Why Patencies of Femoropopliteal Bypass Grafts with Distal End-to-End Anastomosis are Comparable with End-to-Side Anastomosis

Abstract

Objective: Despite the theoretical favourable hemodynamic advantage of end-to-end anastomosis (ETE), femoropopliteal bypasses with distal ETE and end-to-side anastomosis (ETS) have comparable clinical patencies. We therefore studied the effects of different in vivo anastomotic configurations on hemodynamics in geometrically realistic ETE and ETS in vitro flow models to explain this phenomenon.

Methods: Four ETE and two ETS models (30° and 60°) were constructed from in vivo computed tomography angiography data. With flow visualization physiological flow conditions were studied.

Results: In ETS, a flow separation and recirculation zone was apparent at anastomotic edges with a shifting stagnation point between them during systole. Secondary flow patterns developed with flow deceleration and reversal. Slight out of axis geometry of all ETE resulted in flow separation and recirculation areas comparable to ETS. Vortical flow patterns were more stable in wider and longer bevelled ETE.

Conclusion: Primary flow disturbances in ETE are comparable to ETS and are related to the typical sites where myointimal hyperplasia develops. In ETS, reduction of anastomosis angle will diminish flow disturbances. To reduce flow disturbances in ETE, the creation of a bulbous spatulation with resulting axial displacement of graft in relation to recipient artery should be prevented.

Introduction

Midterm and late failure of femoropopliteal bypass grafts are mainly due to stenosis at the distal anastomosis as a result of myointimal hyperplasia (MIH).¹⁾ The development of MIH is complicated and a multitude of factors have been implicated including graft material, suture line stress, compliance mismatch, cellular and humoral factors, blood flow rate and flow patterns. Wall shear stress (WSS) is considered one of the most important factors. MIH is associated with areas of low WSS and/or areas of high spatial or temporal oscillations in WSS. Low WSS is usually found in areas of stagnant or low flow, flow separation and recirculation.²⁾

The geometry of the anastomosis is an important determinant of flow patterns and WSS distribution.³⁾ It has been shown that the angle of an end-to-side (ETS) anastomosis can affect the local flow pattern.^4–6) In vitro studies have also demonstrated altered flow patterns in ETS anastomoses with a patent proximal outflow segment.^7–10) Unfavourable flow patterns associated with the ETS anastomosis may be avoided by using an end-to-end (ETE) anastomosis configuration. A distal ETE anastomosis in a femoropopliteal bypass provides a physiologically and anatomically more correct configuration than an ETS anastomosis – since there is no anastomosis angle or proximal outflow. In an extensive in vitro study carried out in 1960, Szilagyi, et al.¹¹⁾ showed the favourable hemodynamics of ETE anastomoses. Sottiurai, et al.¹²⁾ demonstrated in an animal study that MIH formation was reduced in distal ETE anastomoses.

In a retrospective clinical analysis comparing distal ETE and ETS anastomoses in femoropopliteal bypasses with long term follow up, no statistically significant improvement was found in favour of ETE anastomosis patency as well as ischemic complications.^13–17) In a prospective multicenter clinical study,¹⁸⁾ in which 328 femoropopliteal bypasses were randomized for ETE or ETS anastomosis and followed for three years, primary patency did also not differ between the two types of anastomosis. Major amputation, however, occurred more frequently after failure of an ETE anastomosis (n = 20) than in ETS anastomosis (n = 9).

Although animal and in vitro studies suggest a better patency with ETE anastomosis, clinical data do not seem to support this assumption. We therefore aim to analyse the hemodynamic flow patterns in distal ETE and ETS anastomosis to gain a better insight in factors influencing bypass patency. For this purpose we used ETE and ETS in vitro models derived from three-dimensional computed tomography (CT) data of in vivo femoropopliteal bypass grafts.

Materials and Methods

Model preparation

Transparent silicone rubber models were constructed using a rapid prototyping process.¹⁹⁾ Three-dimensional AngioSpiral CT data of in vivo distal above the knee anastomoses of polytetrafluoroethylene (PTFE) femoropopliteal bypasses were converted into planar symmetrical three dimensional computer models. Four variations of ETE anastomoses (i.e., wide, normal, narrow and short) and two variations of ETS anastomoses (i.e., 30° and 60° angles) were constructed. Solid epoxy resin models were manufactured using laser stereolithography to create negative silicone rubber split moulds. These negative moulds were used to create positive models out of low melting-point metal alloy casts that were then embedded in transparent silicone rubber blocks. After curing of the silicone rubber, the alloy was melted out and the models were finally prepared for flow visualization (Fig. 1).

Fig. 1.

Two examples of three-dimensional AngioSpiral CT images of in vivo distal anastomoses of PTFE femoropopliteal bypasses (top images; ETS and ETE). Line drawings are planar symmetrical three dimensional computer models converted from corresponding CT images. Two variations of ETS anastomoses models (i.e., 30° and 60° angles) and four variations of ETE anastomoses models (i.e. wide, normal, narrow and short) were constructed and are shown as photographs. CT: computed tomography; ETE: end-to-end anastomosis; ETS: end-to-side anastomosis.

Set-up

All models were placed in a flow rig in which the outflow resistances could be accurately set. Pulsatile flow was obtained by superimposing an oscillatory component generated by a servo-controlled piston pump upon the steady flow from the constant head tank. The flow waveform at the inlet of the model was measured using an ultrasound transit-time flow meter (HT 107,Transonic Systems, Inc., Ithaca, New York, USA)

The circulating fluid was an aqueous solution of 40 wt% of glycerol, 0.1% neutrally buoyant 75 µm polystyrene particles (Eastman Kodak company, Rochester, New York, USA) and surfactant (0.01% sodium lauryl sulphate). A 100 µm filter was placed in the reservoir to remove any aggregates.

Accurate measurement of flow in the model anastomosis in orthogonal planes required intense illumination and exact positioning. The anastomosis was therefore illuminated with planar light provided by a 16 mW Helium-Neon laser equipped with a cylindrical lens line-generator and mounted on a micro positioner. A 75 mm focal length plano-convex lens was employed to concentrate the sheet of light to a thickness of app. 0.5 mm. The flow patterns were imaged with CCD monochrome video camera (Hitachi KP-M1E/K, Hitachi Kokusai Electric Inc., Tokyo, Japan) and recorded on S-VHS videotape. Photographs were captured at specific time points in the cardiac cycle as shown in Fig. 2.

Fig. 2.

Resting and exercise flow waveform during one cardiac cycle. Capital letters correspond to the specific measurements during the cycle: A, start cardiac flow cycle; B, mid-acceleration phase of systole; C, near peak systole; D, mid-deceleration phase of systole; E, end of systole; and F, mid diastole.

Flow conditions

All models were studied under steady and pulsatile flow conditions. The pulsatile flow waveforms used were derived from actual duplex scans in patients several weeks after femoropopliteal bypass operations. Average volume flow rate, the ratio of distal to proximal outflow and pulsatility index (PI) in resting and post-exercise conditions were used as standard settings for the in vitro flow experiments. Impedance in each individual anastomosis and for all different circumstances was set by adjusting the outflow aperture valve and by adjusting the height of the outflow fluid column. After setting the flow system, the pulsatility index and flow spectrum was identical to the in vivo circumstances. In the resting condition, mean flow rate was set to 250 ml/min with PI of 6.5 at 60 beats per minute. In the exercise condition, volume flow was set to 400 ml/min, with PI of 1.7 at 75 beats per minute. The ratio of distal to proximal outflow was 2:1 in the ETS anastomosis models in case of a patent proximal outflow. Occlusion of the proximal outflow artery was simulated by cross clamping the proximal outflow tube at 10 cm from the anastomosis.

Results

Figure 3 shows the flow patterns observed in: (1) the ETE model with normal configuration, (2) the 30° ETS model and (3) the 30° ETS model with 2:1 flow division ratio, all with resting and exercise flow conditions. Flow waveforms are depicted in Fig. 2.

Fig. 3.

Flow patterns observed in the ETE model (normal configuration), 30° ETS model and 60° ETS model with 2:1 flow division ratio with resting and exercise flow conditions at various points during the cardiac cycle (see Figure 2). Flow direction is from left to right. ETE: end-to-end anastomosis; ETS: end-to-side anastomosis.

ETS anastomosis – 30°

At the start of the flow cycle in resting flow conditions (Fig. 2A and 3A), flow velocities were low throughout the anastomosis as indicated by the short particle path lines. In mid-acceleration phase of systole (Fig. 2B and 3B), flow was predominantly laminar with low velocity at the graft hood and flow separation at the heel of the anastomosis. The stagnation point on the artery floor formed approximately one third of the length of the anastomosis from the heel. At peak systole (Fig. 2C and 3C), recirculation regions were present at the graft hood and in the recipient artery distal to the toe and proximal to the heel. High flow velocity in the graft was directed towards the artery floor and was diverted into the recipient artery where strong helical flow was generated in the distal and proximal outflow segments. During the mid-deceleration phase (Fig. 2D and 3D), flow in the upper part of the anastomosis was laminar with no evidence of flow separation or low velocity at the hood. The fluid in the proximal outflow came to rest and the flow separation at the toe increased occupying the top half of the artery. At the bottom half of the anastomosis, no coherent flow structures could be recognised. The stagnation point on the floor moved approximately 4 mm distally from its position in mid-acceleration. At end of systole (Fig. 2E and 3E), retrograde flow was observed in the graft. This reversal of flow was observed in the artery proximal to the anastomosis, whilst in the distal segment, fluid particles came to rest. This flow reversal caused entrainment of fluid in the anastomosis and forming a brief vortex towards the distal part of the anastomosis. Towards the mid diastole (Fig. 2F and 3F), flow from both artery segments returned into the anastomosis forming a slow anticlockwise vortex the graft hood. A schematic presentation of the observed specific flow patterns in the 30° ETS anastomosis is shown in Fig. 4A and 4B.

Fig. 4.

Schematic presentation of various flow patterns at different cardiac cycle moments. A, flow patterns in the 30° ETS model during systole; B, flow patterns in the 30° ETS model during diastole; C, flow patterns in the ETE model during systole; D, flow patterns in the ETE model during diastole. Systole corresponds with phase ‘B’ and diastole corresponds with phase ‘E’ as depicted in Figure 2. CL-A, proximal central lumen line; CL-B, distal central lumen line; ETE: end-to-end anastomosis; ETS: end-to-side anastomosis; CL: central lumen line. OoA, out of axis.

At the start of the cardiac cycle in exercise flow conditions (Fig. 2A and 3A), laminar flow was observed in the graft and distal outflow, whereas low velocity was present at the hood. Fluid particles came almost to a stop in the proximal outflow segment. The same flow pattern continued throughout mid-acceleration phase (Fig. 2B and 3B), except for a flow division on the anastomosis floor and increased flow at the proximal outflow. After peak systole (Fig. 2C and 3C), the stagnation point has moved distally. At this point, a recirculation zone has developed at the hood and flow separation regions were observed at the heel and toe. Helical flow was present in both the proximal and distal outflow segments of the artery. Flow separation in the outflow segments increased during deceleration (Fig. 2D and 3D). A large vortex formed at the heel of the anastomosis. Between the end of systole (Fig. 2E or 3E) and mid diastole (Fig. 2F and 3F), a vortex occupied the whole anastomosis and diminished in size and strength and finally shifts towards the anastomosis floor. Fluid velocities in the proximal outflow decreased, whereas flow became laminar again at the distal outflow.

ETS anastomosis – 60°

All flow patterns in the 60° ETS model were similar as in the 30° ETS model. Both in exercise and in resting conditions these flow patterns were more pronounced in the 60° ETS model.

ETS anastomises (30° and 60°) – occluded proximal outflow

Flow patterns in ETS anastomoses with an occluded proximal outflow artery were less complex and more stable. Vortex formation in the bulbus was seen for a very short period in mid-acceleration. In case of exercise conditions, areas of recirculation and re-attachment increased although the basic flow patterns were comparable to the flow patterns observed in systole.

ETE anastomosis – standard size

At the start of the cardiac cycle in resting flow conditions (Fig. 2A and 5A), the velocities were low throughout the anastomosis and flow remained laminar as the velocity increased during the acceleration phase (Fig. 2B and 5B). At peak velocity (Fig. 2C or 5C), the jet-like inflow into the anastomosis led to flow separation and recirculation. This was observed more at the bottom than at the top of the anastomosis. A helical flow pattern developed at the bottom of the arterial outflow. This became more pronounced during deceleration (Fig. 2D and 5D) whereas the flow separation regions in the anastomosis also became more extensive. At the end of systole (Fig. 2E and 5E), the particles were essentially stationary, although recirculation occurred at the top and bottom of the anastomosis. As in the ETS anastomosis, between end of systole (Fig. 2E and 5E) and end diastole (Fig. 2F and 5F), the direction of flow in the artery and graft was retrograde. This led to flow separation and anticlockwise recirculation at the top of the anastomosis. This recirculation region increased into a larger but weaker vortex that gradually moved towards the anastomosis floor before dissipating. A schematic presentation of the observed specific flow patterns in the ETE anastomosis is shown in Fig. 4C and 4D.

The flow patterns in the exercise flow condition are similar to those in the resting flow condition during the systolic phase. Due to the higher flows, flow separation regions on both sides of the anastomosis, as seen from the top, were more extensive and were present throughout the cardiac cycle. The helical flow in the artery appeared stronger, particularly during the deceleration phase (Fig. 2D and 5D).

ETE anastomosis – bulbous size

The effect of bulbous size of the anastomosis on the flow patterns in resting flow conditions is shown in Fig. 5. We present only the flow patterns during systole since the most important changes occurred during this part of the cardiac cycle. The flow was laminar in all models during mid-acceleration (Fig. 2B and 5B). At peak systole (Fig. 2C and 5C), flow separation regions at the top and bottom of the anastomosis were largest in the wide anastomosis and smallest in the short and narrow end-to-end models. Helical flow was observed in the wide, standard and narrow models but not in the short bevelled anastomosis. During the deceleration phase (Fig. 2D and 5D), the helical flow became more pronounced and the flow separation regions increased slightly in size in all models. In the short bevelled model, there was evidence of a small region of helical flow at the bottom of the artery.

Fig. 5.

Flow patterns observed in four different ETE models with resting and exercise flow conditions at various points during the cardiac cycle (see Figure 2). Four types of ETE are shown: wide, normal, narrow and short. Flow direction is from right to left. Lengths of arrows are qualitative. ETE: end-to-end anastomosis.

Discussion

Experimental studies have shown that local hemodynamic flow patterns are strongly influenced by anastomosis geometry. The majority of these studies have focused on flow patterns in ETS anastomosis.^20–22) Although ETE anastomoses are theoretically considered to have better patencies, clinical studies lack the evidence for this assumption. By using high-resolution CT angiography, six different but representative ETS and ETE models were constructed from in vivo anastomosis and used to investigate local hemodynamics in this study.

In both ETS and ETE anastomoses, flow separation and recirculation areas were present and more pronounced during resting condition (high pulsatility index; PI = 6.5) as compared to exercise conditions (low pulsatility index; PI 1.7). Physiological reversal of flow direction in diastole resulted in more complex flow patterns. A shifting stagnation point at the floor of the anastomoses with acceleration and deceleration was only apparent in ETS anastomoses. These observations were more pronounced in ETS anastomoses with angle of 60° than those with 30° angle. In ETE anastomoses, no stagnation point was apparent and secondary flow patterns were less complicated when compared with ETS anastomoses.

In areas of a shifting stagnation point, i.e., only in ETS anastomosis, the graft or vascular wall is exposed to changing flow directions that result in oscillatory wall shear stress (WSS).²⁰⁾ At locations of flow separation, recirculation and flow disturbances, relatively high and/or low velocities may result in corresponding low mean WSS or high WSS gradients. Although poorly understood, abnormal WSS distribution is related to MIH formation. The areas of flow separation, recirculation and stagnation zone correspond well to the previously identified three zones prone for myointimal hyperplasia.^12,23) In areas of low flow, i.e., low WSS, aggregation and accumulation of platelets near the walls and/or humoral factors may stimulate endothelial cells and smooth muscle cells to form MIH.²⁴⁾ In areas of high flow, however, sufficient washing of these factors directly reduces the endothelial response and thus the formation of MIH. As can be deduced from these flow visualization studies, low WSS is predominantly apparent during the diastolic phase of the cardiac cycle. MIH is likely to form during that phase. In addition, WSS is dependent on flow velocity and flow direction, which is directly influenced by the anastomosis geometry. As a consequence, changing anastomosis geometry to increase laminar flow and sufficient washout of blood during diastole may eventually reduce MIH formation.

In ETS anastomosis, a patent proximal outflow artery influences the flow patterns and contributes to flow disturbances in the proximal artery itself. This may lead to increased WSS and MIH, eventually leading to stenosis and occlusion of the proximal outflow artery.²⁵⁾ This natural modelling ultimately leads to an ETE anastomosis.

In ETE anastomosis, flow separation and recirculation was seen at the hood and bottom during deceleration at the end of systole, due to the out of axis geometry, being more pronounced in the longer and wider bevelled anastomoses. Also recirculation areas were seen resulting in low flow along the lateral anastomosis walls especially in the wide ETE models. A stabilizing vortex in the bulb of the anastomosis was identified during reversal of flow due to the typical out of axis connection of graft and artery. This vortex, which maintained fluid circulation along the anastomosis walls, was observed throughout most of diastole. The magnitude and stability of the vortex was dependant of the size of the anastomosis bulb.

This study was designed to study typical flow patterns and the effect of different anastomosis shapes on those flow patterns. This rendered interesting qualitative data to identify high risk areas for MIH formation. Wall shear stress maesurements are possible with flow visualisation but for accurate maesurements we need to determine a mean value within longer visualisation periods at regular intervals along the vascular wall. Particle density and visualisation time in this study were not adequate for accurate quantative WSS measurements.

From this study of in vitro flow visualization of in vivo vascular anastomoses we found several high-risk areas for MIH formation in ETS anastomosis: (1) low WSS areas caused by flow separation and recirculation at the toe and at the heel of the anastomoses, and (2) oscillating WSS area due to a shifting stagnation area at the bottom of the anastomosis. We also showed that reducing the angle in ETS anastomoses would improve the flow patterns by reducing the extent and complexity of secondary flow patterns thereby mitigating MIH formation. Furthermore, a patent proximal outflow artery contributes to flow disturbances in the anastomosis and in the proximal outflow artery, which may lead to stenosis formation and occlusion of the proximal outflow artery itself. In the ETE models, the main features of the flow patterns can be accounted for by the slight out of axis geometry of graft and recipient artery and the bulbous shape that results from the bevelled anastomosis. The flow separation and recirculation areas at the edges of the anastomoses during deceleration at the end of systole are comparable to the areas in ETS anastomoses. In short bevelled anastomoses these features were less predominant then in the longer and wider bevelled anastomoses.

We conclude from this study that high-risk areas for MIH formation are present in ETS anastomosis and cause remodelling towards an ETE anastomosis in the proximal outflow artery. Since ETE anastomosis also have comparable flow patterns as compared to ETS anastomosis, and thus the same high risk areas for MIH formation, we postulate that these features cause comparable clinical patencies between ETS and ETE anastomosis. Theoretically, the ideal ETE anastomosis should be made with a graft connected exactly in line with the recipient artery of matching diameter with edgeless suture line. In surgical practice, however, this is difficult to achieve. One should therefore endeavour to construct as straight a surgical ETE anastomosis as possible to obtain optimal hemodynamic performance and prevent MIH formation.

Disclosure Statement

None of the authors have any conflict of interest.