Vascular Access for Dialysis Technics and Results with Newer Met,hods
Alfred Morgan, MD, Boston, Massachusetts Michael Lazarus, MD, Boston, Massachusetts
Long-term hemodialysis was made practical by the Quinton-Scribner shunt, introduced in 1960 [I]. Details of the original shunt have been modified since then but not its principle, which is to use arteriovenous flow through external tubing to maintain patency of peripheral, adjacent, arterial, and venous cannulas. Shunts are still frequently used and have important advantages, but their liabilities include relatively frequent clotting and infection and considerable inconvenience to the patient. The Cimino fistula [z] offers an alternative to the arteriovenous shunt. It is a subcutaneous, peripheral, arteriovenous anastomosis; with it, blood flow to the dialyzer can be obtained by venipuncture. Many variations in shunts and fistulas have been described. Collectively they are the primary methods for access to the circulation. They are usually performed as outpatient procedures with local anesthesia and are satisfactory in the majority of patients. A secondary group of technics are used when primary procedures cannot be. They are the Thomas femoral shunt [3], an external shunt on the femoral vessels (Figure l), and a variety of internal arteriovenous fistulas, made with either transplanted autogenous veins (Figure 2) or prosthetic material. Secondary procedures are not From the Department of Surgery, Harvard Medical School, Peter Bent Brigham Hospital, Boston, Massachusetts. This work was supported in part by a contract from the National Institute of Arthritis. Metabolism, and Digestive Diseases (NOI-AM-4-2216). Reprint requests should be addressed to Alfred Morgan, MD. 721 Huntington Avenue, Boston, Massachusetts 02115. Presented at the Fifty-Fifth Annual Meeting of the New England Surgical Society, Waterville Valley, New Hampshire, September 26-28. 1974.
432
commonly needed, comprising only 9.8 per cent of all, operations carried out for vascular access in our institution in the last five years; however, they do entail hospital admission and some risk. They are also an imperfect solution to the access problem. This report describes results with three: Thomas femoral shunts (TFS), saphenous vein grafts (SVG), and Sparks mandril grafts (SMG) implanted for dialysis. Our experience with other secondary methods, such as the bovine xenograft used in six patients, is too small for significant comparisons to be made. Material and Methods In the five year period from September 1, 1969 to September 1, 1974, 114 patients had either an SVG, TFS, or SMG at the Peter Bent Brigham Hospital. Seventy-one of these patients were female and forty-three were male. Thirty-nine had more than one secondary access operation. Mean age for the group was 47.9. Twenty-seven of the patients had received and lost one or more renal allograft. The assigned diagnosis in fiftytwo patients was chronic glomerulonephritis (this includes patients with end stage kidney disease whose diagnosis was not firmly established). Other diagnoses were polycystic kidney disease (twelve patients), Kimmelsteil-Wilson disease (twelve), chronic pyelonephritis (ten), nephrosclerosis (ten), cortical necrosis after acute renal failure (three), and metabolic kidney disease such as gout and oxallosis (four); there were six miscellaneous diagnoses. In another five, chronic renal failure was due to collagen-vascular disease. Duration of dialysis, the time from initiation of dialysis to the patient’s first secondary access procedure, was found to be distributed bimodally. Fifty-one had been on dialysis for a year or less. For the others, duration of dialysis ranged from one
TheherkanJolunafdsurgary
Vascular Access for Dialysis
to eight and a half years. The median duration was two and three quarter years. Duration of utility, or access survival time, was calcu lated for each operation performed. These data are shown as current life tables, using Chiang’s method [4]. The life table gives probability of access survival for each thirty day period during which the shunt or fistula is used. It is the fairest way to compare operations. Patients who die or who have transplantation are considered to have withdrawn from the group. The access was considered a failure when it was no longer used for dialysis, for whatever reason. Removal of any TFS, except after transplantation, was defined as a failure, even though some were removed prophylactically after another access route had been established. Thrombosis, if successfully treated by aspiration declotting of a TFS or thrombectomy of an SMG or SVG, did not indicate failure if use of the access continued. Implantation technics are straightforward vascular surgical technics, and all have been described by their originators [3,5,6]. Nevertheless, some points appear to have special importance and deserve re-emphasis. The femoral prostheses have been inserted according to Thomas’ recommendations. After exposure of the femoral vessels, a linear arteriotomy is made, usually in the common femoral artery. The prosthesis is sutured into the defect. The venous limb of the shunt is similarly placed in the femoral vein. (Figure 1.) We believe that points of importance in insertion of the femoral shunt include preservation of a fibrofatty bridge of tissue between the two limbs of the shunt, as emphasized by Thomas. This has considerably facilitated the eventual removal of the device. Stab wounds for the shunt tubing should be placed so that the velour-covered portion is 1 cm below skin level, When closer, it will eventually protrude and become a source of sepsis rather than a barrier to it. An antistaphylococcal antibiotic is used prophylactically for insertion of a Thomas femoral shunt; its efficacy is not established. Revision and removal of a Thomas shunt are more challenging than insertion. The prosthesis stimulates an impressive fibrous reaction; dissection can be time-consuming. Revisions are necessary when a pannus of thrombus covers the intimal side of the Dacron@ patch and begins to intrude on the lumen of the tubing. This usually cannot be cleared with a Fogarty catheter. Catheterization of these shunts is made difficult by the shelf formed at the junction of the patch and tubing. Turning and twisting a bent Fogarty catheter sometimes gets it through, but not always. Revisions are performed by making a semicircular incision on the vessel lateral to the insertion of the patch. It is turned back, cleared of adherent pannus, and resutured. When a Thomas femoral shunt is removed, a vein patch of the artery is usually required. It is easier to bury some residual prosthesis, but there are potential consequences, as illustrated in case I. Both saphenous vein grafts and Sparks mandrils are tunneled subcutaneously, usually forming a U-shaped
Volume 129, April 1975
Figure 1. The Thomas femoral shunt. After exposure of the vessels an appHque patch is sewn into inckkns in the common or sqerfkiai femoral vein and artery. The tubing is brought out through stab wounds_ loop on the forearm or anterior thigh. An effort is made to insert the tunneling instrument very superficially; it is easy to insert the grafts so deeply that venipuncture is difficult and needle trauma likely. The mandril is a flexible Silastic? rod covered by two layers of knitted Dacron. During maturation, the fabric is incorporated into a fibrous tube, forming a composite material with good handling qualities. A six week maturation period has been recommended. One graft explored at five weeks was found insufficiently matured for anastomosis. We have not allowed extra maturation time in diabetic patients. For SMGs and SVGs, selection of anastomotic sites has been dictated by both anatomic factors and patient preference. Anatomically, the availability of a suitable vein has usually been a controlling factor. Preoperative venography has not been employed, the rationale being its demonstrated lack of utility in evaluating vein size before femoropopliteal bypass. A palpable femoral pulse has always been present when the femoral vessels above completely occluded the superficial femoral arteries. Ischemic injury occurred only once, as described in case III. Possible sites for venous anastomosis in the forearm are the cephalic and the basilic systems, when they are
Figure 2. The saphenous vein graft can be implanted in the arm, as shown here, in the groin, retainfng the native saphenofemoral junctkn, or in other sites when vessel pairs can be found for anastomosis.
433
Morgan and Lazarus
6 SURVIVAL
(o,xj
;OrK.
SHUNTS
0.2
I
SAPHENOUS
GRAFTS
1 *0
III
I 6
I
I
I
30 :Y
I
I
11 18
1
I1
24
1
I 30
PERIODS lx)
Figure 3. Combined Iife tab/e curves for 7homas femoral shunts (twenty-seven), Sparks mandril graft? (thkiyseven), and saphenous vein grafts (ninety-nine). 6 SURVIVAL
(,,.J
1.0
Figure 4. Life table curves for saphenous grafts in arm and leg; signlfkance of the differences seen does not exceed p = 0.05 for any pair of points. patent; or the venae commitantes
when the superficial
venous system has been thrombosed or multiply interrupted by previous Scribner shunts. The preferred site, in the arm, is the basilic vein above the joint line of the elbow. Only once (case II) has it not been possible to find a vein adequate for anastomosis. Results Duration of Utility. Figure 3 shows life table data for the three methods. After 180 days, the probability of continued utility is significantly greater for the femoral shunt than for either alternative. Twenty-seven femoral shunts were implanted. Ten of the patients with shunts are dead or have had transplantation, seven shunts failed, and ten are in use. This apparent superiority must be balanced against a high incidence of sepsis.
434
I
Fourteen patients experienced one or more episodes of sepsis severe enough to require antibiotics or drainage. Sepsis caused four of the seven failures; in one patient it was the cause of death. The other TFS failures were due to phlebitis. No embolic episodes were recorded. The longest and largest experience with secondary access procedures has been with the SVG. Ninety-nine were done; twenty-six are in use, fifty-two failed, and twenty-one patients have died or have had transplantation. Fifteen of the failures occurred during the first thirty days and have to be attributed to poor technic or poor patient selection. Specific causes were identifiable in ten patients: four required early ligation for wound sepsis; two had irreversible clotting after cardiac arrest; and six had a tiny greater saphenous vein. Later failures were distributed between sepsis, needle trauma, and cicatrization. Three grafts were ligated for mycotic aneurysm. Two pulsating hematomas at sites of needle puncture required surgical closure of the defect. The influence of patient factors and site on SVG survival was undramatic. There was no significant relation to age. Grafts in the forearm had a slightly greater early closure rate at thirty and sixty days (p
Alfred Morgan, MD, Boston, Massachusetts Michael Lazarus, MD, Boston, Massachusetts
Long-term hemodialysis was made practical by the Quinton-Scribner shunt, introduced in 1960 [I]. Details of the original shunt have been modified since then but not its principle, which is to use arteriovenous flow through external tubing to maintain patency of peripheral, adjacent, arterial, and venous cannulas. Shunts are still frequently used and have important advantages, but their liabilities include relatively frequent clotting and infection and considerable inconvenience to the patient. The Cimino fistula [z] offers an alternative to the arteriovenous shunt. It is a subcutaneous, peripheral, arteriovenous anastomosis; with it, blood flow to the dialyzer can be obtained by venipuncture. Many variations in shunts and fistulas have been described. Collectively they are the primary methods for access to the circulation. They are usually performed as outpatient procedures with local anesthesia and are satisfactory in the majority of patients. A secondary group of technics are used when primary procedures cannot be. They are the Thomas femoral shunt [3], an external shunt on the femoral vessels (Figure l), and a variety of internal arteriovenous fistulas, made with either transplanted autogenous veins (Figure 2) or prosthetic material. Secondary procedures are not From the Department of Surgery, Harvard Medical School, Peter Bent Brigham Hospital, Boston, Massachusetts. This work was supported in part by a contract from the National Institute of Arthritis. Metabolism, and Digestive Diseases (NOI-AM-4-2216). Reprint requests should be addressed to Alfred Morgan, MD. 721 Huntington Avenue, Boston, Massachusetts 02115. Presented at the Fifty-Fifth Annual Meeting of the New England Surgical Society, Waterville Valley, New Hampshire, September 26-28. 1974.
432
commonly needed, comprising only 9.8 per cent of all, operations carried out for vascular access in our institution in the last five years; however, they do entail hospital admission and some risk. They are also an imperfect solution to the access problem. This report describes results with three: Thomas femoral shunts (TFS), saphenous vein grafts (SVG), and Sparks mandril grafts (SMG) implanted for dialysis. Our experience with other secondary methods, such as the bovine xenograft used in six patients, is too small for significant comparisons to be made. Material and Methods In the five year period from September 1, 1969 to September 1, 1974, 114 patients had either an SVG, TFS, or SMG at the Peter Bent Brigham Hospital. Seventy-one of these patients were female and forty-three were male. Thirty-nine had more than one secondary access operation. Mean age for the group was 47.9. Twenty-seven of the patients had received and lost one or more renal allograft. The assigned diagnosis in fiftytwo patients was chronic glomerulonephritis (this includes patients with end stage kidney disease whose diagnosis was not firmly established). Other diagnoses were polycystic kidney disease (twelve patients), Kimmelsteil-Wilson disease (twelve), chronic pyelonephritis (ten), nephrosclerosis (ten), cortical necrosis after acute renal failure (three), and metabolic kidney disease such as gout and oxallosis (four); there were six miscellaneous diagnoses. In another five, chronic renal failure was due to collagen-vascular disease. Duration of dialysis, the time from initiation of dialysis to the patient’s first secondary access procedure, was found to be distributed bimodally. Fifty-one had been on dialysis for a year or less. For the others, duration of dialysis ranged from one
TheherkanJolunafdsurgary
Vascular Access for Dialysis
to eight and a half years. The median duration was two and three quarter years. Duration of utility, or access survival time, was calcu lated for each operation performed. These data are shown as current life tables, using Chiang’s method [4]. The life table gives probability of access survival for each thirty day period during which the shunt or fistula is used. It is the fairest way to compare operations. Patients who die or who have transplantation are considered to have withdrawn from the group. The access was considered a failure when it was no longer used for dialysis, for whatever reason. Removal of any TFS, except after transplantation, was defined as a failure, even though some were removed prophylactically after another access route had been established. Thrombosis, if successfully treated by aspiration declotting of a TFS or thrombectomy of an SMG or SVG, did not indicate failure if use of the access continued. Implantation technics are straightforward vascular surgical technics, and all have been described by their originators [3,5,6]. Nevertheless, some points appear to have special importance and deserve re-emphasis. The femoral prostheses have been inserted according to Thomas’ recommendations. After exposure of the femoral vessels, a linear arteriotomy is made, usually in the common femoral artery. The prosthesis is sutured into the defect. The venous limb of the shunt is similarly placed in the femoral vein. (Figure 1.) We believe that points of importance in insertion of the femoral shunt include preservation of a fibrofatty bridge of tissue between the two limbs of the shunt, as emphasized by Thomas. This has considerably facilitated the eventual removal of the device. Stab wounds for the shunt tubing should be placed so that the velour-covered portion is 1 cm below skin level, When closer, it will eventually protrude and become a source of sepsis rather than a barrier to it. An antistaphylococcal antibiotic is used prophylactically for insertion of a Thomas femoral shunt; its efficacy is not established. Revision and removal of a Thomas shunt are more challenging than insertion. The prosthesis stimulates an impressive fibrous reaction; dissection can be time-consuming. Revisions are necessary when a pannus of thrombus covers the intimal side of the Dacron@ patch and begins to intrude on the lumen of the tubing. This usually cannot be cleared with a Fogarty catheter. Catheterization of these shunts is made difficult by the shelf formed at the junction of the patch and tubing. Turning and twisting a bent Fogarty catheter sometimes gets it through, but not always. Revisions are performed by making a semicircular incision on the vessel lateral to the insertion of the patch. It is turned back, cleared of adherent pannus, and resutured. When a Thomas femoral shunt is removed, a vein patch of the artery is usually required. It is easier to bury some residual prosthesis, but there are potential consequences, as illustrated in case I. Both saphenous vein grafts and Sparks mandrils are tunneled subcutaneously, usually forming a U-shaped
Volume 129, April 1975
Figure 1. The Thomas femoral shunt. After exposure of the vessels an appHque patch is sewn into inckkns in the common or sqerfkiai femoral vein and artery. The tubing is brought out through stab wounds_ loop on the forearm or anterior thigh. An effort is made to insert the tunneling instrument very superficially; it is easy to insert the grafts so deeply that venipuncture is difficult and needle trauma likely. The mandril is a flexible Silastic? rod covered by two layers of knitted Dacron. During maturation, the fabric is incorporated into a fibrous tube, forming a composite material with good handling qualities. A six week maturation period has been recommended. One graft explored at five weeks was found insufficiently matured for anastomosis. We have not allowed extra maturation time in diabetic patients. For SMGs and SVGs, selection of anastomotic sites has been dictated by both anatomic factors and patient preference. Anatomically, the availability of a suitable vein has usually been a controlling factor. Preoperative venography has not been employed, the rationale being its demonstrated lack of utility in evaluating vein size before femoropopliteal bypass. A palpable femoral pulse has always been present when the femoral vessels above completely occluded the superficial femoral arteries. Ischemic injury occurred only once, as described in case III. Possible sites for venous anastomosis in the forearm are the cephalic and the basilic systems, when they are
Figure 2. The saphenous vein graft can be implanted in the arm, as shown here, in the groin, retainfng the native saphenofemoral junctkn, or in other sites when vessel pairs can be found for anastomosis.
433
Morgan and Lazarus
6 SURVIVAL
(o,xj
;OrK.
SHUNTS
0.2
I
SAPHENOUS
GRAFTS
1 *0
III
I 6
I
I
I
30 :Y
I
I
11 18
1
I1
24
1
I 30
PERIODS lx)
Figure 3. Combined Iife tab/e curves for 7homas femoral shunts (twenty-seven), Sparks mandril graft? (thkiyseven), and saphenous vein grafts (ninety-nine). 6 SURVIVAL
(,,.J
1.0
Figure 4. Life table curves for saphenous grafts in arm and leg; signlfkance of the differences seen does not exceed p = 0.05 for any pair of points. patent; or the venae commitantes
when the superficial
venous system has been thrombosed or multiply interrupted by previous Scribner shunts. The preferred site, in the arm, is the basilic vein above the joint line of the elbow. Only once (case II) has it not been possible to find a vein adequate for anastomosis. Results Duration of Utility. Figure 3 shows life table data for the three methods. After 180 days, the probability of continued utility is significantly greater for the femoral shunt than for either alternative. Twenty-seven femoral shunts were implanted. Ten of the patients with shunts are dead or have had transplantation, seven shunts failed, and ten are in use. This apparent superiority must be balanced against a high incidence of sepsis.
434
I
Fourteen patients experienced one or more episodes of sepsis severe enough to require antibiotics or drainage. Sepsis caused four of the seven failures; in one patient it was the cause of death. The other TFS failures were due to phlebitis. No embolic episodes were recorded. The longest and largest experience with secondary access procedures has been with the SVG. Ninety-nine were done; twenty-six are in use, fifty-two failed, and twenty-one patients have died or have had transplantation. Fifteen of the failures occurred during the first thirty days and have to be attributed to poor technic or poor patient selection. Specific causes were identifiable in ten patients: four required early ligation for wound sepsis; two had irreversible clotting after cardiac arrest; and six had a tiny greater saphenous vein. Later failures were distributed between sepsis, needle trauma, and cicatrization. Three grafts were ligated for mycotic aneurysm. Two pulsating hematomas at sites of needle puncture required surgical closure of the defect. The influence of patient factors and site on SVG survival was undramatic. There was no significant relation to age. Grafts in the forearm had a slightly greater early closure rate at thirty and sixty days (p
