Br. J. Surg. Vol. 66 (1979) 23-28
Arteriovenous grafts for vascular access in haemodialysis R . C. K E S T E R * SUMMARY
The outcome of 68 arteriovenous grafts placed in 46 patients requiring haemodialysis was studied over a period of 3.5 years. The biological grafts included autogenous saphenous vein, modified bovine carotid artery and human umbilical cord vein allograft, whereas the synthetic grafts comprised Sparks Dacron mandril, expanded reinforced polytetrafluoroethylene and knitted Dacron velour. These subcutaneous grafts were arranged as looped or straight configurations in the forearm or thigh. Of 59 grafts evaluated in patients with end-stage renal failure, only 48 per cent of the forearm grafts performed well, compared with 85 per cent of the thigh grafts. Although only 38 per cent of the looped grafts were successful, 78 per cent of the straight grafts functioned satisfactorily. Synthetic grafts suffered less serious complications than the commercial biological grafts.
MAINTAINING easy access to the bloodstream is both the keystone and Achilles heel of long term haemodialysis in end-stage renal failure. The most popular means of vascular access for close on two decades have been the external arteriovenous or Scribner shunt (Quinton et al., 1960) and the internal arteriovenous or Cimino-Brescia fistula (Brescia et al., 1966). Unfortunately, these primary procedures to obtain access to the circulation may fail sooner or later because of thrombosis, infection or inadequacy of the blood vessels. When all the forearm and ankle vessels have been obliterated by repeated primary procedures, then continued intermittent haemodialysis has been made feasible through the development of secondstage procedures, which include the implantation of subcutaneous arteriovenous grafts. This paper records the results obtained in one centre using a variety of grafts for continued haemodialysis in problem patients. Patients and methods During the period of study, extending from June 1974 to December 1977, 68 arteriovenous grafts were implanted subcutaneously in 46 patients. The group comprised 19 men and 27 women, aged between 15 and 64 years, with an average age of 41. Apart from 5 cases with disseminated cancer of bronchus or bladder, all patients had end-stage renal failure. The 5 exceptions were scheduled for magnesium depletion in an attempt to control their spreading cancer. The indications for establishing a subcutaneous arteriovenous graft are listed in Table Z. The number of previous first-stage vascular access procedures undertaken in each patient varied between none and 9, with a mean of 3.7. The types and number of vascular grafts used in the study are listed in Table ZZ, All grafts had an internal diameter of about 6 mm. The synthetic prostheses used were the Sparks Dacron mandril (American Hospital Supply UK Ltd) (Fig. l), expanded microporous reinforced polytetrafluoroethylene (PTFE) o r Gbre-Tex (W. L. Gore UK Ltd) (Fig. 2) and knitted Dacron velour or Vasculour-D (USCI International Ltd) (Fig. 3). The biological grafts used were reversed autogenous saphenous vein, dialdehyde-treated human umbilical cord vein allograft
Table I: INDICATIONS FOR AN ARTERIOVENOUS GRAFT Cimino-Brescia fistula impossible when forearm vessels: (a) obliterated by previous infusions, shunts or fistulas (6) indiscernible because of obesity or cachexia (c) unsuitable because of small calibre Inadequate development of veins by arteriovenous fistulas Scrjbner shunt not feasible when ultimate fistula fails Imminent kidney transplant unlikely when ultimate fistula or shunt fails. Table 11: ARTERIOVENOUS GRAFTS IMPLANTED JUNE 1974 TO DECEMBER 1977 Sparks mandril (Dacron) Expanded reinforced PTFE (Gore-Tex) Knitted Dacron velour (Vasculour-D) Autogenous saphenous vein Human umbilical cord vein allograft (HUCVAG) Bovine carotid artery Ficin-digested Gluteraldehyde-treated Antibiotic-treated Total
7 8 6 14
4 5 5
or HUCVAG (Gene-Tec Scientific Ltd) (Fig. 4), ficin-digested bovine carotid artery graft or Surgikos Artegraft (Johnson and Johnson Ltd) (Fig. 5), gluteraldehyde-treated and antibiotictreated bovine carotid artery grafts (Yorkshire Regional Tissue Bank) (Figs. 6, 7). In preparation of the two lastmentioned grafts, fresh bovine carotid arteries collected from the local abbatoir were treated in alternative ways: the first group was immersed in 0 5 per cent phosphate-buffered gluteraldehyde solution for 48 h and then stored in the solution. The second group was immersed in Medium 199 (Gibco Bio-cult Ltd) in which was dissolved gentamicin 4 mg/ml, methicillin 10 mg/ml, erythromycin 6 mg/ml, mycostatin 2500 units/ml and fetal calf serum to a concentration of 10 per cent. The arteries were stored in the solution at 4 "C for no longer than 5-6 weeks. If the vessels of either forearm were judged to be inadequate, then the graft was placed in the thigh. In the forearm and thigh, the graft was usually arranged as a straight configuration, but when poor distal forearm vessels did not permit a straight configuration, then a loop based on the cubital fossa was fashioned. It was decided to place all the commerc!al bioprostheses in the thigh. For a forearm loop, the brachial artery was bridged to the widest vein in the cubital fossa, the loop extending to the distal forearm. For a straight forearm configuration an appropriate vessel in the distal third of the forearm was connected to an antecubital vessel. In the thigh the common femoral artery was usually connected to the superficial femoral vein in the distal part of Hunter's canal. During the early phase of the study, a few thigh loops based on the femoral vessels were established, the loops extending to mid-thigh level. The vascular suture used was 5/0 or 6/0 Prolene (Ethicon Ltd). Usually an interval of 10-14 days was allowed before a graft was cannulated for haemodialysis, so that the surgical wounds could heal. Many of the patients underwent haemodialysis at their homes.
* University Department of Surgery, St James's University Hospital, Leeds.
R. C. Kester
Fig. 1. Sparks mandril: two tubes of knitted Dacron supported by an inner silicone rod of 6 mm internal diameter.
Fig. 4. Human umbilical cord vein allograft, 6 mm internal diameter.
Fig. 5. Ficin-digested bovine carotid artcry graft (Surgikos Artegraft). Fig. 2. Gore-Tex: expanded reinforced PTFE, 6 mm internal diameter.
Fig. 3. Vasculour-D :knitted Dacron velour, 6 mm internal diameter.
Fig. 6. Gluteraldehyde-treated bovine carotid artery graft, 6 mm internal diameter.
Arteriovenous grafts for vascular access
Results Patency and ,function (Table ZII) Primary failure of a graft was defined as the inability of the prosthesis to support even a single dialysis session because of thrombosis, inadequacy or sepsis. Autogenous saphenous vein grafts: Two looped and 2 straight grafts thrombosed during the first postoperative week. A solitary thigh loop and a forearm loop were patent for 3 and 13 months respectively, and supported an average of 90 dialysis sessions. The 4 straight grafts in the forearm did worse, remaining patent for about 3 months, supporting about 62 dialysis sessions. HUCVAG: Two grafts thrombosed at 1 and 2.5 months respectively, and were found to be too degenerative to salvage. Persistent sepsis of one graft prevented its use for haemodialysis, and it was excised after 5 months because of secondary haemorrhage. Two further grafts became aneurysmal after 8 and 11 months respectively and were excised. One graft remains trouble-free at 12.5 months, supporting about 102 dialysis sessions. Bovine carotid artery grafts (jkin-digested grafts, Surgikos Artegraft: The 4 grafts achieved a mean Fig. 7. Antibiotic-treated bovine carotid artery graft, 6 mm patency of 12.3 months, supporting about 129 dialysis internal diameter. sessions. One patient died of exsanguinating haemorrhage from a cannulation site after 11 months and a second graft had been patent for 5.5 months at the after nearly 4 months and they have supported some time of the patient’s death. The third graft became 30 dialysis sessions. One forearm loop graft suffered aneurysmal and then thrombosed by 11 months. The primary failure due to inadequate arm vessels. fourth graft remained patent for 16.5 months and Assessment of graft function: The success of each then had to be excised because of uncontrollable arteriovenous graft was assessed by the arbitrary criterion of whether the graft supported a minimum bleeding from cannulation sites. Gluteraldehyde-treated grafts: Two straight grafts in of 24 dialysis sessions. The grafts implanted in the 5 the forearm thrombosed at 9 weeks and 5 months cancer patients and in those patients who died during respectively. Two thigh grafts remain open after 8.5 the first 2 postoperative months were not included, leaving 59 grafts to be evaluated (Table ZV). Whereas months of function. Antibiotic-treated grafts: Two patients in the group only 48 per cent of forearm grafts were successful, 85 died in the early postoperative period. A forearm per cent of thigh grafts performed well. Straight grafts looped graft and 2 thigh grafts remain open after a enjoyed a measure of success of 78 per cent, compared minimum of 5 months, and have each supported with 38 per cent of looped grafts. about 60 dialysis sessions. Sparks Dacron mandril: Five of the 11 grafts throm- Complications (Table V) bosed at an early stage and never functioned. None Graft thrombosis: Thrombectomy was unsuccessful of the arm or thigh loop grafts remained patent in the three clotted HUCVAGs because the grafts were beyond 2 months. Two of the 3 thigh straight grafts too degenerative. Thrombectomy was successful in one bovine Artegraft, but not in a thrombosed remain trouble-free and open after 23 months. Expanded PTFE grafts (Gore-Tex): Two of the 8 fore- gluteraldehyde-treated graft placed in the forearm. arm straight grafts remain patent and functional at Only 3 of 6 attempts at thrombectomy performed in 5 8 and 9 months respectively. Three primary failures of the Sparks mandril loop grafts succeeded. None of 6 thrombectomies in 4 forearm Gore-Tex occurred within 14 days of operation because of poor arterial inflow. The 5 functioning grafts supported an grafts achieved patency, but 9 such attempts with 5 average of 78 dialysis sessions. Four of the 5 forearm thigh grafts established continued patency in all but looped grafts functioned for an average of 3 months, one graft. but only 1 remains patent. On the other hand, the 9 Deep vein thrombosis: One patient with a thigh thigh straight grafts functioned best of all, with a Gore-Tex graft and another with a thigh Sparks duration of patency varying from 4.5 to 17 months mandril graft developed ipsilateral iliofemoral venous (mean patency of 10 months). Four grafts remain thrombosis. open, while 2 were removed because of infected Venous outflow problems: Five thigh grafts required cannulation sites and a further graft was excised reduction of their blood flow by partial ligation because of brisk bleeding from cannulation sites. The because of (a) gross unremitting oedema of the leg occlusion of one graft at 4.5 months was most (1 Gore-Tex and 2 bovine Artegrafts); (6) high probably related to the extensive atheromatous disease venous pressure interfering with dialysis (1 of the femoral artery. The ninth graft was patent for HUVCAG); (c) increased venous return producing cardiac failure (1 bovine Artegraft). 6.5 months at the time of the patient’s death. Knitted Dacron velour grafts ( Vasculour-D): All 6 Arterial embolism: One patient with a femorofemoral grafts implanted in the thigh in straight configuration arterial ‘jump’ graft of Gore-Tex developed painful red have given trouble-free service. They remain patent areas on the sole of the foot. Later, because of severe
R. C. Kester
Table 111: CONFIGURATIONS AND OUTCOME O F ARTERIOVENOUS GRAFTS FOR HAEMODIALYSIS Range of Mean Mean no. Primary patency patency dialysis Graft No. Configuration failure (mth) hth) sessions Saphenous vein HUCVAG Bovine carotid Ficin-digested Gluteraldehyde-treated Antibiotic-treated Sparks mandril Gore-Tex Vasculour-D
4 4 6 leg
Loop Straight Straight
4 leg 5 2 arm 3 leg 3 arm 5 leg 3 leg 8 arm 5 arm 9 leg 1 arm 6 leg
Straight Straight Loop Straight Loop Loop Straight Straight Loop Straight Loop Straight
0 0 1 0 3 1 1 3 1 0 1 0
5.5-16.5 2.25-10 1-6.5
0.3-1 0.3-2 1-233 0-14 0.3-8.5 4.5-17 0 2.5-4.5
4 3 6.7
90 62 65
12.3 6.9 3.8 3.8 0.5
129 50 78 38 0 6 246 78 33
15.2 4.4 3 10 0 3.8
89 0 31
Table IV: SUCCESSFUL FUNCTION OF ARTERIOVENOUS GRAFTS IN PATIENTS WITH RENAL FAILURE Arm Leg LOOP Straight Loop Straight Graft Saphenous vein Bovine carotid Ficin-digested Gluteraldehyde-treated Antibiotic-treated HUCVAG Sparks mandril Gore-Tex
. . 0 -
0 0 0 . .
0000 00 00 000.0. 00. 000000 000 000000
Table V: COMPLICATIONS O F ARTERIOVENOUS GRAFTS LISTING THE NUMBERS O F GRAFTS AFFECTED Bovine carotid artery Ficin- Gluteraldehyde- AntibioticSaphenous Sparks vein HUCVAG digested treated treated mandril Gore-Tex. Vasculour-D __-
No. of grafts Thrombectomy DVT Venous outflow Embolism Arterial ‘steal’ Bleeding Haematoma Sepsis Aneurysm
8 0 0 0 0 0 0 0 0 0
6 3 0
1 0 0 1
0 1 2
4 1 0 3 0 1 2 0 0 1
bleeding from cannulation sites, excision of the graft was required. Arterial ‘steal’: One patient with a HUCVAG, with obliterated ankle vessels, subsequently experienced severe ischaemia of the foot, presumably brought on by an arterial ‘steal’ phenomenon. A below-knee amputation was performed. Bleeding: Secondary haemorrhage from a graft is a very serious problem, because 4 thigh grafts complicated in this manner (HUCVAG, bovine Artegraft, Sparks mandril and Gore-Tex) had to be excised. One further patient with a bovine Artegraft suffered a fatal exsanguinating haemorrhage. Sepsis: Sepsis along the track of the graft was ineradicable, and 1 thigh Sparks mandril graft and 2 thigh Gore-Tex grafts had to be excised.
5 I 0 0 0 0
0 0 0 0
5 0 0 0 0 0 0 0 0 0
I1 5 1 0
Aneurysm formation : Four bioprostheses (3 HUCVAG and 1 bovine Artegraft) developed multiple aneurysms and the grafts were removed thereafter.
Discussion In end-stage renal failure, when repeated first-stage procedures such as the Scribner shunt or CiniinoBrescia fistula have failed to accomplish vascular access and early kidney transplantation is unfeasible, then second-stage procedures are employed to permit continued dialysis. These may include interposed grafts between suitable artery and vein; translocation of the superficial femoral artery to a subcutaneous position (Brittinger et al., 1970); brachial or proximal radial artery-cephalic vein fistula (Cascardo et al., 1970); and femoral vein and artery catheterization by
Arteriovenous grafts for vascular access the Seldinger technique (Shaldon et al., 1961). Larger arteriovenous shunts are now available, such as the Thomas shunt or large vessel applique (Thomas, 1970) and the Allen-Brown shunt (Extracorporeal Medical Specialties Incorporated). With the rapid development of new biological and synthetic graft materials for vascular replacement, not surprisingly, the implantation of a subcutaneous arteriovenous graft has proved to be a popular method of achieving vascular access. Considering first the biological grafts, haemodialysis through autogenous saphenous vein loops in the forearm was first introduced by May and colleagues in 1969. A failure rate of 20-25 per cent in the first 3 months of implantation has been observed, while only 50 per cent of grafts remain patent by 1 year (Foran et al., 1975). Although individual grafts may function for nearly 4 years (Zerbino et al., 1974), many wide-bore vein grafts eventually fail due to intimal fibrosis and aneurysmal dilatation (Foran et al., 1975). Although the bovine carotid artery graft was originally developed for arterial bypass surgery (Rosenberg et al., 1966), Chinitz and co-workers (1972) were the first to use the xenograft as an arteriovenous conduit to facilitate haemodialysis. The graft is prepared by digesting the adventitia and elastica proteins with ficin, a protease found in fig juice. The remaining collagenous structure is not only strengthened by tanning with dialdehyde starch, but is also rendered essentially non-antigenic. A 1-year patency rate of about 70 per cent is reported (Foran et al., 1975; Rolley et al., 1976), but there is a striking prevalence of needle site infection, bleeding episodes, aneurysms and poor salvageability. Because of a low patency rate of 42 per cent, a sepsis rate of 7 per cent and an aneurysm incidence of 5 per cent, Dale and Lewis (1976) have abandoned using the xenograft for arterial reconstruction. The mean patency rate of some 12 months for the grafts in the present study is in keeping with the other series, but serious complications such as aneurysm formation, massive bleeding and venous outflow problems have made further use of the Artegraft impracticable in this centre. There has been a sustained interest in the preservation of vascular tissue using two main methods: gluteraldehyde treatment and antibiotic treatment. For instance, the Hancock ‘stabilized gluteraldehydeprocess-treated’ porcine aortic valve (Bowman et al., 1973) has been implanted in many patients with successful function for up to 6 years (Dexter, 1977), and such valves are now available commercially (Edwards Labs). Several investigators have evaluated fresh homograft aortic valves treated with antibiotic cocktails, and these valves have also enjoyed a favourable outcome (Angel1 et al., 1973). Preservation of bovine carotid arteries collected from the local slaughterhouse is based on the long term studies of Dexter at the Yorkshire Regional Tissue Bank, Wakefield. Despite its shorter shelf-life, the antibiotictreated graft is preferred to the gluteraldehyde-treated artery because it is more compliant, softer and easier to handle. Preparation of the HUCVAG is broadly similar to that of the Surgikos Artegraft, whereby, following dilatation of the central umbilical vein, the Wharton’s jelly component of the cord is strengthened by tanning either with dialdehyde starch (Mindich et al., 1977) or with gluteraldehyde (Dardik et al., 1976).
The reports by the two latter groups of the outcome of these homografts are not persuasive, as they give an early thrombosis rate of 40-74 per cent, accompanied by a salvage rate of only 50 per cent. These two factors, in addition to the high cost of the graft and the risk of aneurysm formation, have led to suspension of further implantation in this centre. I n meeting the challenge to provide vascular access in a problem patient, synthetic grafts provide creditable alternatives to the biological grafts. One example, the Sparks mandril, consisting of two layers of knitted Dacron tubing surrounding a flexible inner rod, is buried subcutaneously (Sparks, 1972). When the inner rod is withdrawn after a minimum period of 6 weeks, ingrowth of fibrous tissue has converted the mandril into a vascular prosthesis. Although initially the concept appeared attractive, interest in the mandril has now waned, because of the long maturation period and the incidence of graft closure and aneurysm (Hallin and Sweetman, 1976). The present study confirms their finding that successful long term function can be achieved, provided that the Sparks mandril is anastomosed to wide-calibre vessels. Early evaluation of expanded reinforced microporous PTFE (Gore-Tex) has shown a patency rate of 86 per cent for over 400 grafts implanted for haemodialysis (Gore Associates Inc., 1977). In the present study, only one thigh graft remained irretrievably closed, probably because of an atheromatous femoral artery. Arteriovenous fistulas and grafts are prone to thrombose if hypotension is incurred after a dialysis session. Thrombectomy is most easily carried out in a Gore-Tex graft, and it is followed by long term patency, provided the inflow and outflow of blood is adequate. Because these requisites were deficient in the arm grafts, salvage attempts here were of little avail. Dacron velour (Vasculour-D) grafts are warpknitted from texturized yarn to give loops perpendicular to the flow surface, similar to velvet, producing a fabric of intermediate porosity. These characteristics promote fibrous tissue ingrowth, a well-formed neo-intima and excellent rates of patency in arteriovenous grafts (Flores et al., 1973; Lindenauer et al., 1974). Regarding the three types of synthetic graft, although the Sparks mandril is by far the cheaper, Gore-Tex and Vasculour-D do not require a prolonged maturation period after implantation and also both have received more favourable reports. In conclusion, conspicuous factors emerge following consideration of the outcome of these 68 arteriovenous grafts for haemodialysis. The first outstanding fact is that a graft placed in straight configuration has a better outcome than a looped graft. The second feature is that a graft located in the thigh enjoys greater success than one in the arm. In general, the patients preferred the thigh location because both hands were then freed for cannulation manoeuvres. Finally, the synthetic grafts incurred less serious complications than the commercial bioprostheses currently available. Acknowledgements I am deeply thankful to Mr Frank Dexter, Director of the Yorkshire Regional Tissue Bank, Pinderfields Hospital, Wakefield, for advice and for the supply of bovine carotid grafts. I am very grateful to Mrs Alice Halliwell for typing the manuscript.
R. C. Kester
References et al. (1973) Antibiotic sterilisation of aortic homografts. NZ Med. J . 77, 31-35. BOWMAN F. o., HANCOCK w. D. and MALM J. R. (1973) A valvecontaining Dacron prosthesis. Arch. Surg. 107, 724-728. BRESCIA M. J., CIMINO J. E., APPEL K. et al. (1966) Chronic hemodialysis using venipuncture and a surgically created arteriovenous fistula. N . Engl. J . Med. 275, 1089-1092. BRllTINGER W. D., STRANCHM., HUBER W . e t d . (1970) 16 months experience with the subcutaneously fixed superficial femoral artery for chronic haemodialysis. Proc. Eur. Trans. Dial. Assoc. 7, 408-412. CASCARDO s., ACCHIARDO s., BEVEN E. G. et al. (1970) Proximal arteriovenous fistulae for haemodialysis when radial arteries are unavailable. Proc. Eur. Trans. Dial. Assoc. 7, 4246. CHINITZ J. L., YOKOYAMA T., BOWER R. et al. (1972) Self-sealing prosthesis for arteriovenous fistula in man. Trans. Am. Soc. Arrif. Intern. Organs 18, 452-455. DALE w. A. and LEWIS M. R. (1976) Further experiences with bovine arterial grafts. Surgery 80, 71 1-721. DARDIK H., I B R A H I M I. M., SPRAYREGEN s. et al. (1976) Clinical experience with modified human umbilical cord vein for arterial bypass. Surgery 79, 618-624. DEXTER F. (1977) Personal communication. FLORES L., DUNN I., FRUMKIN E. et al. (1973) Dacron arteriovenous shunts for vascular access in hemodialysis. Trans. Am. SOC.Art$. Intern. Organs 19, 33-37. FORAN R. F., SHORE E. H., LEVIN P. M. et al. (1975) Bovine heterografts for hemodialysis. West J. Med. 123, 269-274. GORE ASSOCIATES INC. (1977) SUmi7lUry Of Clinical Evaluation Data for Gore-Tex Vascular Grafts. HALLIN R. w. and SWEETMAN w. R. (1976) The Sparks mandril graft. A seven-year follow-up of mandril grafts placed by Charles H. Sparks and his associates. Am. J. Surg. 132, 221-222. ANGELL W . W., WUERFLEIN R. D., CHUN C. W .
et al. (1974) VelOLIr vascular prostheses. Trans. Am. Soc. Artif. Interu. Organs 20, 314-318. MAY J., TILLER D., JOHNSON J. et al. (1969) Saphenous vein arteriovenous fistula in regular dialysis treatment. N . Engl. J . Med. 280, 770-771. MINDICH B., SILVERMAN M., ELGUEZABEL A. et a]. (1977) Human umbilical cord vein for vascular replacement: preliminary report and observations. Surgery 81, 152-160. QUINTON w. E., DILLARD D. H. and SCRIBNER B. H. (1960) Cannulation of blood vessels for prolonged hemodialysis. Trans. Am. SOC.Art$. Intern. Organs 6, 104-1 13. ROLLEY R. T., STERIOFF s. and WILLIAMS G. M. (1976) Arteriovenous fistulas for dialysis using modified bovine arteries. Surg. Gynecol. Obstet. 142, 700-704. ROSENBERG N., MARTINEZ A., SAWYER P. N. et al. (1966) Tanned collagen arterial prosthesis of bovine carotid origin in man. Ann. Surg. 164, 247-256. SHALDON S., CHIANDUSSI L. and HlGGS B. (1961) Haemodialysis by percutaneous catheterisation of the femoral artery and vein with regional heparinisation. Lancet 2, 857-859. SPARKS c. H. (1972) Silicone mandril method of femoropopliteal artery bypass. Am. J. Surg. 124, 244-249. THOMAS G. I. (1970) Large vessel applique arteriovenous shunt for hemodialysis. A new concept. Am. J . Surg. 120, 244248. ZERBINO V. R., TlCE D. A., KATZ L. A. et a]. (1974) A 6-year clinical experience with arteriovenous fistulas and bypasses for hemodialysis. Surgery 76, 1018-1023. LINDENAUER S. M., WEBER T. R., MILLER T. A.
Paper accepted 26 June 1978.