Hemodynamic and Mechanical Factors

A wide variety of hemodynamic and mechanical factors, including high- and low-flow velocities, high and low wall shear stress, and mechanical compliance mismatch, have been implicated in the formation of intimal hyperplasia.^35–37 The effects of flow velocity on the subsequent development of intimal hyperplasia have been studied in a variety of models. In general, low blood velocity tends to favor the development of neointimal hyperplasia. For example, in a canine carotid vein interposition model, segments with low-flow velocities developed significantly thicker intimal layers³⁸; however, there are exceptions to this rule. In a high-flow renal artery–to–vena cava anastomosis, significant intimal thickening was documented by electron microscopic examination 3 months after formation.³⁹ Similar intimal lesions have also been noted in arteriovenous fistulas (AVF) constructed for hemodialysis access.⁴⁰ This may be a result of the compliance mismatch noted between the arterial and venous systems. In a monkey iliac AVF model, 6 months after construction no increase in intimal thickening occurred on the experimental side.⁴¹ In that study, although flow rate and velocity were markedly increased on the side of the AVF, the calculated wall shear stress was equal on both sides. This equality of shear stress, despite a significant difference in flow velocities, was the result of a twofold increase in the vessel lumen diameter. The vessel wall regulates its diameter to produce a physiologic level of shear stress at the lumen. The endothelium regulates diameter by producing both vasoconstrictors such as endothelin, and vasodilators such as nitric oxide (NO). These factors also affect VSMC proliferation and when chronically present, cause remodeling of the vessel to maintain the new diameter without vasoactivity.

In other primate studies, arteriovenous flow has been shown to inhibit neointimal thickening in highly porous polytetrafluoroethylene (PTFE) grafts.⁴² Return to normal flow causes thickening of these lesions. In a rodent model of arterial injury, low flow caused increased intimal hyperplasia.⁴³ These and many other studies demonstrate that intimal hyperplasia is increased by low shear and inhibited by high shear. Regions of low wall shear stress also may stimulate intimal proliferation by increasing the time for lipid transport.^35,38 The results of postmortem studies have shown that early atherosclerotic lesions occur more commonly in areas of low wall shear stress.⁴⁴ In the carotid artery, atherosclerosis starts at the outer wall of the bulb, where shear forces are low and oscillating.^45,46

Compliance mismatch is also an important hemodynamic factor in the production of anastomotic intimal hyperplasia.⁴⁷ Compliance is defined as the percentage of radial change per unit pressure and is a useful index of vessel wall distensibility to a pressure force. Although experimental and clinical studies have shown that conduits with compliance values approaching that of the native artery have better patency, none of these studies controlled for differences in graft surfaces. In one study, femoropopliteal autografts made from glutaraldehyde-treated carotid arteries had better patency when fixation produced a more compliant graft.⁴⁸ However, other studies have shown that the creation of stiffer, less compliant grafts by treating arteries with glutaraldehyde results in markedly diminished patency rates.^47,49 Although the compliance of a vein graft at the time of implantation is similar to that of a native artery, according to the results of one study the compliance values actually remained within the normal range for a median follow-up of 33 months.⁵⁰ Textile and fabric prostheses, in contrast, are relatively noncompliant. Furthermore, 4 months after implantation, a significant loss of compliance is noted in both polyester fabric and PTFE grafts.⁵¹

For this reason, some authors have suggested imposing a short interposition of autogenous vein between the native artery and prosthetic grafts.⁵² These modifications are largely intended to decrease the compliance mismatch between the prosthetic graft and the native artery. They include vein cuffs, vein patches, vein boots, and AVF at the distal anastomosis (Figure 55-5).⁵³ A prospective randomized trial to examine the efficacy of Miller vein cuffs in improving lower extremity PTFE graft patency showed that vein cuffs did not improve patency rates in bypasses to the above-knee popliteal artery.⁵⁴ However, cuffed below-knee popliteal artery grafts had a 45% 3-year patency rate versus a 19% patency rate in uncuffed grafts (p = 0.018).⁵⁵ Similarly, the use of vein patches, such as those developed by Taylor and Linton, at the distal anastomosis yields better long-term patency rates versus prosthetic material alone.^56,57 Although there have been no randomized, prospective trials to date, retrospective reviews have demonstrated improved patency of vein-patched PTFE versus PTFE alone when used for below-knee bypasses. Taylor and colleagues reported a 5-year primary patency of 54% for 83 infrapopliteal PTFE grafts using his vein patch modification.⁵⁶ Although there is no control arm to this study, previously reported patency rates with PTFE alone are significantly worse.¹ A review of 145 patients who had undergone either above- or below-knee bypasses using autologous vein, unmodified PTFE, and Linton patch–modified PTFE found that above-knee bypasses had no statistical difference in 1- and 3-year patency rates between the three groups.⁵⁸ However, the cumulative patency rates for 1, 3, and 5 years after below-knee bypasses were 93%, 75%, and 75% for the autologous vein group (n = 30); 74%, 52%, and 42% for the unmodified PTFE group (n = 37); and 93%, 93%, and 93% for the Linton patch–modified PTFE group (n = 16), respectively. They speculated that the improved results over unmodified PTFE bypass grafts were a result of improved compliance mismatch between the prosthetic graft and the targeted native artery. The study was limited by the small number of Linton patch–modified PTFE grafts used compared with the autologous vein and PTFE groups. Finally, a retrospective analysis of outcomes of patients with AVF-modified PTFE grafts versus PTFE alone to the infrapopliteal arteries showed that the 2-year patency rates were significantly better for grafts with the AVF modification (23% vs. 5% for the PTFE alone grafts, p < 0.05).⁵⁹ The AVF, originally described by Ascher and colleagues, is created by first mobilizing an adjacent vein and transecting it.⁶⁰ The vein is then anastomosed to the artery in an end-to-side fashion. A veinotomy is created near the AVF and the ePTFE is anastomosed, creating the AVF modification at the distal anastomosis (see Figure 55-5). Despite these surgical improvements, patency rates remain poor, especially when the distal outflow is at or below the level of the popliteal artery, primarly as a result of the formation of neointimal hyperplasia.

FIGURE 55-5 Surgical techniques using vein interposition for creation of distal anastomoses. A, Standard end-to-side anastomosis; B, Linton patch; C, Taylor patch; D, Miller cuff; E, vein boot; and F, arteriovenous fistula.

Platelets

Platelets have long been known to play a central role in the reaction of a vessel wall to injury. To date, most research in this area has focused on the activation of platelets by the injured endothelium as the major factor in the development of intimal hyperplasia. Denudation of the arterial wall exposes the subendothelial matrix, which leads to adherence and subsequent activation of platelets. The luminal surface quickly passivates by adsorption of plasma proteins, limiting the period of platelet aggregation to less than 24 hours. Platelet adherence requires the interaction of subendothelial collagen, a platelet membrane glycoprotein receptor (GP Ib), plasma von Willebrand factor, and fibronectin. After adherence, platelets undergo a morphologic change, stretching to cover the exposed surface. Activated platelets release adenosine diphosphate and activate the arachidonic acid pathway to release thromboxane A2. Both of these factors lead to platelet aggregation. Recruitment of platelets requires the rapid expression of the platelet membrane receptor complexes GP IIb and GP IIIa, both of which promote platelet aggregation through the binding of circulating fibrinogen. Platelet adhesion and granule release also lead to a parallel acceleration of the coagulation cascade. This activation of clotting pathways, combined with high local concentrations of fibrinogen mediated by binding to the GP IIb/IIIa complex, creates a fibrin protein network that further stabilizes the aggregated platelet plug. Once platelet aggregation is initiated by the pathways mentioned earlier, its further formation is actively inhibited by an intact endothelium. Thus, a damaged endothelium not only initiates platelet activation but also impairs its inhibition.

Platelet-derived growth factor (PDGF) is secreted along with other granule constituents, including platelet factor 4, thromboglobulin, and thrombospondin. PDGF is a cationic protein with a molecular weight of 28 to 31 kDa. It comprises two subunits (α and β). Physiologically, PDGF functions as both a chemoattractant and a mitogen for VSMC and fibroblasts. Because it binds with high affinity to VSMC, some have suggested that PDGF may attract the VSMC from the media into the intima, bind to them, and stimulate their proliferation. Interestingly, evidence has shown that the platelet may not be the only source of this protein. PDGF (α and β subunits) is produced by human umbilical vein and saphenous vein endothelial cells. In fact, a large increase in PDGF production can be measured in injured endothelial cells. Also, both α-subunit and β-subunit messenger RNA (mRNA) have been noted in fresh endarterectomy specimens obtained during carotid surgery.⁶¹ VSMC themselves also produce PDGF-like activity in response to arterial injury, and VSMC from human atheroma contain mRNA for the PDGF α subunit.⁶² Taken together, these findings may explain how proliferation continues after re-endothelialization occurs. PDGF may be released by both platelets and endothelial cells, causing the activation and migration of VSMC and myofibroblasts, which then secrete additional PDGF, leading to proliferation. Finally, demonstrating the importance of PDGF to the arterial injury response, inhibition of PDGF using antibodies or antisense oligonucleotides in animal models of arterial injury has been shown to inhibit neointimal hyperplasia by 60% to 80%.^63–65

Inflammatory Cells

Inflammatory cells are intimately involved in the arterial response to injury. Electron microscopic studies after balloon catheter intimal injury show that leukocytes attach to the de-endothelialized surface of an arterial lumen within minutes.^17,18,66 This includes polymorphonuclear neutrophil (PMN), monocytes, and lymphocytes. In addition to adhering to the damaged endothelium, these cells even penetrate it in some instances. These adherent leukocytes are paramount to the secretion of many cytokines and growth factors that influence the subsequent events in neointimal hyperplasia (Table 55-2).^67–69 The central role of leukocyte chemotaxis in the arterial injury response has been demonstrated through several novel models of inflammation. One model that supports the role of inflammation in the development of neointimal hyperplasia consists of delivering leukocyte-derived myeloperoxidase in the presence of its substrate hydrogen peroxide (H₂O₂). The infusion of myeloperoxidase and H₂O₂ into an isolated rat common carotid artery without mechanical injury elicited the development of neointimal hyperplasia.⁷⁰

TABLE 55-2 Cellular Sources of Growth Factors and Cytokines Involved in the Arterial Response to Injury

In another model, an endotoxin-soaked thread was placed on half of a rat femoral artery to produce an inflammatory response.⁷¹ This technique consistently caused a significant leukocyte infiltration, which occurred only on the treated side of the vessel. Histologic examination performed 14 days later revealed nonuniform neointimal lesions in which proliferating VSMC were located exclusively on the side of the lumen adjacent to the treated half of the arterial wall. Clearly, these animal models suggested a strong association between inflammation and neointimal hyperplasia. The mechanisms controlling this relationship are a target for therapeutic intervention.

PMN adhesion to the surface of endothelial cells is controlled by several complex glycoproteins located on the surface of both endothelial and white blood cells. Together, these binding molecules constitute a sophisticated communication system. Two endothelial cell adhesion molecules involved in neutrophil binding have been well characterized: endothelial leukocyte adhesion molecule-1 (ELAM-1) and intercellular adhesion molecule-1 (ICAM-1). ELAM-1 is either unexpressed or remains intracellular in the inactive endothelial cell. However, after activation by a variety of different cytokines, this adhesion complex is rapidly induced and appears on the membrane surface of the stimulated endothelial cell. ICAM-1 is located in small amounts on the surface of inactive endothelial cells. After activation by either injury or a variety of stimulating agonists, the expression of this binding complex is significantly increased. ICAM-1 is the binding ligand for the CD11a/CD18 receptor on the leukocyte membrane.

Histochemically, the adhesion molecules located in the surface membranes of the white blood cells responsible for leukocyte binding to endothelial cells can be separated into three related heterodimers. Each of these heterodimer protein complexes are composed of an α and a β subunit. The α subunit differs among the three glycoprotein complexes (CD11a, CD11b, or CD11c), whereas the β subunit remains constant (CD18). The CD11a/CD18 complex is found on the surface of all white blood cells and mediates the attachment of unstimulated PMNs to stimulated endothelial cells. This binding probably occurs through an interaction with the ICAM-1 receptor that, as mentioned, is expressed on the luminal surface of activated or injured endothelial cells. The binding of the CD11a/CD18 complex to the ICAM-1 receptor may also play a role in the cytokine-induced transendothelial migration of PMNs. The CD11b/CD18 adhesion complex is referred to as either “Mac-1” or the “C3b complement receptor.” This glycoprotein has been implicated in the adhesion of chemotactically stimulated PMNs and controls several cellular functions such as aggregation and cytotoxicity. The third heterodimer complex, CD11c/CD18, exists on both PMNs and monocytes, but its role in binding to endothelial cells remains undetermined.

During inflammatory states, the attachment of PMNs to the involved endothelium is greatly increased, primarily because of the upregulation and enhanced expression of the binding glycoproteins described earlier. A variety of substances are capable of stimulating this enhanced PMN adherence. Interleukin (IL) 1, tumor necrosis factor-α, lymphotoxin, and bacterial endotoxins (lipopolysaccharide) all increase the production of both ELAM-1 and ICAM-1 on the surface of affected endothelial cells. In addition, PMN activation can be stimulated by several substances that are released secondary to inflammation. Examples include the complement factors C5a and C3b, which stimulate PMN chemotaxis and phagocytosis, respectively. Likewise, PMN adhesion and chemotaxis can also be stimulated by interleukin-1, xanthine oxidase, PDGF, and the lipid mediators leukotriene B and platelet-activating factor. Finally, tumor necrosis factor-α is an important mediator of PMN phagocytosis and can lead to increased lysosomal enzyme release.

After activation and adhesion to the damaged vessel lumen, leukocytes migrate into the arterial wall. In one study, 42 days after a denuding injury, leukocytes were shown to penetrate the arterial media and were seen deep within the hyperplastic lesions.⁷² Several different mechanisms trigger this leukocyte migration. VSMC and macrophages elaborate potent chemotactic factors for leukocytes, which could sustain continued leukocyte recruitment.⁷³ Supporting this notion, serum containing medium that was conditioned from VSMC stimulated leukocyte migration. Other chemotactic agents for leukocytes include PDGF and, in some reports, factors released by fibroblasts. VSMC removed from atherosclerotic plaques expressed ICAM-1, suggesting a possible connection between these medial cells and leukocyte recruitment and activation.

The mechanism by which leukocytes may initiate the formation of intimal hyperplasia after penetrating the injured arterial wall remains unknown. Again, several pathways have been suggested. After a denuding endothelial injury and the deposition of inflammatory cells, a variety of inflammatory products may be elaborated. These include chemotactic factors, growth factors, complement components, and enzymes. One of the most studied substances is monocyte- and macrophage-derived growth factor (MDGF). MDGF is a well-known stimulator of VSMC and fibroblast proliferation. This growth factor is similar and may be identical to PDGF. Thus, the stimulation of VSMC proliferation is one mechanism by which inflammatory cells may contribute to the formation of neointimal hyperplasia. A second possibility involves the production of lysosomal degradation enzymes. Activated leukocytes secrete several potent proteases capable of degrading collagen, basement membranes, and other important extracellular structural proteins. One example of a PMN-derived enzyme that has been implicated in peri-inflammatory extracellular damage is myeloperoxidase. Liberation of these destructive enzymes into the wall of an injured vessel may weaken the extracellular matrix. This “loosening” of the vessel wall may facilitate the migration of VSMC from the medial layer toward the lumen. Finally, leukocytes may act directly at sites of vessel injury to extend endothelial injury. PMNs can produce oxygen-free radicals through the action of the (nicotinamide adenine dinucleotide phosphate [NADPH]) oxidase system present on their membranes. The toxic substances elaborated by these activated PMNs, including superoxide anion, hydrogen peroxide, and hydroxyl radicals, can damage endothelial cells and alter capillary permeability. With PMN activation, marginally injured endothelial cells bordering a lesion may be destroyed, increasing the magnitude of the damage to the vessel wall. This further exposure of the subendothelial layer allows more platelet cell adherence, aggregation, and activation, as well as the recruitment of more leukocyte mediators and thus the stimulation of a continued cycle of inflammatory injury.

In addition to luminal entry, the adventitia has been recognized as an important site of inflammation after arterial injury (Figure 55-6). The accumulation of B cells, T cells, and macrophages in the adventitia has been documented in numerous studies.^74–77 Okamoto and colleagues⁷⁸ characterized macrophage localization to the adventitia after porcine coronary artery angioplasty and which receptors are important to their recruitment. Macrophages present in the adventitia were significantly increased at day 1 after injury, peaked at day 3, and then slowly decreased back to baseline at day 14. Interestingly, the number of adventitial macrophages was larger than the number found in either the media or the intima. In addition, macrophage accumulation preceded cell proliferation by 48 to 72 hours. Remarkably, they also found that macrophages had a nearly twentyfold greater presence than neutrophils. Early and sustained expression of E-selectin, P-selectin, and VCAM-1 by the vasa vasorum indicated that these receptors were important in allowing the macrophages to localize to the adventitia. Macrophages have also been noted to phenotypically differentiate into myofibroblasts after arterial injury and contribute to the development of neointimal hyperplasia.⁷⁹ Bayes-Genis and coworkers⁷⁹ demonstrated that 14 days after arterial injury to the pig coronary artery, 42% of the neointimal cells expressed the macrophage-specific antigen SWC3, and 9% of the cells co-expressed SWC3 and alpha smooth muscle actin (αSMA). Finally, macrophage localization and activation has been found to rely on angiotensin II (Ang II)–induced upregulation of CCL2 and IL-6. Tieu and associates was recently able to quantify the degree to which Ang II increases the macrophage population in the adventitia and found a 5.3-fold increase in populations of mice treated with Ang II.⁸⁰ They also documented the importance that CCL2 plays in Ang II–induced macrophage localization because the absence of CCR2, a CCL2 receptor, in CCR2^−/− mice significantly attenuated adventitial macrophage accumulation. The cytokine IL-6 was shown to be important to the activity of Ang II because IL-6^−/− mice adventitia did not show the same propensity to recruit macrophages as the IL-6^+/+ mice after Ang II stimulation.

FIGURE 55-6 Arterial cross sections of immunohistochemical staining for (A) monocytes/macrophages (red staining) and (B) lymphocytes (brown staining) 2 weeks after arterial balloon injury in the rat carotid artery. Note that the majority of infiltrates are found in the adventitia.

Growth Factors

The stimulation of cellular proliferation is the final common pathway of all postulated mechanisms, leading to the development of intimal hyperplasia. The view that the vascular wall is a complex integrated organ, complete with its own endogenous local autocrine system, is gaining increasing support. In this theory, neointimal hyperplasia is postulated to result from an imbalance of these local hormonal systems. This could be caused by an excess of stimulatory molecules; alternatively, cellular proliferation may result from the absence or reduction of inhibitory hormones.

Angiotensin II, PDGF, and MDGF are all examples of growth factors believed to be involved in the formation of hyperplastic neointimal lesions; each has been reviewed in detail earlier. bFGF, a smooth muscle cell mitogen that is not secreted, is released from damaged VSMC and acts in an autocrine fashion to stimulate proliferation of adjacent, viable cells. This mitogen is largely responsible for the proliferative response after balloon injury.⁶ Denudation of endothelium with a fine wire that does not damage the endothelium results in much less VSMC activation.⁸¹ Lindner and associates¹⁹ observed that administration of bFGF after an injury increased VSMC proliferation from 11.5% to 54.8%. Their finding of an equivalent increase in proliferation after wire loop injury demonstrates that bFGF can act as a mitogen for undamaged medial cells. bFGF administered to normal, nondenuded arteries has no effect on VSMC growth. bFGF is also a potent stimulator of endothelial cells and, accordingly, stimulates angiogenesis. After arterial injury, bFGF is important for stimulating endothelial cell growth to completely cover injured regions that would normally remain denuded. Cuevas and colleagues demonstrated that the direct local infusion of bFGF into either normal adventitia or injured media results in proliferation of both vasa vasorum and VSMC.⁸²

TGF-β is mostly known for its role in stimulating extracellular matrix deposition. More recently, however, it is now understood to have a role in negatively regulating endothelial cell activation, VSMC migration, and macrophage and lymphocyte accumulation.⁸³ Early studies demonstrated that TGF-β was expressed after arterial injury.²⁴ Furthermore, application of TGF-β to the arterial wall using gene therapy or intraluminal delivery stimulated the development of neointimal hyperplasia.^23,84,85 Inhibition of TGF-β using antibodies inhibited neointimal hyperplasia.^86–88 These early studies focused on the stimulation of procollagen and fibronectin, as well as effects on remodelling. However, identification of SMAD3 because the downstream target of TGF-β has shed more light on the precise role of TGF-β. SMAD3 knockout mice exhibited significantly more neointimal hyperplasia after injury compared with the wild type controls. Specifically, they exhibited more cellular proliferation, and bone marrow transplantation studies revealed that the neointima that developed in the SMAD3 knockout mice was derived from resident cells, not circulating cells.⁸⁹

Renin-Angiotensin System

The classic view of the regulatory function of the renin-angiotensin system is that it is primarily an endocrine-based system designed for the homeostatic control of hemodynamic and electrolyte balance. In response to low perfusion pressures, renin is released by renal tissue and circulates in the plasma, where it cleaves angiotensinogen, produced by the liver, into angiotensin I. Angiotensin I is converted into active Ang II by angiotensin-converting enzyme (ACE), located primarily in the pulmonary vasculature. Finally, Ang II exerts its homeostatic hemodynamic effects via specific Ang II receptors located in peripheral vascular arterial beds. This traditional view of the renin-angiotensin system has been revised, and a much more complex concept has emerged. In this new description, the primary site of Ang II production is not the pulmonary vasculature but rather local sites within the affected tissues themselves. This portion of the entire renin-angiotensin axis has been referred to as the tissue renin-angiotensin system.

The concept that active Ang II in the vascular wall is synthesized locally, and not delivered via the systemic circulation, was originally suggested by Swales and Thurston.⁹⁰ In this report, the authors noted that the amount of angiotensin antiserum required to inhibit endogenous angiotensin effects in sodium-loaded rats could not be explained solely on the basis of systemic production. In addition, several investigators had noticed residual renin-like activity after bilateral nephrectomy in animal studies.⁹¹ More convincing evidence for the existence of a locally active vascular renin-angiotensin system, operating independently of the classic systemic circuit, has been provided by several studies. Angiotensinogen mRNA, the only known precursor to the angiotensin peptides, has been detected in several extrahepatic vascular tissues, including the aorta.^92,93 Immunohistochemical studies have demonstrated renin in cells throughout the vascular wall.⁹⁴ Both VSMC and endothelial cells can synthesize renin in vitro.^95,96 Also, mRNA coding for renin has been identified in human VSMC.^91,97 Finally, ACE is located primarily on the luminal surface of vascular endothelial cells.⁹⁸ Thus, the normally functioning vascular wall possesses all the necessary components for the independent local production of Ang II.

The physiologic function of locally produced Ang II is an area of continued controversy. Ang II receptors have been identified on vascular endothelial cells, VSMC, and circulating platelets.^91,99–101 Ang II stimulation of endothelial cell–bound receptors leads to the secretion of prostacyclin and possibly endothelium-derived relaxant factor.^102,103 Both of these substances cause VSMC quiescence and relaxation. Activation of the angiotensin receptors located directly on the VSMC themselves causes the opposite effect. Campbell-Boswell and Robertson¹⁰⁴ reported that Ang II stimulates the proliferation of human VSMC in vitro. Geisterfer and associates¹⁰⁵ found that the protein content of these VSMC increased by 20% when stimulated by Ang II for 4 days. Further, this stimulation was abolished by the Ang II receptor antagonist saralasin. Finally, the Mas protooncogene, located on the surface of medial VSMC, increased mitogenic activity when stimulated and has been identified as a functional Ang II receptor.¹⁰⁶ Therefore, one explanation for the physiologic function of a locally active renin-angiotensin system is the autocrine balance of vascular wall metabolic activity and tone.

This view has led many investigators to study the effect of the local production of Ang II on the subsequent development of intimal hyperplasia. In this hypothesis, denudation of the arterial endothelium disrupts the balance of the local renin-angiotensin system and allows the anabolic and mitogenic effects of Ang II on the medial VSMC to proceed unchecked. Another possible mechanism for the promotion of hyperplastic neointimal growth by the local production of Ang II involves the activation of platelet metabolic pathways. As mentioned previously, human platelets possess specific binding sites for Ang II, and platelet activation and the subsequent release of PDGF may play a role in the development of neointimal hyperplasia.¹⁰¹ In a 1990 study by Swartz and Moore,¹⁰⁷ Ang II enhanced both collagen-induced platelet aggregation and the production of thromboxane A2. Finally, the expression of PDGF by activated VSMC is upregulated in the presence of Ang II.¹⁰⁸