and superficial veins (SV), 0D venous valves (VV), micro-circulation, and inflow and outflow boundary conditions (BC). The length and radius of the 1D elements are not to scale (geometrical parameters of all 1D segments can be observed in Table 1). The z-positions indicate the locations where the flows are determined as explained in Section 2.8.

Table 1

Geometrical parameters of the different vessels 16 as depicted in Figure 2.

Vessel	Numbering (Figure 2)	Radius [mm]	Length [cm]	Configuration (Figure 2)
artery	AR1	2.5	34	A,B,C
deep vein	DV1	1.5	34	A
	DV2	1.5	4	B
	DV3	1.5	24	B,C
	DV4	1.5	4	B
	DV5	1.5	2	C
	DV6	1.5	2	C
superficial vein	SV1	3.5	2	C
	SV2	1.5	2	C
	SV3	1.5	24	C
	SV4	1.5	2	C
	SV5	3.5	2	C
venous valves	VV1	1.5	1	B,C
	VV2	1.5	1	B,C
	VV3	1.5	1	C
	VV4	1.5	1	C
perforating vein	PV#-S	0.5	1	C
	PV#-V	0.5	1	C
	PV#-D	0.5	1	C

The four perforating veins contain a valve (PV#-V), a deep (PV#-D), and a superficial (PV#-S) vein of which the parameters are noted separately.

Table 2

Constant model parameters.

Symbol	Value	Unit	Description	Equation	Reference
ρ	1.050	103kg·m−3	Blood mass density	(1),(12)	30
g	9.81	m·s−1	Gravitational acceleration	(1)	-
ν	0.5	-	Poisson’s ratio	(1)	30
E	1.6	106Pa	Arterial Young’s modulus	(5)	30
pref,art	13	103Pa	Arterial reference pressure	(5)	29
b	10	-	Constant Shapiro’s tube law	(6)	19
c		-	Constant Shapiro’s tube law	(6)	19
	1.37	-	Fitting constant	(8)	-
	-2.53	-	Fitting constant	(8)	-
	3.02	-	Fitting constant	(8)	-
B	0.108	-	Fitting constant	(9)	-
	1.28	-	Fitting constant	(9)	-
	-1.49	-	Fitting constant	(9)	-
	2.03	-	Fitting constant	(9)	-
γ	4	-	Fitting constant	(10)	-
η	4.05	10−3Pa·s	Dynamic blood viscosity	(12)	31
Δp	12	103Pa	Pressure drop over micro-circulation	(16)	32
τRC	2	s	RC-time	(17)	29
pmax	20	103Pa	Maximal extravascular pressure	(18)	7
pin	13.4	103Pa	Inlet pressure boundary condition	-	-
pout	667	Pa	Outlet pressure boundary condition	-	-
p0	0	Pa	Extravascular pressure in windkessel elements	-	-

2.1 1D pulse wave propagation:arteries and veins

The relation between pressure and flow in the large arteries and veins are based on the one-dimensional equations of mass and momentum balance 11,28: where C is the compliance per unit length, ptr is the transmural pressure, t is the time, q is the flow, and z is the axial coordinate. Additionally, in the momentum balance equation, A = A(ptr) is the cross-sectional area, is the velocity in z-direction averaged over the cross-sectional area, p is the intravascular pressure, is the radius, and ρ is the density of the blood. The wall shear stress is given by τ, and g=g·e is the contribution of gravitational acceleration in the z-direction, where g is the gravity vector. The wall shear stress and non-linear convection term are estimated using the approximated velocity profile of Bessems et al. 29. The gravity term is included in the derivation of the velocity profile. The resulting wall shear stress and non-linear convection term are as follows: where η is the dynamic viscosity, ζ is the dimensionless core thickness, and δ1 a function of ζ. To solve the aforementioned equations, a constitutive law relating cross-sectional area and pressure is needed for both arteries and veins.

2.2 Constitutive law for arteries

To capture the mechanical behavior of the arterial wall, the system of equations is completed with a linear A,p-relation for the arteries: where Aref,art is the cross-sectional area at reference pressure pref,art, and C is the linearized compliance per unit length around reference pressure pref,art. The compliance is determined using thin-walled-cylinder theory for a linear isotropic elastic material: where is the reference radius, ν is the Poisson’s ratio, h≈a0/10 is the vessel wall-thickness 30, and E is the Young’s modulus.

2.3 Constitutive law for veins

As the internal venous pressure is very low, the influence of the extravascular pressure on the mechanical behavior of the venous cross-sectional area is significant, and even negative transmural pressure can be reached, resulting in venous collapse. As the mechanical behavior of the venous wall is non-linear in this transmural pressure range, the proposed relation for the arteries can no longer be used. Therefore, Shapiro 19 derived the following relation to capture the collapse of the veins based on in vitro measurements. where K is the bending stiffness, and b and c are constants that determine the shape of the tube law. To obtain a relation that captures the behavior described by Equation 6 but cast in an A,p-relation, Equation 6 is approximated by the following expression, which has a continuous first order derivative. where Aref,ven is the reference cross-sectional area at zero transmural pressure, and p*=ptr/K is the dimensionless pressure. The functions f+ and f− are fits of the positive and negative pressure part of the original tube law of Shapiro 19, and h(p*) is a scaling function. where B, , , , , , , and γ are fitting constants determining the shape of the A,p-relation. Venous compliance is calculated as the derivative of A with respect to ptr. Superficial veins differ from deep veins as they are less influenced by a muscle contraction, which is included in the model by applying pex only to the deep veins. Figure 3 shows a comparison between this approximation and the Shapiro tube law.

Figure 3

A,p-relation according to Shapiro tube law: Equation 6(-) and the fit: Equation 7(-·).

2.4 0D venous valves

The different states of the venous valves result in changes in resistance to flow. This is included in the model using a diode with an extra inertial term to represent the inertia of the blood. The pressure–flow relation within the valve region is defined as follows: where L is the inertance and R is the resistance, defined to be equal to a Poisseuille resistance in open state and very large in closed state. where the effective cross-sectional area Aeff is where leff=5a0 is an effective length, representing the geometrical region in which the inertial forces are dominant due to the presence of the valve, and cross-sectional area A0 is taken equal to the reference cross-sectional area of the connecting venous elements. Opening occurs on a positive transmural pressure difference and the closure on negative flow.

2.5 0D micro-circulation

To capture the pressure drop and the storage capacity of the micro-circulation. a windkessel element consisting of two resistances Rwk,i (i = 1,2), and a compliance Cwk connected to the extravascular pressure p0 is used, described by where the first resistance Rwk,1 is set equal to the characteristic wave impedance of the connecting arterial element given by The total resistance is defined by the ratio of the pressure drop over the micro-circulation Δp divided by the mean baseline flow Baseline flow is defined as of a population averaged femoral artery flow of 2.78 mL/s 33, because only one of the six deep veins is included in the geometry. When the superficial system is included, of the femoral artery flow is used in order to maintain the same baseline flow in the deep vein. Finally, the compliance of the micro-circulation is computed based on a representative value of the time constant τRC, defining the decay of the pressure wave in diastole 11 The extravascular pressure of the micro-circulation is set to p0.

2.6 0D boundary conditions

The arterial inlet and venous outlet are both connected to windkessel elements with the same form as that of the micro-circulation. To avoid reflections of the higher order harmonics, the resistance connected to the 1D element is calculated using Equation 15, the second resistance is set equal to the first one, and the compliance is again calculated using Equation 17. It is noted that the windkessel elements are not included to represent the vascular tree attached to the reduced circulation, and therefore, the true value of the second resistance is of minor importance. The extravascular pressure of the inlet and outlet boundary condition is set to p0. Inlet and outlet boundary conditions are set to a constant pressure pin and pout, respectively, to simulate the pressure gradient between the arterial and venous circulation.

2.7 Muscular extravascular pressure

The extravascular pressure generated during the muscle contraction is applied to the deep venous elements and not to the artery and superficial vein, as the embedded artery is assumed to be less influenced due to its high intravascular pressure. Furthermore, the influence of muscle contraction on the superficial vein is assumed to be negligible compared with the influence on the deep vein due to the location of the superficial vein in the skin. The prescribed extravascular pressure is assumed to be where pmax is the maximal value of extravascular pressure, k(t) the temporal course of the extravascular pressure, and m(z) the spatial course of the extravascular pressure. The spatial and temporal variations in extravascular pressure are shown in Figure 4, and the full equations of k(t) and m(z) are given in the Appendix A.

Figure 4

Extravascular pressure of the deep veins is increased to simulate a muscle contraction. Plot A and B show the spatial m(z) and temporal k(t) courses of extravascular pressure as applied to the deep venous elements, respectively (see Appendix A for the full equations of m(z) and k(t)). The gray areas in the spatial plot indicate the location of the venous valves.

2.8 Numerical implementation and post-processing

The governing equations described in Section 2.1–2.7 were solved numerically using the finite element package SEPRAN (Ingenieursbureau SEPRA, Leidschendam, The Netherlands) with the reduced complexity method as described by Kroon et al. 27. A second-order backward difference scheme was used for time discretization with timestep dt = 1ms. The trapezium rule was used for spatial integration with dz = 10 mm for arterial and superficial venous elements and dz = 1 mm for deep venous elements, as the spatial transmural pressure gradient can be much higher in the deep venous elements due to the prescribed muscle contraction. Pre-processing and post-processing was performed using MATLAB R2012b (MathWorks, Natick, MA, USA).

To compare the muscle pump function in the different configurations, several parameters were defined. First, the additional volume that refills the deep vein from the micro-circulation during the relaxation, called perfusion volume VPE, was calculated as a measure of the increase in perfusion. The increase in venous return during the contraction phase is captured in the additional volume that leaves the deep vein, called venous return VVR. Finally, the reflux volume VRE was calculated as a measure for the volume that refills the veins from the proximal side of the vein, decreasing the effective venous return. VPE, VVR, and VRE were calculated using the following relations: where qart is the arterial flow into the micro-circulation (at position z1 in Figure 2), qart,bl is the arterial flow just before the onset of the contraction (t = 1 s), qven is the venous flow at the end of the deep vein (at position z2 in Figure 2), and qven,bl is the venous flow just before the onset of the contraction (t = 1 s). Furthermore, to determine the time needed for the deep vein to refill during relaxation, an estimate for the refilling time tfill was determined from the start of relaxation (t = 3 s) until the volume of the deep vein reached 95% of the baseline volume. Finally, to measure the decrease in arterial inflow due to contraction, the change in arterial inflow Δqart was calculated relative to the arterial baseline flow qart,bl. As the computational domain only covered a small part of the calf circulation, relative changes of the previously defined parameters were used to provide an indication of the magnitude of changes between different configurations. Additionally, deep and superficial venous inflows were determined at location z3 and z4, respectively (Figure 2). The perforating inflow was calculated as the sum of the flows at locations z5.

2.9 Simulations and analysis

To demonstrate that the model is able to capture the different aspects of the muscle pump effect, the following comparisons were made:

Venous collapsibility: Collapse was demonstrated using a reduced geometry without valves in the supine position as in Figure 2A.

Effect of venous valves: To demonstrate that the venous valves prevent backflow in both the contraction and relaxation phases, a muscle contraction was simulated in a reduced configuration with and without valves (see Figure 2B and A, respectively), both configurations in the supine position.

Effect of hydrostatic pressure: The reduced configuration including valves was simulated in the supine and upright positions to examine the influence of hydrostatic pressure (Figure 2B).

Effect of superficial system: The complex venous system was compared with the reduced geometry in the upright position (see Figure 2C and B, respectively) to investigate the influence of the superficial system on the valve dynamics and the hemodynamics.

3 Results

The hemodynamics during a muscle contraction are reported using the different model configurations shown in Figure 2. The results are reported in four sections which examine the course of the deep venous collapse, the effect of venous valves, the effect of hydrostatic pressure, and the importance of the superficial system, respectively.

3.1 Venous collapsibility

The time and spatial courses of venous collapse during a muscle contraction are studied in the configuration without venous valves (Figure 2A). The temporal and spatial plots of the cross-sectional area of the deep vein at several axial positions and time points, respectively, are shown in Figure 5. For clarity, the temporal plots (Figure 5C and E) are split into the distal and proximal regions, and the spatial plots (Figure 5D and F) are split into time points during contraction and relaxation.

Figure 5

Venous cross-sectional area in the reduced configuration without valves (Figure 2A) over time (C and E) and over the z-coordinate (D and F) with the corresponding extravascular pressure on top (B and A, respectively). The temporal plots are divided in the distal (C) and proximal (E) courses of the cross-sectional area. The spatial plot are divided in contraction (D) and relaxation (F) plots. The different lines represent different spatial and temporal points, which correspond to the following symbols in the diagonal plots (A and B) of the extravascular pressure: temporal plots of the distal (C, in B) and proximal (E, in B) part of the deep vein, and spatial plots of the contraction (D, in A), and relaxation (F, ∘ in A) phase.

During the contraction phase in the temporal plot ( , Figure 5C and E), it can be observed that the collapse starts at the proximal side, as the blue and black curves in the proximal plot show a higher gradient than the green and orange curves in the distal plot. Furthermore, the uncollapsed proximal and distal parts of the vein increases in the cross-sectional area during the contraction phase, which is best observed in Figure 5D while comparing all curves to the black curve (before contraction). In the subsequent relaxation phase, the vein refills from the distal side.

3.2 Effect of venous valves

The influence of venous valves on venous hemodynamics during muscle pump activation was assessed by comparing the configuration without venous valves to the one with venous valves (Figure 2A and B, respectively). The results from the two simulations are compared in Figure 6, which reports the extravascular pressure and the state of the proximal (VV2) and distal (VV1) valves (Figure 6A), the arterial inflow (Figure 6B), venous outflow (Figure 6C), and venous volume (Figure 6D). Gray regions in Figure 6B and C indicate the extra volume due to perfusion VPE, venous return VVR, and reflux VRE for which the numerical values are reported in Table 3.

Figure 6

The muscle pump effect is shown in the reduced geometry in the case valves are excluded (blue) and included (red) (Figure 2A and B). In plot A, the course of the extravascular pressure and the state of the proximal and distal valves are shown, where the closed state is denoted with a solid line. Plots B and C show the arterial inflow and venous outflow, respectively. The gray areas indicate the extra perfusion (VPE), venous return (VVR) and reflux volume (VRE) of which the values can be found in Table 3. Plot D shows the volume inside the deep vein.

Table 3

Venous return (VVR), perfusion (VPE), and reflux volumes (VRE), refilling time tfill and the relative decrease in arterial inflow during the muscle contraction Δqart as indicated in Figure 6, 7, and 9 and explained in Equation 20, 19, and 21.

	VVR[mL]	VPE[mL]	VRE[mL]	tfill[s]	Δqart[%]
No valves, supine (Figure 2A)	0.87	0.10	0.46	1.69	106
Valves, supine (Figure 2B)	0.88	0.17	0.00	2.62	97
Valves, upright (Figure 2B)	0.92	0.26	0.00	2.45	97
Complex venous system (Figure 2C)	0.85	0.16	0.00	1.54	6

Figure 7

The muscle pump effect is shown in the reduced configuration including valves (Figure 2B) in the supine (blue) and upright (red) positions. In plot A, the course of the extravascular pressure and the state of the proximal and distal valves are shown, the closed state is denoted with a solid line. Plots B and C show the arterial inflow and venous outflow, respectively. The gray areas indicate the extra perfusion (VPE) and venous return (VVR) of which the values can be found in Table 3. Plot D shows the perfusion pressure.

Figure 9

The muscle pump effect is shown with the absence and presence of the superficial venous system (Figure 2B and C). In plot A, the course of the extravascular pressure and the state of the proximal and distal valves are shown, the closed state is denoted with a solid line. Plots B and C show the arterial inflow and venous outflow, respectively. The gray areas indicate the extra perfusion (VPE), venous return (VVR), and reflux volume (VRE) of which the values can be found in Table 3. Plot D shows the volume inside the deep vein.

During the initial phase of the muscle contraction , increased venous outflow is observed in Figure 6C. In the second part of the contraction , the flow decreases and stabilizes under constant extravascular pressure . The extra volume venous return VVR is only slightly increased after the inclusion of venous valves. Along with increased venous outflow, a decrease in arterial inflow of 97% is observed during the contraction in the case where valves are included. When no valves are present, arterial flow decreases below zero to −6% of baseline arterial flow, resulting in backflow. In Figure 6A, VV1 is observed to close during part of the contraction.

In the subsequent relaxation phase (t > 3s), arterial flow increases, resulting in an extra perfusion volume VPE of 0.17 and 0.10 mL with and without valves, respectively. VV2 closes, and the venous flow reduces to zero when valves are included, resulting in a zero reflux volume VRE. Without valves, negative venous outflow results in a reflux volume VRE of 0.46 mL that refills the vein from the proximal side. Figure 6D compares the volume of the deep vein over time in both cases and reveals a faster refilling if no valves are included, quantified by the measurement of the refilling time without valves, tfill=1.69 s, and with valves, tfill=2.62 s (Table 3).

3.3 Effect of hydrostatic pressure

The effect of hydrostatic pressure was studied by comparing the hemodynamics in the supine and upright positions, with valves included in both cases (Figure 2B). The results from the two simulations are compared in Figure 7, which reports the extravascular pressure and the state of VV2 and VV1 (Figure 7A), the arterial inflow (Figure 7B), venous outflow (Figure 7C), and perfusion pressure (Figure 7D). Gray regions of Figure 7B and C indicate the extra volume due to perfusion VPE and venous return VVR for which numerical values are reported in Table 3.

During the contraction phase , similar flow patterns are observed in supine and upright positions, with a small increase of 0.04 mL in the venous return volume VVR in the upright position. In the subsequent relaxation phase (t > 3s), the arterio-venous perfusion pressure (Figure 7D) is elevated, while VV2 is closed, with a maximal difference of 2.3 kPa. This increased perfusion pressure is accompanied with an increase in arterial inflow and a 0.09 mL increase in perfusion volume VPE. The venous refilling process is slightly faster in the upright position as it reaches a 95% refilling in 2.45 s instead of 2.62 s (Table 3).

Figure 8 shows the spatial variation of internal venous pressure and cross-sectional area at different time points during the relaxation phase in upright position. The pressure gradient over the whole length of the vein is observed to be equal to −ρg at t = 6.0 s.

Figure 8

Spatial pressure and cross-sectional area at different point in time during the relaxation in upright position (Figure 2B). The gray areas represent the venous valves.

3.4 Effect of superficial system

The influence of the superficial venous system was studied by comparing the hemodynamics in a reduced and deep-superficial configuration (including one superficial and four perforating veins). Both models included valves and were simulated in upright position (Figure 2B and C, respectively). The results from the two simulations are compared in Figure 9, which reports the extravascular pressure and the state of VV2 and VV1 (Figure 9A), the arterial inflow (Figure 9B), deep venous outflow (Figure 9C), and total venous volume (Figure 9D). Gray regions of Figure 9B and C indicate the extra volume due to perfusion VPE and venous return VVR for which numerical values are reported in Table 3.

During the contraction phase , arterial inflow remained high when the superficial system was included (Δqart=6%), compared with a large reduction when only the deep system was included (Δqart=97%) (Figure 9 and Table 3). The venous return during the relaxation phase is slightly reduced with VVR=0.85mL in the deep-superficial configuration and VVR=0.92mL in the reduced configuration. In the subsequent relaxation phase, a faster refilling is observed in the more detailed configuration, quantified by the measurement of the refilling time in the reduced configuration, tfill=2.45s and including the superficial system, tfill=1.54s (Figure 9D and Table 3). The perfusion volume in the deep-superficial configuration is decreased by 0.10 mL compared with the reduced configuration.

Figure 10 shows detailed venous dynamics for the deep-superficial configuration, reporting the state of one perforating valve and the extravascular pressure (Figure 10A), the deep venous inflow (Figure 10B), superficial venous inflow (Figure 10C), and total perforating venous inflow (Figure 0D).

Figure 10

The muscle pump effect is shown with the absence and presence of the superficial venous system (Figure 2B and C). In plot A, the course of the extravascular pressure and the state of the proximal and distal valves are shown, where the closed state is denoted with a solid line. Plot B, C, and D show the deep venous, superficial, and total perforating venous inflow, respectively.

During the contraction phase , deep venous inflow decreases and superficial inflow increases. In the subsequent relaxation phase superficial inflow decreases, deep inflow increases, and perforator inflow increases. The perforating valve is closed during the whole contraction phase and alters state continuously after the relaxation .

4 Discussion

This study examined the main mechanisms of the muscle pump effect using a 1D pulse wave propagation model that includes venous collapsibility, gravity, and venous valves. Different configurations of the model were used to investigate the effect of venous valves, hydrostatic pressure, and the superficial venous system on the valve dynamics and hemodynamics during calf muscle contraction. The results presented in Section 3 show the details of the hemodynamics observed within each of the model configurations. The following discussion compares these results with other publications which report specific features of the venous behavior along with opportunities for further refinement of the modeling approach.

In the current study, muscle contraction was simulated by increasing extravascular pressure on the deep veins, resulting in a collapse developing from the proximal towards the distal side of the vein. This behavior is in agreement with the fast collapse near the outlet of the tube observed in the simulations and in vitro experiments of Marchandise et al. 21, which can physically be explained by the lower hydrostatic pressure at the proximal end. Collapse results in an increased venous outflow with a peak value of five times the baseline flow. These results are in line with the seven-fold flow augmentation observed numerically and in vivo by Fullana et al. 26 in the common femoral vein. The results presented in Section 3 support the in vivo observations of Crisóstomo et al. 34, that venous flow velocity augmentation is not significantly different between patients with venous valve insufficiency and healthy subjects. This can be considered by comparing results between the models with and without valves, where the absence of valves represents the extreme case of valve incompetence, which is often a feature of chronic venous disease. The increased arterial flow observed during the refilling shows the same behavior as observed by Fullana et al. 35: initial rapid increase, quasi-steady filling resulting in an almost linear decay in flow, and a final stage, where the flow goes back to its baseline value. The spatial and temporal courses of collapse are in line with previous studies.

The main effects of including venous valves were observed in the relaxation phase. Without valves, the refilling of the vein occurred from both the distal and proximal sides. Refilling from the proximal side resulted in a reflux volume of 0.46 mL, 53% of the extra venous return volume generated during the muscle contraction. The consequence of single-sided refilling, when valves are included, is an associated increase in both refilling time tfill and perfusion volume VPE. We found in this study that the proximal valve is essential to the efficiency of the muscle pump, while the distal valve does not significantly impede arterial backflow, as the high resistance of the micro-circulation prevents this, both with and without valves.

Including hydrostatic pressure in upright position increased the arterial inflow during the relaxation phase, as a result of the increased perfusion pressure. Although one might logically relate the increased perfusion pressure to hydrostatic pressure shielding by the proximal valve, the observed increase of perfusion pressure of 2.3 kPa cannot be accounted for by the hydrostatic pressure of the 5cm segment of vein proximal to the valve (0.5 kPa). Studying the spatial pressure distribution in the vein (Figure 8) reveals a pressure drop over the closed proximal valve higher than the contribution of the hydrostatic column alone. This phenomenon was also reported by Zervides et al. 22 in a numerical study and observed in vitro by Raju et al. 36. This pressure drop is associated with fluid redistribution from the proximal to distal section due to gravitational loading, lowering the pressure below the proximal valve, and thereby increasing the pressure drop over the valve. For the internal deep venous pressure, this means that it keeps decreasing until the end of the relaxation due to the low filling and the removal of the extravascular pressure. While filling progresses, the deep venous pressure increases until the proximal valve opens. Furthermore, the spatial pressure plot shows a pressure gradient along the vein, that is larger than the hydrostatic pressure gradient in the initial phase of the refilling (t < 4s), which is caused by the resistance of the collapsed area. A similar phenomenon is described by Young and Tsai 37,38 while studying the pressure drop over a stenosis. Our study suggests that the total pressure drop along the deep vein is determined by the state of refilling and the resistance of the collapsed area.

The most important effect of including the superficial venous system can be seen in the changes in the arterial flow during muscle contraction. In the configurations with only a deep venous system, a decrease in arterial inflow of 97% was observed during the muscle contraction, whereas arterial flow was only reduced by 6% when the superficial system was included in the model. The division of arterial flow between the deep and superficial veins (Figure 10) changes significantly during muscle contraction, with the deep venous inflow decreasing to zero and a large increase in superficial venous flow. This can be explained by the increased resistance of the deep vein as it collapses, with the superficial vein remaining circular with low resistance. In the subsequent relaxation phase, the flow in the superficial vein is observed to approach zero, whereas the deep venous inflow receives most of the arterial flow, due to the lower pressure in the deep vein. Although the perfusion volume decreases while including the superficial system, the refilling time is 37% shorter. The alteration of open and closed state of the perforating valve in the late relaxation phase seems a numerical artifact but might also be explained physiologically. During the late relaxation, the deep and superficial veins have a similar filling and therefore a similar pressure, which means the pressure difference between deep and superficial easily changes sign and therefore easily opens or closes the valve. Furthermore, it has to be noticed that these fast alterations in valve state have no influence on the general flow dynamics. Summarizing, the superficial system is found to be of importance for maintaining calf perfusion during the contraction phase and decreasing the refilling time in the relaxation phase.

The collapse course of the deep vein is determined by its A,p-relation, which is currently based on the work of Shapiro 19 and the bending stiffness of Müller and Toro 16. Using ex vivo or in vivo pressure and cross-sectional area measurements, as performed by Bassez et al. 39, could provide a more realistic relationship, resulting in a more accurate prediction of the volume shifts. However, we expect that effects of an experimentally determined tube law would not be significant for the qualitative study of the muscle pump. The course of the collapse could also be improved while using a more realistic course of extravascular pressure, which is now included as a smooth sinus function in order to have a clear distinction between the contraction and relaxation phases. An electromyogram of a muscle contraction could be used to provide a more realistic course of the extravascular pressure 40. Apart from the pressure originating from the muscles, the arteries, running in parallel with the veins, are also known to slightly compress the veins, which experience the arterial pulsatility as an extravascular pressure 41. However, this pressure is expected to be much lower than the extravascular pressure originating from the muscle contraction.

A second important part of the model are the valve elements, which are currently modeled as a resistance and an inertance, and open with a positive pressure gradient and close with negative flow. More advanced models could be used to capture the detailed physiology of venous valve dynamics 25. Including a continuous valve cross-sectional area would allow slow opening and closing, and enable modeling of valve regurgitation 24. The distal valve was shown to be of minor importance in impeding arterial backflow due to the high resistance of the micro-circulation. However, it has to be noticed that the current configuration is a simplified representation of the calf vasculature. Inclusion of a parallel branch representing the foot vasculature could increase the importance of the distal valve by blocking backflow towards the foot during the contraction. Furthermore, in this study, only a small number of valves were included to demonstrate the effect of valves at the proximal and distal positions. Anatomical data 10 suggests a higher number of valves in the calf veins. The inclusion of extra venous valves may result in an increased perfusion pressure if a full model of the circulation is considered, whereas this effect is expected to be small when only the calf is modeled.

The current model used a constant pressure boundary condition on both the arterial and venous sides, thereby neglecting the arterial and venous pulsatility. Preliminary model results showed that the inclusion of a pulsatile arterial boundary condition did not propagate significant pulsatility to the venous system and did not alter the mean flow dynamics. As venous pulsatility is very variable due to the large influence of breathing and right atrium contraction 42,43, a pulsatile venous boundary was not implemented. Therefore, development of a closed-loop model is necessary to investigate the influence of pulsatility on the muscle pump effect.

Although a maximal 1.5-fold increase in arterial flow is observed in the relaxation phase, it is much smaller than the up to seven-fold increase observed in vivo after a 4-s contraction of the calf muscle by Leguy et al. 33. However, the increase in arterial inflow during the first 5 s is known to be a combination of the muscle pump and rapid vasodilation 44, of which the latter is currently not included. Extending the model with regulation of the micro-circulation could help in better predicting the arterial flow during the relaxation phase. Previous modeling studies have successfully included local cerebral auto-regulation 45,46 including myogenic, metabolic, neurogenic, and endothelial regulation mechanisms, which could be adapted to simulate changes in calf vascular tone.

The muscle pump effect is not the only mechanism of importance for studying the fluid shift on orthostatic stress. Considering the importance of other regulatory mechanisms, such as the baroreflex and the effect of venous return on cardiac output during tilt 7, requires a more global circulatory model. Olufsen et al. 46 developed a closed-loop model to study the effect of tilt on hemodynamics while including global autonomic and cerebral auto-regulation. Furthermore, Heldt et al. 47 studied the effect of tilt and lower body negative pressure using a closed-loop lumped parameter model including the arterial and cardiopulmonary baroreflex. However, both models use a lumped parameter approach with a limited number of compartments. These models do not take into account pressure and flow wave propagation phenomena or collapsibility of the veins as reported in this paper. In future research, we aim to develop a global, closed-loop 1D pulse wave propagation model of the total circulation, including regulation mechanisms such as the baroreflex and local auto-regulation, to gain further insight into the mechanisms responsible for reducing fluid shifts during tilting.

A further advantage of a closed-loop model is that it allows the influence of the full hydrostatic column to be considered. To demonstrate the influence of this effect in the current model (reduced configuration with and without valves in the upright position), a ρgh term (with h = 1.0m) was added to the pressure at the inlet and outlet (pin and pout, respectively). The results of these simulations, shown in Figure 11 and Table 4, further highlights the importance of the venous valves as the reflux volume VRE increased to 67% of the venous return volume VVR (Figure 11 and Table 4). The abrupt increase of venous outflow at the end of the relaxation phase is most probably caused by the fact that, in this pressure regime, a small increase in the cross-sectional area is accompanied by a large increase in pressure. Furthermore, a higher perfusion was observed in the relaxation phase while the perfusion pressure increased.

Figure 11

The muscle pump effect is shown in the reduced geometry in case valves are excluded (blue) and included (red) with the full hydrostatic column on the boundary conditions (Figure 2A and B). In plot A, the course of the extravascular pressure and the state of the proximal and distal valves are shown, where the closed state is denoted with a solid line. Plots B and C show the arterial inflow and venous outflow, respectively. The gray areas indicate the extra perfusion (VPE), venous return (VVR), and reflux volume (VRE) of which the values can be found in Table 3. Plot D shows the perfusion pressure.

Table 4

Venous return (VVR), perfusion (VPE) and reflux volumes (VRE), refilling time tfill and the relative decrease in arterial inflow during the muscle contraction Δqart as indicated in Figure 11 and explained in Equation 20, 19s and 21.

	VVR[mL]	VPE[mL]	VRE[mL]	tfill[s]	Δqart[%]
No valves, upright, full pH (Figure 2A)	0.87	0.08	0.58	1.10	44
Valves, upright, full pH (Figure 2B)	0.86	0.35	0.00	1.77	44

Apart from studying the muscle pump effect, the current model has potential application to specific clinical conditions. Inclusion of leaking valves could inform understanding of the effect of valve insufficiency on venous hemodynamics. Furthermore, cuff compression is used to promote venous return in particular conditions, the model could be used to study different compression patterns to improve cuff efficiency, as suggested by Simakov et al. 48. A series of contractions is expected to result in lower filling of the veins, as a second contraction can start before the vein is fully refilled. This lower filling of the veins results in a lower pressure in the vein, maintaining a higher perfusion pressure. This lower venous pressure also results in a smaller pressure difference with the surrounding tissue, and therefore less edema is expected. Modeling a series of contractions therefore necessitates inclusion of filtration mechanisms and local auto-regulation.

5 Conclusion

A 1D pulse wave propagation model has been developed, which allows the study of lower limb hemodynamics during muscle pump activation. The model is able to predict the increase in venous return during muscle contraction. As the proximal valves close during the relaxation phase, reflux is prevented, which without valves resulted in a loss of 53% of effective venous return. Furthermore, the shielding function of the valves increased the perfusion in the relaxation phase. Finally, inclusion of the superficial venous system demonstrates the role of the superficial veins in maintaining arterial inflow during muscle contraction and decreasing refilling time by 37% during relaxation.