Mechanical Factors Associated With the Development of Intimal and Medial Thickening in Vein Grafts Subjected to Arterial Pressure
A Model of Arteries Exposed to Hypertension
Abstract Arteries exposed to sustained hypertension undergo a moderate degree of intimal thickening and a marked amount of medial thickening. Autogenous veins that are used as bypass grafts undergo similar histological changes. In this study autogenous vein grafts were used as an indirect model of arteries exposed to sustained hypertension. It was hypothesized that it is not pressure per se but rather mechanical changes brought about by exposure to increased pressures that act as a stimulus inducing histological changes. Exposure to arterial pressure increases the following nine mechanical factors: deformation in the circumferential, longitudinal, and radial directions; stresses in each of these three directions; pulsatile deformations and pulsatile stresses; and flow velocity. All of these mechanical changes occur simultaneously. Accordingly, a three-step algorithm was devised to separate each of the nine mechanical factors and correlate them with histological changes. Three sequential experimental studies were performed in 38 dogs following the algorithm. These experiments demonstrated that intimal thickening was best correlated with low flow velocity, a correlate of low shear stress, whereas medial thickening was best correlated with deformation in the circumferential direction.
The large arteries of hypertensive subjects exhibit thickened walls,1 hypertrophy of medial smooth muscle cells, and increased connective tissue content.2 3 4 5 These histological changes narrow the lumen and are associated with increased viscoelasticity of the vessel wall.6 The precise stimulus for the structural changes in these vessels is uncertain, although there are data to show that at least some of them are associated with increased pulse pressure.7 This is consistent with the observation that cyclic stretching of smooth muscle cells in vitro stimulates the synthesis of connective tissue components.8
An interesting indirect model of the responses of arteries to hypertension may be found in the histological responses of autogenous veins when they are used as arterial bypass grafts. After exposure to arterial pressures, these vessels undergo intimal hyperplasia and medial thickening. However, it is unlikely that arterial pressure itself elicits these responses. It is more likely that pressure and flow produce deformations and stresses in the vessel wall and that these act internally as stimuli. One may specify nine mechanical factors that occur in the wall of pressurized vein that might act as stimuli.9 There are three static deformations: increased circumference (Fig 1⇓, factor 1), increased length (Fig 1⇓, factor 3), and increased radial deformation (ie, decreased wall thickness; Fig 1⇓, factor 5). In opposition to these deformations, the vessel wall generates increased stress in the circumferential direction (Fig 1⇓, factor 2), in the longitudinal direction (Fig 1⇓, factor 4), and in the radial direction (Fig 1⇓, factor 6). Stresses in the circumferential and longitudinal directions are tensile, whereas stress in the radial direction is compressive. In addition, vein grafts subjected to elevated pressure may undergo pulsatile deformations and increased pulsatile stresses (Fig 1⇓, factors 7 and 8). Finally, blood driven by arterial pressure produces flows at different velocities. This is important because velocity is directly related to the blood-intima shear stress (Fig 1⇓, factor 9). Shear stress may be computed as
where TW is the shear stress, μ is the viscosity of blood (0.035 poise), Q is the velocity of blood flow (in milliliters per second), and r is the inner radius of the vessel. Shear stress cannot be meaningfully calculated unless flow is well ordered and stable. The mathematical methods used to compute each of the mechanical factors listed above are given elsewhere.10
Several investigators have reported that intimal hyperplasia developing in vein grafts correlates with low flow velocity.11 12 13 Others have correlated it with circumferential deformation,14 circumferential wall tension,14 or circumferential stress.15 16 However, all of the nine factors illustrated in Fig 1⇑ occur simultaneously, and morphological changes may occur in response to any or all of them. Because of this simultaneity, an algorithm was developed to experimentally separate the nine mechanical factors shown in Fig 1⇑ as possible stimuli for the development of intimal hyperplasia and medial thickening in autogenous vein grafts. The experimental algorithm is shown in Fig 2⇓.17
Mongrel dogs 20 to 24 kg in weight were used for these experiments. Care of the animals complied with the Guide for Care and Use of Laboratory Animals (National Institutes of Health publication 85-23, revised 1985). The animals were preanesthetized with 25 mg/kg of thiamylal sodium and maintained under inhalation anesthesia with 0.8% halothane and 30% nitrous oxide and oxygen, and sterile surgery was performed. Incisions were made over both femoral arteries. In some experiments, incisions also were made over both common carotid arteries. The femoral arteries and veins were isolated from the inguinal ligament to a point 10 cm distal. In some animals, the carotid arteries and jugular veins were exposed for 8 to 10 cm. Proximal and distal control of the vessels was obtained with vascular clamps. The veins then were ligated distally and proximally, and a 6- to 7-cm segment of vein was removed, rinsed with saline, and implanted immediately as a reversed vein graft. End-to-side anastomoses were constructed using continuous 7-0 polypropylene sutures. End-to-end anastomoses were constructed using interrupted 7-0 polypropylene sutures.
Step 1 Experiments
In 14 dogs, the femoral arteries were bypassed bilaterally using autogenous femoral vein grafts as illustrated in Figs 3⇓ and 4⇓. The artery on one side was ligated; this caused all femoral flow on that side to pass through the graft at high velocity. The artery on the opposite side was left patent; this permitted flow to be divided between the vein graft and the native artery. As a result, flow velocity through the vein graft on the side where the artery was not ligated was lower than on the side where the artery had been ligated. No thrill was palpable over the grafts at the time of surgery, indicating that there was not frank turbulence. However, minor degrees of flow disturbance may have been present at and beyond the anastomosis and at the flow dividers. Flows were measured with an electromagnetic flowmeter at the time of vein graft implantation. These measurements indicated that total femoral flows on the two sides were equal and that flow through the vein grafts was approximately doubled on the side in which the femoral artery was ligated. However, flow measurements were discontinued when it was found that it was not possible to obtain flow measurements with the electromagnetic flowmeter when the grafts were excised, as this would require extensive dissection through scar with possible injury to the grafts.
A portion of the vein graft on both sides of the animal was restricted in dimension by placing a Marlex cuff over a portion of the length of the graft (Figs 3⇑ and 4⇑). The cuff was placed so that when the vessel was distended by arterial pressure the restricted region of the vein did not exceed its usual dimensions at normal venous pressures. The cuff was placed proximally around the vein grafts bilaterally in odd-numbered animals and distally around the vein grafts bilaterally in even-numbered animals. The portions of the vein without cuffs were able to distend without constraint. Experimental data were used as follows. (1) Since total flow on the two sides of the animal was equal, flow velocity must by physical principles have been higher through the restricted region of the graft on the side on which the artery was ligated, since all flow was required to pass through the graft. Conversely, flow velocity was lowest through the distended region of the graft where the native artery was left patent and flow was divided. In this case, only part of the flow passed through the vein graft. On both sides, flow velocity was higher through the portion of the vein where diameter was restricted by the cuff than through the unrestricted, diameter-distended portion of the graft. These differences in flow velocity would apply even if there were turbulence because a given volume of blood was required to pass through series-connected vessels that were of different diameters. Because entering and exiting volumes must be equal, the flow velocity must have been higher through the restricted area than through the wider restricted area of the grafts (Bernoulli’s principle). (2) Circumferential deformation is measured by increased vessel diameter and clearly was highest in the dilated, unrestricted areas. (3) Mean circumferential stress (ςθ) is given by
where PT is transmural pressure, ri is internal radius, and h is wall thickness. In the experiments shown in Figs 3⇑ and 4⇑, circumferential deformation and circumferential stress were largest in the distended portion of the grafts but were only slightly elevated in the portions of the grafts where the cuff limited distension. The mesh bore much of the circumferential load in the restricted regions. (4) Radial deformation is reflected by narrowing of wall thickness. Narrowing was greatest in the regions of the graft that were distended. (5) Mean radial compressive stress (ςr) is equal to one half the transmural pressure, ie, ςr=PT/2. Therefore, radial stress was greatest in the portions of the grafts wrapped by the cuff. In these regions, the vein grafts were pressurized from within the lumen and constrained from without by the mesh.
Step 2 Experiments
In 13 dogs, 3- to 5-cm-long segments of vein were implanted bilaterally as end-to-end interposition grafts replacing a segment of excised carotid artery (Fig 5⇓). After construction of the anastomoses, two sterile, rigid, polyethylene rods were placed parallel to the graft on one side of the animal. The rods were sutured to the pressurized veins to deliberately overextend the grafts by 10% to 15%. Greater lengthening caused tearing of the grafts. Actual measured extension was 12.6±0.8%. This increased the longitudinal deformation and longitudinal stress in the wall of the graft on the one side of the animals. Longitudinal deformation is produced by forced stretching of the vessel. The stretched vessel wall exerts increased longitudinal stress in response to extension. Longitudinally extending the vessel also slightly increased radial deformation by causing thinning of the wall. It also slightly decreased circumferential deformation and decreased circumferential stress because extension slightly reduced the radius.
Step 3 Experiments
Reversed femoral veins were implanted bilaterally as end-to-end interposition grafts in 11 dogs, replacing a segment of excised carotid artery (Fig 6⇓). After the anastomoses were constructed, a 1-cm Marlex band was wrapped around the carotid artery proximal to the vein graft on one side but not on the other. This Marlex band was secured to narrow the lumen of the artery sufficiently to decrease pressure in the vein grafts but maintain diastolic pressure at at least 50 mm Hg. Measurements were obtained distal to the grafts during surgery with a fluid-filled, 1-in, 20-gauge needle connected to a Statham pressure transducer. The Marlex band was used to lower the pressure slightly in the veins to separate circumferential deformation from circumferential stress. Veins are extremely distensible at pressures up to 35 to 50 mm Hg, but they become virtually rigid at pressures above this range.17 Because diastolic pressure in the vein graft was maintained above 50 mm Hg, the diameters of the vein grafts on the two sides were equal (equivalent circumferential deformations). Systolic and diastolic pressures measured at the time of surgery were 103±6 and 74±6 mm Hg in the graft on the side without the proximal carotid stenosis and 79±3 and 61±2 mm Hg in the grafts with the proximal carotid stenosis, respectively. Thus, diastolic pressure was maintained above 50 mm Hg, although pulse pressure was reduced on the banded side. In summary, the Marlex band on the one carotid artery decreased systolic, diastolic, mean, and pulse pressures in the vein grafts but maintained diastolic pressures at greater than 50 mm Hg. As a result, the stenosis on the carotid artery used in the step 3 experiments provided equal circumferential deformations bilaterally but unequal circumferential stresses (equation 2).
All of the dogs were killed 3 months after surgery; the grafts were excised, perfusion-fixed at 100 mm Hg with 10% formaldehyde embedded in paraffin, sectioned, stained with Verhoeff’s elastica, and prepared for histological examination. The histological responses of the grafts on the two sides were correlated with the mechanical factors listed in Fig 1⇑. Sections of the midpoint of the vein grafts were studied, not the anastomotic regions. Drawings of these figures were used to determine the cross-sectional areas of the intima and the media with a calibrated grid planimeter. Statistical comparisons were made with paired t tests.
Step 1 Experiments: Flow Velocity
Fig 3⇑ gives the cross-sectional areas of the intima (mean±SEM) for each region of the grafts for 14 animals. Comparisons of the cuffed versus adjacent noncuffed regions showed that intimal thickening was greater on both sides in the distended regions, where the grafts were subjected to low flow velocity, than in the regions constrained by the cuffs, where the grafts were subjected to high flow velocity (P<.05). This was found whether the femoral artery was ligated (0.87 versus 0.48 mm2) or was left patent (0.78 versus 1.41 mm2). Comparison of the distended regions across the two sides (0.87 versus 1.41 mm2) also disclosed that intimal thickening was greater (P<.05) on the side where the artery was left patent (low flow velocity) compared with the side where the artery was ligated (high flow velocity). Thus, without exception, intimal thickening was greater in graft regions exposed to low flow velocity than in regions exposed to high flow velocity. These data also show that intimal thickening was not consistently correlated with static deformations or static stresses but instead was best associated with low flow velocity.
Fig 4⇑ gives the cross-sectional area values for the media in each portion of the graft for the same 14 animals. Comparison of the cuffed and contiguous noncuffed portions showed that medial thickening was almost twice as great in the distended regions as in the regions limited by the cuffs. This was true for the side on which the artery was ligated (high flow velocity, 1.53 versus 2.91 mm2) as well as the side on which the artery was left patent (low flow velocity, 1.79 versus 2.71 mm2). These differences were statistically significant (P<.05). However, comparison of the distended regions across the two sides (2.91 versus 2.71 mm2) disclosed no statistically significant differences. Similarly, comparison of the cuffed regions across the two sides (1.53 versus 1.79 mm2) showed no significant differences. Therefore, unlike intimal thickening, medial thickening was associated with graft deformations or static wall stresses, not with flow velocity. However, it was not possible from those data to identify the direction of the deformation or stress associated with medial thickening. To make this identification, step 2 and step 3 experiments were performed.
Step 2 Experiments: Longitudinal Extension
Step 2 experiments were performed in 13 animals to determine the effects of deformation and stress in the longitudinal direction. Grafts were constructed in an end-to-end fashion with the graft on one side extended longitudinally 12.6±0.8% compared with the pressurized but unstretched graft on the contralateral side. Extending the vein also caused thinning of the wall and therefore increased radial deformation (wall thinning). Longitudinal stretch also produced increased longitudinal stress. Fig 5⇑ demonstrates that longitudinal hyperextension did not produce increased intimal thickening. The cross-sectional areas were 0.76±0.13 mm2 on the stretched side versus 0.67±0.20 mm2 on the unstretched side. The difference was not statistically significant. From these data, it may be concluded that intimal hyperplasia was not associated with increased longitudinal deformation (increased longitudinal stretch) or increased radial deformation (wall thinning).
Fig 5⇑ provides the cross-sectional areas of the media of the grafts in the 13 animals. Medial areas were 5.83±0.62 mm2 on the stretched side versus 6.02±0.71 mm2 on the unstretched side. These areas were not significantly different from one another. Therefore, one may conclude that neither intimal nor medial thickening was associated with deformation or stress in the longitudinal direction or deformation in the radial direction. This suggests that medial thickening is associated with deformation or stress in the circumferential direction. Further clarification required step 3 experiments.
Step 3 Experiments: Circumferential Deformation Versus Stress
Step 3 experiments involved reversed femoral veins anastomosed end-to-end as interposition grafts in the carotid artery (Fig 5⇑). A calibrated Marlex band was placed proximally around the carotid arteries on one side of the animals to lower the pressure delivered to the graft but maintain diastolic pressure at greater than 50 mm Hg. Fig 7⇓ shows pressure-diameter relations from an earlier study for 30 native dog veins and dog vein grafts.18 These curves illustrate that both native veins and vein grafts exhibit marked distensibility at pressures up to 50 mm Hg, with very little distensibility and little change in diameter at pressures greater than 50 mm Hg. Therefore, the calibrated decrease in pressure produced by the Marlex band in the step 3 experiments (Fig 6⇑) permitted the vein grafts on the two sides to be at equal diameters, ie, equal deformations. However, because the pressure was decreased on the side with the carotid stenosis, the circumferential stress was decreased on that side. This occurs because stress is dependent on both pressure and radius (Equation 2). Statistical comparison demonstrated that intimal thickening for the nine animals in which the grafts remained patent was not significantly different on the two sides (1.30±0.34 versus 1.08±0.37 mm2). Therefore, it may be concluded that intimal hyperplasia was not associated with circumferential stress. However, as shown in Fig 4⇑, medial thickening was greatest where circumferential deformation was largest. From these data and those of previous experiments (Figs 3⇑ and 5⇑), one may conclude that intimal thickening is best correlated with low flow velocity and not with increased deformation or static stresses in any direction in the vessel wall.
Fig 6⇑ also shows that the cross-sectional area of the media of the vein grafts was 4.02±0.97 mm2 on the side with the proximal stenosis compared with 3.85±0.52 mm2 on the nonstenotic control side. These medial areas were not significantly different. Therefore, medial thickening was similar in the presence of equivalent circumferential deformations, despite the fact that the circumferential stress was decreased on the side with the stenosis. From these data and those of previous experiments (Figs 4⇑ and 5⇑), it may be concluded that medial thickening of vein grafts was best associated with deformation in the circumferential direction (increased diameter) and not with circumferential stress. Fig 5⇑ shows that medial thickening was not associated with deformation or stress in the longitudinal direction.
During construction of the stenosis of the proximal carotid artery as shown in Fig 6⇑, the pulse pressure (systolic minus diastolic pressure) was reduced from 29 mm Hg on the nonstenotic side to 18 mm Hg on the side with the carotid band. Measurements made at the time the grafts were harvested showed that the pulse pressure was reduced to a similar extent. It was 24 mm Hg on the unbanded side and 15 mm Hg on the side with the carotid band. Pulse pressures are directly related to pulsatile stresses (Equation 2). Because medial thickening was not significantly different on the two sides, it may be concluded that medial thickening is best associated with mean deformation and not with pulsatile pressures or pulsatile stresses. Conclusions really cannot be drawn regarding the role of pulsatile deformations in vein grafts because the vein wall is virtually nondistensible at pressures above 50 mm Hg (Fig 7⇑). Therefore, little pulsatile motion occurs with or without a proximal band as long as the diastolic pressure remains greater than 50 mm Hg.
The present data suggest that intimal and medial thickening in vein grafts are responses to two separate stimuli: intimal thickening occurs in response to low flow velocity and, presumably, decreased shear stress, whereas medial thickening occurs in response to deformation in the circumferential direction. This is consistent with the observation in experimental animals that intimal thickening can be inhibited by treatment with aspirin and dipyridamole, whereas medial thickening in the same grafts remains unchanged.18
Several studies have reported that intimal thickening in veins and prosthetic grafts is associated with low blood flow velocity,9 11 12 13 whereas other studies suggest that it is associated with circumferential wall tension14 or wall stress.14 15 16 However, exposing a vein graft or an artery to elevated pressures subjects it simultaneously to deformations and stresses in all three directions. The present study attempted to systematically separate these simultaneous factors. Flow velocity is directly related to low blood-artery shear stress, a factor increasing the probability of platelet adherence to the intima. Low shear stress also alters the mechanical and biochemical environment of the endothelium.19 Shear stress is directly related to blood flow velocity but cannot be computed accurately unless flow is well ordered and has developed stable velocity profiles. For this reason, no attempt was made in the present experiments to actually compute shear stresses in the different regions in the vein graft. However, the relative velocities are predictable through these series-coupled vessel segments, as dictated by Bernoulli’s principle.
Medial thickening occurs to the greatest extent in regions of the vessel subjected to increased circumferential deformation, ie, increased diameter. In vitro studies of arterial smooth muscle cells demonstrate that oscillating motion increases the synthesis of connective tissues,8 but it is not clear from those experiments whether the mean stretch around which pulsations oscillate or the pulsatile deformation itself is the effective stimulus. The present experiments suggest that medial thickening is best associated with increased mean deformation in the circumferential direction.
These experiments in vein grafts may be viewed as an indirect model of that which occurs in arteries subjected to hypertension. Studies in hypertensive animals treated with a variety of antihypertensive agents demonstrate that the structural changes that occur in the media of these arteries best correlate with increased pulse pressure.7 This may be explained by an important difference in the compliance characteristics of veins and arteries. Veins are virtually rigid at pressures greater than 50 mm Hg (Fig 7⇑), whereas arteries exhibit gradually decreasing compliance at pressures up to 200 mm Hg.10 As a result, veins subjected to increased arterial pulse pressures experience almost no change in dimensions. Conversely, arteries subjected to increased arterial pulse pressures undergo graded deformations, depending on the mean pressure and diameter and the magnitude of the pressure pulse. This is consistent with the finding that medial thickening in vein grafts depends on deformation in the circumferential direction.
Gariepy J, Massoneau M, Levenson J, Heudes D, Simon A, and the Groupe de Prévention Cardio-vasculaires en Médicine du Travail. Evidence of in vivo carotid and femoral wall thickening in human hypertension. Hypertension. 1993;22:111-118.
Warshaw DM, Mulvany MJ, Halpern W. Mechanical and morphological properties of arterial resistance vessels in young and old spontaneously hypertensive rats. Circ Res. 1979;45:250-259.
Wolinsky H. Responses of the rat aortic media to hypertension: morphological and chemical studies. Circ Res. 1970;26:507-519.
Owens GK, Schwartz SM. Alterations in vascular smooth muscle mass in the spontaneously hypertensive rat. Circ Res. 1982;51:280-289.
Bandick NR, Sparks HV. Viscoelastic properties of the aorta of hypertensive rats. Proc Soc Exp Biol Med. 1971;134:56-60.
Christensen KL. Reducing pulse pressure in hypertension may normalize small artery structure. Hypertension. 1991;18:722-727.
Leung DYM, Glagov S, Mathews MB. Cytolic stretching stimulates synthesis of matrix components by arterial smooth muscle cells in vitro. Science. 1976;191:475-477.
Dobrin PB. Vascular mechanics. In: Shepherd JT, Abboud FM, eds. Handbook of Physiology. Section 2, Vol 3: The Cardiovascular System. Washington DC: American Physiological Society; 1983:65-102.
Faulkner SL, Fischer RD, Conkle DM. Effects of blood flow rate on subendothelial proliferation in venous autografts used as arterial substitutes. Circulation. 1975;51-52(suppl I):I-163-I-172.
Kohler T, Kirkman T, Kraiss L, Zierler BK, Clowes AW. Increased blood flow inhibits neointimal hyperplasia in endothelialized vascular grafts. Circ Res. 1991;69:1557-1565.
Zwolak RM, Adams MC, Clowes AW. Kinetics of vein graft hyperplasia: association with tangential stress. J Vasc Surg. 1987;51:126-136.
Dobrin PB. Mechanical factors associated with the development of intimal hyperplasia with respect to vascular grafts. In: Dobrin PB, ed. Intimal Hyperplasia. Georgetown, Tex: RG Landes Co; 1994:85-110.
Evans LV, Sumpio BE. The role of mechanical forces in vitro in the development of intimal hyperplasia. In: Dobrin PB, ed. Intimal Hyperplasia. Georgetown, Tex: RG Landes Co; 1994:111-140.