Biomechanics and Mechanobiology of Saphenous Vein Grafts

In this review, we consider how the mechanical environment regulates the physiological and pathological remodeling of saphenous veins (SV) when they are used as vascular grafts. There is a large body of literature on how different aspects of the mechanical environment regulate the remodeling of various arteries and veins. While we draw upon studies of vessels other than SV, whenever possible, we base our discussion on studies of SV for several reasons. First, there already are reviews of the role of mechanical forces in vascular remodeling [1–4] and vascular disease [5,6] including several that focus on vascular grafts [7,8]. Instead of contributing another review to a more general topic that has already been well covered, we chose to focus on a more specific subarea that, to our knowledge, has not been reviewed comprehensively. Second, while considering the results from studies of various vessels as a whole is an appropriate way to uncover reoccurring themes, we believe that it is also important to consider the unique aspects of various vessels and vascular beds. Third, mechanically induced remodeling of the SV is involved in clinically important conditions including varicose veins, peripheral revascularization and especially coronary artery bypass grafts (CABG).

Saphenous veins used in CABG experience environments dramatically different than that of their native environment. While it is clear that some aspects of these changes in environment trigger remodeling in the veins, in some cases, there is not a consensus on which environmental stimuli elicit which aspects of remodeling. The primary goal of our review is to contribute to a better understanding of what are the major environmental stimuli for different forms of remodeling by SV. Given this focus, we do not concentrate on specific cellular or molecular level mechanisms, though we do include references to other reviews that contain these topics.

The use of the SV [9] and internal mammary artery (IMA) [10] for CABG were established in the 1960s. Over the subsequent decades, the use of other conduits has been explored including the radial [11,12] and gastroepiploic [13,14] arteries but the SV and IMA continue to be used in the vast majority of bypasses. In CABG patients, the IMA is routinely used to bypass the left anterior descending (LAD) artery and the SV to bypass other coronary arteries. The conventional technique to harvest the SV, now referred to as open or direct-vision harvesting, consists of making a long linear incision that can run most of the length of the leg. Complications of open harvesting can include postoperative pain, infection, and poor healing. To minimize these complications, clinicians have transitioned from a single incision over the entire length of the vein to a series of smaller interrupted bridging incisions. By 2008, ∼70% of SV isolations were performed using endoscopic techniques [15]. Once the vein is removed, it is distended (typically with heparinized saline) and leaks, either from side branches or small tears, are repaired either with sutures or small clips. The vein is visually inspected for defects, such as significant varicosities or areas that are felt to be clinically of “poor quality.” These defects are typically discarded or utilized on less critical bypasses if additional higher quality vein segments are not available. The vein is typically then stored in heparinized saline until it is needed.

Data from the Society of Thoracic Surgeons' National Adult Cardiac Database, which collects data from ∼80% of all hospitals that perform CABG in the U.S., reported that from 2008 to 2016, there were 150,000 CABG procedures per year [16]. It is standard practice to use the left IMA for CABG to revascularize the left anterior descending artery. According to the Society for Thoracic Surgery's data for the first quarter of 2011, 95.0% of CABG patients received a left IMA graft. Most CABG patients (94%) received multiple bypasses with an average number of grafts of 3.2 per operation. While some patients receiving multiple bypasses receive all-arterial revascularization utilizing the right internal mammary artery, the radial arteries, and rarely the gastro-epiploic artery, the majority do not. For example, in the first quarter of 2011, right IMA were used in only 4.8% of CABG patients and radial arteries were used in only 4.8% of CABG patients. Since the average number of grafts is about three per patient, of which one is almost always the left IMA, that leaves about two grafts per patient. Only a small fraction of the patients receive arterial grafts other than the left IMA, so on average there are about 2 SV grafts per CABG patient. Thus, a large fraction of the ∼175,000 patients receiving CABG in the U.S. each year receive one or more SV grafts.

Given the difference between how patency is defined and assessed in different studies as well as the presence of many uncontrolled variables, we suggest that conclusions regarding patency rates are better based on consistent trends seen across multiple studies opposed to what is seen in one or several studies. In their recent review of SV graft failure, Harskamp et al. present data from 14 clinical studies on SV failure [17]. On average, patency rate per graft was ∼95% at 1 month, ∼85% at 1 year, ∼70% at 5 years, and ∼60% at 10 years [17]. In contrast, the patency rates of the IMA are much greater with a number of studies reporting IMA patency rates of ∼90% at ∼10 years [18,19].

While 10 year patency values of ∼60% for SV and ∼90% for IMA accurately reflect the patency of these vessels as they are typically used, failure to consider these values in context may be potentially misleading. Since failure of the LAD coronary artery has profound negative impact on clinical outcomes [20], the vast majority (>90% prevalence) of the LAD grafts use IMA while the IMA is seldom (<5% prevalence) used to bypass other coronary arteries. Thus, the more appropriate comparison is how the patencies of the IMA and SV compare when used to bypass the same coronary artery. The 10-year patency of SV grafts to the LAD is ∼70%, which is better than the ∼60% when it is used to bypass other coronary arteries (right coronary artery or circumflex) but still less than the ∼90% for IMA bypass of the LAD [19]. Several clinical studies comparing the patency of SV and IMA as bypass coronary arteries other the LAD found that the SV and IMA had similar patency. In summary, IMA is superior to the SV for the bypassing the LAD but not necessarily for bypassing the other coronary arteries. Thus, it may be preferable to consider the IMA as being particularly well suited for grafting to a target vessel that is already more likely to maintain patent grafts and not necessary to consider the IMA as an inherently better graft for all coronary arteries.

In this section, a concise review of the remodeling of SV grafts is presented with the goal of providing background for a subsequent discussion of factors regulating the remodeling. A number of thorough reviews of vein graft remodeling or vein graft disease are available [21–29]. It is common to divide the pathological remodeling of grafted SV into three major stages: “thrombosis,” “intimal hyperplasia (IH),” and “atherosclerosis.” These stages are related to the major modes of graft failure at early (∼first month), intermediate (∼first year) and late (after several years) time points, respectively, [22,21,27]. The process associated with each name is not limited to that stage, however, as illustrated below.

While SV do not experience atherosclerosis in their native environment and diseased veins (e.g., varicose veins) are typically not used in CABG, this does not mean that the SV is pristine before harvest. Intimal thickening due to phlebosclerosis is common even in vessels that appear macroscopically normal [30,31]. The degree of this thickening in a typical human SV used for CABG (Fig. 1(a)), however, typically is small. The average intimal thickness and outer diameter of SV harvested for CABG were 0.064 mm and 4.2 mm, respectively [32] and only in rare cases, ∼1% of SV harvested for CABG, is there >50% stenosis [33].

Within a day of implantation, local fibrin deposits are evident on the intima, potentially due to denuding or dysfunction of the endothelium. At this early time point, occlusive thrombosis is rare but within the first month following implantation, approximately 5% of SV grafts fail, primarily due to thrombosis. Early thrombosis is attributed to damage to the endothelium, poor surgical technique, and poor distal runoff [27]. In the longer term, thrombosis can be remodeled to fibrosis, which can continue to fully occlude the SV graft [26] or be recanalized by the body allowing partial flow [27]. Within a week to month, intimal thickening is histologically evident, though not to a degree that would influence profusion [34].

Intimal hyperplasia, which is characterized by the proliferation and migration of smooth muscle cells (SMCs) from the media into the intima, is the major disease process from ∼1 month to several years after implantation [22,27]. While intimal hyperplasia contributes to the narrowing of SV grafts with ∼25% decrease in luminal diameter typical 1 year after implantation (Fig. 1) [27,35,36], it is unlikely that IH directly leads to complete occlusion [37]. Graft failure in the first several years is likely due to thrombosis in grafts already partially narrowed by intimal hyperplasia.

Atherosclerosis accounts for most stenosis and graft failure after 3–5 years. While this is commonly called late graft failure, it is noteworthy how quickly vein graft disease typically progresses. In contrast to typical atherosclerosis in arteries, which takes multiple decades to progress, many grafted SV transition from healthy and functional grafts to diseased and stenotic grafts within several years.

Atkinson and coworkers characterized the long-term morphological changes in 117 SV grafts from 56 cadavers that had received SV grafts on average 5 years prior. All grafts exhibited pathological remodeling with 25 exhibiting atherosclerosis, 66 fibrointimal proliferation, and 26 with fibrosis, and total occlusion. Thus, even among the grafts that would be considered patent by most standards (e.g., 18% of grafts had less than 25% stenosis and 36% had 25–50% stenosis in this study), some degree of pathological remodeling is present [26].

The observations that human SV do not develop significant atherosclerosis in their native venous environment coupled with the rapidity that they develop intimal hyperplasia following grafting into the arterial circulation suggests that factors associated with the isolation and preparation of saphenous veins and/or differences between the venous and arterial environments contribute to disease progression. At least in an animal model, harvesting and grafting of veins back into the venous circulation does not stimulate IH while grafting into the arterial circulation does, suggesting exposure to some aspect of the arterial environment is essential for the development of IH [38]. In Sec. 3, changes in the environment experienced by a vein as it is surgically removed from the venous circulation, prepared for grafting, and placed in the arterial circulation are described and summarized (Table 1). Only once all of the changes in the environment are covered are the potential impacts of these changes on the vein graft considered in Sec. 4. The decision to separate the discussion of the changes in the environment and the potential impacts of these changes is motivated by a belief that it is important to first consider all the potential factors that might be responsible for an observed change before one attempts to attribute a given stimulus to a given change.

The SV is vascularized by a vasa vasorum [39]. Before harvesting, the SV's vasa vasorum is fed by branches from the femoral artery with at least some of the outlets from the vasa vasorum empting into the lumen of the SV [39]. Thus, excising the SV disrupts the flow of arterial blood to the vein's vasa vasorum, which can be further reduced when the highly vascularized perivascular and adventitial tissue are stripped during conventional preparation of the veins (Fig. 2) [40,47]. Innervation of the SV is also disrupted upon excision and further disrupted during conventional surgical preparation [40].

While quantitative assessments of the mechanical loads placed on the SV during various harvest procedures are not available, one can imagine that open harvest of the SV under direct vision might potentially allow for gentler handling of the vessel than permitted with endoscopic harvest. Following harvest of the SV, the conventional technique is to remove perivascular tissue from the adventitial surface. In contrast, a “no-touch” technique, first reported in 1996, consists of harvesting the vein along with the surrounding tissue and to not remove the perivascular tissue prior to grafting (Fig. 2) [48]. Use of the no-touch technique could potentially alter the mechanical environment experienced by the SV in three ways. First, reducing the surgical manipulation of the SV might reduce forces applied to the SV and thereby might reduce initial injury to the vessel. Second, relative to those isolated and prepared using conventional technique, SV isolated using no-touch technique are less likely to spasm (i.e., exhibit excessive contraction). In conventional techniques, SV are distended with saline to overcome this contraction and also to identify any leaks. One of the original papers describing this technique recommended inflation to 200–300 mmHg of pressure [49]. Consistent with this recommendation, one study of 200 SV prepared for CABG reported inflation pressures 342±37 mmHg [32]. Since the no-touch technique reduces vasospasm, manual distention with pressure is less necessary, thereby reducing the exposure of the SV to supraphysiological pressures during preparation. Third, the perivascular tissue might provide a mechanical support to the SV once grafted into the arterial circulation, thereby reducing the distension of the SV due to exposure to arterial blood pressure (similar to the extravascular supports covered in Sec. 6.1).

A vessel is exposed to mechanical loads resulting from hemodynamic forces initially acting on the intima and tethering forces acting on the adventitia (Fig. 3). Once applied to the vessels, these forces can be transmitted through the thickness and along the length of the vessel. When excised from the venous circulation and installed in the arterial circulation, SV are exposed to dramatically different mechanical conditions. Some of the resulting stresses increase dramatically while other stresses potentially decrease.

When blood flows through the lumen of a vessel, velocity gradients result in a fluid shear stress, which exerts a tangential force on the endothelium or intima. Though more complex equations or numerical simulations can capture the contributions of pulsatile flow, non-Newtonian rheological properties of blood, vessel wall motion, and complex patient-specific geometries, to a first approximation, the flow-induced shear stress on the endothelium (τ_wall) is τ_wall = 4μQ/πr³. Given that τ_wall is a function of the local blood flow rate (Q) and vessel radius (r), both of which vary dramatically between vessels, it is noteworthy that over much of the arterial circulation (i.e., from aorta to arterioles), the average wall shear stress is relatively constant (∼10 dyn/cm²) and throughout much of the venous circulation (i.e., venules to vena cava), the wall shear stress is about an order of magnitude less [50].

Based on flow and diameter measurements taken on 252 SV grafts conducted on 197 patients, Isobe et al. calculated that the average flow-induced wall shear stress ranged from a minimum of 2.1 ± 0.3 dyn/cm² for SV grafted to the posterolateral branch to a maximum of 3.6 ± 0.6 dyn/cm² for ones grafted to the LAD artery [51]. Isobe et al. noted that due to the smaller diameter of the left IMA, when they were grafted to the left anterior descending artery, τ_wall was 13.8 ± 1.1 dyn/cm², almost four times that for SV grafted to the same coronary arteries [51]. Thus, the SV may experience a modest increase in average wall shear stress when grafted into the arterial circulation, but since it has a larger diameter than the host artery, it will not experience typical arterial levels of shear stress. The beating of the heart generates pulsatile blood flow. As blood moves away from the heart through the elastic to the muscular arteries, the extent of the pulsatility is diminished and essentially eliminated as it passes through the microcirculation. Saphenous veins grafts are connected directly to the aorta so they are exposed to large cyclic changes in the magnitude, and in some regions direction of blood flow. Regions of disturbed flow may occur in the coronary artery near the anastomosis with the SV, due to the abrupt changes in luminal areas and the angle of the union between the SV and the coronary artery [52], a topic discussed in more detail in Sec. 6.2.

Smooth muscle cells within the vessel wall can experience flow-induced shear stress as the result of two distinct processes. Disruption of the intima, which can occur during surgical preparation of SV or after vascular injury in vivo, can potentially expose underlying SMC to flow-induced shear stress similar to that experienced by the intact endothelium [53]. Alternatively, transmural flow through the vein wall results in flow-induced shear stress on SMC across the thickness of the vessel wall. The volumetric flow rate of transmural flow is only a tiny fraction of the blood carried by the vessel, but due to very small pores through the vessel wall, the average magnitude of the shear stress due to transmural flow for arteries is predicted to be in the order of 1 dyn/cm² and potentially larger near fenestrations due to the channeling of flow [54–56]. Thus, at least in some vessels, flow-induced shear stress experienced by SMC due to transmural flow can be of the same general magnitude as that experienced by the endothelium due to blood flow. When SV are moved from the venous to arterial circulation, the transmural pressure and hence the driving force for transmural flow, increases several fold. In parallel, disruption of the endothelium could reduce the resistance to transmural flow. In rabbit and rat aorta, denuding the endothelium roughly doubles transmural flow [57,58] and should have a proportional effect on flow induced shear stress. In human vessels, where the wall thickness is much greater, one might anticipate that the presence of the endothelium might have a smaller impact on transmural flow, but the combination of an increased driving force and potentially reduced resistance would be expected to result in a large increase of flow-induced shear stress on SMCs in SV grafted into the arterial circulation.

In resting humans in the supine posture, the pressure within the long SV is ∼5–10 mmHg [43,59]. Pressures are larger in standing individuals (∼75 mmHg) or individuals exercising on a stationary bike (∼35 mmHg) [43]. Thus, human SV routinely experience pressures larger than “textbook values” of venous pressure and fluctuations in pressure greater than 50 mmHg. Grafting into the arterial circulation subjects the SV to pressures significantly greater than its native environment, roughly doubling the mean pressure. In addition, while the magnitude of the variations in pressure may be similar between the native and grafted environment, the frequency is much greater in the arterial circulation. Especially, in the early postoperative period, hypotension (e.g., systolic pressure <80 mm Hg or diastolic pressure <40 mm Hg) can occur. While not thoroughly studied, one can imagine this hypotension resulting in transient stasis or low flow.

Veins exposed to arterial pressure exhibit increases in circumferential stress (σ_θ). For example, if one considers a simplified case and applies a thin-wall approximation, the law of Laplace (σ_θ = Pr/t where P is transmural pressure, r is inner radius, and t is wall thickness) reveals that increasing P from venous to arterial pressure directly increases σ_θ. The increased pressure also distends the vessel, thereby increasing its inner radius (r) and decreasing its wall thickness (t). Both of these geometric changes contribute to an increase in σ_θ beyond that due to the increase in pressure. As an upper bound for the increase in circumferential stress, one can consider a vein as its pressure changes from near the minimum of the wide range of pressures it would experience in its native position to a pressure typical of the arterial circulation (∼10 and ∼100 mmHg, respectively). For excised porcine SV subjected to an increase in pressure from 10 to 100 mmHg, the calculated σ_θ increased approximately 15-fold [60,42]. Thus, pressure-induced geometric changes resulted in a 50% greater increase in σ_θ than would a ten-fold increase in pressure acting on a rigid tube. As discussed in detail in Sec. 4.2.4, similar to those from pigs, human saphenous veins also exhibit only modest (<10%) changes in diameter when pressurized from 10 to 100 mmHg. Thus, to a first approximation, the fold increase in σ_θ for a human SV used for CABG will follow the fold change in pressure and pressure-induced changes in geometry will have a significantly smaller effect. Since human saphenous veins routinely experience pressures of ∼75 mmHg in their native environment, as a first approximation, one could estimate that their maximum circumferential strains roughly double after grafting into the arterial circulation.

In addition to hemodynamic forces, arteries and veins are exposed to significant forces exerted by the surrounding tissue on the adventitia. For example, motion of surrounding tissue can cause vessels to bend or flex, which has been shown in ex vivo models to alter vessel function [61]. In the heart, the compliance of coronary arteries is affected by the passive myocardium and myocardial contraction, and tethering prevents collapse of large blood vessels under compression [62]. In addition, tethering creates axial tension that is transmitted along the length of the vessels and results in significant axial strains and stresses. Axial strains are often measured by the axial stretch ratio, λ, which is the current axial length of the vessel/the axial length of the vessel under no axial load. Arteries in vivo have stretch ratios typically ranging from 40% to 65% [63]. Though this ratio is relatively constant in many situations, it can vary depending on the anatomical position of the artery [64] and decreases significantly with age [65,66] and in arteries that have undergone significant remodeling (e.g., as occurs in collateral formation [67]). The stresses generated by the axial stretch can be considerable, in some cases, on the same order of magnitude as that generated in the circumferential direction by blood pressure. While no studies were found quantifying the axial stretch ratio or axial stress in human saphenous veins prior to harvest for CABG, the clinical observation that there is little to no shortening of the SV after harvest suggests a correspondingly small axial stretch in its native position.

For CABG, the lengths of grafts are chosen such that they reach from the two suture points with little to no externally applied load (e.g., λ∼1.0). Since the heart is compressed when on cardiopulmonary bypass (i.e., when the graft length is chosen) and its volume increases at the termination of cardiopulmonary bypass and potentially later if cardiac failure or distension occurs, “most surgeons therefore tend to err on the side of long vein grafts” [68]. There are several surgical techniques designed to take up excessive slack and prevent the grafts from kinking, but placing tension on the grafts, which would increase the stretch ratio, is avoided [68]. Following grafting, the SV is subjected to cyclic axial loading with the normal beating of the heart as well as the potential for additional axial loading if cardiac hypertrophy occurs.

Blood vessels in the body have differences in their ability to distend with increasing transmural pressure. Several closely related indices are used to quantify this property including compliance (absolute or percent change in volume per change in pressure) or diametrical compliance (absolute or percent change in diameter per pressure). For both arteries and veins, compliance is a function of pressure with compliance decreasing with pressure. As summarized by Ghista and Kabinejadian [7], the compliance of a vein is typically much greater than that of an artery at low pressures, but at high (arterial) pressures, the compliance of a vein decreases to that of an artery. Tai et al. measured the compliance of various conduits used for vascular reconstruction in an ex vivo pulsatile perfusion system. For excised human SV, at low mean pressure (30 mmHg) the compliance was more than an order of magnitude greater than its value at a high mean pressure (100 mmHg; 21.0 versus 1.5%/100 mmHg, respectively, Fig. 4) [44].

Comparing compliances for conduits at typical arterial pressures (mean pressure of 100 mmHg) reveals that the compliance for human iliac artery (2.6 ± 0.8%/100 mmHg) tended to be greater than that for saphenous vein (1.5 ± 0.4%/100 mmHg), Dacron (1.9 ± 1.2%/100 mmHg) or polytetrafluoroethylene (PTFE) (0.9 ± 0.1%/100 mmHg) [44]. In contrast, a compliant polyurethane graft was more compliant (8.1±0.7%/100 mmHg) [44]. In addition, Walden et al. [69] found greater diametrical compliance in human vessels (6.0 and 4.6% per 100 mmHg, for human femoral arteries and saphenous veins, respectively) than Dacron or PTFE (2.0 and 1.5% per 100 mmHg) grafts. Thus, when a saphenous vein is grafted to a coronary artery, the compliance will decrease dramatically within the grafted vein due to the increase in local blood pressure and there will likely be a compliance mismatch at the anastomosis.

Vessel compliance can alter the mechanical environment in at least three ways. First, as already noted, grafting a saphenous vein into the arterial circulation greatly reduces its compliance (thereby increasing its stiffness). Second, the difference in compliance between the graft and artery can lead to stress concentrations at the interface between the graft and artery. Using a computational model, Ballyk et al. [70] found that noncompliant Dacron grafts led to significant increases in stress distributions surrounding suture sites, especially in end-to-side anastomoses (Fig. 5) Third, wall compliance can alter the fluid flow profile and thereby alter flow-induced wall shear stress. Also, using a computational model, Steinman et al. [71] found that the flow-induced wall shear stress profiles for rigid and distensible grafts were similar though the magnitude of wall shear stress can vary more significantly at specific locations such as the toe, heel, and bed (Fig. 6).

In addition to the above studies that consider the possibility that compliance is an independent factor contributing to graft patency, Davies et al. demonstrated that vein compliance measured noninvasively prior to harvest correlated with the extent of hyperplasia assessed after the vein was harvested. Compliance of the human SV with moderate to severe hyperplasia was roughly half of that for the SV with no to mild hyperplasia [72]. Thus, preoperative measurement of compliance could potentially be used to identify prior to harvest the relatively small subset of human SV (13% in this study) with marked pre-existing IH [72]. Since pre-existing IH might suggest a propensity to future graft stenosis, vein compliance might be a useful tool for assessing the suitability of SV for grafting and thereby influence the decision to perform CABG with SV versus other options.

Typically, endoscopic harvest of SV is facilitated by insufflation of carbon dioxide to distend the tissues, thereby assisting in the dissection. While no data on how the local influx of carbon dioxide alters the dissolved gas composition and pH in and around the SV was found, these changes might be dramatic given that often the anesthesia team must compensate for local carbon dioxide insufflation by adjusting the ventilator to limit the systemic build-up of CO₂ in the patient.

After implantation into the arterial circulation, the SV may experience a physiochemical environment different than its native environment. While several of the parameters routinely measured in a typical blood gas analysis including pH, pCO₂, and bicarbonate level are statistically different between arterial and venous blood, the magnitude of these differences is small, on the order of 10% (Table 1). In contrast, the pO₂ in arterial blood patients without respiratory ailments is more than double that of their venous blood (81 versus 33 mmHg, Table 1). Since patients undergoing CABG surgery inspire pure O₂ (i.e., FiO₂ = 100%), their arterial pO₂ (362±142 mmHg) is more than triple their normal values. In addition, patients often experience systemic biochemical changes following CABG surgery. For example, stress hyperglycemia—common with surgery and often aggressively managed early on—can result in huge fluctuations in blood glucose levels that can last for weeks/months postoperative. In addition, acute protein calorie malnutrition and the associated hypoproteinemia might also significantly impact protein binding and delivery to local tissue beds.

Given the large number of changes in the environment experienced by the SV as it is harvested from the venous circulation and then following implantation into the arterial circulation, an important question is, “which changes are major stimuli leading to vein graft remodeling and disease?” In this section, we attempt to address this question while recognizing that vein graft remodeling is a complex process and different aspects of vein graft remodel may be triggered by different environmental signals or that a single form of remodeling may have multiple signals. Below, we consider the potential impacts of various changes in an SV's environment roughly in the chronological order that they occur during CABG.

Relative to traditional open isolation, endoscopic isolation reduces the morbidity associated with vein harvest and several donor site complications [73]. Prevent IV was a large multicenter trial designed to test the effect of edifoligide, a short double-stranded oligonucleotide decoy, on SV graft patency. In this study, endoscopic harvest was an independent predictor of graft failure, death, myocardial infarct, and repeat revascularization [15], though it should be noted that patients were not randomized with respect to vein harvest protocol. The reason for the apparent inferior performance of SV harvested endoscopically is not known, but increased potential for “traction injury” during endoscopic harvest has been suggested [74]. Alternatively, since endoscopic isolation is better suited for isolating the portion of the SV above the knee (which has a larger diameter than the portion below the knee), this approach may bias toward the use of SV with larger diameters, which have poorer patency [75].

Several studies report higher patency for SV grafts harvested by a no-touch technique compared to those harvested with traditional techniques [47,76]. In their best-evidence review on the topic, Sepehripour et al. concluded that there is “angiographical evidence of superior graft patency when the no-touch SV harvesting technique is employed.” They noted, however, that supporting evidence comes from “small-scale studies conducted mainly in a single center” and that “it would be reassuring to see similar results replicated elsewhere” [77].

The reasons for the apparently higher patency for SV harvested using no-touch techniques are not obvious. Dashwood et al. have reviewed how traditional surgical preparation of the SV disrupts its vasa vasorum and suggested that “damage to the (human saphenous) vein's vasa may result in wall graft hypoxia and subsequent neointima formation” [40]. The most direct evidence for the link between disruption of the vasa vasorum and IH is the observation that occluding the advential vasa vasorum in nongrafted arteries results in intimal hyperplasia [78,79]. While disruption of the vasa vasorum of an artery decreases the pO₂ in the arterial wall, it is unclear whether human SV used for CABG experience a decrease in pO₂ due to the disruption of their vasa vasorum, an increase in pO₂ due to exposure to arterial blood on their luminal surface, or whether these two effects largely cancel each other out or create large spatial gradients. Removal of the perivascular tissue during SV harvest preparation may have effects independent of changes in pO₂. The richly vascularized adventitia contains endothelial cells that could release soluble factors such as nitric oxide (NO) that regulate vein function and potentially vein graft disease. Consistent with this view, seeding donor endothelial cells within collagen foam placed around the adventitia of an artery reduces balloon injury-induced IH [80] and the induction of highly vascularized neo-adventitia reduces IH in pig SV used as interpositional carotid artery grafts [81,82]. The potential contribution of the sympathetic innervation on SV physiology and the impact that disrupting innervation during SV harvest and preparation may have on vein graft disease has been reviewed [83]. While the structural differences between SV prepared by the conventional and no-touch techniques have been relatively well characterized, it appears that the biomechanical properties of the vessels have not been assessed. In light of the proposed use of exogenous extravascular supports to potentially improve the patency of SV grafts (Sec. 6.2), it would be interesting to explore how removal of the adventia during conventional preparation alters the pressure–diameter relationship of the veins.

An additional factor that might contribute to the improved patency of SV harvested using a no-touch technique is that it avoids manual pressurization of the human SV, which can damage or alter cellular viability and function and appears to permanently increase the vessel diameter—all factors which could potentially reduce patency.

A number of studies show that pressurization of human SV can denude the endothelium [84–88]. A study by Viaro et al. illustrates that even briefly distending vessels ex vivo with relatively modest pressures can impact the endothelium. Pressurizing human SV at 300 mmHg for as short as 15 s resulted in nearly complete denudation of the endothelium as detected by immunostaining for eNOS [85]. Exposure to 200 mmHg for 15 s decreased endothelial coverage to approximately half of that in unpressurized controls, though these differences did not reach statistical significance [85]. In contrast, SV pressurized to 100 mmHg had endothelial coverage very similar to unpressurized vessels. Even if not physically removed, lower distension pressures can impact endothelial morphology [85]. Consistent with the report by Viaro et al., Chester and coworkers observed with electron microscopy denudation of regions of the endothelium in human SV distended with 300 mmHg but not in those distended with 100 mmHg [86]. In veins distended with 100 mmHg, however, there was lifting and rounding of individual endothelial cells [86]. Consistent with disruption of the endothelium, distension of human SV with pressures greater than the patients' blood pressure decreased prostacyclin and NO release [89]. Given the role of the endothelium in regulating vasoactivity, maintaining a nonthrombogenic intima, and regulating SMC proliferation, its denudation or damage could be expected to promote thrombosis and intimal hyperplasia.

Distension of human SV with as little as 150 mmHg reduced the adenosine triphosphate content of the vessel, which likely reflects metabolic damage to SMC, and greater pressures further reduced adenosine triphosphate content [90]. Distention of human SV increased SMC proliferation by approximately ten-fold in SV subsequently cultured ex vivo [84], though whether this was due to direct action on the SMC or secondary to injury to the endothelium is not known. In a separate study, distension of human SV with 180 mm Hg resulted in a dramatic increase in propidium iodide staining of cells in the vessel walling, suggestive of increased cell death [87].

Since the endothelium and smooth muscle cells play essential roles in vasoactivity, one would anticipate that mechanically distending SV could alter vasoactive responses. Consistent with this notion, human SV distended with 100–300 mmHg of pressure exhibited increased agonist-induced and KCl-induced contractions and force generation relative to undistended veins [91]. In contrast, exposure to 600 mmHg, which the authors used to mimic “uncontrolled manual distention under operating room conditions,” inhibited contraction and force generation in response to both signals [91].

Acute distension of human SV upregulates the levels of vascular cell adhesion molecule-I (VCAM-1) and intercellular cell adhesion molecule (ICAM-1), both mediators of monocyte adhesion to the endothelium, toll-like receptors, which are expressed early in inflammatory pathways, and scavenger receptors, which are involved in the update of modified low-density lipoprotein by macrophages [92,93]. While the fold increase in upregulation varied between molecules from several fold to greater than 20 fold, for each molecule, there was a trend toward higher upregulation with higher pressure [93]. Relative to control human SV not exposed to distention, neutrophil adhesion to human SV exposed to 300 mmHg distending pressure was greater than that to SV that had not been distended [92].

Zhao et al. assessed pressure diameter relationships for SV with and without prior manual distention of approximately 300 mmHg [41]. Relative to those with prior manual distention, SV segments without manual distension had smaller outer diameters at all pressures tested (Fig. 7). The circumferential stress–strain curves for distended vessels were shifted to the left relative to those for nondistended vessels indicating that that manual distention resulted in stiffer vessels. Assuming that the observed increase in diameter with distension is permanent, it would result in decreased flow-induced shear stress and increased circumferential stresses following implantation into the arterial circulation—both factors that could potentially decrease patency as discussed in Sec. 4.3.

To our knowledge, clinical trials have not directly assessed the impact of using pressure to distend the vessel during surgical preparation. Using a porcine model, Angelini et al. investigated how distending grafts with 600 mmHg during surgical preparation impacted their acute function as interpositional carotid grafts. Relative to SV that were not distended, 2 h after grafting, distended SV retained less of their endothelium (38% versus 98%) and had increased platelet and leukocyte adhesion to exposed subendothelium [94]. One to five weeks after grafting, distended porcine SV had inferior patency relative to nondistended SV (64% versus 96%, respectively) [94]. A subsequent study by the same group studied the impact of no touch versus traditional surgical preparation using an ex vivo organ culture model [88]. Compared to SV isolated using a no-touch procedure, those exposed to surgical preparation initially exhibited reduced endothelia coverage and medial cell viability. After 2 weeks of ex vivo culture, the lumen of surgically prepared veins had re-endothelialized but exhibited greater SMC proliferation and intima thickening than cultured SV not subjected to surgical preparation.

Arteries and veins remodel (i.e., change their mass, dimensions, and or composition) in response to the local hemodynamic forces generated by blood flow and pressure as well as tethering forces applied by adjacent tissue on the vessel resulting in axial tension. This concept, proposed by Thoma more than 100 years ago to explain the remodeling of the embryonic vasculature [95], has been clearly demonstrated in the perinatal and postnatal period as well [2,96–101]. In adults, hemodynamic-induced remodeling of the vasculature can occur in physiological situations such as adaptation to exercise [102], but often is associated with pathological conditions. The seminal observations of Glagov of human atherosclerotic coronary arteries suggest that coronary arteries change their size and shape in an effort to compensate for plaque accumulation so as to maintain luminal area [103]. Though this form of remodeling is likely advantageous, other forms of mechanically induced remodeling clearly are not, or at best result in a mixture of desirable and undesirable outcomes (e.g., remodeling that occurs with pulmonary hypertension [104] and systemic hypertension [101]). Similarly, the mechanically induced remodeling of veins [105] grafted into the arterial circulation might be a mixture of desirable and undesirable outcomes. For example, the medial thickening that is commonly observed in vein grafts could be an adaptation to the higher arterial pressures. This adaptation might minimize the increase in circumferential stress despite the increase in pressure. In contrast, pathological remodeling consisting of intimal hyperplasia and inward eutrophic remodeling can reduce luminal area [106].

To explore the role of different aspects of the mechanical environment on vein graft remodeling, investigators have used various systems including (1) clinical studies where different aspects of remodeling in human SV grafts are correlated with various mechanical stresses, (2) various animal models, (3) ex vivo organ culture models, and (4) cell culture model. In general, as one progresses through this list, the ability to control and characterize the mechanical environment increases. At the same time, the model systems with better control of the mechanical environment also are less representative of a human SV used for CABG. For example, organ cultures models where human or animal veins are cultured ex vivo allow good control of aspects of the mechanical environment, but may potentially introduce confounding variables not present in vivo. In vitro studies, which often consist of a monolayer culture, are exposed to well-defined mechanical stimuli and are a convenient method to explore the mechanistic details, but translating these observations to the complex in vivo settings is challenging. Thus, there is no “best” model system, but ideally data from various types of systems would support the same general conclusions.

Clinical Studies. Within the arterial system, atherosclerosis is not uniformly distributed but instead is preferentially localized to bifurcations and curves such as the aortic arch. The site-specific nature of atherosclerosis within the arterial system has led a number of investigators to suggest that hemodynamic forces, especially flow-induced shear stress, influence the development of the disease. More specifically, since atherosclerosis tends to not occur at areas of high average shear stress (e.g., the flow divider of a bifurcation or the outer radius for a curved artery), but does localize to areas of low average shear stress (e.g., the outer side of a bifurcation or the inner radius of a curve), some investigators have suggested that exposure to high average shear stress is atheroprotective while regions exposed to low average shear stress are prone to atherosclerosis [107]. Other work suggests that considering only the average magnitude of the flow-induced shear stress is inadequate [108] and that other properties of the fluid flow such as highly variable shear stress [71,109], flow reversals [110], or stress phase angle [111] are important.

Based on their observation that flow-induced shear stress in human SV grafts was ∼25% of that for the better performing IMA, Isobe et al. suggested that flow-induced shear stress accounted for differences in patency. Consistent with this notion, Shah et al. found that larger diameter SV grafts, which would yield lower flow-induced shear stress, were associated with lower graft patency than smaller diameter SV grafts [75]. Similarly, the clinical observation that SV grafted to coronary arteries with poor distal outflow have reduced patency compared to those with greater outflow suggests that exposure to low flow, and potentially the accompanying low flow-induced shear stress, correlates with vein graft disease. Similarly, competitive flow (i.e., ability of blood to still flow through the portion of the coronary artery being bypassed and therefore reduced flow through the graft) decreases the patency of grafts though this effect is more pronounced in artery grafts than SV grafts [112,113].

Animal models. In an attempt to isolate specific aspects of the mechanical environment that stimulate specific forms of vein remodeling, Dobrin grafted femoral or jugular veins into femoral or carotid arteries of dogs and surgically manipulated the graft environment [114]. In this work, intimal hyperplasia correlated with low shear stress [114]. Similarly, in rabbit models where flow was altered using ligations, low wall shear stress was correlated with increased IH [115,116]. To study the role of flow disturbances in IH, Sunamura et al. installed canine interpositional vein grafts into the femoral artery that had diameter mismatch between graft and host artery. When a saphenous vein was used, which has a smaller diameter than the artery, intimal thickening was found distal to the distal anastomosis (i.e., downstream of the graft in the artery). When a jugular vein was used, which has a larger diameter than the artery, intimal thickening was found distal to the proximal anastomosis (e.g., in the upstream portion within the vein graft; Fig. 8). Thus, in one case, the thickening occurred in the artery and another case in the vein, but in both cases, the thickening occurred downstream of the flow expansion. Flow visualization confirmed disturbed flows downstream of the expansions [117]. In canine, saphenous veins used as bypass grafts, again intimal thickening co-localized with regions of disturbed flow [118].

Organ culture. Under the appropriate culture conditions, blood vessels cultured ex vivo remain viable for at least several weeks [119,120]. During ex vivo culture, veins can remodel in response to changes in the mechanical environment [42,60,120,121]. To expose human SV to laminar fluid flow, Porter et al. [120] pinned human SV to the inside of a silicone tube with the luminal surface facing away from the wall of the tube. Vessels cultured under static conditions experienced the greatest neointimal thickness. Exposure to venous levels of flow-induced shear stress reduced the neointimal thickness by approximately half while exposure to arterial shear stress completely blocked neointimal thickening. A limitation of the systems used by Porter et al. is that since the vein segment is pinned to the inside of tube, the vein is not exposed to transmural pressure and the associated circumferential strains. Gusic et al. cultured intact porcine SV in an ex vivo perfusion system that allowed for independent control of volumetric flow rate and pressure [42,121]. Increasing pressure while maintaining constant volumetric flow increased intimal thickening, similar to that seen by others with cultured human SV [122]. Since increasing pressure increases inner diameter of the perfused SV, which in turn decreases flow-induced shear stress, it was initially unclear if the observed intimal thickening was due to increased pressure, or secondary to a decrease shear stress. To distinguish between these possibilities, pressure was increased while volumetric flow rate was increased to maintain constant flow-induced shear stress. Pressure had no independent effect on IH. Instead, flow-induced shear stress correlated with IH even as pressures were varied from venous to arterial levels (Fig. 9(a)).

Cell culture. There is a wealth of data on the responses of cultured endothelial cells to well-defined fluid flow. Such data are summarized in a number of comprehensive reviews that document scores of endothelial responses to fluid flow [5,123]. Several of these reviews specifically consider the relative importance of the type of flow in these responses [124]. In general, these data are consistent with the notion that relative to no flow or low flow conditions, exposure to steady flow with an arterial level of flow-induced shear stress increases the production of compounds thought to be atheroprotective such as NO [125] and prostacyclin [126], while decreasing potentially atherogenic compounds such as ET-1 [127]. In the same systems, flow reversals and disturbed flow result in a further shift toward an atherogenic phenotype [111,128,129].

Summary. Human SV grafts experience lower flow-induced shear stress than better performing IMA grafts. SV grafts with greater flow due to better distal outflow or reduced competitive flow have better patency. These clinical correlations suggest that greater levels of flow-induced shear stress improve graft patency. This assertion is supported by animal and ex vivo studies that show that higher flow-induced shear stresses correlate with reduced intimal hyperplasia while regions exposed to disturbed flow exhibit greater IH. In vitro studies suggest a number of molecular mechanisms by which endothelial cells exposed to flow might reduce atherosclerosis.

No studies directly evaluating effects of flow-induced shear stress on SMC in vein grafts were found. Several studies have investigated the impact of flow-induced shear stress on SMC cultured as a monolayer and subjected to flow in a parallel plate flow chamber. Under these conditions, increased flow-induced shear stress increased the release of NO [53]. Functionally, flow decreases the proliferation of SMC in monolayer culture [130], which if also occurred in vivo could potentially limit intimal hyperplasia. A review by Shi and Tarbell contains a more complete discussion of the potential role of flow-induced shear stress on SMC in various aspects of vascular biology [131] but at this time, its role, if any, in vein graft failure is poorly understood.

Hypertension is a risk factor for many forms of cardiovascular disease. Atkinson and coworkers correlated the type of remodeling observed in SV grafts with various clinical parameters and noted that increased blood pressure was associated with increased prevalence of grafts with fibrointimal proliferation, while elevated serum cholesterol was associated with increased prevalence of atherosclerotic grafts [26]. In addition to changes in the mechanical environment, hypertension results in complex changes in the biochemical environment that very likely affect vascular structure and function independent of changes in pressure. For example, Milesi et al. [132] found that human SV that had not been used as arterial grafts from hypertensive patients had different mechanical and vasoactive properties than those from normotensive patients. These differences were not due to the pressures the SV had experienced in vivo, “since veins were certainly not exposed to high pressures during the hypertensive state” [132] (i.e., hypertension does not alter venous blood pressure). Instead, the authors conclude that the observed changes are not due to altered pressure but “phenomenon affecting the entire vascular system” [132]. Thus, it is difficult to determine if the observed correlation between hypertension and IH in SV grafts is stimulated by changes in the mechanical or biochemical environment associated with hypertension. If we limit our consideration to the mechanical effects of exposing a grafted SV to arterial pressure, the increased hydrostatic pressure could act indirectly through pressure-induced transmural flow (Sec. 4.3.1.2) or pressure-induced wall stress or strain (Sec. 4.3.1.4) or directly on the cells and tissue as discussed below.

In studies of the direct effects on hydrostatic pressure, care should be taken to eliminate or account for pressure-induced interstitial flow, stresses, strains, and changes in gas solubility. Relative to many other mechanical stimuli, the direct effects of hydrostatic pressure on cells are less well studied though several reviews covering the topic are available [133,134]. In cultured endothelial cells, exposure to constant or cyclic hydrostatic pressure alters proliferation, apoptosis, morphology, as well as the expression of genes and surface markers [135]. The large majority of studies on the effects of pressure on smooth muscle cells are from nonvascular tissues. Extrapolation from studies showing that cyclic hydrostatic pressure stimulates the proliferation of bladder SMC [136] suggests that elevated hydrostatic pressure could potentially stimulate vascular SMC hyperplasia. No studies of the direct effect of hydrostatic pressure (e.g., opposed to transmural pressure) on intact vessels were found.

Several authors have suggested that the elevated pressure (or pulsatility) in the arterial circulation subjects the vein to excessive circumferential stresses (or cyclic strains), which it does not typically experience, thereby mechanically injuring the vein and initiating a cascade of events that leads to IH [23,137,138]. This hypothesis is conceptually attractive, as it is easy to imagine how an increase in circumferential stress might mechanically damage a grafted vein. This coupled with the long standing hypothesis that atherosclerosis is a response to injury [139] provides a conceptual framework coupling a significant change in the mechanical environment to vein graft disease. Though conceptually attractive, direct evidence supporting a role circumferential stress or strain is limited. As part of their study of grafted canine veins summarized in Sec. 4.3.1.1, Dobrin et al. concluded that increased deformation of the wall in the circumferential direction best correlated with medial thickening and not IH. Similarly, porcine SV cultured ex vivo exhibited a strong correlation between mean transmural pressure and wall area (Fig. 9(b)) but not IH. Another possibility is that the increased pulsatility in the arterial system, which exposes SMCs in the vessel wall to cyclic strain, might contribute to graft failure. Though we are not aware of in vivo or ex vivo studies that assess this idea, in SV SMCs cultured on deformable substrates both ³H-thymidine and cell number were increased after application of cyclic stretch, indicating increased cell proliferation while cyclic stretch of IMA SMC cultures resulted in no change in cell proliferation [140].

Since harvested human SV exhibit minimal shortening when excised (indicating minimal axial loading in situ) and clinical practice is to install grafts without axial tension, it does not appear that grafted veins experience significant changes in average axial length upon installation as CABG. Even if there is no change in average length, the beating of the heart will cyclically load and unload a grafted saphenous vein. To our knowledge, the impact of cyclic axial loading on vessels has not been studied, but steady changes in axial loading are an established regulator of growth, remodeling, and vasoactivity. We briefly summarize several of these studies below in hopes that they might motivate (1) closer consideration of the potential role of axial loading in vein graft function, (2) studies of vein (as opposed to arterial) responses to axial loading, and (3) studies of cyclic axial loading on vessels.

Though it has long been known that the tethering forces transmitted to vessels on their adventitial surfaces can generate significant axial stresses and strains, the ability of these forces to influence vascular remodeling has only recently been appreciated. Stark implanted a tissue expander under the leg adductor muscle of a rat and slowly inflated it over a period of days [141]. Adjacent saphenous arteries and veins were permanently elongated. A second group, headed by Langille, demonstrated mechanically induced remodeling when carotid arteries of adult rabbits were subjected to a step increase in longitudinal strain [96]. These carotid arteries showed rapid elongation (a calculated ∼20% increase in unloaded length within 7 days) with an associated 15-fold and >50-fold increase in endothelial cell and smooth muscle cell proliferation, respectively, as well as increased deposition of elastin and collagen. In contrast, rabbit carotid arteries did not decrease their unloaded length when their axial stretch ratio was partially removed, but instead developed tortuosity [142].

Arteries cultured ex vivo with elevated axial load exhibit increased unloaded length, biaxial mechanical properties, cell proliferation, matrix content, total mass, and MMP levels [143–149]. Interestingly, there is interplay between axial loading and other types of mechanical loads to determine vascular response. For example, the amount of flow through the vessel modulated the amount the vessels remodeled in response to axial load [145]. In the presence of physiological levels of axial load, arteries exhibited pressure-induced remodeling but when the arteries were axially unloaded, this response was inhibited [150]. Taken together, these studies illustrate that axial load is an important regulator of axial remodeling, a topic reviewed more comprehensively by Humphrey et al. [3]. In addition to remodeling to chronic changes in axial loading, acutely altering axial loading by as little as 5% can alter basal and pharmacologically induced vasoactivity [151–153].

The impact of compliance on graft patency is unclear with some studies suggesting a role for compliance while others suggest that compliance is not a factor. One possible explanation for this uncertainty is that it is difficult to vary graft compliance without varying other properties that could impact patency. Abbott and coworkers were some of the early advocates of the hypothesis that compliance mismatch is detrimental for graft patency. This hypothesis was initially based on their observations that synthetic grafts such as Dacron and PTFE had both lower compliances and lower patencies than autologous vessel grafts [69]. Abbott [69] and others [8] recognized that there are many differences between a synthetic and native vessel besides compliance that could potentially impact graft patency. In an effort to determine if compliance mismatch alone impacts patency, Abbott et al. [154] used glutaraldehyde treatment to vary the compliance of carotid arteries used as interposition graft autologous grafts. After 12 weeks, grafts with compliances similar to host arteries had a patency rate double that of grafts with 60% less compliances [154]. While these results are consistent with their hypothesis, one cannot neglect the possibility that treatment with glutaraldehyde decreases graft patency independently of changing stiffness.

Other authors using other models found no support for the notion that compliance mismatch is an independent factor contributing to graft patency. Wu et al. noted that with time, thrombi deposited on the intima of a graft remodel and the resulting intimal thickening can appear macroscopically and microscopically similar to intimal hyperplasia. Since synthetic grafts are more thrombogenic than native vessels, it is possible that the greater intimal thickening typically seen on synthetic grafts is due to their increased thrombogenicity and not due to their decreased compliance. To test this idea, they implanted autologous grafts and externally supported Dacron grafts with a compliance ∼0.3% of a native artery in dogs with naturally occurring low thrombotic potential. All of the native and ∼90% of the synthetic grafts were patents after 6 or 12 months. Among the patent grafts, there were no difference in anastomotic neointimal thickness between compliant (natural) and noncompliant synthetic grafts, which led the authors to conclude that compliance mismatch does not cause host arterial intimal hyperplasia at the anastomotic interface [155]. Okon et al. [91] also reported evidence of compliance mismatch alone not being a significant cause of IH or decreased patency in dogs. In order to achieve compliance mismatch, a segment of iliac artery was banded with a noncompliant plastic tube, which fixed the diameter of the vessel. This method allowed for no disturbance of surface properties (e.g., thrombogenicity) as the tube was placed on the outside of the vessel.

Taken together, the above animal studies suggest that the compliance mismatch between synthetic and host artery might potentially contribute to reduced patency, but the evidence is not compelling. It should be noted that the compliance mismatch between a synthetic graft and host artery (three-fold to 300-fold) is much greater than the < one-fold difference between SV and arteries so it is possible that even if compliance mismatch is a significant cause of intimal hyperplasia in synthetic grafts, it may play little to no role in SV grafts used for CABG.

While the potential effects of compliance on vascular grafts is typically considered in terms of the difference between the graft and host artery, it might be useful to consider potential impacts from the temporal change in compliance within the saphenous vein as it moves from the lower pressure venous circulation to the higher pressure arterial circulation and the accompanying ∼ten fold decrease in compliance. Change in material stiffness, which would be inversely related to compliance, regulates the function of many cell types including endothelial [156–158] and smooth muscle cells [159,160].

As summarized in Table 1, SV experience dramatic changes in both its mechanical and chemical environments upon grafting into the arterial circulation. While many authors consider the impacts of the changes in mechanical environment, the potential impacts of the changes in the chemical environment are largely ignored. Many studies have reported that when SV are cultured ex vivo in various organ culture models, they rapidly develop IH. These vessels are typically cultured in a mechanically static environment, which suggests that that factors other than the mechanical environment present during the ex vivo culture stimulate IH. Since SV are normally cultured using typical cell culture conditions, they are subjected to an atmosphere of 5% CO₂ with a balance air. This yields a pO₂ of 140 mmHg, which is greater than the 95 mmHg typical of the arterial circulation and 40 mmHg of the venous circulation. To explore the effect of pO₂ on intimal hyperplasia, Joddar et al. cultured porcine SV at various pO₂ values for 1 week [161]. Veins cultured with venous pO₂ had intimal areas indistinguishable from that of freshly isolated SV, but veins cultured with greater pO₂ showed a dose dependent increase in intimal area and cellular proliferation rate (Fig. 10). In contrast, culturing a porcine femoral artery at arterial pO₂ for 1 week did not influence intimal area or other aspects of its histological appearance. The ability of elevated levels of pO₂ to induce intimal thickening was a robust observation, which occurred (1) in the presence or absence of serum in the culture medium [161], (2) in SV perfused under conditions to mimic a venous mechanical environment or static conditions [161], or (3) in human and porcine saphenous veins [161,162]. SV cultured at arterial pO₂ have elevated markers of oxidative stress and treating the saphenous veins with the exogenous antioxidant N-acetyl-cysteine or upregulating endogenous antioxidant enzymes blocks intimal hyperplasia in human saphenous veins cultured ex vivo at arterial pO₂ [163]. Taken together, these studies suggest that exposure to arterial levels of pO₂ is a potent stimulus of intimal hyperplasia in saphenous veins.

Based on the evidence that aspects of the mechanical environment contribute to vascular disease, a number of groups have explored different approaches to alter the mechanical environment with a goal of increasing vein graft patency. The efforts can be broadly divided into two major approaches: extravascular supports and altering the graft geometry.

External supports placed around the adventitial surface of grafted veins, sometimes called an external sheath or stent, have been proposed as a means to alter the mechanical environment, thereby improving the patency of grafted vein. These supports can be composed of synthetic polymers [82,164–167], metal [168,169], or even fibrin glue [170,171]. External supports reduce IH in various animal models including rodents [172], rabbits [167], sheep [168,173], and pigs [82] as well as reduce early vein graft injury in human SV cultured ex vivo [164,170,171].

While these studies are consistent in that they all suggest that external supports reduce IH, they vary in their explanation of why this happens. It was suggested by many authors, at least initially, that these external supports are effective because they prevent the veins from overextending, thereby decreasing the circumferential stress and cyclic strain [81,174–179]. The studies that most clearly link the ability of external supports to reduce mechanical loading and to minimize IH in grafted veins utilized rabbit or rat models employing veins other than the SV [174–177]. These veins, however, lack much of the robustness of human SV, distend significantly when exposed to arterial pressure, and therefore experience very large increases in stress after grafting. For example, Moore et al. estimated 140-fold increases in circumferential stress following grafting into the arterial circulation [174]. Perhaps not surprisingly given the very large increase in circumferential stress, ∼60% of the cells in the media die within 1 day [176]. The presence of an external support around the rat vein graft prevented cell death [176]. In contrast to the veins used in these studies, we estimated that the maximal circumferential stress for a human SV approximately doubles after grafting (Sec. 3.3.1.4) and massive cell death is not a feature of human SV remodeling following CABG (Sec. 2.4). Thus, it is difficult to speculate on how conclusions from these specific animals model might apply to human SV grafts but it would be premature to conclude the circumferential stress does not play a role. Other animal studies suggest external supports might reduce IH by reducing disturbance to flow [172].

The results of some studies using external sheaths for grafted pig SV are difficult to interpret in light of mechanical effects. Specifically, in some studies, snug-fitting external sheaths, which would be most effective at reducing circumferential stress (or reducing disturbed flows), stimulated IH instead of inhibiting it. Only loose-fitting “over-sized” macroporous sheaths, which do not restrict the expansion of the vein graft in response to arterial pressure and are therefore ineffective at reducing circumferential stress or cyclic strain, inhibited IH [81,177]. These results lead the researchers to speculate that formation of the neoadventitia and the associated vascularization that forms between the vessel and the over-sized sheaths and not the changes in the mechanical environment inhibited IH [177]. The authors of these studies proposed several mechanisms by which the neoadventitia could inhibit neointima formation, including reducing oxidative stress in the vessel wall [82].

Based on the promising results from animal studies with loose fitting vascular sheath, a small randomized clinical trial was conducted to compare patency of SV grafts with and without Dacron sheaths. Unfortunately, the sheaths performed extremely poorly—all SV with sheaths failed due to thrombosis while all grafts without sheaths remained patent [166]. The authors suggested that sheath rigidity or dimensions resulted in it kinking [166], which is relatively easy to imagine occurring for a relatively rigid tube placed on a beating heart. A recent clinical trial used a metallic expandable extravascular support designed to be flexible while resisting kinking. After 1 year, supported and control grafts had similar failure rates (30% and 28%, respectively) but supported grafts had a modest decrease in average intimal area [180]. Computational fluid dynamics based on patient specific data suggests that relative to control grafts, supported grafts have similar average flow induced wall shear stress but lower oscillating shear index and fewer regions of low wall shear [169].

There is evidence that properties of blood flow including wall shear stress and especially the presence of disturbed flow can regulate intimal thickening in grafted veins (Sec. 4.3.1.1). One factor contributing to the flow disturbances in SV grafts is their diameter mismatch with host vessel (Fig. 8). Diameter mismatch could potentially be reduced by surgeons selecting regions of the SV with smaller diameters to use for grafting. Such veins have been shown to have better patency [75]. Alternatively, as already discussed in Sec. 6.1, extravascular supports can prevent pressure from distending the vein or even constrict it and some authors attribute the ability of these supports to reduce IH to their effects on blood flow. Since SV are used as bypass (opposed to interpositional) grafts during CABG, disturbed flow also occurs downstream of the artery to vein interface. Altering the angles that grafts make with the host vessels or changing other geometric properties can reduce the flow disturbance. A number of variations on this theme have been evaluated. For details of this promising approach, readers are referred to comprehensive reviews [7,181].

Despite growing enthusiasm for total arterial revascularization strategies in CABG, the SV remains a cornerstone for surgical coronary revascularization. In the search for improving short- and long-term patencies, investigators continue to explore factors that impact SV graft function. It is clear that aspects of the mechanical environment are among the important factors but there are major gaps in the knowledge. In some cases, the impacts of a mechanical factor can be dramatic and well established such as the structural and functional damage to SV due to pressure distension (Sec. 4.2). Yet even in this case, to our knowledge, there is no clinical study on the impact of unregulated manual pressure distension on patency and the practice is still widely used. In other cases, the potential impacts of the mechanical environment are subtle. For example, clinical data suggest that SV isolated using endoscopic techniques have inferior patency than those isolated using open isolation (Sec. 4.1.1). Maybe this is because of subtle mechanical damage to the SV as it is extracted endoscopically or maybe it is because endoscopic isolation is better suited for the portion of the SV with a larger diameter (which will have lower flow-induced shear stress after grafting). Identifying why SV harvested endoscopically have lower patency would be an important step toward improving their patency to that of traditionally harvested SV. In some other cases, mechanical factors may play an important role, but in concert with nonmechanical signals from the arterial environment. For example, SV exposed to arterial pO₂ develop robust IH (Fig. 10). The mechanical environment can modulate this pO₂-induced IH with SV cultured in a venous mechanical environment developing more IH that SV cultures with an arterial environment (Fig. 9). A major challenge is to convert our knowledge in these and other areas summarized in this review into improved clinical outcomes. Toward this end, extravascular supports and controlling graft geometry (Sec. 6) are two promising approaches.

• American Heart Association (Grant Nos. 15GRNT2579003 and 17GRNT33700288).

• National Science Foundation Directorate for Engineering (Grant No. CMMI-1334757).