Comparison of Arteriovenous Grafts in the Thigh and Upper Extremities in Hemodialysis Patients

Materials and Methods

Results

We analyzed the outcomes of all grafts placed in UAB dialysis patients during the 3.5-yr period from January 1, 1999 to June 30, 2002. A total of 409 grafts were placed: 346 in the upper extremities and 63 in the thighs. Thus, thigh grafts accounted for approximately 15% of all grafts placed during the study period. Table 1 summarizes the clinical characteristics of the patients who received grafts. Patients who received a thigh graft had a similar age, race, and gender distribution to those who received upper extremity grafts but were significantly less likely to have diabetes. Not surprising, patients who received a thigh graft had a significantly higher number of previous vascular accesses than those who received an upper extremity graft. Moreover, the mean time on dialysis at the time of access placement was >2 yr longer for thigh grafts than for upper extremity grafts. Approximately one third of patients with an upper extremity graft had had a previous ipsilateral dialysis catheter. The frequency of vascular disease was not significantly different between patients who received thigh grafts and those with upper extremity grafts, although peripheral vascular disease and cerebrovascular disease tended to be less frequent in the former group. Using multiple variable logistic regression, the likelihood of having a thigh graft was independently predicted by only two factors: absence of diabetes (odds ratio, 2.51 [1.36 to 4.66] and years on dialysis (odds ratio, 1.09 [1.02 to 1.15] per year on dialysis).

The graft outcome was indeterminate in 17 cases (approximately 4% of the total) as a result of either early death or transfer to a dialysis unit outside of UAB within 1 wk of graft placement. The remaining 392 grafts were evaluated for subsequent outcomes. Twenty-eight graft placements (6.8% of the total) were technical failures (they could not be placed because of inadequate vessels during the surgical exploration or clotted immediately after the surgical anastomosis was completed). The technical failure rate was approximately twice as high for thigh grafts, as compared with upper extremity grafts (12.7 versus 5.8%; P = 0.046). Seven patients with mild peripheral vascular disease had a thigh graft placed after careful clinical assessment by the surgeon. No patient developed ischemic changes in the limb after thigh graft placement.

The intervention-free survival was similar for thigh and upper extremity grafts (Figure 1A). The median survival was 3.9 mo for thigh grafts and 3.5 mo for upper extremity grafts (hazard ratio, 0.91; 95% confidence interval, 0.67 to 1.25). Thrombosis-free survival was also comparable between the two graft locations (Figure 1B). The median survival was 5.7 mo for thigh grafts and 5.5 mo for upper extremity grafts (hazard ratio, 1.01 [0.72 to 1.43]). Finally, the cumulative survival was similar for thigh grafts and upper extremity grafts (Figure 2A). The median cumulative survival was 14.8 mo for thigh grafts and 20.8 mo for upper extremity grafts (hazard ratio, 1.11 [0.74 to 1.69]). When technical graft failures were excluded, the median cumulative survival was 27.6 mo for thigh grafts and 22.5 mo for upper extremity grafts (hazard ratio, 0.91 [0.56 to 1.49]; Figure 2B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. (A) Intervention-free graft survival comparing patients with thigh grafts with those with upper extremity grafts; P = 0.55 for the comparison between the two survival curves. (B) Thrombosis-free graft survival comparing patients with thigh grafts with those with upper extremity grafts; P = 0.94 for the comparison between the two survival curves. Technical failures and grafts that were never useable for dialysis are included in the analysis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Cumulative graft survival comparing patients with thigh grafts with those with upper extremity grafts, with inclusion of technical failures (A), or after exclusion of technical failures (B); P = 0.62 and 0.72 for the comparison between the two survival curves in the two panels, respectively.

A comparison of the survival curves for thigh and upper extremity grafts was repeated after statistical adjustment for the following factors: age, gender, race, diabetes, peripheral vascular disease, coronary artery disease, cerebrovascular disease, congestive heart failure, years on dialysis, and history of previous ipsilateral dialysis catheter. There was no difference between intervention-free survival, thrombosis-free survival, or cumulative graft survival, whether technical failures were included or excluded (Table 2).

Patients who received thigh grafts were much less likely to have diabetes than those who received upper extremity grafts (Table 1). It is possible that the outcomes of thigh grafts are actually worse than that of upper extremity grafts, but the difference is obscured by the lower prevalence of diabetes among patients who received a thigh graft. To evaluate this possibility, we compared the cumulative graft survival between patients with and without diabetes. The median cumulative survival was 22.5 and 19.4 mo, respectively (P = 0.79).

There were a total of 344 graft-years of follow-up in the present study. To maintain graft patency required a mean of 1.77 interventions, including 0.52 angioplasties, 1.04 thrombectomies, and 0.20 surgical revisions per graft-year. The frequency of salvage procedures was not significantly different between thigh and upper extremity grafts, although thigh grafts tended to undergo fewer angioplasties and more thrombectomies (Table 3).

In the course of the study, 171 grafts failed permanently, 146 (85%) as a result of thrombosis and 25 (15%) as a result of infection. Seven thigh grafts and 18 upper extremity grafts failed as a result of infection. Access loss as a result of infection tended to be higher for thigh grafts than for upper extremity grafts (11.1 versus 5.2%; P = 0.07). The infection rate was 0.13 per patient-year for thigh grafts and 0.07 for upper extremity grafts.

Discussion

We observed similar outcomes of thigh and upper extremity grafts in a well-defined hemodialysis population with prospective tracking of outcomes. This was true, whether one compared intervention-free survival (Figure 1A), thrombosis-free survival (Figure 1B), or cumulative graft survival (Figure 2A) between graft sites. Moreover, the graft outcomes did not differ when technical failures were excluded from the analysis (Figure 2B). Because this was not a randomized study, patients who received a thigh graft differed in several ways from those who received an upper arm graft (Table 1). However, even after adjusting for other demographic and clinical factors in the multivariable model, there was no significant difference in outcomes between thigh and upper extremity grafts (Table 2). Finally, with careful preoperative screening, no patient developed limb ischemia after placement of a thigh graft.

The long-term outcome of thigh grafts has been reported previously only in retrospective studies (Table 4). Cumulative graft survival in the present study was similar to that found by Khadra et al. (11) but lower than that in three other series (7,9,12⇓⇓). In a retrospective comparison, Zibari et al. (8) reported a mean patency of 1.5 yr for thigh grafts, as compared with 1.9 yr for upper extremity grafts. Most of the retrospective studies do not state whether technical failures or graft loss to infection were included in calculation of cumulative survival. Exclusion of these events would suggest an overly optimistic graft outcome. Thus, for example, Tashjian et al. (12) observed an 83% thrombosis-free survival at 2 yr (censoring for loss as a result of infection) and a 76% infection-free survival at 2 yr (censoring for loss as a result of thrombosis). Combining the two events would yield an overall cumulative graft survival of 59% at 2 yr.

Diabetes was less common among patients who received thigh grafts than in those with upper extremity grafts. This difference is likely due to the shorter dialysis patient survival in individuals with than without diabetes (21) and the longer duration on dialysis before a patient would be considered for a thigh graft (Table 1). In other words, fewer dialysis patients with diabetes live long enough to exhaust all options for upper extremity vascular access. Nevertheless, the difference in frequency of diabetes was unlikely to be a confounding variable affecting the relative survival of thigh and upper extremity grafts, given the similar cumulative graft survivals that we observed in patients with and without diabetes.

Both thigh and upper extremity grafts required a high frequency of elective angioplasty, thrombectomy, and surgical revision to maintain their long-term patency for dialysis (Table 3). The major cause of graft failure is thrombosis as a result of occlusion of the venous anastomosis, draining vein, or central vein (22). Myointimal hyperplasia is the initial pathophysiologic event leading to graft stenosis. In the absence of prophylactic measures to prevent myointimal hyperplasia, most grafts develop progressive stenosis. When the stenosis is not detected and corrected in a timely manner, grafts typically clot and require thrombectomy to restore their patency. A number of studies have found that an ongoing surveillance program for graft stenosis and timely intervention with angioplasty or surgical revision can substantially reduce (but not eliminate) the frequency of graft thrombosis (13,23–26⇓⇓⇓⇓). In the present study, thigh grafts tended to have less frequent angioplasties and more frequent thrombectomies than did upper extremity grafts (Table 3). These differences suggest that clinical monitoring for graft stenosis may be less sensitive for thigh grafts than for upper extremity grafts, although this specific question has not been addressed in the medical literature.

Because of their proximity to the groin, thigh grafts might be thought to be more prone to infection than upper extremity grafts. In fact, we found a trend (P = 0.07) of higher risk of infection with thigh grafts. The frequency of thigh graft infection that we observed was in the low end of that observed in several previous retrospective studies (Table 4). In a retrospective comparison, Zibari et al. (8) reported an infection in 12.5% of thigh grafts versus 16.7% of upper extremity grafts.

In conclusion, the performance of thigh grafts is comparable to that of upper extremity grafts. Despite a significantly higher technical failure rate, intervention-free survival, thrombosis-free survival, and cumulative survival were similar between thigh and upper extremity grafts. The thigh graft infection rate of 0.13 per patient-year that we observed is far lower than that reported for tunneled dialysis catheters (0.7 to 2.0 per patient-year, or 2.0 to 5.5 per 1000 patient-days) (27–32⇓⇓⇓⇓⇓). The thigh graft thrombosis rate of 1.58 per patient-year compares favorably to that reported for tunneled dialysis catheters (3.0 to 3.2 per patient-year, or 8.1 to 8.8 per 1000 patient-days) (28,33⇓). The cumulative thigh graft survival rate at 1 yr (62%) is substantially higher than the corresponding 9% rate reported for tunneled dialysis catheters (34). Finally, the median dialysis blood flow delivered by grafts is substantially higher than that obtained with catheters, thereby enhancing dialysis adequacy (6). For all of these reasons, placement of thigh grafts should be considered a viable option among hemodialysis patients who have exhausted all options for a permanent vascular access in both upper extremities and is certainly far preferable to the long-term use of a tunneled dialysis catheter. Because our results reflect the outcomes at a single institution, they may not generalize to all dialysis centers.