Stent vs. Angioplasty for Treatment of Thrombosed AV Grafts: Long-Term Outcomes

HYPOTHESES

1. The primary patency of grafts following thrombectomy and angioplasty is limited by the high re-stenosis rate.
2. Stent deployment reduces the likelihood of re-stenosis at the site of the stenotic lesion, thereby prolonging primary graft patency following thrombectomy, and decreasing or delaying the need for future interventions.
3. The cost savings arising from stent deployment (decrease in subsequent access procedures) outweigh the additional cost of the stent itself.

AIMS

1. To evaluate whether stent placement in thrombosed grafts with underlying stenosis at the venous anastomosis results in longer primary graft patency, as compared with angioplasty alone.
2. To determine whether stent placement in thrombosed grafts with underlying stenosis at the venous anastomosis reduces the subsequent cost of access procedures in excess of the cost of the stent.

To accomplish these goals, we will conduct a randomized clinical trial of hemodialysis patients who are referred for treatment of thrombosed A-V grafts. The patients will be randomized to be treated with thrombectomy + stent deployment or thrombectomy + angioplasty. The primary outcomes will be primary (unassisted) graft patency. The secondary outcomes will be secondary (assisted) graft patency and the cumulative cost of all access-related procedures, hospitalizations, and complications.

SIGNIFICANCE AND BACKGROUND

Most patients with end-stage renal disease undergo hemodialysis thrice weekly to optimize their survival, minimize medical complications, and enhance their quality of life. A reliable vascular access is a critical requirement for providing adequate hemodialysis. The ideal vascular access would be easy to place, ready to use as soon as it is placed, deliver high blood flows indefinitely, and be free of complications. None of the existing types of vascular accesses achieve this ideal. Among the three types of vascular access currently available, A-V fistulas are superior to A-V grafts, which in turn are superior to dialysis catheters. Recognizing the relative merits of the vascular access types, the NKF-DOQI guidelines recommend placement of A-V fistulas in 50% of hemodialysis patients, A-V grafts in 40%, and dialysis catheters in 10% [19]. The actual current distribution of vascular accesses among prevalent hemodialysis patients in the United States is 25-30% fistulas, 45-50% grafts, and 25% dialysis catheters [20, 21]. Vascular access procedures and their subsequent complications represent a major cause of morbidity, hospitalization and cost for chronic hemodialysis patients [22-26]. Over 20% of hospitalizations in hemodialysis patients in the United States are access-related, and the annual cost of access morbidity is close to $1 billion [25].

A-V grafts are more prone to recurrent stenosis and thrombosis than are fistulas, and require multiple radiologic or surgical interventions to ensure their long-term patency for dialysis [26]. A report from UAB observed that the first salvage procedure to maintain graft patency (thrombectomy, angioplasty, or surgical revision) was required in 29% of grafts at 3 months, 52% at 6 months, 77% at 12 months, and 96% at 24 months [5]. Most grafts required multiple interventions over time to maintain their patency for dialysis. A mean of 1.22 interventions were required to maintain access patency per graft-year, including 0.51 thrombectomies, 0.54 angioplasties, and 0.17 surgical revisions. Secondary graft survival (time from initial placement to permanent failure, regardless of number of interventions) has ranged from 59 to 87% at one year, and 50 to 73% at two years [27-35]. A number of studies have examined the association of demographic and clinical factors with long-term graft survival, but no consistent clinical predictor of poor graft outcome has emerged. Windus et al [27] observed lower graft survival in diabetics as compared with nondiabetics. In contrast, three other studies observed no significant difference in graft survival between diabetic and nondiabetic patients [5, 35, 36]. Several investigators have reported a lack of association between patient age or sex and long-term graft survival [5, 35-37]. Windus et al [27] found that older age predicted lower graft survival in non-diabetics, but not in diabetic patients. Graft survival is similar between upper arm and forearm grafts [5, 29]. Finally, hypoalbuminemia has been associated with shorter A-V graft survival in 2 reports [5, 36].

The major cause of graft failure is thrombosis due to underlying stenosis of the venous anastomosis, draining vein, or central vein [10, 38]. When the stenosis is not detected and corrected in a timely fashion, grafts typically thrombose. Clotted grafts require thrombectomy by either Radiology or Surgery, most commonly in association with correction of the underlying stenosis. If graft patency cannot be restored, it becomes necessary to construct a vascular access at a new anastomotic site. Intervention-free graft patency is considerably lower following thrombectomy than after elective angioplasty [10]. A comparison of graft outcomes after radiologic interventions at our institution found that the primary (unassisted) patency was only 30% at 3 months for clotted grafts, as compared with 71% for patent grafts undergoing elective angioplasty [10]. Not surprisingly, graft survival was worse if there was a residual stenosis after the angioplasty. However, even in the subset of patients with no residual stenosis after the intervention, the primary 3-month patency was still lower in clotted grafts, as compared with patent grafts undergoing elective angioplasty (median survival, 2.5 vs 6.9 months).

Because graft patency is much worse following thrombectomy than after elective angioplasty, the current state of clinical access management involves an ongoing surveillance program to monitor for evidence of hemodynamically significant graft stenosis, and referral of such patients for elective angioplasty or surgical revision. The rationale for this approach is that correction of graft stenosis in a timely fashion will decrease the risk of graft thrombosis and prolong the survival of the graft [19]. Substantial clinical research has been directed at evaluating the predictive values of several noninvasive screening tests to identify hemodynamically significant graft stenosis, so that the patients can be referred for angioplasty or surgical revision before the graft has a chance to clot. These surveillance methods have included: dynamic dialysis venous pressures, static dialysis venous pressures, recirculation, Doppler ultrasound, ultrasound dilution methodology, and access blood flow (Reviewed in [39]). Graft stenosis can also be predicted by aggressive clinical monitoring (looking for decreases in Kt/V, prolonged bleeding at graft needle sites, and abnormalities of graft inspection and auscultation [1, 40]. A number of observational studies have reported substantial (50-60%) reductions in the frequency of graft thrombosis using a variety of surveillance methods [40-43]; none have eliminated this problem. Four randomized studies evaluating the efficacy of access surveillance on graft outcomes have found that surveillance is an excellent tool for identifying hemodynamically significant graft stenosis. However, the higher frequency of preemptive angioplasty in the surveillance group did not appear to translate into a reduction in graft thrombosis or prolongation of graft survival [44-47]. This discrepancy suggests that preemptive angioplasty may not be effective in producing sustained improvement of the stenotic lesion. Stent deployment may improve the durability of graft angioplasty and decrease the variability between operators in graft patency following radiologic interventions. Thus, it is possible that more frequent use of stents might enhance the value of graft surveillance.

Graft stenosis occurs as a consequence of aggressive myointimal hyperplasia, which occurs most commonly at the venous anastomosis [48]. Vascular injury resulting from the angioplasty may actually accelerate the process of myointimal hyperplasia, thereby resulting in early re-stenosis [49]. Endoluminal stents, by forming a rigid scaffold at the venous anastomosis, may slow the encroachment of the area of myointimal hyperplasia into the vascular lumen, thereby limiting the magnitude of recurrent stenosis. Thus, use of stents may be of utility in preventing restenosis following angioplasty. Stent placement has been attempted for treatment of grafts in which angioplasty results in suboptimal technical success or if the stenosis recurs rapidly. A number of small series have reported the outcomes of stent placement for vascular access with refractory stenosis [50-56]. Unfortunately, these studies have suffered from several methodologic limitations, including retrospective data collection, absence of a suitable control group, combining patent and thrombosed grafts, combining stents placed at a variety of stenotic sites, and combining grafts with fistulas.

A small, randomized study comparing stents with conventional angioplasty found no difference in primary graft patency following the intervention [54]. However, this study enrolled a mixture of clotted grafts and patent grafts, and the stenotic lesions were at a variety of locations, limiting the interpretation of the findings. A recent, uncontrolled study reported the outcomes of clotted grafts undergoing thrombectomy, as well as stent placement at the venous anastomosis [50]. In this more homogeneous group of grafts, the primary graft patency was 63% at 6 months. Although there was no matched control group treated with angioplasty alone, the unassisted graft survival was far superior to that reported in several series (11 to 34% at 6 months)[10, 57-61]

RESEARCH DESIGN AND METHODS

Design: This will be an open-label, prospective, randomized clinical trial comparing the outcomes of thrombosed AV grafts treated with mechanical thrombectomy and angioplasty (current standard of care), as compared with stent deployment after mechanical thrombectomy and angioplasty.

Subjects: The subjects for this study will be recruited from the University of Alabama at Birmingham (UAB) Nephrology practice. UAB provides medical care for approximately 500 hemodialysis patients, under the supervision of 12 full-time nephrologists. We average 150 graft thrombectomy procedures annually at our institution.

Methods:

Screening: After obtaining informed consent, the clinic chart will be reviewed for inclusion and exclusion criteria. The subject will undergo a screening evaluation including a history and physical prior to the intervention.

Randomization: All subjects with a recent thrombosed AV graft who meet the inclusion and exclusion criteria and consent to participation in the study will be randomized to either the angioplasty arm or the angioplasty plus stent deployment arm. Patients will be taken to the interventional suite in the Interventional Radiology department. Randomization will occur only after ascertaining that that graft flow has been restored and that there is a >50% stenosis at the venous anastomosis. Randomization will be accomplished by unsealing a sequentially numbered opaque envelope that contains the randomization allocation for that subject.

Angioplasty Arm: These patients will undergo the standard of care protocol: mechanical thrombectomy plus angioplasty of the stenosis at the venous anastomosis.

Stent Deployment Arm: These patients will undergo the same protocol as the angioplasty arm but at the end a covered stent (wallgraft) will be deployed at the stenotic lesion of the venous anastomosis.

Procedures:

Angioplasty Arm: All patients diagnosed with a thrombosed graft will undergo mechanical thrombectomy within 48 hours of diagnosis in conjunction with angioplasty of the underlying stenotic lesion. The grafts are initially accessed with a single needle at the arterial limb of the graft. A glide wire is passed up to the central vessels and the needle exchanged for a 6-French catheter sheath. Mechanical thrombectomy is achieved with a Trerotola device. A second 6-French sheath is placed in the venous limb of the graft, and a glide wire passed into the arterial circulation. A Fogarty balloon is passed through the wire beyond the arterial anastomosis and pulled back to dislodge the clot. An anterograde and retrograde angiogram of the graft is performed to assess patency and to look and grade the stenotic lesions. Lesions at the venous anastomosis with at least 50% stenosis are considered hemodynamically significant. An angioplasty balloon is placed and inflated at the level of the stenotic site.

Stent Arm: The procedure will be identical to that followed in the angioplasty arm. However, after the stenotic lesion has been angioplastied, the appropriate Bard Fluency-cPTFE Encapsulated Nitinol Stent will be deployed.

All patients will receive 3000 to 4000 units of heparin during the procedure.

All patient receive conscious sedation with Fentanyl and Versed, unless allergic to either one

In both randomized groups, a final angiogram of the graft will be performed, and the residual stenosis at the site of the angioplasty quantified. In addition, intra-graft and systemic blood pressures will be measured upon completion of the intervention. These pressures will be measured directly through a disposable pressure transducer. The ratio of graft to systemic systolic pressure will be calculated. This ratio has been previously shown to be predictive of subsequent primary graft patency [1, 10].

Subsequent Intervention for Either Arm: Subsequent graft interventions will be determined by the clinical judgment of the patient's nephrologists, independently of the study intervention. Such intervention may include mechanical thrombectomy with angioplasty if the graft re-thromboses; referral for diagnostic fistulogram with possible angioplasty if there is clinical suspicion o graft stenosis; surgical revision if the graft has a stenosis that is not amenable to radiologic intervention; and placement of a dialysis catheter if graft patency cannot be restored. All these events will be tracked prospectively. Completeness of information about subsequent access interventions will be optimized by a cross-check with the Division of Nephrology prospective, computerized, access database maintained by our Access Coordinators [40].

Endpoints. The primary end point will be the primary patency of the graft (time from the initial thrombectomy to the next graft intervention (angioplasty, thrombectomy, or surgical revision).

The secondary end points are:

1. Secondary patency, (time from thrombectomy to permanent graft failure, regardless of number of subsequent salvage procedures)
2. Total cost of access procedures and access complications per year of followup.

Follow-up Period: Subjects will be followed for a period of 2 years from the date of randomization. Prospective data will be collected on (1) all subsequent access procedures (angioplasty, thrombectomy, surgical revision, or placement of dialysis catheter), (2) all access-related hospitalizations (those due to an access complication or non-access related hospitalizations in which an access procedure is performed), and (3) all access-related complications, e.g., catheter-related bacteremia or metastatic infections.

Statistical Analysis:

Sample size: Power calculations were performed by the statistician (Jill Barker, Ph.D.) on the basis of our preliminary data from the outcome of thrombosed grafts at UAB. The estimated sample of 130 patients, evenly divided between the two groups, was sufficient to detect a tripling in median graft survival from 1 month on the control arm to 3 months on the stent arm, assuming an exponential distribution, 3 years of accrual, 1 year of followup, two-sided significance level of 0.05, and power of 0.80.

Analysis: The statistical analysis will be done in collaboration with Jill Barker, Ph.D. (Consultant). Baseline patient characteristics between the 2 groups will be compared by unpaired Student t-tests or Chi-square analysis, as appropriate. Statistical analysis will be performed on an intent-to-treat basis. Primary and secondary graft survivals curves will be generated by the Kaplan-Meier method. The differences in graft survival between groups will compared by log rank test. In addition, the association between baseline clinical characteristics and graft survival will be analyzed by univariate and multi-variable regression analysis using the Cox proportional hazards model. The costs of all subsequent access-related procedures, hospitalizations, and complications will be determined using Medicare reimbursement rates in Alabama. The total cost of access care between groups will be compared by unpaired Student t-tests.