Low-Intensity Warfarin Is Ineffective for the Prevention of PTFE Graft Failure in Patients on Hemodialysis: A Randomized Controlled Trial MARK A. CROWTHER,* CATHERINE M. CLASE,* PETER J. MARGETTS,* JIM JULIAN, † KIM LAMBERT,* DENISE SNEATH, ‡ RYUTA NAGAI, ‡ SARAH WILSON,* and ALISTAIR J. INGRAM* *Department of Medicine and † Department of Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada; and ‡ Division of Nephrology, Hotel Dieu Hospital, St. Catharine’s, Ontario, Canada Abstract. Polytetrafluoroethylene (PTFE) dialysis grafts in pa- tients with end-stage renal disease (ESRD) are prone to throm- botic failure. The objective of this multicenter, randomized, double-blind, placebo-controlled clinical trial was to determine if warfarin reduces the risk of failure of PTFE dialysis grafts. Patients with ESRD and newly placed PTFE grafts were stud- ied at community and academic dialysis centers in Southwest- ern Ontario. Patients were allocated to receive warfarin or matching placebo, with the warfarin administered to achieve a target INR of 1.4 to 1.9. Time to graft failure was the main outcome measure. A total of 107 patients (56 allocated to warfarin) were randomized. The time-to-event analysis re- vealed no significant difference in the likelihood of graft survival between the two groups (odds ratio, 1.76 in favor of placebo; 95% confidence interval, 0.72 to 4.34). Six major bleeds occurred in five patients allocated to warfarin com- pared with none in the patients who received placebo (P = 0.03). In conclusion, low-dose warfarin was associated with an excess of clinically important major bleeding in patients with ESRD enrolled in this study. Furthermore, low-intensity, monitored-dose warfarin does not appear to prolong PTFE graft survival. Access complications account for 30% of hemodialysis patient hospital admissions (1), and it has been estimated that 14% of total Medicare end-stage renal disease (ESRD) expenditures relate to the treatment of complications of vascular access (2). Mechanical malfunction of access due to thrombosis accounts for much of this morbidity and resource utilization; prevention and treatment of thrombosis is consequently of major clinical and economic importance. Given its effectiveness for the pre- vention of thrombosis in other clinical settings, warfarin was used extensively in the first years of dialysis in an effort to prevent thrombotic failure of external arteriovenous shunts. However, despite reducing the risk of thrombosis in Scribner shunts (3,4), use of warfarin fell into disfavor because very high rates of serious hemorrhage were reported (5). Currently, most North American hemodialysis patients use a polytetrafluoroethylene (PTFE) graft as their primary blood access (6 – 8). These grafts are subject to a high rate of failure attributable to acute thrombosis. To reduce the risk of throm- botic failure, many nephrologists prescribe anticoagulants. To our knowledge, there are no randomized clinical trials to sup- port this widespread clinical practice, particularly the use of warfarin, to prevent thrombotic complications in PTFE grafts. To address this knowledge gap, particularly to determine whether warfarin is effective for the prevention of PTFE graft thrombosis, we performed a randomized trial, allocating pa- tients with newly placed PTFE hemodialysis access grafts to receive either warfarin administered to achieve an INR of 1.4 to 1.9 or matching placebo. The primary efficacy outcome was the time to thrombotic graft malfunction, and the primary safety outcome was the frequency of hemorrhage. Materials and Methods Study Subjects and Randomization This study was a multicenter, randomized, double-blind, placebo- controlled trial in which patients with newly placed (incident) PTFE grafts for hemodialysis access were eligible to participate. Two uni- versity-based teaching hospitals and a community dialysis unit par- ticipated. Consenting patients meeting inclusion and exclusion criteria (Table 1) were allocated, using a concealed computer-generated ran- domization scheme, to receive warfarin or matching placebo in ran- dom permuted blocks of four (DuPont Pharma, Mississauga, Canada). Patients were stratified by clinical center, use of antiplatelet therapy at the time of randomization, and whether the graft was placed in a patient currently receiving hemodialysis or in anticipation of the need for dialysis. Treatment and Follow-Up Study drug was initiated within 7 d of surgery. INR monitoring was initiated on day 3 after starting study drug and was continued accord- ing to a predefined protocol. INR was determined using Thromborel Received January 29, 2002. Accepted June 3, 2002. Correspondence to Dr. Alistair J. Ingram, 708-500 Charlton Ave. East, Ham- ilton, Ontario. Phone: 905-521-6151; Fax: 905-521-6153; E-mail: ingrama@mcmaster.ca 1046-6673/1309-2331 Journal of the American Society of Nephrology Copyright © 2002 by the American Society of Nephrology DOI: 10.1097/01.ASN.0000027356.16598.99 J Am Soc Nephrol 13: 2331–2337, 2002