The development of vascular protheses in the early 1950s and of cardiac prostheses in the early 1960s extended surgical treatment to diseases that had been previously beyond this field and set the stage for the development of the artificial assist devices that are now at the forefront of surgical research. Although room for improvements in materials and design remains, the clinical use of these devices has reduced the mortality associated with ruptured aortic aneurysm from 100 per cent to 30 per cent, and that of symptomatic aortic stenosis from 100 per cent to 5 per cent.

In general, grafts of lower porosity are stiffer, while higher porosity grafts are softer. These are important considerations when treating patients receiving heparin or when operating on friable or extensively diseased vessels. The use of the wrong graft in any particular situation can cause life-threatening complications. The two most commonly used prosthetic materials are Dacron and polytetrafluoroethylene (Teflon). Dacron grafts come in knitted and woven varieties: the former are of high porosity and are easier to sew, but leak when the vascular clamps are removed or until clot forms in the graft interstices. Woven grafts are of low porosity, and more difficult to sew but bleeding through the graft is minimized or eliminated. Teflon grafts are not porous and do not leak but tend to be less compliant and are therefore more difficult to sew than Dacron.

It was originally thought that a high porosity (knitted) graft was necessary to allow endothelial cells and their subendothelial matrix to invade and adhere to the inner graft surface, reducing thromboembolism and increasing patency. Although prosthetic vascular grafts in canine and non-human primate models may develop a neoendothelium, re-endothelialization occurs only sporadically or not at all in man. The patency of woven and knitted grafts was compared in a study in which bifurcation grafts, one limb of which was knitted and one limb woven, were implanted in 143 consecutive patients with aortoiliac atherosclerosis or aneurysms. There was no difference in patency rates between the woven and knitted limbs during observation periods of 1 month to 2 years.

Knitted grafts are designed for abdominal and peripheral vascular procedures. In general, they are easily sewn to blood vessels because of their soft, compliant nature. The compliance of the graft allows a snug fit, even when the native vessel is heavily calcified, and minimizes bleeding at anastomotic sites. However, removal of the clamps is followed by bleeding through the graft until clot forms in the interstices. Blood loss can be considerable, and use of these grafts is contraindicated in patients receiving heparin, due to the likelihood of uncontrollable haemorrhage. A modification of the standard knitted graft is the double velour graft (Medox) which incorporates a velour inner and outer pile to enhance the incorporation of clot in the graft interstices. Dacron grafts of less than 5 mm in diameter should not be used since they are associated with a high incidence of thrombosis. Such grafts are therefore not suitable for the treatment of obstructive disease below the knee.

Woven grafts are designed specifically for use in patients who require systemic heparinization during graft insertion. They are thus ideally suited for replacement of the thoracic aorta. Graft stiffness is not usually a problem when reconstructing the aorta after resection of fusiform or sacular aneurysms, as the resection margins tend to be thick and fibrous, and incorporate sutures and graft well. In patients with ascending thoracic aortic dissection, however, the aortic tissues are extremely thin, tenuous, and friable. Attempts to sew a stiff, non-compliant graft into these tissues are often fraught with further tearing, sometimes to the point of avulsion of a suture line. On the other hand, low porosity is a requirement since patients undergoing surgery for Type A dissections are on cardiopulmonary bypass and often develop severe pre- or intraoperative coagulopathies. A satisfactory compromise has been developed by Medox in the form of the low porosity Veri-Soft woven graft. In this case, the Dacron yarn is soft enough to allow good flexibility, but the weave is tight enough to prevent haemorrhage through the interstices. The Veri-Soft graft is effective as a replacement for the ascending aorta; however, we also toughen the aorta with glutaraldehyde to ensure the stability of the aortic tissues prior to performing the anastomosis.

All grafts, whether knitted or woven, can be preclotted, thereby eliminating the problem of haemorrhage through the graft. Although routinely preclotting of Veri-Soft woven grafts is not required before their use in the thoracic aorta, this may be necessary when a thoracic dissection is superimposed upon coagulopathic states such as chronic aspirin therapy. If only heparinized blood is available, the graft can be soaked in thawed fresh frozen plasma and then autoclaved. Although this results in some stiffening, it guarantees a water-tight graft.

The smooth, non-adherent, non-wettable surface of Teflon theoretically does not allow platelets and fibrinous material to adhere; grafts of smaller diameter can therefore be placed in peripheral vessels. An additional theoretical advantage of Teflon is that it is non-porous to blood elements, and bleeding through the graft does not occur. However, Teflon grafts are non-compliant, and bleeding through suture holes and gaps between sutures is common. Bleeding through suture holes can be minimized by using a smaller diameter needle and Gore-Tex suture material that expands when wet. While Dacron grafts are pleated, Teflon grafts are not, and they tend to kink more easily. However, some Teflon grafts are now being manufactured with external ribbing to help prevent kinking.

The superiority of Teflon over Dacron regarding long-term patency is controversial. There is no difference in the development of intimal hyperplasia at anastomotic sites or in the patency of aortic bifurcation grafts made from the two materials. However, there appears to be a somewhat higher patency rate for Teflon grafts when used in the axillofemoral position or in the femoral-popliteal-tibial position. Neither Teflon or Dacron is comparable in patency to reversed saphenous vein.

New graft developments include the intraluminal graft, the composite graft, human umbilical vein grafts, endothelial cell seeded grafts, and collagen-impregnated grafts.

The intraluminal graft was designed for use in the thoracic aorta and to aid in keeping aortic cross-clamp times to a minimum. The ends of the graft are firm. The prosthesis is inserted into the aorta and occlusively ligated in place with ties placed around the aorta at the proximal and distal ends of the graft. It is always more difficult than it first appears to insert the graft into the aorta, and a graft substantially smaller in diameter than the aorta should be used. Severe atherosclerosis or tree barking associated with luetic aortitis makes it difficult to obtain a tight fit between the graft ends and the aorta, and leaking often occurs through the areas of ligation. In patients with thoracic dissection,tying down the ties can tear the aorta.

Composite grafts are woven Dacron grafts that are sewn to either a porcine or mechanical valve. These are used for operations on the right ventricular outflow tract in children or to the ascending aorta in adults. In children it is common to use a porcine valve conduit to minimize thromboembolism, although the incidence of degradation of these valves in children is high. In adults, a composite graft using a mechanical valve (usually a St Jude valve) is always recommended to eliminate the problem of valve degradation and the associated difficulty of re-replacement. The St Jude valve conduit must be autoclaved before use.

Glutaraldehyde-stabilized human umbilical vein has been used as a conduit for grafting small, peripheral vessels. In a study of 218 patients undergoing lower limb revascularization, the 3-year patency rate of human umbilical vein was comparable to that of polytetrafluoroethylene and considerably lower than that of saphenous vein. Human umbilical vein has, therefore been not widely used as a bypass conduit for small vessels.

There is a considerable body of literature describing techniques for seeding Dacron or polytetrafluoroethylene grafts in vitro with endothelial cells. Clinical trials have not been performed, but results in animals indicate that clot formation is inhibited in such endothelialized prostheses. At the present time, there are many problems that make this approach impractical including the time required to endothelialize the graft and rejection problems which occur if the patient's own endothelial cells are not used. Nevertheless, this approach may represent an advance in prosthetic graft development for future use in small vessel reconstruction.

Impregnation of Dacron grafts with collagen eliminates bleeding through interstices and the need for preclotting; they are only weakly immunogenic.

The decision to use any particular heart valve is often based on personal clinical experience. Although mechanical valves offer satisfactory haemodynamic performance, the associated risks of thromboembolism have promoted the development of biological valves which, by partially decreasing the non-biological–biological interface, are less thrombogenic. Several clinical studies have shown porcine valves are less thrombogenic than are mechanical valves. However, biological valves themselves are also associated with problems, the most ominous of which is tissue failure. The risk of failure in the biological valves must therefore be weighed against the risks of malfunction, thromboembolism, and anticoagulation-related haemorrhage associated with mechanical valves. Porcine valves are obtained from freshly slaughtered pigs and are treated so that the final product is a non-living valve, mounted on a Dacron stent. Upon removal from the pig, the endothelial cells quickly die and are easily removed by washing. The immunogenic subendothelial elastin layer is removed by treating the valve with the enzyme ficin. The sole remaining layer is collagen, which has sufficient species homogeneity to prevent rejection. Since native collagen lacks the strength to withstand arterial pressures over millions of cycles it is strengthened by cross-linking with glutaraldehyde. The valve is then sewn to the stent and stored in 0.065 per cent glutaraldehyde solution.

The Starr Edwards mechanical valve has undergone many modifications since its introduction in 1961. Its current configuration consists of a silastic ball, stainless steel struts, and a Dacron sewing ring. The Starr Edwards valve has the longest history of continuous use of any valve, and it remains a mainstay of cardiac surgery 30 years after its initial introduction. However, an inherent problem with ball and cage design valves is the high profile of the cage and tertiary orifice obstruction by the ball.

The Starr Edwards valve has three orifices (Fig. 1) 1658: the primary orifice is the orifice through which the blood must go when leaving the left ventricle. The secondary orifice is the orifice subtended by the angle made from the ball in the open position and the primary orifice. The height of the cage determines the area of the secondary orifice. In a properly designed valve, the areas of the primary and secondary orifices will be equal. The size of the tertiary orifice, between the perimeter of the ball in the open position and the walls of the aorta, cannot be controlled by valve design, and this can cause obstruction to blood flow. Insertion of a valve with a large primary orifice will produce a small tertiary orifice due to the size of the ball; conversely, a large tertiary orifice will be associated with a small primary orifice. Because of the dual problems of tertiary orifice obstruction and high cage profile, the Starr Edwards valve should not be used in the aortic position in patients with narrow aortic roots, or in the mitral position in patients with small left ventricular chambers.

To obviate the problems associated with ball and cage design valves, the so-called low profile valves were developed. The most commonly used of these are the St Jude bileaflet valve, made from pyrolight carbon with a Dacron sewing ring, and the Bjork-Shiley monoleaflet valve, which has a pyrolight carbon disc, stainless steel trapping ring, and Dacron sewing ring. There is no statistically significant difference in long-term clinical outcome of patients treated with each of the three types of mechanical valves.

A longitudinal analysis of biological and mechanical valves inserted at the Yale-New Haven Hospital over an 11-year period was unable to demonstrate a clear advantage to the generalized use of either of mechanical or biological valves. However, specific valves are best suited for specific indications.

Mechanical valves rarely develop mechanical failure. If clinical problems develop a year or so after implanting a mechanical valve, the most common types of malfunction are tissue ingrowth, causing primary orifice obstruction, leaflet jamming, or ball jamming. Such incidents occurred in 10 of 510 mechanical valves followed for 10 years. Over the same period 606 biological valves were inserted: 57 developed torn leaflets with insufficiency, and 11 developed calcified leaflets with stenosis. Therefore, the possibility of valve failure must be an important consideration when deciding upon the use of a tissue valve. Tissue valves fail more rapidly in younger patients than they do in older children: more than 90 per cent of such valves inserted in children fail within 5 years. The failure rate decreases from 7 per cent/year in patients aged 18 to 30 to 2 per cent/year in patients aged 70 to 80. In the actuarial analysis presented in Fig. 2 1659, patients receiving biological valves were arbitrarily divided into groups above and below 50 years of age: there was an obvious effect of age on valve failure rate. However, division into groups above and below 60 years age also showed significant differences in failure rate. Freedom from failure can, therefore, be shown to decrease in a linear fashion with increasing age, but there may be no clearcut age above which freedom from failure becomes negligible. Nevertheless, biological valves may provide a relatively safe management option in patients in whom anticoagulation may be difficult or hazardous and who frequently have other medical disorders that may require surgical intervention.

Although valve failure is not an important consideration for mechanical valves, thromboembolism is. Figure 3 1660 shows the actuarial analysis of all valves in which thromboembolism, including thrombosed valves, was documented. Although the incidence of thromboembolism was significantly lower in biological valves, its sequelae were just as serious as in the mechanical group. For example, there were four fatal cerebral vascular accidents in 13 patients with biological valves and six in 37 patients with mechanical valves. All thrombosed valves (one biological and seven mechanical), resulted in death of the patient.

There is no difference between outcome in patients receiving biological and mechanical valves when followed over a 10-year period (Fig. 4) 1661. When total morbidity and valve related mortality are considered, biological valves are associated with fewer problems over the first 5 years (p < 0.01), while mechanical valves causes fewer problems over the last 5 to 6 years of the study (p < 0.001). However, over the entire study, there was no significant difference between the groups.

Improvements in mechanical valves require the development and use of less thrombogenic materials. The tissue failure problem with biological valves may be eased somewhat by the introduction of low pressure-fixation valves. Fixation under low pressure has the advantage of helping to preserve the tertiary conformation of collagen molecules. Thus closure energy is partially dissipated through disruption of non-covalent bonding interactions, which are able to reform during opening rather than through disruption of covalent bonds that can never reform because of the non-living nature of the tissue.

Low pressure fixation valves are FDA approved and are available for the mitral position but, at time of writing, not for the aortic position. The laxity of the tissue allows impingement upon the aortic struts during systole and holes are worn in the leaflets.

Cantelmo NL, Quist WC, Lo Gerfo FW. Quantitative analysis of anastomotic intimal hyperplasia in paired Dacron and PTFE grafts. J Cardiovac Surg 1989; 30: 910–15.

Carpentier A, et al. Continuing improvements in valvular bioprostheses. J Thoracic Cardiovasc Surg 1982; 83: 27–42.

Geha AS, Stansel HC, Cornhill JF, Kilman JW, Buckley MJ, Roberts WC. Late failure of porcine valve heterografts in children. J Thoracic Cardiovasc Surg 1979; 78: 351–64.

Gossetti B, et al. The salvage of the lower limbs: choice of the prosthetic material in femoro-distal and axillo-femoral bypass. Ital J Surg Sci 1988; 18: 69–73.

Hammond GL, Geha AS, Kopf GS, Hashim SW. Biological versus mechanical valves. J Thoracic Cardiovasc Surg 1987; 93: 182–98.

Herring MB, Dilley R, Jersild RA, Jr., Boxer L, Gardner A, Glover J. Seeding arterial prostheses with vascular endothelium. Ann Surg 1979; 190: 84–90.

Karner J, Schemper M, Teleky B, Kretschmer G, Piza F, Polterauer P. Aorto-Y-bifurcation graft: Dacron versus PTFE. Preliminary results of a randomized prospective study. Int Surg 1988; 73: 218–20.

Kopf GS, Geha AS, Hellenbrand WE, Kleinman CS. Fate of left-sided cardiac bioprosthetic valves in children. Arch Surg 1986; 121: 488–90.

Magilligan DJ, Lewis JW, Heinzerling RH, Smith D. Fate of a second porcine bioprosthetic valve. J Thoracic Cardiovasc Surg 1983; 85: 362–70.

Miyamura H, Kanazawa H, Hayashi J, Egushi S. Thrombosed St. Jude medical valve prosthesis in the right side of the heart in patients with tetralogy of Fallot. J Thoracic Cardiovasc Surg 1987; 94: 148–50.

Robicsek F, et al. Indium 111-labelled platelet deposition in woven and knitted Dacron bifurcated aortic grafts with the same patient as a clinical model. J Vasc Surg 1987; 5: 833–7.

Sottiurai VS, Yao JST, Flinn WR, Batson RC. Intimal hyperplasia and neointima: An ultrastructural analysis of thrombosed grafts in humans. Surgery 1983; 93: 809–17.

Stinson EB, Griepp RB, Bieber CP, Shumway NE. Aortic valve allografts for mitral valve replacement. Surgery 1975; 77: 861–7.

Stinson EB, Griepp RB, Oyer PE, Shumway NE. Long-term experience with porcine aortic valve xenografts. J Thoracic Cardiovasc Surg 1977; 73: 54–63.

Szuman JE, Mensah OW, Lorkiewicz Z. Use of the intraluminal vascular prosthesis for the repair of anuerysm of the ascending aorta. Thoracic Cardiovasc Surg 1986; 34: 362–4.

The Canadian Multicenter Hemashield Study Group. Immunologic response to collagen-impregnated vascular grafts: A randomized prospective study. J Vasc Surg 1990; 12: 741–6.

Vasseur B, Hammond GL. New technique for repair of ascending thoracic aortic dissections. Ann Thoracic Surg 1989; 47: 318–9.

Weinberg CB, Bell E. A blood vessel model constructed from collagen and cultured vascular cells. Science 1986; 231:397–400.

Weisel RD, Johnston KW, Baird RJ, Drezner AD, Oates TK, Lipton IH. Comparison of conduits for leg revascularization. Surgery 1981; 89: 8–15.

Yates SG, et al. The preclotting of porous arterial prostheses. Ann Surg 1978; 188:611–22.

Zacharias RK, Kirkman TR, Clowes AW. Mechanisms of healing in synthetic grafts. J Vasc Surg 1987; 6: 429–36.