Ann Thorac Surg 2001;72:782-787
© 2001 The Society of Thoracic Surgeons
Original article: cardiovascular
Effect of distal graft anastomosis site on retrograde perfusion and flow patterns of native coronary vasculature
Lin-rui Guo, MDa,
David A. Steinman, PhDc,e,
Byung C. Moon, FRCS(C)a,b,
Wan-Kei Wan, PhDd,
Richard J. Millsap, BSd
a Division of Cardiovascular Surgery, London Health Sciences Center, London, Ontario, Canada
b Department of Surgery, The University of Western Ontario, London, Ontario, Canada
c Department of Medical Biophysics, The University of Western Ontario, London, Ontario, Canada
d Department of Chemical & Biochemical Engineering, The University of Western Ontario, London, Ontario, Canada
e Imaging Research Lab, The John P. Robarts Research Institute, London, Ontario, Canada
Accepted for publication April 25, 2001.
Address reprint requests to Dr Moon, London Health Sciences Centre, Rm C110, 370 South St, London, ON, N6B 1B8, Canada
e-mail: byungchoo.moon{at}lhsc.on.ca
 |
Abstract
|
|---|
Background. To select the site of a target vessel for distal anastomosis surgeons use different approaches. Some try to place the graft as close to the stenosis as possible, whereas others routinely anastomose the graft onto the distal portion. In this latter case the proximal portion and its tributaries are perfused from the graft in a retrograde rather than an antegrade fashion. The aim of this study was to investigate the effect of local hemodynamics associated with the different location of distal anastomoses on flow patterns in the proximal native artery and its branches.
Methods. Computational fluid dynamic and in vitro model studies were carried out in a control model composed of a straight tube (host) with a 45E side branch and models in which the proximal end of the host had various degrees of stenosis; a 45E end-to-side "graft" anastomosis was introduced either proximal (upstream) or distal (downstream) to the branch.
Results. Placing the graft proximal to the branch largely preserved the flow patterns that were seen in the control model. Placing the graft distal to the branch, however, introduced an extensive region of relatively stagnant flow in the native vessel near the branch. Such regions are known to promote thrombus formation that could ultimately lead to occlusion of the retrograde portion of the host vessel.
Conclusions. This study suggests that, although often less convenient surgically, long-term outcome of coronary artery bypass grafting may be improved by placing grafts in the most proximal portion of the native vessel, as close to the occlusion or stenosis as possible for better preservation of a proximal artery and its branches.
|
Introduction
|
|---|
Coronary artery bypass grafting is one of the major effective modalities in the treatment of coronary artery disease; however, its long-term efficacy is still affected by the ongoing diseases of native coronary arteries and the unsatisfactory long-term patency of bypass conduits, especially saphenous vein grafts [1, 2]. Studies have shown that control of risk factors such as normalizing the levels of blood glucose and lipid and managing hypertension can curtail the progression of coronary artery disease [3]. To achieve long-term patency of the conduit, multiple arterial grafts have been proposed [4, 5].
Left internal mammary artery (LIMA) to left anterior descending artery (LAD) remains the most important graft placed in coronary artery bypass procedures. The conventional approach is to place the LIMA graft wherever the LAD itself is most accessible, usually in the distal portion of its course. In this case, however, the proximal native vessel and branches are perfused in an unphysiologic, retrograde manner. An alternative approach is to place the LIMA at a more proximal portion of the LAD, with the goal of maintaining antegrade perfusion of the vessel and branches distal to the stenosis; however, this method is associated with tedious dissection to expose this portion of LAD, along with the added risk of increased hemorrhage or even entering the ventricular cavity. The goal of this study was to determine how the placement of the graft affects the flow dynamics in the native vessel and branches, and how this might ultimately impact the patency of the proximal LAD and its tributaries. To achieve this goal we carried out in vitro (dye washout) and computational fluid dynamic (CFD) studies in an idealized end-to-side anastomosis model in which the position of the graft relative to a native vessel branch and stenosis was varied.
|
Material and methods
|
|---|
Models
The models used for the present study are shown in Fig 1. They are based on the idealized 45-degree end-to-side anastomosis commonly used in studies of bypass graft hemodynamics in which the host and graft vessels are assumed to be cylindrical and of an equal diameter [6, 7]. For our studies we assumed a native/graft diameter of 3.3 mm, representative of the normal human coronary artery [8]. A single cylindrical branch was attached to the wall opposite to the graft at a distance of 5 host vessel diameters (ie, 16.5 mm) proximal or distal to the graft mouth, and oriented at 45 degrees toward the distal native vessel. (Hereafter these placements are referred to as the "downstream" graft model and "upstream" graft models, respectively, referring to the position of the graft relative to the branch.) A branch diameter was 1.65 mm (ie, half of the native artery diameter), representative of a branch near the origin of the LAD [8]. A "native" model, consisting of the native and branch vessels only, was used to establish the normal hemodynamic state of the native vasculature in the absence of a graft or stenosis.
View larger version (27K):
[in this window]
[in a new window]
|
Fig 1. (a) Schematic of the downstream and upstream graft configurations illustrating the anticipated directions of blood flow. LAD = left anterior descending coronary (host) artery; LIMA = left internal mammary artery graft. (b) Corresponding finite element models, identifying the flow rate (Q) variables.
|
|
Computational studies
Computational fluid dynamic studies were carried out using well-validated finite element software [9]. The native and two graft geometries were constructed using a commercial computer-aided design and mesh generation package (DDN/TETRA; Icem Engineering, Bellevue, WA). Mesh refinement studies carried out under steady flow conditions were used to optimally place the finite elements to maximize the accuracy of the computed velocity fields. Owing to the symmetry of the models, it was necessary only to mesh and analyze half of the geometry (see Fig 1b).
Steady flow studies were carried out at a total flow rate of 2.25 mL/s, corresponding to a Reynolds number of 250 (assuming a blood viscosity of 3.5 cPoise [10]). Pulsatile flow studies were carried out using a representative human LAD flow rate waveform [11], having mean and peak flow rates of 2.1 and 4.1 mL/s, respectively, corresponding to mean/peak Reynolds numbers of 230/460. For all studies the total flow rate entering the model (ie, Qgraft + Qstenosis) was fixed; an increase in the flow rate through the stenosis was offset by a decrease in the flow through the graft. Because of the inability of our CFD solver to model turbulent flow, it was not possible to directly simulate the effects of a severe stenosis (although direct simulation was possible in the in vitro studies, as described below). However, the reduction of flow through the proximal native vessel that is expected with increasing stenosis severity was indirectly modeled by varying the division of flow through the graft (Qgraft) and proximal native vessel (Qstenosis). Steady and pulsatile flow studies carried out with Qgraft/Qstenosis = 100:0 (ie, a fully occluded proximal native vessel), 90:10 (representative of a severe stenosis), 75:25, and 60:40 (representative of moderate to severe stenosis). Unless otherwise noted, flow exiting the branch was fixed at 25% of the total flow rate. Further details of the flow modeling are provided elsewhere [7].
In vitro dye studies
Three models, representing the downstream, upstream, and native configurations, were constructed at the University of Western Ontario machine shop. The models were scaled up from the CFD models approximately 4x using 12.7-mm (0.5-inch) diameter clear acrylic rigid tubing. Long (> 30 cm) straight sections were attached to the graft and branch inlets to facilitate the development of parabolic flow into the model. A schematic diagram of the experimental setup described below is shown in Fig 2.
Steady flow studies were carried out a Reynolds number of 250 to allow direct comparison with the steady flow CFD studies. Valves were used to control the total flow rate and graft to stenosis flow division. The total flow rate from a constant pressure head reservoir was adjusted to 2.5 mL/s to account for the larger diameter and lower viscosity of water (ie, 1 cPoise) than was used for the CFD studies. Flow resistors, composed of an adjustable clamp placed around 16-mm diameter polypropylene tubing, were placed on the native and branch outlets to provide the specified 75:25 host to branch flow division. For studies in which the graft to stenosis flow ratio was 100:0 (ie, a fully occluded stenosis), the stenosis valve was closed. For studies with stenosis inflow, removable stenosis sections of 50%, 75%, and 95% area reductions were alternately placed at the proximal native entrance, and the stenosis flow was controlled rate (independently of the area reduction chosen) by adjusting the stenosis valve.
Two 10-mL dye injection syringes with 14-gauge needles were used for the dye injection. Water and glycol-based red and blue supermarket food dyes were used as a flow tracer. One milliliter of food dye was added to 10 mL of water. Two milliliters of red dye was then injected using a 14-gauge needle into the polypropylene pipe 40 cm upstream from the anastomosis site. A similar syringe was used to inject another 2 mL of blue dye 2.5 cm upstream from the stenosis site. The injections occurred at approximately the same time.
An 8-mm video camera (Model CCD-F77; Sony Electronics, San Antonio, TX) was placed 45 cm above the anastomosis site, and filming was begun just prior to the dye injection. Video data were transferred to a Pentium II 400 Mhz computer with a video capture card. Video acquisition was stopped when the dye had disappeared from the model.
|
Results
|
|---|
As shown in Fig 3, the location of the graft relative to the branch had a marked effect on the local hemodynamics in the native vessel and branches. Specifically, the section of the native vessel between the graft and branch experienced slow flow for the downstream graft placement that was largely absent when the graft was placed upstream of the branch. The direction of flow toward the branch in the downstream graft model was also in the opposite sense to both the upstream graft and native models. Furthermore, the upstream graft placement largely preserved the native flow patterns seen at the branch itself; for the downstream graft placement, velocities were skewed toward the opposite wall. These differences were also seen for pulsatile flow, not only at the peak flow rate shown in Figs 3d–f, but also throughout the cardiac cycle. Not shown are steady flow CFD studies in which branch outflow was reduced from 25% to 12.5%, and for which the distance between the graft and branch was reduced by a factor of 2. In both cases, the marked differences in the flow patterns between the upstream and downstream graft configurations described above were maintained.
View larger version (44K):
[in this window]
[in a new window]
|
Fig 3. The effect of graft placement on computed center-plane velocities for the case of a fully occluded stenosis. Computational fluid dynamic (CFD) studies with steady flow. (a) downstream model; (b) upstream model; (c) native model. Pulsatile flow at peak flow rate (see inset flow waveform in f): (d) downstream model; (e) upstream model; (f) native model. Velocity magnitude (in cm/s) is indicated by both color and vector length; flow direction is indicated by vector orientation. Note the different contour levels and vector lengths used for the steady (a–c) and pulsatile (d–f) velocity fields.
|
|
The introduction of competing flow from a partially occluded stenosis had little overall effect on the results described above, namely that the downstream graft placement always resulted in much slower flow in the native vessel between the graft and branch than that for the upstream graft placement. As shown in Fig 4, however, the nature of the slow flow induced by the downstream graft placement was affected by the amount of stenosis inflow. Specifically, when Qstenosis < Qbranch, blood flowed toward the branch from the graft, as evidenced by the computed velocity vectors in Fig 4a and the presence of red (graft) dye in Fig 4d. When Qstenosis > Qbranch, the opposite was seen, with flow directed toward the graft. When Qstenosis = Qbranch, complete stagnation of the proximal native vessel was observed. This condition is most apparent in Fig 4e, which was acquired several seconds after dye injection had ended. Such behavior was also observed in the pulsatile CFD studies, and for in vitro studies in which the stenosis flow rate was maintained, but 50% and 90% stenosis severities were introduced to induce lesser and greater degrees of stenosis jetting and turbulence, respectively.
View larger version (41K):
[in this window]
[in a new window]
|
Fig 4. Effect of stenosis inflow on center-plane velocity field of the downstream model. Computational fluid dynamic (CFD) studies with steady flow: (a) stenosis inflow = 10%; (b) stenosis inflow =25%; (c) stenosis inflow = 40%. In vitro dye: (d) stenosis inflow = 10%; (e) stenosis inflow = 25%; (f) stenosis inflow = 40%. Refer to Fig 3c for the contour levels and vector lengths used for (a)–(c). Images of the dye studies were acquired shortly after dye injection had ceased.
|
|
|
Comment
|
|---|
In clinical practice, most cardiac surgeons agree that a graft should be anastomosed onto the most healthy and reasonable sized (> 1.5 mm in diameter) segment in the target vessel. The proximal portion is always larger in size and frequently healthy enough for the anastomosis. However, the proximal portion of the vessel is often buried in the epicardial fat and even in the myocardium. Thus, some surgeons routinely anastomose a graft to the distal portion far downstream from the stenosis, but which is easily seen through the epicardium. As a result, a long segment of native artery and its tributaries are located between the stenosis and the distal anastomosis, which receive blood from the graft in a retrograde fashion and from a jet flow of the stenotic native vessel in a prograde fashion if it is partially occluded. The principles of fluid dynamics suggest that such flow patterns may induce disturbed flow, decreased flow velocity, and even stagnation. Our results confirmed that these effects occurred in the proximal portion of the vessel. The computational and in vitro flow studies showed that a downstream or distal placement of the graft altered the flow patterns and results in the stagnant or recirculating blood flow at the proximal artery and its branches, as compared with the native state. These changes may explain the reocclusion of the proximal native vessel by two possible mechanisms. First, it is known that endothelial cells will reorient when flow conditions are altered [12], and thus it is possible that endothelial dysfunction and intimal hyperplasia may be initiated at native vessel branches because of the reorientation of the wall shear stresses. Second, the presence of stagnant or recirculating blood flow provides an ideal environment for thrombus formation [13], particularly when platelets are activated, as they are exposed to suture-lines and atherosclerotic plaques. Thrombotic reocclusion of the proximal native vessel may therefore be promoted by the reduced flow rates and velocities introduced by the downstream graft placement. Ultimately, reocclusion of the native vessel would cut off flow to any branches, and hence the perfusion to the myocardium may be compromised.
Our studies also suggest that introduction of a graft to a partially occluded vessel may in fact exacerbate the conditions for thrombotic occlusion when the branch outflow is approximately equal to the stenosis inflow for downstream graft placement. In this case our model studies predicted almost completely stagnant flow. Although complete stagnation is unlikely to happen in practice, because the ratio of branch to stenosis flow is likely to vary over the cardiac cycle (as discussed below), it does suggest that the presence of competing flow from a partially occluded stenosis may actually enhance thrombotic reocclusion compared with the case of a fully occlusive stenosis.
Our model is admittedly highly idealized compared with the in vivo situation. We have ignored such geometric factors as unequal graft to host diameter (eg, relevant to vein grafts), tapering of the coronary artery, the presence of multiple branches, and the pulsation and motion of the vessels themselves. We have also used fluid dynamic parameters (ie, Reynolds/Womersley numbers and flow waveform dynamics) that are appropriate for the coronary circulation rather than other vascular territories (eg, peripheral vasculature), and have ignored the fact that the flow division between the stenosis and graft is likely to vary throughout the cardiac cycle. Inclusion or variation of such factors would certainly alter the details of the observed flow patterns (eg, size of recirculation zone distal to toe, inertia of flow emanating from the graft mouth) as has been previously demonstrated by a number of studies [6, 7, 14, 15]. However, in all cases the gross fluid dynamic features that characterize flow in an end-to-side anastomosis—for example, skewing of velocities toward the bed, helical flow in the distal host, slowly recirculating or stagnant flow in the (occluded) proximal host—are maintained. Thus, we are confident that the marked differences we have observed between the flow patterns of the upstream and downstream graft models would not be masked had we varied the vessel geometries or flow parameters.
Although we were initially surprised by the similarity between the steady and pulsatile flow results, this finding is almost certainly an inevitable result of the prescribed (and constant) constant flow divisions among the branches. This similarity is unlikely to be the case in vivo, as the relative impedances of the different vessels will vary during systolic contraction and diastolic relaxation of the host vessel. Modeling such effects would be highly complicated, however, as little data are available on the impedances (or even flow divisions) of the different branches as a function of cardiac phase. Similarities between the steady and pulsatile flow patterns also result from the relatively low flow pulsatility of the coronary circulation: for our study the Womersley number was 2.3.
We have demonstrated that the graft distal anastomosis site may play a role in the long-term success of coronary bypass operations. Although preliminary, our studies suggest that grafts should be placed upstream of as many branches as possible to more closely reestablish the native hemodynamics and minimize the length of native vessel prone to thrombotic reocclusion. Further studies are, however, required to test this hypothesis directly. Retrospective studies of angiograms could identify those patients with postoperative occlusions, and these could be compared with measurements of anastomosis-stenosis distance and the number of collateral branches in the proximal host region observed preoperatively. Alternatively, prospective studies comparing the outcome of patient or animal populations with each graft type could help elucidate these mechanisms. We note, however, that if the distal anastomosis site is ultimately shown to affect the patency of the native vasculature, only time will tell whether the additional risk associated with implantation of the graft close to the occlusion outweighs the benefits to the patient.
|
Acknowledgments
|
|---|
R.J. Millsap was supported by a London Health Sciences Center research grant no. LHRF7514 (BCM). D.A. Steinman acknowledges the support of the Heart & Stroke Foundation of Canada Research Scholarship.
We thank Jaques Milner for assistance in constructing the finite element models.
|
References
|
|---|
-
Campeau L., Enjalbert M., Lesperance J., et al. The relation of risk factors to the development of atherosclerosis in saphenous-vein bypass grafts and the progression of disease in the native circulation. A study 10 years after aortocoronary bypass surgery. N Engl J Med 1984;311:1329-1332.[Medline]
-
Lytle B.W., Loop F.D., Cosgrove D.M., Ratliff N.B., Easley K., Taylor P.C. Long-term (5 to 12 years) serial studies of internal mammary artery and saphenous vein coronary bypass grafts. J Thorac Cardiovasc Surg 1985;89:248-258.[Abstract]
-
Watts E. Prevention of coronary heart disease. Lancet 1987;1(8533):636.[Medline]
-
Fiore A.C., Naunheim K.S., Dean P., et al. Results of internal thoracic artery grafting over 15 years: single versus double grafts. Ann Thorac Surg 1990;49:202-208.[Abstract/Free Full Text]
-
Lytle B.W., Blackstone E.H., Loop F.D., et al. Two internal thoracic artery grafts are better than one. J Thorac Cardiovasc Surg 1999;117:855-872.[Abstract/Free Full Text]
-
Fei D.Y., Thomas J.D., Rittgers S.E. The effect of angle and flow rate upon hemodynamics in distal vascular graft anastomoses: a numerical model study. J Biomech Eng 1994;116:331-336.[Medline]
-
Ethier C.R., Steinman D.A., Zhang X., Karpik S.R., Ojha M. Flow waveform effects on end-to-side anastomotic flow patterns. J Biomech 1998;31:609-617.[Medline]
-
Ding Z., Biggs T., Seed W.A., Friedman M.H. Influence of the geometry of the left main coronary artery bifurcation on the distribution of sudanophilia in the daughter vessels. Arterioscler Thromb Vasc Biol 1997;17:1356-1360.
-
Ethier C.R., Steinman D.A., Ojha M. Comparisons between computational hemodynamics, photochromic dye flow visualization and magnetic resonance velocimetry. In: Xu X.Y., Collins M.W., eds. The haemodynamics of arterial organs—comparison of computational predictions with in vivo and in vitro data. Southampton, UK: WIT Press, 1998.
-
Nichols W.W., O’Rourke M.F. McDonald’s blood flow in arteries, 3rd ed. Philadelphia: Lea & Febiger, 1990.
-
Ku D.N., Zhu C. The mechanical environment of the artery. In: Sumpio B.E., ed. Hemodynamic forces and vascular cell biology. Austin, TX: R.G. Landes, 1993.
-
Nerem R.M., Alexander R.W., Chappell D.C., Medford R.M., Varner S.E., Taylor W.R. The study of the influence of flow on vascular endothelial biology. Am J Med Sci 1998;316:169-175.[Medline]
-
Slack S.M., Cui Y., Turitto V.T. The effects of flow on blood coagulation and thrombosis. Thromb Haemost 1993;70:129-134.[Medline]
-
Moore J.A., Steinman D.A., Prakash S., Johnston K.W., Ethier C.R. A numerical study of blood flow patterns in anatomically realistic and simplified end-to-side anastomoses. J Biomech Eng 1999;121:265-272.[Medline]
-
Ethier C.R., Prakash S., Steinman D.A., Leask R., Couch G., Ojha M. Steady flow separation patterns in a 45 degree junction. J Fluid Mech 2000;411:1-38.
Top
Abstract
Introduction
