The .hemodynamics of thrombus formation in arteries Kenneth Ouriel, MD, Carlos Donayre, MD, Cynthia K. Shortell, MD, Catherine Cimino, AAS, Joan Donnelly, BA, David Oxley, MD, and Richard M. Green, MD, Rochester, N.T. Alterations in arterial blood flow are thought to predispose to thrombus formation, but the exact relationships have not been fially elucidated. The effect o f varying blood flows on the accumulation o f thrombotic material within arteries was investigated, with use o f shear rate as an index o f flow across the luminal surface. Partially denuded rabbit aortas were perfused with fresh nonanticoagulated human blood for 3 minutes, with an in vitro recirculating apparatus, Indium i l l - l a b e l e d platelets, and fibrinogen I 125. Shear rates ranged from zero to 1500 sec-~, correlating with the hemodynamics o f various segments o f the human arterial tree. A significant correlation was observed between shear rate and platelet deposition, ranging from 5.2 + 2.8 x 106 platelets/cm 2 o f vessel surface area at zero shear to a m a x i m u m o f 64.7 -+ 8.3 x 106 platelets/cm 2 at a shear rate o f 1500 sec(F = 5.01, p < 0.05). Fibrin deposition paralleled that o f platelets, ranging from 28.2 - 7.6 ~ g / c m 2 at zero shear to 354.11 -+ 6 2 . 7 wg/cm 2 at a shear rate o f 1500 sec -~ (F = 5.91, p < 0.05). These results suggest that shear rate is a m o s t important determinant o f platelet and fibrin deposition on altered arterial surfaces. (J VASC SURG
Thrombus formation on arterial lesions encompasses interactions between blood cells, principally platelets, and the coagulation proteins. The process generally requires an altered vessel wall surface, such as an atherosclerotic lesion. Thrombus growth may be accelerated by flow disturbances, by procoagulants, and by an increased platelet count and hematocrit. Many investigators have attempted to mimic the sequence of events occurring during the development of an intraluminal thrombus by obsetwing the process in in vitro perfusion systems. Prior studies have defined shear :rate as an important hemodynamic variable determining thrombus formation and have drawn clinical correlations from the laboratory data. 1-3 Previous experimental models ofthrombogenesis have frequently neglected the role of the coagulation system by using citrated or heparinized blood as the perfusate.4,s Other investigators have relied on someFrom thc Departmentof Surgery,The Universityof Rochester, Rochester,New York. Supportedby grantno. HL 40889 fromthe NationalInstitutesof Health. Presented at the Fifth AnnualMeeting of the Eastern Vascular Society,Pittsburgh, Pa., May 2-5, 1991. Reprint requests: KermethOuriel,MD, Departmentof Surgery, The Universityof Rochester, 601 ElmwoodAve., Rochester, NY 14642. 24/6/33157
what subjective light microscopic evaluation of platelet and fibrin deposition. 1,6 The present study was undertaken in an effort to define the role of shear rate in the development of intraluminal thrombus, with use of radiolabcled platelet and fibrinogen deposition as objective means for assessing thrombus growth. An in vitro experimental model with nonanticoagulated human blood was designed to more closely approximate the properties of the bloodintimal interactions occurring within various segments of the human arterial tree. MATERIAL AND M E T H O D S Perfusion system
An in vitro whole blood recirculating perfusion system was constructed by use of an all plastic (silicone and polyvinylchloride plexiglas) circulation pathway to minimize platelet and coagulation cascade activation. Rabbit aortas were harvestcd and partially deendothelialized with passage of balloon catheters. The aortas vcere everted, placed on plexiglas rods (average diameter, 0.5 cm), and positioned within plexiglas annular perfusion chambers (average outer diameter, 0.9 cm). Fresh blood was withdrawn from 20 healthy human volunteers. Volunteers had a platelet count within the normal range, a normal activated dotting time, and no history of nonsteroidal antiinflamma757
Journal of VASCULAR SURGERY
Ourid et al.
tory drug use. N o differences were observed in the frequencies of blood types or the mean platelet counts between experimental groups. Indium 111 (~HIn)labeled platelets from another volunteer and fibrinogen I 125 (Amersham Corp., Arlington Heights, Ill.) were added in trace amounts to the blood. The blood (40 ml) was placed in the perfusion system and perfused in a recirculatory fashion across the aortic segments for a period of 3 minutes. A peristaltic pump was used to maintain the desired flow rate, controlling the rate with a previously calibrated ultrasonic flow probe. Twenty aortic segments were perfused at shear rates ranging from zero to 1500/sec, corresponding to the flow characteristics observed throughout the arterial tree and capillary network. A second set of 28 rabbit aortic segments were perfused at shear rates ranging from 34 to 479 sec -1 for 5 minutes with citrated human blood. Radiolabeled platelets were added to 14 of the perfusions, and platelet and fibrin deposition were determined with radiotracer techniques. A matched set comprising the remaining 14 segments were perfused without addition of radiolabel and were fixed, stained with a modified Wright-Giemsa technique, embedded in plastic, and sectioned ultra thin. The sections were examined by light microscopy, evaluating 100 points along the circumference of the luminal surface in a random fashion. The results were correlated with the gamma counted segments to validate the radiolabeling technique for determining platelet deposition. The aortic segments were removed from the perfusion chambers at the conclusion of the perfusion period. The segments and a 1 ml specimen of the whole blood perfusate were separately counted in a dual channel gamma counter with distinct 11~In and I 125 windows. The platelet count in the perfusate was determined with hemochromocytometric techniques, and the fibrinogen concentration was measured with a spcctrophotometric assay. A comparison of the counts of radiotracer with the platelet count and fibrinogen concentration provided a means for detcrmining platelet and fibrinogen deposition on the aortic segments as follows: Platelets deposited = (Vessel lllIn counts) x (Perfusate platelet count) (Perfusate Hqn counts) x (Vessel surface area) Fibrin deposited = (Vessel I 125 counts) x (Perfusate fibrinogen concentration)
(Perfusate 1 125 counts x (Vessel surface area)
Statistical analysis The data were analyzed by use of analysis of variance (ANOVA) techniques. Correlations were performed with Pearson's coefficient. Significance was assumed when the two-tailed p value was less than 0.05. Values are expressed as mean + SEM. RESULTS Blood was perfused across the injured aortic surfaces at shear rates of zero, 100, 500, 1000, and 1500 sec-1. The perfusions were undertaken at flow rates designed to approximate the flail range of shear rates observed in the arterial tree.7,8 Scanning electron micrographs of the injured aortic segments before perfusion revealed spotty disruption of the endothelial lining with exposed basement membrane in many areas. There were signs of endothelial cell damage in areas where the intimal lining was preserved, with loss ofmicrovilli, development of cellular edema, and randomization of cellular longitudinal axis. Scanning electron micrographs of the perfused segments revealed formation of thrombi diffusely over the luminal surface. This deposition appeared to be most pronounced at areas of complete endothelial denudation. Microscopic analysis documented the thrombus composition to be platelets, fibrin, red cells, and a few white blood cells. The ratio of platelets to red cells subjectively increased with increasing shear rates (Fig. 1). Light microscopic sections of rabbit aorta perfused with citrated blood were studied to validate the radiolabeled technique for determining platelet deposition. An increase was observed in the percent surface area covered with platelets with increasing shear rate (F = 4.92, p < 0.05), from 9.9% to 39.7% as the shear rate was increased from 34 sec -1 to 479 sec -1 (Fig. 2). The number of platelets per square centimeter of vessel surface increased from 2.43 + 0.29 x 1 0 6 to 7.03 + 0.44 x 1 0 6 in matched samples (F = 11.2, p < 0.005). Excellent correlation was observed between the percent of the vessel surface covered with platelets as assessed with the light microscope and the number of platelets on the surface as assessed with the radiolabeled technique (r = 0.94,p < 0.001). Platelet deposition was remarkably low in the zero shear rate (static blood) group, averaging 5 . 2 ± 2 . 8 X 1 0 6 platelets/cm 2 of vessel surface area (Fig. 3). Deposition increased significantly with increasing shear rate, to a maximum of 64.7 + 8.3 x 106 platelets/cm 2 at a shear rate of 1500 sec -~ (F = 5.01,p < 0.05). Fibrin deposition paralleled that of platelets (Fig. 4), ranging from
Volume 14 Number 6 December1991
Hemodynamics of thrombus formation in arteries 759
Fig. 1. Scanning electron micrograph (original magnification × 1500) ofperfused (shear rate 100/see) rabbit aorta, illustrates activated, adherent platelets over the deendothelialized surface, an occasional red cell, and fibrin strands (arrows). 50
Shear Rate per sec
Fig. 2. Relationship between platelet deposition and shear rate in citrated perfusions, quantified by light microscopy (percent surface coverage) and radiolabeled techniques (millions platelets per square centimeter). 28.2 + 7.6 ~g/cm 2 at zero shear rate to 354.1 + 62.7 ~g/cm 2 at a s]hear rate of 1500 sec -1 (F = 5.91, p < 0.05). The largest increase in both platdet and fibrin deposition occurred between a zero shear rate and a shear rate of 100 sec -1, corresponding to the flow characteristics observed in static blood and large arteries, respectiw:ly. Platelet deposition increased 506%-+ 97% and fibrin deposition increased 440% + 87% between these two shear rate levels. DISCUSSION
Arterial thrombosis occurs in a variety of clirfical situations and is frequently accompanied by cata-
strophic infarction in the tissue supplied by the involved artery. Thrombus formation is most often associated with atherosclerotic plaque, however, it may also occur overlying vessel wall inflammatory process such as the autoimmune and infectious vasculopathies. A common histopathologic theme in acute myocardial infarction is fresh luminal mural thrombosis on a ruptured atherosclerotic plaque. The peripheral corollary of this clinicopathologic paradigm is the acutely thrombosed atherosderotic lower extremity artery and the highly stenotic carotid bifurcation lesion progressing to acute internal carotid occlusion. These processes frequently result in
Journal of VASCULAR SURGERY
760 Ouriel et aI.
4-" E c,o
Shear Rate per sec
Fig. 3. Relationship between platelet deposition and shear rate, nonanticoagulated blood, radiolabeled techniques. 400 E 300
Shear Rate per sec
Fig. 4. Relationship between fibrin deposition and shear rate, nonanticoagulated blood, radiolabeled techniques. irreversible ischemia in the form of cardiac arrest, lower extremity gangrene, and stroke. The present study was directed at an investigation of the relationship of hemodynamic factors, specifically vessel wall shear rate, to the formation of intravascular thrombus. Baumgarmer and Sakariassen 1 were pioneers in the development of an in vitro model of vessel wall thrombogenesis, with the institution of a perfusion circuit and annular perfusion chamber to precisely control the hemodynamic parameters involved in this process. These investigators have repeatedly demonstrated a correlation between shear rate and platelet deposition on vascular surfaces, with increasing platelet attachment and aggregation as the flow across the luminal surface is augmented. It is interesting that the same investiga-
tors documented an inverse relationship between shear rate and fibrin deposition, with decreasing fibrin attachment as blood flow was increased,s'9 These studies were done with anticoagulated blood in many instances, and the fibrin deposition was determined by use of light microscopic techniques. The findings of the present study corroborate the published results of previous investigators with respect to the direct relationship between platelet deposition and shear rate. Although our work was done in a relatively thrombogenic system, the model was designed to parallel the clinical paradigm of hypercoagulable blood perfused across a thrombogenie surface. Our observations explain the clinical observation of white platelet thrombus at altered vessel surfaces within areas of high flow rate, for
Volume 14 Number 6 December 1991
example, adherent to a fresh carotid endarterectomy site. The attacbxnent of cells and aggregate,; on vascular surfaces is governed by the rate at which these elements bind to the surface compared to the rate at which they are removed) ° The affinity of the particle (platelet, red cell, leukocyte) to the surface relative to the shear forces imposed on the particle at the surface determines whether arrival or removal of the particular partMe will predominate. Typically, platelets form strong attachments to altered vascular luminal surfaces through their interactions with collagen and platelet glycoprotein la and lb. Moreover, once a platelet attaches to the surface cohesive interplatelet interaction through fibrinogen molecules and glycoprotein IIb/IIIa result in platelet aggregation and :formation of a platelet thrombus. Therefore, the rate of platelet arrival (deposition) overshadows the rate of platelet departure in the instance of the altered arterial ~luminal surface:. In contrast to platelets, red cells have tittle affinit~/for altered vascular surfaces. Few active mechanisms exist for red cell deposition in this situation. The rate of arrival of red cells to the surface parallels their rate of departure, and hence thrombi over damaged arterial surfaces tend to be composed primarily of platelets. We have been unable to confirm the findings of previous investigators with respect to fibrin attachment to the altered arterial luminal surface. Our observation of increasing fibrin deposition "with increasing shear rate is in contradistinction to the work of those investigators who observed an inverse fibrin-shear rate relationship) ,4 One explanation for this apparent parado x may relate to the method of quantification of fibrin deposition. Other investigators have implemented morphometric analyses in the determination of fibrin deposition, manually counting the percentage of the vascular surface covered with fibrin strands. We initially used this light microscopic method, but abandoned the technique when we were unable to accurately define fibrin strands when the surface contained massive numbers of platelets and red cells. It is possible that the vast platelet deposits on vessels perfused at high slhear rates may hide the fibrin strands from microscopic view, rendering increasingly spurious fibrin counts at higher shear rates. The findings of the present study partially explain the clinical observation of white thrombus in arteries and red thrombus in veins. The rapidly flowing blood within diseased arteries predisposes to increased platelet deposition and the development of plateletrich white thrombus. Superficially, one might expect
Hemodynamics of thrombusformation in arteries
the slower flow in the venous system to be associated with platelet-poor thrombus and a red clot. However, it is likely that venous thrombi originate in areas of endothelial damage. Platelets will adhere actively at these locations, in contrast to the passive adherence of red cells. Thus, even in areas of low shear, the arrival/departure ratio for platelets will exceed that of red cells and a white thrombus will be generated. The explanation for white versus red thrombus may not relate as much to shear rate as it does to whether one is observing the initial causative process or the prograde and retrograde propagated thrombus. The initial thrombus may always be primarily a platelet thrombus and will be white. The propagated thrombus only develops after the initial thrombus completely occludes the lumen and induces cessation of flow in the vessel. This propagated thrombus then comprises only the blood elements present in the column of blood between the initial offending thrombus and the next large collateral side branch. The red cells vastly oumumber the platelets in whole blood, therefore the column of blood comprising the propagated thrombus will always be red. In summary, the formation of thrombus on altered arterial surfaces is highly dependent on the flow characteristics at the blood-intimal interface. Shear rate appears to be a most important determinant of thrombus growth, with increasing platelet and fibrin deposition as shear rate increases from zero to 1500 sec-L These findings suggest that relative stasis alone is not associated with a propensity for thrombus formation and may help explain the thrombotic propensity of small diameter vascular grafts and the observed differences in the color of thrombus in distinct segments of the vascular tree. It is hoped that an increased understanding of the pathophysiologic mechanisms underlying thrombogenesis will lead to the development of new therapeutic interventions directed at the modulation of the process. REFERENCES
1. Baumgarmer HR, Sakariassen KS. Factors controlling thrombus formation on arterial lesions. Ann NY Acad Sci 1985; 454:162-7. 2. Badimon L, Badimon JJ, Galvez A, et al. Influence of arterial damage and wall shear rate on platelet deposition. Arteriosclerosis 1986;6:312-20. 3. Turitto VT, Weiss HJ, Banmgartner HR, et al. Cells and aggregates at surfaces. Ann NY Acad Sci 1987;516:45367. 4. Tijburg PNM, Ijsseldijk MJW, Sixma JJ, de Groot PG. Quantification of fibrin deposition in flowing blood with peroxidase-labeled fibrinogen. Arteriosclerosis Thromb 1991; 11:211-20.
Journal of VASCULAR SURGERY
Ouriel et al.
5. Turritto VT, Baumgarmer HR. Platelet interaction with subendothelium in flowing rabbit blood: effect of blood shear rate. Microvasc Res 1979;17:38-54. 6. Sakariassen KS, Banga JD, de Groot PG, SLxmaJJ. Comparison of platelet interaction with subendothdium of human renal and umbilical arteries and the extracellular matrix produced by human venous endothelial cells. Thromb Haemost 1984;52:60-5. 7. Lewis P, Psaila JV, Davies WF. Measurement of volume flow in the human common femoral artery using a duplex ultrasound system. Ultrasound Med Biol i986;I2:777-84. 8. Levenson JA, Peronnean PA, Simon A, Safaaar ME. Pulsed Doppler: determination of diameter, blood flow velocity, and
volumic flow of brachial artery in man. Cardiovasc Res 1981;15:164-70. 9. Inauen W, Baumgartner HR, Bombeli T, et al. Dose- and shear rate-dependent effects of heparin on thrombogenesis induced by rabbit aorta subendothelium exposed to flowing human blood. Arteriosclerosis 1990;10:607-15. 10. Sakariassen KS, Fressinaud E, Girma JP, et al. Role ofplatelet membrane glycoproteins and Von Willebrand factor in adhesion of platelets to subendothelium and collagen. Ann NY Acad Sci 1987;516:52-65.
Submitted May 13, 1991; accepted Aug. 15, 1991.
DISCUSSION Dr. Charles Franco (Newark, N.J.). I thank the authors for focusing our attention on a topic that is so fundamentally important to our daily practice. I think some details o f the experimental design need to be clarified. The annular perfusion chamber has been a useful tool for investigating the relationship between thrombus formation and hemodynamic factors for the last 20 years or so. Most of the prior studies used morphometric methods on arterial segments totally denuded of their endothelium. The model presented today uses "partial deendothelialization" o f the rabbit aorta. Were the authors able to quantify the percentage of surface area denuded? What method of control was used to verify that your technique reliably and consistently produced aortic segments with similar degrees of deendothelialization? Since wall shear stress is a tangential drag force produced by blood moving across the endothelial surface, did the authors investigate the effects of increased shear rates on the injured endothelial surface? Specifically, did increased shear rates lead to further endothelial injury and an increase in subendothelial surface area available for platelet adhesion? Shear rate as defined by the HagenPoiseuille equation is directly proportional to blood flow and also blood viscosity and inversely proportional to the cube of the radius. This relationship assumes laminar flow. Could the authors please describe what the flow characteristics are in this model? If shear rate was modified by changes in flow rate, what velocities were achieved and was laminar flow maintained or did the flow become turbulent at the higher flow rates? H o w was shear rate calculated, and what methods were used to validate the results with regard to flow in your laboratory? Since fresh whole blood was obtained from different sources, was there any significant difference in perfusate viscosity and was this a factor in your calculations? Turitto et al. studied the importance of the radius or the distance between the core and the circumference of the perfusion chamber. They found that a variation in vessel wall thickness as little as .02 mm may lead to significant variations in calculated shear rate depending on
the dimensions of the perfusion chamber. Was this factor accounted for in your calculations? The authors have chosen to study the effects of shear rates from 0 to 1500 inverse seconds on platelet deposition because it was felt that this range represents the flow characteristics observed in various parts of the arterial tree under normal circmnstances. They have observed an increase in platelet deposition throughout this range. However, there does not appear to be any significant difference between shear rates of 100 and I000 inverse seconds. Could you please speculate on why this is so. Furthermore, it may not be valid to extrapolate these results to the much higher shear rates that would be typical of stenotic lesions. Others have found that initial rates of platelet adhesion increased with shear rate, but the slope of the curve begins to decrease beyond 650 inverse seconds, and actually becomes negative beyond 2600 inverse seconds. Using morphometric methods they were able to differentiate between platelet adhesion to the subendothelial surface and platelet aggregation with thrombus formation. They found that platelet aggregation continued to increase with rising shear rates. Although the use of isotope labeling is an elegant technique, it is an indirect method that cannot differentiate between platelet adhesion and platelet thrombus formation, and the authors must be careful in the conclusions drawn from their results. On the other hand, radioisotope techniques may be more accurate in determining fibrin content for the reasons cited by the authors, and this avenue of invesngation should be pursued. Another question I have is why the validation studies were performed using citrated blood and only for shear rates between 34 and 479 inverse seconds, particularly since the series of experiments using fresh whole blood showed no significant changes in platelet deposition between i 0 0 and 1000 inverse seconds. Was there any platelet activation or fibrin consumption during the process of collecting the blood? At what temperature was the blood stored and perfused?
Volume 14 Number 6 December 1991
The question of greatest concern to me is fundamental to the conclusions that can be drawn from these experiments. A positive correlation has been made between shear rate and platelet deposition. However, to increase shear rate, flow was increased. Platelet adhesion is dependent on contact with the vessel wall. Increased blood flow should increase the delivery of platelets and platelet contact with the vessel wall. Can the authors comment on whether or not platelet delivery was constant at every shear rate? If platelet delivery differed, what was the relative importance of shear rate versus flow in platelet deposition? If platelet delivery increased with increasing flow rates, then we must be careful in extrapolating these results to the clinical situation. In an area of stenosis, the flow velocity and shear rates will increase, but the volume flow and, therefore, platelet delivery will not increase. Dr. Mark Fil~inger (Syracuse, N.Y.). I have one question. With regard to increasing your shear rates, I presume that you did it by increasing your pump speed, and as a function of mechanical pumps with increasing speed, there is a known effect, certainly at least with red cells, that it increases injury. Did you use any controls where you had altered pump speed and had similar shear rates on the vessel? Dr. Ouriel. With regard to Dr. Fillinger's concern about the high mechanical pump speed causing damage to the blood components, previous data generated with our model documented an absence of red cell damage with flow rates up to 1000 ml/min, equivalent to a shear rate of 1500 sec -~ within the annular chamber. Virtually no free hemoglobin was generated, and platelet aggregation was maintained at baseline levels below these flow rates. Dr. Franco asked several questions relating to the design and validity of the experimental model. The percent of endothelial denudation was not quantitated in the present study. Previous work in our laboratory with scanning and light microscopic en face views documented a mean of 8% to 10% denudation, a value that did not vary substantially from experiment to experiment over a wide range of shear rates. Blood was drawn from volunteers in
Hemodynamics of thrombus formation in arteries 763
a nonanticoagulated, noncitrated state and placed immediately in the perfusion apparatus at 37 ° C. This process has produced little platelet activation as assessed morphometrically and little fibrinogen consumption as assessed chemically. We have assumed laminar flow and Newtonian properties in our model. Viscosity was not measured directly, but the low variability of hematocrit levels suggests that this variable was not likely to be confounding. The shear rates used were calculated with standard formulas outlined in our previous publications, with use of near exact dimension measurements within the annular chamber, which varied less than 5%. By use of these methods of calculation, the shear rates corresponded to relatively low blood velocities where turbulence was not likely to occur.
Dr. Franco suggested that platelet deposition did not increase significantly between shear rates of 100 and 1000 sec -~, but a closer examination the plots presented will reveal an increase in deposition by a factor of 1.5 over this range. It may not be valid to extrapolate our results to the highert shear rates observed in critically stenotic lesions, and this may be the topic of future studies. Validation studies were performed with citrated blood over a smaller range of shear rates because these investigations were done at an earlier stage of our work. We used these data because it seemed unlikely that significant correlations between the radiolabeled and morphometric data performed with cita'ated blood at slightly lower shear rates would not change significantly under different conditions. Concerns over the pathophysiology of increased platelet deposition with increasing shear rate are interesting, and we agree with the premise of increased platelet presentation to the injured endothelium as the most likely origin for the observed platelet-shear rate relationship. Shear rate rather than blood flow is perhaps our best correlate of platelet delivery to the luminal surface. It is likely that flow rate per se has no direct effect on platelet deposition, and this contention had been borne out by studies where flow was varied at a constant shear and vice versa.