bs_bs_banner

Copyright © 2013 International Center for Artificial Organs and Transplantation and Wiley Periodicals, Inc.

Blood Flow in Hemodialysis Catheters: A Numerical Simulation and Microscopic Analysis of In Vivo-Formed Fibrin *Thabata Coaglio Lucas, †‡Francesco Tessarolo, *Victor Jakitsch, §Iole Caola, ¶Giuliano Brunori, †‡Giandomenico Nollo, and *Rudolf Huebner *Department of Mechanical Engineering, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil; †Department of Industrial Engineering, BIOtech Centre on Biomedical Technologies, University of Trento; ‡Healthcare Research and Innovation Program (IRCS), Bruno Kessler Foundation; §Section of Electron Microscopy, Department of Medicine Laboratory, Azienda Provinciale per i Servizi Sanitari di Trento; and ¶Department of Nephrology, S. Chiara Hospital, Trento, Italy

Abstract: Although catheters with side holes allow high flow rate during hemodialysis, they also induce flow disturbances and create a critical hemodynamic environment that can favor fibrin deposition and thrombus formation. This study compared the blood flow and analyzed the influence of shear stress and shear rate in fibrin deposition and thrombus formation in nontunneled hemodialysis catheters with unobstructed side holes (unobstructed device) or with some side holes obstructed by blood thrombi (obstructed device). Computational fluid dynamics (CFD) was performed to simulate realistic blood flow under laminar and turbulent conditions. The results from the numerical simulations were compared with the fibrin distribution and thrombus architecture data obtained from scanning electron microscopy (SEM) and two photons laser scanning microscopy (TPLSM) on human thrombus formed in catheters removed from patients. CFD showed that regions of

flow eddies and separation were mainly found in the venous holes region. TPLSM characterization of thrombi and fibrin structure in patient samples showed fibrin formations in accordance with simulated flux dynamics. Under laminar flow conditions, the wall shear stress close to border holes increased from 87.3 ± 0.2 Pa in the unobstructed device to 176.2 ± 0.5 Pa in the obstructed one. Under turbulent flow conditions, the shear stress increased by 47% when comparing the obstructed to the unobstructed catheter. The shear rates were generally higher than 5000/s and therefore sufficient to induce fibrin deposition. This findings were supported by SEM data documenting a preferential fibrin arrangement on side hole walls. Key Words: Central venous catheter— Hemodialysis—Computational fluid dynamics—Thrombus—Electron microscopy—Two photon laser scanning microscopy.

Although arteriovenous fistulas are recommended as the preferred choice of vascular access for hemodialysis patients, many patients dialyze using central venous catheters (CVCs) (1,2). Catheters guide the venous blood toward the hemodialysis machine and return it to the venous system in a safe and reliable way (2–4). The residence time for a

nontunneled catheter in the vein should theoretically not exceed 2 weeks (3). However, in clinical practice, catheters may remain within a patient’s vein up to 2 months to ensure sufficient lead time for fistula maturation. During the residence time, blood thrombi and fibrin sheaths can develop, impairing lumen patency and catheter performance (4). Nontunneled catheters are generally designed with side holes facing different directions on the catheter wall. These side holes may affect the flow pattern, shear stress distribution, flow separation and recirculation, thus providing a critical hemodynamic environment. Blood velocity and pressure fluctuations close to the catheter surface could cause a transition from a laminar

doi:10.1111/aor.12243 Received July 2013; revised September 2013. Address correspondence and reprint requests to Dr. Thabata Coaglio Lucas, Bioengineering Laboratory, Department of Mechanical Engineering, Federal University of Minas Gerais, Antônio Carlos Avenue 6627, 31270901 Belo Horizonte, Minas Gerais, Brazil. E-mail: [email protected] Artificial Organs 2014, 38(7):556–565

BLOOD FLOW IN HEMODIALYSIS CATHETERS to a turbulent flow regime and promote the onset of the blood clotting process with thrombus formation in the catheter holes. Microscopic structure and composition of venous thrombi have been previously reported (5,6), and we recently showed that blood thrombi formed in a hemodialysis CVC present a complex structure and are mainly composed of fibrin and erythrocytes (7). A fibrin sheath commonly develops within 24 h after CVC insertion at the region where the catheter contacts the vessel wall, and may completely enclose the proximal portion of the CVC in 5 to 7 days (8,9). Fibrin conformation is remodeled by shear stress and mechanical stimuli, which alter the structural integrity of the thrombus. Hemodynamic forces are modulators of fibrin structure and result in the development and progression of the thrombus. Numerical models based on computational fluid dynamics (CFD) allow the investigation of hemodynamic factors that lead to thrombus formation and progression. Given the microscopic threedimensional architecture of catheter thrombi, scanning electron microscopy (SEM) has been reported as a powerful technique for characterizing the surface structure (10–12). SEM has been applied to understand the extent of fibrin network formation and the organization of fibrin fibers over the thrombus and in the CVC holes (10). The same technique can provide information on the alignment of fibrin fibers in proximity to the CVC holes and tip, reflecting blood flow patterns in these crucial catheter regions. Two photon laser scanning microscopy (TPLSM), an optical technique based on fluorescence induced by two photon absorption, has recently been proposed as an additional tool for characterizing blood thrombi in a fully hydrated state without the need to section the sample (7,8). Recently, we reported the potential of TPLSM in imaging the fibrin structure of thrombi in CVCs removed from patients (6). All these methods allow a thorough investigation of the local blood flow and dynamic conditions in relation to fibrin deposition and blood thrombus formation resulting from the CVC presence in the vein. To provide further insights into the role of blood flow and catheter design in modulating thrombus formation, this work compared the blood flow in nontunneled hemodialysis catheters with and without obstruction of some side holes, and analyzed the influence of shear stress and the shear rate in fibrin conformation. The results obtained by CFD were discussed in the light of the fibrin patterns obtained by SEM and TPLSM on a set of nontunneled hemodialysis CVCs removed from patients.

557

MATERIALS AND METHODS Three-dimensional geometry and computational mesh A realistic CVC model, similar to the clinical device MedCOMP/HEMO-CATH (Harleysville, PA, USA), was designed in SolidWorks (SolidWorks, Inc., Concord, MA, USA). The CVC design was characterized by a dual-lumen system, arranged in a single shaft 2 mm in diameter and 20 cm in length. Figure 1a shows the realistic dual-lumen CVC that contains one outflow lumen with three venous outlet holes and a second inflow lumen with four arterial inlet holes.The internal cross-sectional areas of both the arterial and venous lumens were 2.93 mm2. Catheter holes were labeled as showed in Fig. 1. In the model of the unobstructed CVC, each hole was completely pervious. In the model for the obstructed CVC, venous holes #2 and #3 and arterial hole #6 were completely obstructed by a thrombus-type occlusion (red areas in Fig. 1b). In the obstructed CVC model, also the venous and arterial lumens were partially obstructed, reducing the original cross-sectional area to 2.04 and 1.33 mm2, respectively. According to this configuration, the measurements of flow parameters were obtained for holes #1, #4, #5, #7, and #8. A realistic model of the right and left internal jugular veins and

FIG. 1. The realistic model of an unobstructed (a) and obstructed (b) hemodialysis CVC used in the CFD simulations. The numbers indicate the labels as referred in the text: venous hole tip (#1), three venous side holes (#2, #3, #4), four arterial side holes (#5, #6, #7, #8). The red areas in the obstructed CVC indicate the shape of the simulated CVC thrombi, obstructing arterial hole (#6) and venous holes (#2 and #3). Artif Organs, Vol. 38, No. 7, 2014

558

T.C. LUCAS ET AL.

FIG. 2. Design and boundary conditions of the CFD model. Three-dimensional geometry of the jugular veins (light gray) and the intravenous portion of the hemodialysis catheter (dark gray). Arrows indicate blood inlets (where velocity data presented in the upper graph were applied) and the blood outlet (where pressure data presented in the lower graph were applied as a boundary condition). The temporal evolution of velocity and pressure during a whole cardiac cycle are reported indicating diastole (D), atrial contraction (A), systole (S), atrial filling (V).

superior vena cava up to the right atrium were obtained by segmentation of CT images from the Visible Human Project (US National Library of Medicine, Bethesda, MD, USA) (Fig. 2).Two hundred axial anatomical images (2048 × 1216 pixels) at 1-mm intervals were used to create the jugular vein geometry. The vein model was imported into SolidWorks and bounded by interpolating polynomial curves. A mesh independency study was carried out performing simulations with coarse to fine mesh sizes until results were no longer influenced. The results exhibited a maximum of 3% difference. The mesh resolution was Artif Organs, Vol. 38, No. 7, 2014

maximized near the catheter and vein walls and consisted of approximately 1.3 × 106 nodes, 5.0 × 106 tetrahedral elements, and 6.3 × 102 pyramidal elements. A boundary layer consisting of eight rows was established, with an expansion factor of 1.2 and a total depth of 1 mm. Boundary conditions and numerical simulation A laminar and turbulent transient flow, k–ω Shear Stress Transport (SST) model, was implemented in ANSYS CFX 12.1 (ANSYS-Fluent, Inc., Lebanon, NH, USA). The SST model was designed to give

BLOOD FLOW IN HEMODIALYSIS CATHETERS highly accurate predictions of the onset and the amount of flow separation under adverse pressure gradients by the inclusion of transport effects into the formulation of the eddy-viscosity. Therefore, this model was substantially more accurate for determining the wall shear stress in the internal hole wall of the catheter. The time-dependent and incompressible Navier-Stokes and Carreau-Yassuda (C-Y) model equations were used for the mathematical description (9,10). The following set of parameters was used for blood analog fluid: μ∞ = 3.45 × 10−3 Pa·s, μ0 = 56 × 10−3 Pa.s, a = 1.25, n = 0.22, and λ = 1.19 s (9), where μ∞ is the viscosity at infinite shear rates, μ0 is the zero shear rate viscosity, λ is the relaxation time in seconds,a and n parameters can be varied to match the shear thinning behavior of blood. Blood viscosity in humans is non-Newtonian.With increasing shear rate, the viscosity decreases. In this work the values of μ∞ and μ0 were obtained from optimum hematocrit in the physiological plasma viscosity. The viscosity increases exponentially with increasing hematocrit. The C-Y model is widely accepted and used to correct the variations in blood viscosity caused by a variation in hematocrit (7,9). Blood density was set equal to 1050 kg/m3. The values for the turbulent kinetic energy,specific rate of turbulence,eddy frequency,and turbulent viscosity were modeled according to Abraham et al. (10). The duration of a cardiac cycle was assumed to be 0.8 s, yielding a heart rate of 75 beats per minute. A second-order backward Euler method was used for the time integration. The doubling time step was performed, and no obvious differences in the results of the mass flow rate in the veins were observed. A constant time step was employed, where Δt = 0.0005 s with 1600 total time steps per cardiac cycle.The waveforms of the jugular blood inlet velocity and atrium pressure are presented in Fig. 2 and were approximated from the experimental curves found in the literature and intravenous color Doppler lines (11). The (A) wave occurs when the atrium contracts, increasing atrial pressure.At the same time, the blood is propelled in a retrograde direction toward the veins.When the tricuspid valve closes, the systole wave (S) occurs. The transitional (V) wave corresponds to atrial overfilling against a closed tricuspid valve, anticipating the opening of the valve in diastole (D). The inlet and outlet waveforms designed for this study showed a correlation between flow velocity in the internal jugular veins and the pressure in the right atrium. The model was initially configured with flow data from a steady-state simulation with an inlet velocity of 0.18 m/s for the right internal jugular vein and 0.20 m/s for the left internal jugular vein. The outlet pressure of the vein was 533.28 Pa. Mass flow in

559

the CVC venous lumen was set to 0.0052 kg/s. The outlet for the CVC arterial lumen had a pressure of 33 330.60 Pa. Parameters were set according to National Kidney Foundation guidelines for hemodialysis catheters (3). The results obtained for the steady-flow calculations were used as the initial values for unsteady flow, and five cycles of calculations were conducted for checking variations in the parameters. A no-slip boundary condition was imposed along the vessel walls. The continuous variables were expressed as means ± SDs. Characterization of thrombus and fibrin structure in CVC removed from patients Samples of real CVC devices with blood thrombi obstructing some of the side holes were obtained from a collection of 23 nontunneled CVCs (MedCOMP/ HEMO-CATH; median insertion time 17 [5–150] days) gathered during a recent observational clinical trial performed from August 2011 to July 2012 at the Nephrology Department of the Trento Hospital in Italy. The study was approved by the Ethics Committee Concerning Research Involving Human Subjects and Clinical Trials (Process 04/2010) of Trento, Italy. Samples were treated as previously reported (6). Briefly, CVCs were rinsed in sterile saline and fixed in 4% buffered formaldehyde immediately after removal from the patient. Segments of the CVC containing a venous catheter hole were considered for analysis by SEM and TPLSM. SEM analysis was conducted after sample dehydration in ascending alcohol solutions, drying in a laminar flow cabinet overnight and sputter-coating with gold. Sample imaging was performed with a XL30 ESEM FEG scanning electron microscope (FEI, Philips,Amsterdam,The Netherlands). Samples for TPLSM were washed for 2 min in deionized water and stained by immersion in 1 mM Rhodamine 6G (Sigma-Aldrich, St. Louis, MO, USA) for 20 min in the dark. Sample imaging was performed with a two photon laser scanning microscope (Ultima IV, Prairie Technologies, Middleton, WI, USA). An ultra-short pulsed laser (Mai Tai Deep See HP, Spectra-Physics, Santa Clara, CA, USA) served as the light source. The excitation wavelength was set to 800 nm, and the fluorescence was collected in the red band (607 ± 20 nm). The images were acquired with a 100× objective lens (NA 1.0, water immersion, Olympus LUM Plan FI, Center Valley, PA, USA) and a dwell time of 50 μs. RESULTS Velocity vectors and streamlines Representative streamlines for the turbulent flow into the vein in the unobstructed and obstructed Artif Organs, Vol. 38, No. 7, 2014

560

T.C. LUCAS ET AL.

FIG. 3. Streamline profiles in the vein and within CVC holes for the unobstructed (left) and obstructed catheter (right). The central part of each image shows a comprehensive view of the vein and the proximal portion of the catheter, where side holes are located. Insets present a closer view of venous (V) and arterial (A) side holes.

catheter shaft at t = 0.4 s (corresponding to the systole) are shown in Fig. 3. The flow pattern was nonuniform and irregular in both the patent and occluded catheter models. A recirculation zone and stagnation point, due to the outlet flow from the venous hole, can be observed in the enlarged view of Fig. 3. However, recirculation eddies with reverse streamlines and changes in direction over time were markedly more evident in the obstructed catheter model. The microscopic characterization of the corresponding area by TPLSM and SEM showed that fibrin fibers were not aligned in a parallel fashion but were frequently arranged in circles following flow disturbances (Figs. 4a and 5a). Fibrin fibers bifurcate and exhibit a change in alignment and arrangement. Twisted fibers were also observed. Wall shear stress and shear strain rate The shear stress in the hole wall #4 under turbulent flows (t = 0.4 s) is shown in Fig. 6a,b for the unobstructed and obstructed catheter models, respectively. Figure 6a,b shows the internal hole wall that was rotated to view peaks in the shear stress that were high on the wall inside the side holes from obstructed catheters. Maximal shear stress values reach 1000 Pa between the edge of the bottom of the hole wall and the wall of the catheter shaft. The average strain rate value in the two-catheter model differed by about 35% in the laminar flow and 38% in the turbulent flow conditions. Arterial hole #8 was subjected to the highest wall shear stress and shear strain since the blood velocity was maximal. Wall shear stress and shear strain rates for the other pervious holes are summarized in Table 1. The SEM micrograph of the internal venous hole wall (Fig. 4c) showed a Artif Organs, Vol. 38, No. 7, 2014

preferential deposition of fibrin in accordance to the areas with highest shear stress indicated in Fig. 6b. Figure 6c,d shows the distribution of the shear strain rate on the external surface of the catheter around venous hole #4 in both the unobstructed and obstructed catheters. Simulations showed the presence of catheter areas subjected to strain rates higher than 5000 and 10 000/s for the unobstructed and obstructed catheter models, respectively. DISCUSSION The computational parameters provided a flow pattern that was in agreement with the microscopic arrangement of the fibrin revealed by SEM and TPLSM on in vivo-formed thrombi. The overall flow patterns within the vein and around the venous side holes contained regions of flow eddy and flow separation in both the obstructed and unobstructed catheter models. Previous studies reported that platelet aggregation preferentially occurs at regions of low shear stress, that is, in flow eddies and in flow separation regions (2,12,13). In the flow eddy zone, the wall shear stress has been reported to remain at a low level (0.0–1.0 Pa) throughout the cardiac cycle due to the reversed flow conditions (14,15). The velocity vectors showed that flow eddies were not present in front of the arterial holes, but were seen downstream of the venous hole. Differently, flow separation and stagnation regions were observed close to the catheter wall, between the arterial holes. These findings suggested that the flow disturbance in the venous hole regions can occur more frequently than in arterial hole regions, thus creating conditions for accelerating the development of blood thrombi (16).

BLOOD FLOW IN HEMODIALYSIS CATHETERS

FIG. 4. SEM images of fibrin formed in vivo on CVCs removed from patients. (a) Fibrin plaque on the catheter surface around a venous side hole. The heterogeneous direction of fibrin strands, with branching points related to the presence of high shear strain rate as shown in Fig. 6d. (b) Unobstructed venous side hole (corresponding to positon #4) with initial deposition of fibrin layer. Fibrin was found preferentially on the hole wall, coherently to the area with highest shear stress indicated in Fig. 6b. (c) Close up view of the initial fibrin layers formed on the wall of the side hole. Original magnification 500× (a), 39× (b), and 500× (c).

561

FIG. 5. TPLSM of fibrin plaque formed in vivo on CVC removed from patients. (a) Fibrin plaque microstructure in an area subjected to high shear strain rate. Fibrin strands and bundles have anisotropic directions and many branching points are present (arrows), forming circles of fibers. (b) Fibrin plaque on the catheter surface around a venous side hole, in the corresponding area where shear strain rate was high, as shown in Fig. 6d. The dashed line indicates approximately the position of the hole rim. Fibrin plaque is thicker and more structured close to the hole. Only sparse fibrin network is present in the areas subjected to low shear strain rate (lower part of the micrograph). Round-shaped structures close to hole rim are white blood cells. Bar is 25 μm.

Artif Organs, Vol. 38, No. 7, 2014

562

T.C. LUCAS ET AL.

FIG. 6. Distribution of shear stress and strain at CVC surface of venous hole #4 (indicated by the arrow). Wall shear stress on the hole profile under turbulent flow for the unobstructed (a) and obstructed (b) catheter scenario. Shear strain rate around the hole rim for the unobstructed (c) and obstructed (d) catheter.

TABLE 1. Shear stress in the internal wall of the side holes and shear strain rate close to the hole rim Laminar flow Wall shear stress (Pa) (Mean ± SD)

Venous hole #4 Arterial hole #8 Arterial hole #7 Arterial hole #5

Turbulent flow

Shear strain rate (per s) (Mean ± SD)

Wall shear stress (Pa) (Mean ± SD)

Shear strain rate (per s) (Mean ± SD)

Unobstructed catheter

Obstructed catheter

Unobstructed catheter

Obstructed catheter

Unobstructed catheter

Obstructed catheter

Unobstructed catheter

Obstructed catheter

87.3 ± 0.2 278 ± 3 70.9 ± 1.5 2.5 ± 0.3

176.2 ± 0.5 284 ± 4 72 ± 2 7.3 ± 0.3

8 047 ± 18 16 000 ± 150 5 302 ± 60 391 ± 39

12 400 ± 7 6 000 ± 150 5 795 ± 78 1 014 ± 32

130.4 ± 0.2 295 ± 3 64 ± 1 2.0 ± 0.3

244.1 ± 0.1 299 ± 3 66.7 ± 1.5 5.5 ± 0.3

6 449 ± 18 11 103 ± 130 3 477 ± 54 272 ± 33

10 497 ± 5 11 090 ± 120 3 794 ± 59 632 ± 40

Note: Spatial distribution was calculated at t = 0.4 s. Wall shear stress and shear strain rate values were averaged on the entire internal wall of the side hole as seen in Fig. 6a,b. Artif Organs, Vol. 38, No. 7, 2014

BLOOD FLOW IN HEMODIALYSIS CATHETERS In addition, the blood flux among the venous holes is separated into downstream high-velocity jets exiting from the three venous holes and low-velocity circulation along the catheter surface wall, as shown in the enlarged view in Fig. 3. These conditions are a plausible explanation for the arrangement of the fibrin plaque on the hole wall, which can be associated with the flow separation region between the three venous holes. In these areas, the fibrin attaches to the catheter wall and propagates along the direction of flow, generating a lateral aggregation close to the hole border and eventually forming a fibrin plaque that can obstruct the hole. To prevent catheter clotting, heparin typically is placed into the catheter at the end of dialysis as a catheter lock (8,10). In spite of the fact that multiple side holes provide good blood flow, they also may allow the catheter lock to seep out, thus increasing the risk of clotting. Seepage of the catheter-locking solution increases the risk of clotting and thrombus formation between the last of the side holes and the catheter tip. Real-time TPLSM imaging of the thrombogenesis process in mice subjected to laser injury of the arterial intima showed that also the neo-formed thrombus perturbs the blood flow (7). Red blood cells and platelets were observed to move slowly in the regions of vortices and were captured and incorporated into the growing thrombus (7). Although we were not able to follow thrombus formation in patients in real time, TPLSM imaging of thrombi formed on CVCs removed from patients showed fibrin plaques that develop in a corresponding region containing flow eddies. Furthermore, the presence of circles and twisted layers in the fibrin plaque that we have observed by both SEM and TPLSM techniques, suggested the presence of eddy flow close to the hole rim (2). In this region, the fibers bifurcate, and alterations in the arrangement of the fibrin in the thrombi were observed also by TPLSM as reported in Fig. 5a. The comparison of flow parameters between unobstructed and obstructed device showed that the shear stress value in the internal wall increased by 50% in venous hole #4 under a laminar regime and by 47% under turbulent flow conditions. It has been reported that higher shear stress can induce conformational changes in fibrinogen adsorbed onto a polymer, thus increasing the binding surface area between the fibrinogen and platelet receptors (17,18). At an elevated shear strain rate, typically in excess of 10,000/s, binding between the glycoprotein receptor of platelets and von Willebrand factor (vWf) is known to be unique, with the capability to recruit fast-flowing platelets at a high shear rate in the absence of any other platelet activators (19). Mareels

563

et al. (20) proposed a numerical study for simulating the blood flux in a CVC with side holes reduced in size. When the diameter of the side holes was halved, the shear rate increased near the side hole walls and on the lumen wall near the side holes. Conversely, in this study, the side holes that were not closed in the obstructed catheter showed an increase of the shear rate on the internal wall of the side hole. Although in this study the diameter was not reduced, the shear rate was particularly high on the internal wall of the side hole that was not closed in the obstructed catheter. We found a shear rate higher than 10000/s around the venous hole #4 in the obstructed catheter model and around the arterial hole #8 under both turbulent and laminar flow conditions. This factor is a plausible explanation for the establishment of additional bonds, leading to elevated fibrin development and subsequent thrombus formation in these holes. In contrast, at shear rates

Blood flow in hemodialysis catheters: a numerical simulation and microscopic analysis of in vivo-formed fibrin.

Although catheters with side holes allow high flow rate during hemodialysis, they also induce flow disturbances and create a critical hemodynamic envi...
785KB Sizes 0 Downloads 0 Views