Factors predictive of failure of Brescia-Cimino arteriovenous fistulas

ORIGINAL ARTICLE

Factors Predictive of Failure of Brescia-Cimino Arteriovenous Fistulas Clark Zeebregts,1 Jan van den Dungen,1 Arjen Bolt, 1 Casper Franssen,2 Eric Verhoeven1 and Reinout van Schilfgaarde1 From the 1Department of Surgery and 2 Department of Internal Medicine, University Hospital Groningen, Groningen, The Netherlands

Eur J Surg 2002; 168: 29–36 ABSTRACT Objective: To evaluate patency rates of Brescia-Cimino
stulas and to
nd out which independent factors were predictors of failure. Design: Retrospective clinical study. Setting: University hospital, The Netherlands. Subjects: 150 consecutive patients (mean age 56 years, range 17–80) who had 153 primary Brescia-Cimino
stulas created during the 5-year period January 1995–December 1999. Main outcome measures: Patency rates calculated by the Kaplan-Meier method and the possible predictive value of 20 different variables assessed by Cox’s proportiona l hazard model. Results: The primary patency rate was 70% at 3 months, and 7 distinct factors were signi
cantly associated with failure of the
stula. The ones with a hazard ratio (HR) for failure greater than 2.5 were: the start of dialysis before creation of the
stula (HR 2.79, p < 0.01), moderate or poor quality of both the artery (HR 2.54, p < 0.01) and vein (HR 3.55, p < 0.001), and postoperative use of acenocoumarol instead of acetylsalicylic acid (HR 3.14, p < 0.01). Conclusion: The major determinants for a successfully created Brescia-Cimino
stula were creation of the
stula before the start of dialysis, as well as good quality of both the artery and the vein. This argues for timely creation of such
stulas in patients with end-stage renal disease and for accurate preoperative examination to establish the quality of the vessels. Key words: haemodialysis, Brescia-Cimino fistula, patency, prognostic factor, multivariate analysis.

INTRODUCTION Patients with end-stage renal disease (ESRD) are dependent on long-term dialysis until transplantation is possible. In the Netherlands, the total number of patients on dialysis was 2091 in 1983, and has more than doubled in the past two decades (total population in 1999: 15.8 million). A further increase in the number of patients on dialysis is expected, mainly as a result of the ageing population (19). Since the 1980s, continuous ambulatory peritoneal dialysis has gained in popularity. Nevertheless, the percentage of patients on haemodialysis has stabilised over the past 5 years at roughly 70% of the patients with ESRD (19). Increasing numbers of patients on haemodialysis have also been noted throughout other parts of the world (2, 8, 17, 25). There is consensus that the distal autogenous arteriovenous
stula is the
rst option for permanent access for haemodialysis (25). This type of
stula, as originally described by Brescia et al. permits easy repeated access to the circulation (1). The BresciaCimino
stula is also easy to create without the need Ó 2002 Taylor & Francis. ISSN 1102–4151

for prosthetic material. In addition, it requires no daily care, permits freedom of activity with unlimited use of the arm, and is usually easy to cannulate (15). However, various postoperative complications may occur, thrombosis and local infection being the most common. The major predisposing factor for thrombosis of the
stula is stenosis, which occurs as a result of myointimal hyperplasia at the anastomotic site, accounting for 80% to 85% of thromboses (9, 27). Other contributory causes of thrombosis include dissection and haematoma as a consequence of puncture, excessive compression of the
stula after dialysis, hypotension, hypovolaemia, or a hypercoagulable state. The primary reported failure rate is 12% to 24% (3, 5, 10, 12, 18). After 6 months, patency rates further diminish to 50%–70%. It is thought that variations in the values reported by different groups of investigators reect the use of different criteria for selection of patients (16). The relatively high failure rates depend largely on patient-associated factors, and therefore insight into their characteristics may help to improve preoperative selection. This study was undertaken to
nd out which patientEur J Surg 168

30

C. Zeebregts et al.

Table I. Causes of primary renal disease Disease

No. (%) of patients

Chronic renal failure, aetiology unknown

8 (5)

Primary glomerular diseases IgA nephropathy Membranous nephropathy Glomerulosclerosis Glomerulonephritis

8 3 7 10

Interstitial nephritis, pyelonephritis, drug-induced nephropathy and urolithiasis Cystic kidney diseases

14 (9) 12 (8)

Other congenital and hereditary kidney diseases Hereditary/familial nephropathy Alport’s syndrome Congenital renal dysplasia Horseshoe kidney Other congenital disorders

3 1 8 1 5

(5) (2) (5) (7)

(2) (1) (5) (1) (3)

Renal vascular disease, excluding vasculitis Renal vascular disease Renal vascular disease caused by hypertension Ischaemic renal disease or cholesterol embolism

17 (11) 11 (7) 3 (2)

Diabetes mellitus

20 (13)

Other multisystem diseases Wegener’s granulomatosis Amyloidosis Henoch-Scho¨nlein purpura Haemolytic uraemic syndrome

8 7 1 2

Others Cortical necrosis

1 (1)

associated factors predicted failure of Brescia-Cimino
stulas to improve our procedure for selection of patients and our perioperative management. We retrospectively reviewed the results of patients who had this procedure in our hospital. The failure rates were calculated and a multivariate analysis was done to identify independent predictors of failure. PATIENTS AND METHODS Primary arteriovenous Brescia-Cimino (BC)
stulas were studied in patients who required permanent vascular access for haemodialysis. Between January 1995 and December 1999, a total of 153 such
stulas were created in 150 consecutive patients at the University Hospital Groningen, Groningen, The Netherlands. There were 95 male and 55 female patients with a mean (SEM) age of 56 (1.4) years (range 17 to 80). The most frequent coexisting factors included hypertension (n = 81, 54%, de
ned as taking antihypertensive drugs and/or having a preoperative systolic blood pressure ¶160 mm Hg and/or a preoperative diastolic blood pressure ¶95 mm Hg), smoking (n = 65, 43%, de
ned as a current smoker or having smoked in the Eur J Surg 168

(5) (5) (1) (1)

past 10 years), hypercholesterolaemia (n = 56, 37%, de
ned as serum cholesterol >6.0 mmol/L or use of medication to control cholesterol concentration, or both), and diabetes mellitus (n = 24, 16%, de
ned as using oral antidiabetic medication or insulin, or both). The causes of renal failure are given in Table I. Preoperative procedure Patients were referred from regional dialysis centres after the preferred form of dialysis in this cohort of patients had been decided to be haemodialysis. A Brescia-Cimino
stula was created if on physical examination the cephalic vein on the radial side of the wrist was visible and palpable with a suf
cient outow tract. Differentiation of occluded arterial segments was accomplished by Allen’s test for occlusion of the ulnar and radial arteries. Additional duplex assessment was made if there was any doubt about the size of the vessels on physical examination. Usually, a patent radial and ulnar artery and a venous diameter of ¶2.5 mm during proximal venous occlusion were regarded as suf
cient. The non-dominant arm was preferred.

Predictors of failure of Cimino fistulas Surgical procedure Operations were done under brachial plexus block or general anaesthesia. Through a lower arm incision at the radial site of the wrist, the cephalic vein and radial artery were dissected from the surrounding tissues. The cephalic vein was transected and dilated with heparinised saline. Severe spasm of the vein was treated with local papaverine solution (2 mg/ml). An arteriotomy was made in the radial artery so that an endvein-to-side-artery anastomosis could be made with a hand-sutured anastomosis using two running 6/0 or 7/0 polypropylene sutures (Prolene1 , Ethicon, Inc, Johnson and Johnson Medical BV, Amersfoort, The Netherlands). The patency of anastomoses was assessed intraoperatively by palpation, auscultation, and hand-held Doppler. The wound was then closed intracutaneously.

End points of the study and de
nitions During operation, the surgeon assessed the vessels macroscopically and the quality of each vessel was classi
ed into one of the following categories: poor, moderate, or good. The features that were considered to qualify the vessels into terms of poor, moderate, and good were for the radial artery: the diameter, presence of atherosclerotic plaques, and pre-existing stenoses. For the vein, the features were the diameter and suf
ciency of the outow tract. Whenever possible, the
stula was allowed to mature for a period of about 6 weeks, by which time the cephalic vein had usually developed suf
ciently to be used for haemodialysis and sustain suf
cient blood ow. End points for outcome were: useless and irreparable
stula (either early or late failure, including insuf
cient development of the
stula at 6 weeks), inability to use this access site for dialysis, need for an intervention to maintain the function of the
stula, discontinuation of dialysis with a functioning
stula (such as after renal transplantation), patients’ choice to change to another form of access, patients’ death (either with a functioning or a non-functioning
stula), or a functioning
stula at the time of data analysis. De
nitions of patency, failure, and revision were those described by others (3). Briey summarised, the primary patency is the percentage of
stulas functioning without any salvage intervention of the total number of
stulas at risk, and the secondary patency is the percentage of
stulas functioning with all salvage interventions included of the total number of
stulas at risk. Failure means inadequate function of a
stula and revision refers to those procedures that do not essentially change the type of the
stula, and in which no prosthetic material is used.

31

Identi
cation of predictors of failure of the
stula Personal and clinical data were collected retrospectively. For follow-up purposes, the locoregional dialysis centres, as well as general practitioners, were contacted. Finally, if needed, the patient or relatives were asked to give additional information. To identify predictors of
stula failure, various preoperative, operative, and postoperative variables were chosen because of their potential relevance to failure. Dichotomous variables included: sex (male or female), smoking history (non-smoker or smoker), serum cholesterol concentration (normocholesterolaemia or hypercholesterolaemia), diabetes mellitus (nondiabetic or diabetic), cause of ESRD (local or systemic disease), history of dialysis (patients who had not been dialysed at the time of creation of the
stula or patients in whom dialysis began before the operation), side of the
stula (left or right), type of anaesthesia (brachial plexus block or general), thickness of the suture (6/0 polypropylene or 7/0 polypropylene) , quality of the artery (good or moderate to poor), quality of the vein (good or moderate to poor), use of intraoperative heparin (patients who were not given an intraoperative heparin bolus 2000–3000 IU intravenously or those who were), surgeon (resident or staff member), and postoperative anticoagulation (acetylsalicylic acid or acenocoumarol). Continuous variables included: age (years), preoperative systolic and diastolic blood pressure (mm Hg), preoperative serum urea (mmol/L) and creatinine (mmol/L) concentrations, and operating time (minutes). Statistical analysis The data are presented as mean (SEM), unless otherwise stated. Primary and secondary patency rates were calculated by the Kaplan-Meier method (11). Patients who died, patients who chose to change to another form of access, or those who had a successful transplant with a functioning
stula were censored. Differences between categorical variables were as-

Fig. 1. Primary patency rate of Brescia-Cimino
stulas calculated using the Kaplan-Meier method. Eur J Surg 168

32

C. Zeebregts et al.

Table II. Predictors of failure of Brescia-Cimino
stulas on univariate analysis. Data are expressed as number (%) unless otherwise stated Fistula Variable Sex Male Female Mean (SEM) age (years) Smoking No Yes Not known Cholesterol Normal High Not known Diabetes mellitus No Yes ESRD Local Systemic Dialysis before operation No Yes Mean (SEM) preoperative systolic BP (mmHg) Mean (SEM) preoperative diastolic BP (mmHg) Mean (SEM) preoperative urea (mmol/L) Mean (SEM) preoperative creatinine (mmol/L) Side of
stula Left Right Anaesthesia Brachial plexus block General Not known Mean (SEM) operation time (min) Thickness of suture 6/0 7/0 Not known Quality of artery Good Moderate or poor Not known Quality of vein Good Moderate or poor Not known Intraoperative intravenous heparin No Yes Not known Surgeon Resident Staff member Postoperative anticoagulation Acetylsalicylic acid Acenocoumarol Not known

Eur J Surg 168

Patent (n = 89)

Occluded (n = 64)

p value

62 (70) 27 (30) 55 (2)

37 (58) 27 (42) 56 (2)

0.13

44 (49) 34 (38) 11 (12)

33 (52) 24 (38) 7 (11)

0.86

52 (58) 33 (37) 4 ( 4)

41 (64) 22 (34) 1 ( 2)

0.63

75 (84) 14 (16)

53 (83) 11 (17)

0.81

50 (56) 39 (44)

29 (45) 35 (55)

0.19

33 56 154 89 21 676

(37) (63) (3) (2) (1) (30)

62 (70) 27 (30) 23 52 14 100

(26) (58) (16) (3)

12 52 151 83 20 664

(19) (81) (4) (2) (1) (36)

42 (66) 22 (34) 22 36 6 109

0.88

0.01 0.54 0.03 0.56 0.79 0.60

(34) (56) ( 9) (5)

0.38

70 (79) 16 (18) 3 ( 3)

48 (75) 13 (20) 3 ( 5)

0.68

62 (70) 14 (16) 13 (15)

35 (56) 21 (33) 8 (13)

0.01

64 (72) 17 (19) 8 ( 9)

23 (36) 35 (55) 6 ( 9)