BRIEF REPORT
Complications
of Dialysis Access: A Six-Year Study
Stephen L. Hill,
MD, FACS,
Antonio T. Donato,
ialysis access in many patients with chronic renal D failure can be tedious, difticult, and oftentimes recurrent. The complications of dialysis access are chronic and potentially life-threatening. We examined the complications of our dialysis access procedures over the past 6 years and have found several interesting characteristics. The complications of dialysis access in the form of thrombosis, infection, and anastomotic stenoses are similar to the complications found in most forms of vascular surgery. However, the dialysis population has several unique characteristics and the formation of an arterial venous fistula has different hemodynamics, both of which give rise to some significant differences in recognition and treatment. The treatment of some complications seen in dialysis access procedures can be guided by principles well established in vascular surgery. However, other complications must be approached in a unique fashion to ensure a good and timely result. In the 6-year period from July 1983 through August 1989, all patients referred for dialysis access were entered into this study. Our first choice in all patients was to perform a Cimino futula. Unfortunately, due to the urgent need for dialysis in many patients, only 5.6% of the patients were candidates for the procedure initially. The remainder of the patients received a polytetrafluoroethylene (PFTE, GoreTex) graft in the arm and underwent dialysis within 3 to 7 days after the operative procedure. All patients were followed from the time of their initial graft placement until either a problem developed with their dialysis or they died. During this period, we performed a total of 158 procedures on 77 patients for dialysis access. There were 41 males and 36 females, with an average age of 56 years. Of the total group, only 10 patients were candidates for a radial cephalic Cimino arteriovenous fistula. Two patients had poor cephalic veins and clots developed. Another two patients developed a stenosis in the vein that had to be revised. Thus, 80% of the patients with a Cimino ftstula continued with a functioning shunt throughout the study. The remaining patients and the remaining procedures were performed with or on PFTE grafts. The vast majority were upper arm grafts with a small number of looped forearm grafts. As might be expected, the primary complications were thrombosis and infection. Infections occurred in eight patients (lO%), all of
From the Department of Surgery, Community Hospital of Roanoke Valley, Roanoke, Virginia. This work was presented at the Virginia Surgical Society, The Homestead, Virginia, May 6,199O. Requests for reprints should be addressed to Stephen L. Hill, MD, 1125 South Jefferson Street, Roanoke, Virginia 24016. Manuscript submitted June 25,1990, and accepted in revised form November 15,199O.
MD, FACS, Roan&, Virginia
whom had a prosthetic graft in place. Four patients with an infected graft were referred to us, and four patients had the graft placed by us. All of the infections occurred more than 6 months after implantation and thus were not due to intraoperative contamination. These patients all presented with fever, leukocytosis, and erythema over the graft; most had positive blood culture results. Thrombosis of the graft occurred in 24 patients (32%) over the 6-year period, and a total of 57 procedures were required to maintain patency of the original shunt or to place a new one. The average length of patency for the prosthetic grafts prior to occlusion was 20 months. Thrombectomy of the graft alone did not provide satisfactory results in most cases. Thrombectomy was initially successful in almost every case, but the graft remained patent for only several hours or several days in most patients. Shunts were maintained for 3 to 15 months in 7 of 24 patients (29%). The remainder of the shunts stayed patent for an average of 8 days before becoming thrombotic again. If the venous anastomosis was revised or a new graft was placed, the new grafts remained patent for much longer periods of time (1 to 2 years). The most common cause of thrombosis was a smooth narrowing of the venous anastomosis caused by pseudointimal hyperplasia. This diagnosis was easily made by intraoperative arteriogram and, once seen, dictated that a new graft or at least a new venous anastomosis was necessary. There was one other interesting characteristic we discovered after reviewing our data. In this dialysis population, there was a small subset of patients (five) in whom many of the complications of infection and thrombosis occurred. This subset was responsible for more than 45% of the re-operative procedures (37) performed for the entire population. The data suggest that these patients are not good candidates for hemodialysis and would be better served by alternate forms of treatment such as continuous ambulatory peritoneal dialysis or transplantation. Most patients with chronic renal failure will never receive a transplant due to medical conditions, age, or absence of a cadaveric or live donor. The formation of an angioaccess, which is not only effective but durable, is of paramount importance in these individuals. Without question, the Cimino radiocephalic arteriovenous listula is the best long-term dialysis access. It is autogenous and relatively easy to construct, and over time, it has proved to be durable, efficient, and extremely reliable [I ,2]. Unfortunately, many patients need acute dialysis, and their renal function is so limited that they cannot wait for the maturation period of 4 to 6 weeks prior to dialysis, which is the length of time needed for the Cimino fistula to begin functioning. Furthermore, the cephalic vein in many patients is not adequate, is missing, or has already become thrombotic because of the number of needle sticks.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 162 SEPTEMBER 1991
265
HILL AND DONATO
Flgure 1. lntraoperative arteriogram showing smooth tapered narrowing of the verKW!3 anastomoslstypicalof pseudointlmalhyperplasia.
In those patients who are candidates for a Cimino radiocephalic tistula, approximately 10% to 15% will experience early failure due to technical factors such as poor vessels, anatomic problems, or segmental stenosis [3]. We found approximately the same results in our small series; however, this high early failure rate should not deter surgeons from attempting to construct a Cimino fistula at some point during treatment. As in other types of vascular surgery, the autogenous in situ vein represents the best solution. When a classic Cimino radiocephalic fistula was not available, we used a brachiocephalic tistula at the antecubital space, which turned out to be effective. When an autogenous fistula was not an option due to restraints of time or anatomy, we found PFI’E to be the best alternative and used it exclusively in this study. The consensus is that PFTE is the best prosthetic to use in dialysis access and has primary patency of 2 to 3 years in 50% to 60% of cases [4,5]. In contradistinction to other types of vascular surgery where inflow and outflow are the critical determinates of the patency and longevity of a graft, these parameters were rarely a problem with arteriovenous shunts. The brachial artery was almost always a good inflow vessel, and the axillary or brachial veins were low resistance beds with good runoff. There was no progression of atherosclerosis as is often the case in grafts placed for peripheral vascular disease. The primary problem with the prosthetic graft appears to be thrombosis. Again, unlike grafts placed for limb salvage or claudication, the problem was not with the runoff bed but almost exclusively with the venous anastomosis. Thrombosis due to intimal hyperplasia at the venous end of the graft was the most common complication causing the graft to fail. This has been described numerous times in the literature [6-81 and has been attributed to mechanical endothelial damage, the shearing effect of blood flow, and the high-pressure pulsatile nature of arterial flow on the venous system [9].
Studies have been done [la] showing that histologically, the stenosis is caused by intimal hyperplasia that extends along the vein for a distance of 2 to 4 cm. The prosthetic graft does not become stenotic but, rather, the runoff vein is overwhelmed by the proliferative hyperplasia. This problem with the venous endothelium has been the primary cause of the shortened patency rate found with prosthetic grafts in the renal dialysis population. Numerous solutions have been suggested concerning management of this problem [5,9,ZU];however, all of them either discuss revision of the venous anastomosis or the placement of a jump graft to extend beyond the hyperplastic endothelium of the venous bed. Because the intimal hyperplasia is so extensive, catheter thrombectomy is rarely successful in these cases. It will restore flow for a short period of time in a majority of cases but, as our experience shows, more than 80% of grafts will fail within the first several weeks after catheter thrombectomy. Consequently, we have directed our treatment technique at the venous anastomosis whenever a graft that has been in place longer than 30 days becomes thrombotic. Catheter thrombectomy is attempted, followed by angiography (Figure 1). If another suitable vein is found in close proximity, the graft is transposed onto that vein. If not, then the venous anastomosis is revised or, more likely, a jump graft is placed to a more proximal portion of the venous anatomy. Oftentimes, this will maintain the patency of the primary graft for an additional 1 to 2 years. Infection was encountered in 10% of our patient population. All patients with infection in our study presented with leukocytosis, fever, and erythema over the graft. In addition, most had positive blood culture results, suggestive of the severity of the infections. These types of infections in the renal dialysis population cannot and should not be treated with local conservative care. We have since implemented a new policy whereby all infected grafts are removed with a venous patch placed on the arterial de-
266
162
THE AMERICAN
JOURNAL
OF SURGERY
VOLUME
SEPTEMBER
1991
feet. Since the adoption of this policy, we have had minimal problems with postoperative sepsis or bleeding. As might be expected, this always necessitates at least two operations for every individual with an infected graft, one to remove the graft and one to place a new angioaccess. Thus, we have shown that in patients undergoing long-term dialysis, the major complications of vascular surgery-infection and thrombosis-are also present. Infection, as in all forms of vascular surgery, can be eradicated only when the graft in its entirety has been removed. Patients with renal failure appear to be at an increased risk of infection due to the repetitive sticks required for dialysis as well as the possible compromise in the immune status of the patient. Thrombosis, however, is not due to poor runoff or acceleration of atherosclerosis as is true of the typical vascular population. It is the hemodynamics of an arteriovenous ftftula with its high flow pressures that possibly contribute to thrombosis. Inflow is rarely a problem, and outflow only becomes a problem because of the microscopic changes occurring at the venous anastomosis. However, it is important to remember that pseudointimal hyperplasia at the venous anastomosis is the most common cause of thrombosis and, consequently, a catheter thrombectomy is only a temporary procedure. The vast majority of patients who present with a thrombotic graft will require a major revision of the ve-
nous anastomosis or placement of a new graft; suc4zess and prolonged patency by means of catheter thrombectomy are the exception rather than the rule. REFERENCES 1. Mandel S, Martin P, Blumoff R, et al. Vascular access in a university transplant and dialysis program. Arch Surg 1977; 112: 1375. 2. Crockett RE. Blood access for hemodialysis. Nephron 1974; 12: 338-54. 3. Enrenfeld W, Grausz H, Wylie E. Subcutaneous arteriovenous tistulas for hemodialysis. Am J Surg 1972; 124: 200-6. 4. Kherlakian GM, Roedersheimer LR, Arbaugh JJ, et al. Comparison of autogenous fistula versus expanded polytetrafluoroethylene graft. Am J Surg 1986; 153: 238-43. 5. Bell DD, Rosenthal JJ. Arteriovenous graft life in chronic dialysis. Arch Surg 1988; 123: 1169-71. 6. Connolly JE, Brownell DA, Levine EF, McCart PM. Complications of renal dialysis access procedures. Arch Surg 1984; 119: 1325-8. 7. Bone GF, Pomajzl MJ. Management of dialysis fistula thrombosis. Am J Surg 1979; 138: 901-6. 8. Palder SB, Kirkman RL, Whittemore AD, ef al. Vascular access for hemodialysis patency rate and results of revision. Ann Surg 1985; 202: 236-9. 9. Wilson SE, Owens ML. Vascular access surgery. Chicago: Yearbook Medical Publishers, 1980: 101-14. 10. Jenkins A, Buist TAS, Glover SD. Medium-term follow up of forty autogenous vein and forty polytetrafluoroethylene (GoreTex) grafts for vascular access. Surgery 1988; 88: 667-72.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 162 SEPTEMBER 1991
267