- Email: [email protected]
A comparison of patency and interventions in thigh versus Hemodialysis Reliable Outflow grafts for chronic hemodialysis vascular access
From the Southern Association for Vascular Surgery
A comparison of patency and interventions in thigh versus Hemodialysis Reliable Outflow grafts for chronic hemodialysis vascular access Evan R. Brownie, MD,a Clark D. Kensinger, MD,a Irene D. Feurer, PhD,b Derek E. Moore, MD, MPH,a and David Shaffer, MD,a Nashville, Tenn Objective: With improvements in medical management and survival of patients with end-stage renal disease, maintaining durable vascular access is increasingly challenging. This study compared primary, assisted primary, and secondary patency, and procedure-specific complications, and evaluated whether the number of interventions to maintain or restore patency differed between prosthetic femoral-femoral looped inguinal access (thigh) grafts and Hemodialysis Reliable Outflow (HeRO; Hemosphere Inc, Minneapolis, Minn) grafts. Methods: A single-center, retrospective, intention-to-treat analysis was conducted of consecutive thigh and HeRO grafts placed between May 2004 and June 2015. Medical history, interventions to maintain or restore patency, and complications were abstracted from the electronic medical record. Data were analyzed using parametric and nonparametric statistical tests, Kaplan-Meier survival methods, and multivariable proportional hazards regression and logistic regression. Results: Seventy-six (43 thigh, 33 HeRO) grafts were placed in 61 patients (54% male; age 53 [standard deviation, 13] years). Median follow-up time in the intention-to-treat analysis was 21.2 months (min, 0.0; max, 85.3 months) for thigh grafts and 6.7 months (min, 0.0; max, 56.3 months) for HeRO grafts (P [ .02). The groups were comparable for sex, age, coronary artery disease, diabetes mellitus, peripheral vascular disease, and smoking history (all P $ .12). One thigh graft (2%) and five HeRO (15%) grafts failed primarily. In the intention-to-treat analysis, patency durations were significantly longer in the thigh grafts (all log-rank P # .01). Point estimates of primary patency at 6 months, 1 year, and 3 years were 61%, 46%, and 4% for the thigh grafts and 25%, 15%, and 6% for the HeRO grafts. Point estimates of assisted primary patency at 6 months, 1 year, and 3 years were 75%, 66%, and 54% for the thigh grafts and 41%, 30%, and 10% for the HeRO grafts. Point estimates of secondary patency at 6 months, 1 year, and 3 years were 88%, 88%, and 70% for the thigh grafts and 53%, 43%, and 12% for the HeRO grafts. There were no differences in ischemic (P [ .63) or infectious (P [ .79) complications between the groups. Multivariable logistic regression demonstrated that after adjusting for follow-up time, HeRO grafts were associated with an increased number of interventions (P [ .03). Conclusions: Thigh grafts have significantly better primary, assisted primary, and secondary patency compared with HeRO grafts. There is no significant difference between thigh grafts and HeRO grafts in ischemic or infectious complications. Our logistic regression model demonstrated an association between HeRO grafts and an increased number of interventions to maintain or restore patency. Although HeRO grafts may extend the use of the upper extremity, thigh grafts provide a more durable option for chronic hemodialysis. (J Vasc Surg 2016;-:1-8.)
Long-term, durable vascular access for hemodialysis (HD) remains a challenging problem for surgeons and interventionists caring for patients with end-stage renal disease (ESRD). The preferred surgical procedures for HD access, either autogenous vein arteriovenous (AV) fistulas or upper extremity prosthetic AV grafts, unfortunately, From the Department of Surgery,a and the Departments of Surgery and Biostatistics,b Vanderbilt University Medical Center. Author conflict of interest: none. Presented at the Fortieth Annual Meeting of the Southern Association for Vascular Surgery, Cancun, Mexico, January 20-23, 2016. Correspondence: Evan R. Brownie, MD, Department of Surgery, Vanderbilt University Medical Center, Medical Center North, Ste CCC-4312, 1161 21st Ave S, Nashville, TN 37232-2730 (e-mail: evan.r.brownie@ vanderbilt.edu). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2016 by the Society for Vascular Surgery. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jvs.2016.04.055
have limited longevity due to venous stenosis, intimal hyperplasia, aneurysm formation, or infection. Central venous stenosis secondary to central venous catheters can preclude future AV access on the ipsilateral extremity due to inadequate flow rates, thrombosis, or arm swelling. Once upper extremity access sites have been exhausted, surgeons have historically resorted to lower extremity AV accesses, typically in the thigh, using prosthetic grafts or autogenous access with saphenous vein or femoral vein superficialization. A new graft, the Hemodialysis Reliable Outflow (HeRO) graft (Hemosphere Inc, Minneapolis, Minn), became available in 2008 after United States Food and Drug Administration approval for catheter-dependent patients or patients with failed or failing upper arm AV accesses due to central venous stenosis.1 The HeRO graft combines a standard prosthetic graft on the arterial end with a catheter on the venous end, implanted with the catheter extending into the right atrium and bypassing the central venous stenosis. Thus, the surgeon now has two options when faced with the chronic ESRD patient who is no longer a candidate for 1
2 Brownie et al
a traditional upper arm AV access, a thigh AV access, or an upper arm HeRO. Although the Kidney Disease Outcomes Quality Initiative and the Society for Vascular Surgery (SVS) guidelines provide an algorithm and consensus for establishing and placing initial permanent upper extremity HD access, there are no guidelines for the management of patients with central venous stenosis or loss of upper extremity access options who need HD.2,3 Tunneled dialysis catheters are associated with increased risk of infection, thrombosis, and increased death compared with AV fistulas or grafts4,5 and, thus, are not an ideal option for maintaining permanent HD access. At our institution, the two most common alternative HD access procedures performed in patients with central venous stenosis not amenable to percutaneous intervention or otherwise unsuitable upper extremity veins are prosthetic femoral-femoral looped inguinal access (thigh) grafts and HeRO grafts. This study compared patency durations and procedure-specific complication rates and evaluated whether the number of interventions to maintain or restore patency differed between thigh and HeRO grafts. METHODS After Institutional Review Board approval under a “Waiver of Consent and/or Authorization,” a singlecenter, retrospective, intention-to-treat analysis was performed of consecutive thigh and HeRO grafts placed between May 2004 and June 2015. To ensure the most potentially representative consecutive sample, grafts that were placed in a given patient after a previous thigh or HeRO graft were specifically included; whether a previous thigh or HeRO graft had been placed was addressed in the statistical analyses. All HeRO grafts were placed during or after 2010. Grafts were identified by a surgeon’s query for Current Procedural Terminology (American Medical Association, Chicago, Ill) codes during the study period. The access modality was selected by the operative surgeon in accordance with Kidney Disease Outcomes Quality Initiative guidelines. The electronic medical record was examined for patient demographics and comorbidities before the index procedure, postprocedural complications, and subsequent operative and endovascular interventions to restore or maintain access patency. All thigh grafts were placed in a loop (femoral arteryto-femoral vein) configuration after a preoperative evaluation to exclude significant lower extremity peripheral arterial disease that included a history to evaluate for claudication and a physical examination evaluating lower extremity pulses. Adjunctive imaging, at the surgeon’s discretion, included any or all of lower extremity arterial duplex ultrasound imaging, ankle-brachial indices, or lower extremity computed tomography angiography. Polytetrafluoroethylene (PTFE) grafts ranging from 6 to 8 mm, at the surgeon’s discretion and depending on the size of the recipient’s vessels, were used as a conduit. HeRO grafts were inserted as previously described.6 In brief, access to the right atrium is obtained in the
JOURNAL OF VASCULAR SURGERY --- 2016
interventional radiology suite by direct central access or chest wall access. The patient is then transported to the operating room with an occlusive dressing over the venous puncture site to secure the placeholder catheter. This catheter is then exchanged for the 5-mm nitinol-reinforced silicone outflow catheter of the HeRO device. An anastomosis is created from the ipsilateral brachial or axillary artery to the 6-mm expanded PTFE inflow component of the HeRO device. The inflow and outflow components of the HeRO device are then tunneled subcutaneously and coupled. Demographic data included age at first or only procedure and gender. Clinical data included the key comorbidities of diabetes mellitus, tobacco use, cerebrovascular disease, coronary artery disease, peripheral arterial disease, and documented central venous stenosis. Additional variables were whether the patient had undergone a prior attempt at peritoneal dialysis, whether listed for kidney transplant, time on dialysis, and the number of prior attempts at permanent HD access. Primary outcomes were primary, assisted primary, and secondary patency, as described by SVS recommended standards for reporting.7 Secondary outcomes were the number of interventions to maintain or restore access patency, ischemic, and infectious complications. Procedures to maintain or restore access patency included surgical thrombectomy, angiography with intended intervention, percutaneous thrombolysis, and balloon angioplasty, stenting, or operative revision. Ischemic complications were defined as steal syndrome, ischemic rest pain, tissue loss, or any of the aforementioned warranting abandonment of the access. Infectious complications included documented graft infection based on admission or discharge diagnosis, details of operative notes and postmortem analysis, persistent bacteremia attributed to graft infection, or any of the aforementioned warranting abandonment of the access. Parametric and nonparametric tests were used for between-group comparisons as appropriate for a particular measure. Kaplan-Meir survival analyses with log-rank tests were used to compare primary, assisted primary, and secondary patency between thigh and HeRO grafts in the intention-to treat (three analyses/end points) and in the sample that excluded grafts with initial failure (three additional analyses/end points). Data encoding for the survival analyses adhered to SVS recommendations. Specifically, primary patency reflected intervention-free graft survival. Events were the first/only intervention to maintain or restore patency, thrombosis (if no intervention was performed), and failure for any reason (if no intervention was performed). Observations without interventions were censored on the date of death, transplantation, or final event-free follow-up. Assisted primary patency reflected thrombosis-free graft survival. Events were the first/only thrombosis or graft failure (with or without thrombosis), and observations were censored as in the primary patency model. Secondary patency reflected graft survival to the time of abandonment. Events were graft failures, and observations were censored as in the primary and
JOURNAL OF VASCULAR SURGERY Volume -, Number -
Brownie et al 3
Table I. Demographic and clinical characteristics Variablea
Table II. Access outcomes
Thigh graft HeRO graft P (n ¼ 43) (n ¼ 33) value
Age at index graft, years 52.2 Male gender, No. (%) 21 Diabetes mellitus, No. (%) 10 Tobacco use, No. (%) 13 Coronary artery disease, 8 No. (%) Peripheral vascular disease, 11 No. (%) Central venous stenosis, No. (%) 14 Prior peritoneal dialysis attempt, 14 No. (%) Listed for kidney transplant, 6 No. (%) Time on dialysis before index 60 procedure, months Prior HD access attempts, No. 3
(13.9) (49) (23) (30) (19)
53.3 22 11 14 12
(13.0) (67) (33) (2) (36)
.74 .16 .44 .34 .12
(26)
8 (24)
1.00
(33) (33)
33 (100) 11 (33)
<.01 1.00
(14)
4 (12)
1.00
(0, 256) 95 (2, 255) .16 (0, 6)
3 (1, 6)
.91
HD, Hemodialysis; HeRO, Hemodialysis Reliable Outflow; thigh, prosthetic femoral-femoral looped inguinal access. a Continuous data are shown as mean (standard deviation) or as median (min, max), and categoric data are shown as indicated.
assisted primary patency models. Intention-to-treat analyses included all grafts. Because the sample included consecutive grafts, some in patients with a history of previous thigh or HeRO grafts, the presence of a previous thigh or HeRO graft was adjusted for in six parallel multivariable Cox proportional hazards regression models (three intention-to-treat and three that excluded initial failures). These analyses tested the likelihood of graft failure using the three patency end points in HeRO grafts (reference thigh) after adjusting for whether a previous thigh or HeRO graft had been placed (reference no previous thigh or HeRO graft). The number of interventions to maintain or restore patency was calculated as the number per graft per year and as the total number of interventions per graft. Between-group comparisons of numbers of interventions were performed for those grafts that did not fail initially. Two statistical approaches were used: (1) the median test compared the number of interventions per graft per year, and (2) multivariable logistic regression tested whether an increased number of interventions per graft, after adjusting for follow-up time in months, were associated with graft type (coded HeRO, reference thigh). Data were analyzed using SPSS 22 statistical software (BM Corp, Armonk, NY), and nondirectional Type I error probabilities of <.05 were considered statistically significant. RESULTS Seventy-six (43 thigh, 33 HeRO) grafts were placed in 61patients (54% male; age 53 [standard deviation] 13 years); of which, 22 were placed after a previous thigh (15 grafts) or HeRO (seven grafts) graft. All of the HeRO grafts were placed in the upper extremities, with
Variablea
Thigh graft (n ¼ 43)
HeRO graft (n ¼ 33)
P value
Follow-up time, 21.2 (0.0, 85.3) 6.7 (0.0, 56.3) months 22.9 (<1.0, 85.3) 10.3 (<1.0, 56.3) Follow-up time,b months Ischemic 3 (7) 1 (3) complications Infectious 11 (26) 7 (21) complications Interventions per 0.6 (0.0, 8.5) 1.5 (0.0, 16.0) year to maintain or restore patency, No.b
.02 .05 .63 .79 .09
HeRO, Hemodialysis Reliable Outflow; thigh, prosthetic femoral-femoral looped inguinal access. a Continuous data are shown as median (min, max) and categoric variables as frequency (%). b Includes 42 thigh and 28 HeRO grafts that did not fail initially.
right atrial access being successfully achieved in 33 of the HeRO grafts (100%) in the interventional radiology suite. Central access for 14 of the 33 HeRO grafts (43%) was gained through chest wall collaterals. Five HeRO grafts (15%) had initial failures. Two of these resulted from an inability to exchange and upsize a placeholder catheter located in a chest wall collateral vein for the 19F sheath introducer for the outflow component of the HeRO graft due to significant central venous thrombosis incompatible with stiff wire retrograde dilation of the access site. One HeRO graft failed to achieve significant flow and was eventually excised, one patient sustained an intraoperative air embolus and the graft was never used, and one patient was lost to follow-up without any documentation of the graft being accessed. One thigh graft (2%) was an initial failure because the patient experienced immediate foot ischemia after the procedure, warranting graft excision the same day. Demographic and clinical data are summarized in Table I. The groups were comparable for sex, age, coronary artery disease, diabetes mellitus, peripheral vascular disease, and smoking history (all P $ .12). Central venous stenosis was documented by venogram before the procedure in a significantly greater proportion of HeRO grafts compared with thigh grafts (100% vs 33%; P < .01). The median time on HD before the index graft placement did not differ between the groups (P ¼ 1.00). Follow-up time, complications, and the number of interventions are summarized in Table II. Grafts deemed initial failures were included in the intention-to-treat analyses and had time-to-event follow-up durations of 0 months. Median follow-up time for secondary patency was longer in the thigh grafts (21.2 vs 6.7 months; P ¼ .02). The three separate intention-to-treat Kaplan-Meier analyses demonstrated that primary (Fig 1), assisted primary (Fig 2), and secondary patency (Fig 3) were significantly longer in the thigh grafts (all P # .01 by log-rank).
4 Brownie et al
Fig 1. Kaplan-Meier primary patency of prosthetic femoralfemoral looped inguinal access (thigh) grafts (n ¼ 43) and Hemodialysis Reliable Outflow (HeRO) grafts (n ¼ 33). Primary patency was significantly longer in thigh grafts through 3 years, with all standard errors (SE) of <10.0%.
This is reflected by the percentage survival point estimates and their standard errors that are summarized in Table III and by the corresponding observed mean patency times. Primary patency at 6 months, 1 year, and 3 years was 61%, 46%, and 4% for the thigh grafts and 25%, 15%, and 6% for the HeRO grafts, and all standard errors of these estimates were <10%. In addition, point estimates of assisted primary and secondary patency at 6 months, 1 year, and 3 years were consistently higher in the thigh grafts, and all standard errors were <10% (Table III). Mean primary patency times (95% confidence interval [CI]) were 15.5 months (10.7-20.2 months) and 7.0 months (3.4-10.6 months), assisted primary patency averaged 39.0 months (27.3-50.7 months) and 10.5 months (6.0-15.0 months), and secondary patency averaged 53.3 months (41.5-65.1 months) and 16.1 months (9.3-22.8 months) for thigh and HeRO grafts, respectively. In the intention-to treat sample, multivariable Cox proportional hazards regression demonstrated that, after adjusting for whether a previous thigh or HeRO graft had been placed (hazard ratio [HR], 1.95; 95% CI, 1.07-3.59; P ¼ .03), primary patency failure was 2.6times more likely in HeRO grafts than in thigh grafts (HR, 2.57; 95% CI, 1.46-4.52; P < .01). Similarly, after adjusting for whether a previous thigh or HeRO graft had been placed (P $ .11), assisted primary patency failure was 4.0-times more likely (HR, 4.04; 95% CI, 2.10-7.78; P < .01), and secondary patency failure was 6.5-times more likely (HR, 6.47; 95% CI, 2.991-3.97; P < .01) in HeRO grafts than in thigh grafts.
JOURNAL OF VASCULAR SURGERY --- 2016
Fig 2. Kaplan-Meier assisted primary patency of prosthetic femoral-femoral looped inguinal access (thigh) grafts (n ¼ 43) and Hemodialysis Reliable Outflow (HeRO) grafts (n ¼ 33). Assisted primary patency was significantly longer in thigh grafts, with standard errors (SE) of <10.0% through 50 months. The vertical reference line indicates the point where the SEs transitioned to being #10.5% in thigh grafts.
The three additional Kaplan-Meier models that excluded the six grafts with initial failure were consistent with the intention-to-treat analyses in that patency durations were significantly longer in the thigh grafts (all P < .05 by log-rank). Primary patency at 6 months, 1 year, and 3 years was 62%, 47%, and 4% for thigh grafts and 30%, 18%, and 7% for HeRO grafts, respectively, and all standard errors for these estimates were #10% (Table III). Consistent with the intention-to-treat analyses, point estimates of assisted primary and secondary patency at 6 months, 1 year, and 3 years were, with the exception of the 3-year primary patency estimates, also generally higher in thigh grafts (Table III). Mean (95% CI) primary patency for thigh and HeRO grafts were 15.8 (11.0-20.6) and 8.2 (4.1-12.3) months, assisted primary patency averaged 39.9 (28.1-51.7) and 12.4 (7.4-17.4) months, and secondary patency averaged 54.5 (42.8-66.4) and 18.9 (11.4-26.4), respectively. In the sample that excluded primary graft failures, multivariable Cox proportional hazards regression demonstrated that after adjusting for whether a previous thigh or HeRO graft had been placed (HR, 1.94; 95% CI, 1.01-3.71; P ¼ .05), primary patency failure was 2.3times more likely in HeRO grafts than in thigh grafts (HR, 2.35; 95% CI, 1.29-4.28; P < .01). Similarly, after adjusting for whether a previous thigh or HeRO graft had been placed (P $ .14), assisted primary patency failure was 3.8-times more likely (HR, 3.78; 95% CI, 1.88-7.60;
JOURNAL OF VASCULAR SURGERY Volume -, Number -
Brownie et al 5
Table II). However, the multivariable logistic regression model, which adjusted for follow-up time and used the metric of total interventions per graft, demonstrated that HeRO grafts were associated with an increased number of interventions to maintain or restore patency. Specifically, the odds ratio for follow-up time in months (0.933; 95% CI, 0.892-0.977; P < .01) reflects the shorter observed total follow-up time for the HeRO grafts. Then, the odds ratio for the number of interventions per graft (1.224; 95% CI, 1.017-1.473; P ¼ .03) indicated that after adjusting for between-group differences in follow-up time, each additional intervention to maintain or restore patency for a particular graft was 1.2-times more likely to be associated with a HeRO graft than with a thigh graft. DISCUSSION
Fig 3. Kaplan-Meier secondary patency of prosthetic femoralfemoral looped inguinal access (thigh) grafts (n ¼ 43) and Hemodialysis Reliable Outflow (HeRO) grafts (n ¼ 33). Secondary patency was significantly longer in thigh grafts, with standard errors (SE) of <10.0% through 41 months. The vertical reference line indicates the point where SEs transitioned to being >10.0% in thigh grafts.
P < .01) and secondary patency failure was 6.4-times more likely (HR, 6.41; 95% CI, 2.79-14.71; P < .01) in HeRO grafts than in thigh grafts. Ischemic complication occurred in three thigh grafts (7%). As mentioned, one patient with steal syndrome in the recovery room underwent immediate resection. Steal syndrome in the second patient improved after graft revision from an 8-mm to 6-mm PTFE graft, and the third patient had a nonhealing ulcer of a pre-existing transtibial amputation stump, but the graft was left in place. Only one patient (3%) of the HeRO group had an ischemic complication, and his graft was removed for steal phenomenon. Overall, there was no difference in ischemic (P ¼ .63) complications between the groups. The number of infectious complications was similar (P ¼ .79) between the thigh and HeRO groups. An associated infection occurred in 11 of 43 thigh grafts (26%), nine of which were resected, one required operative revision, and the final graft was managed with antibiotics for persistent bacteremia attributed to the graft. An infectious complication occurred in seven of 33 grafts in the HeRO group (21%), and five required operative excision. In addition, one patient in the HeRO cohort required operative revision of a bleeding pseudoaneurysm, but the graft remained patent. Among grafts without initial failure, the median number of procedures per graft per year to maintain or restore patency was 0.6 in the thigh group and 1.5 in the HeRO group and did not differ between the groups (P ¼ .22;
With the increasing prevalence of ESRD and longer patient survival on HD, maintaining long-term dialysis access presents an increasing challenge to vascular surgeons. Limited data are available to guide the vascular surgeon on the choice of AV access procedures in patients who have exhausted traditional upper extremity AV fistula or graft options. We reviewed a 10-year experience at a single center to compare outcomes between thigh AV grafts and upper extremity HeRO grafts in this difficult patient population. Several reports have reviewed thigh and HeRO grafts independently. Katzman et al1 initially reported 36 HeRO devices averaging 8.4 months’ follow-up and noted primary and secondary patency at 8 months was 39% and 72%, and an average of 2.5 procedures per year were required to maintain patency. In a larger study of 164 patients from four centers, Gage et al8 reported 1-year primary and secondary patency rates of 41% and 91%, and 1.5 procedures per year were required to maintain patency. Cull et al9 reported similar results in 125 lower extremity AV grafts over a mean of 20 months, with 1-year primary and secondary patency of 34% and 68% and 1.68 interventions per patient per year. Two prior studies directly compared thigh and HeRO grafts. Steerman et al10 was the first and noted 1-year primary and secondary patency of 51% and 60% for thigh grafts compared with 15% and 57% for HeRO grafts, which was not statistically significant. The principal difference observed was significantly fewer interventions to maintain access patency of lower extremity grafts, 2.21 per year in the HeRO and 1.17 per year in the thigh group. The results of Kudlaty et al11 differed, with similar short-term patency rates for the two procedures and no statistically significant difference in the mean number of interventions to maintain patency, 1.1 vs 1.65 per year, in thigh and HeRO grafts. In contrast to previous reports, we found significantly longer primary patency, assisted primary patency, and secondary patency in thigh grafts. In addition, after adjusting for follow-up time, we observed an association between HeRO grafts and an increased number of interventions
JOURNAL OF VASCULAR SURGERY --- 2016
6 Brownie et al
Table III. Kaplan-Meier survival point estimates for Hemodialysis Reliable Outflow (HeRO) and prosthetic femoral-femoral looped inguinal access (thigh) grafts by sample and outcome 6 months Sample and outcome measurea Intention-to-treat (n ¼ 76) Primary patency Assisted primary patency Secondary patency Exclude initial failures (n ¼ 70) Primary patency Assisted primary patency Secondary patency
1 year
3 years
HeRO
Thigh
HeRO
Thigh
HeRO
Thigh
25 (8) 41 (9) 53 (9)
61 (8) 75 (7) 88 (5)
15 (6) 30 (8) 43 (9)
46 (8) 66 (8) 88 (5)
6 (5) 10 (7) 12 (7)
4 (4) 54 (9) 70 (9)
30 (9) 48 (10) 63 (9)
62 (8) 77 (7) 90 (5)
18 (8) 35 (10) 50 (10)
47 (9) 68 (8) 90 (5)
7 (6) 12 (8) 14 (9)
4 (4) 55 (9) 72 (9)
a
Survival estimates are presented as percentages (standard error of the estimate).
to maintain or restore patency. Our study does not address why the thigh grafts remained patent so much longer than the HeRO graft. One possibility is the larger size of the femoral vessels, particularly the venous outflow. It is noteworthy that 43% of the patients in the HeRO cohort obtained central venous access through a chest wall collateral and not direct cannulation of the central venous system, which may partially account for the lower HeRO patency rates we observed. This practice is suggested by Gage et al8 in 2.2% of patients but is not thoroughly described and reflects the comfort of our interventional radiologists with this technique, allowing patients who would otherwise be relegated to catheter dependence an attempt at alternative HD access. Our observation that, after adjusting for follow-up time, HeRO grafts were associated with an increased number of interventions to maintain or restore patency is consistent with Steerman et al10 but differs from the findings of Kudlaty et al.11 Although not statistically significant, our data demonstrated that a median of 0.6 interventions were performed per year to maintain or restore patency of thigh grafts compared with 1.5 interventions per year for HeRO grafts. Additional costs associated with the HeRO graft include the increased cost of the HeRO device itself, fluoroscopy equipment and personnel, and potentially increased operating room time. One decision analysis model that reviewed previously published data showed that the HeRO device is the least costly in patients at the end-stage of dialysis access.12 As health care costs become an increasing concern, the number of interventions to maintain patency over time may function as a surrogate for resource utilization or cost. Although our data show an association between the number of interventions to maintain patency and HeRO grafts, further study will be needed to evaluate the comparative costeffectiveness of these procedures. Historically, a key concern with using the groin as a site for HD access has been the risk of infection. Previous studies of thigh accesses reported infection rates from 9% to 41%,9,13 but the differences of thigh and HeRO grafts in our study and the two prior studies10,11 were not
statistically significant. In addition, we found no difference in ischemic complications. Limb ischemia or steal syndrome were identified in only three patients (7%) in our thigh graft cohort despite 23% of our patients having diabetes mellitus and the occasional use of 8-mm PTFE grafts. This is comparable to the 5% ischemic complication rate reported by the two prior studies comparing thigh and HeRO grafts,10,11 suggesting the utility of an appropriate preoperative screening for peripheral vascular disease. Therefore, our findings do not favor one procedure or the other based on the risk of infectious or ischemic complications alone. The strengths of our study include a relatively larger sample and longer-term follow-up that enabled statistically stable patency estimates through at least 3 years. Inclusion of all sequential HeRO and thigh grafts, regardless of whether a previous graft of either type had been placed, coupled with statistical adjustment for this consideration, allowed for greater statistical power and potentially enhances the generalizability of our findings. Limitations of our study include its retrospective, nonrandomized, single-center design. In addition, our patients reflect a heterogeneous population, including those with central venous stenosis and those with patent central veins who were otherwise unsuitable or in whom upper extremity vascular sites had been exhausted. CONCLUSIONS Maintaining long-term, durable, cost-effective vascular access for HD in patients with central venous stenosis or who have otherwise exhausted upper extremity sites remains a challenge. In a single-center, intention-to-treat analysis during a 10-year period, we found that thigh grafts have significantly longer primary, assisted primary, and secondary patency and that HeRO grafts are associated with an increased number of interventions to maintain or restore patency. We found no significant difference in ischemic or infectious complications between thigh and HeRO grafts. Although HeRO grafts may extend the short-term use of the upper extremity, thigh grafts remain a more durable option for chronic HD.
JOURNAL OF VASCULAR SURGERY Volume -, Number -
AUTHOR CONTRIBUTIONS Conception and design: EB, IF, DM, DS Analysis and interpretation: EB, IF, DM, DS Data collection: EB, CK Writing the article: EB, IF, DS Critical revision of the article: EB, CK, IF, DM, DS Final approval of the article: EB, CK, IF, DM, DS Statistical analysis: IF Obtained funding: Not applicable Overall responsibility: EB
REFERENCES 1. Katzman HE, McLafferty RB, Ross JR, Glickman MH, Peden EK, Lawson JH. Initial experience and outcome of a new hemodialysis access device for catheter-dependent patients. J Vasc Surg 2009;5: 600-7. 2. Vascular Access 2006 Work Group. Clinical practice guidelines for vascular access. Am J Kidney Dis 2006;48(Suppl 1):S176-247. 3. Sidawy AN, Spergel LM, Besarab A, Allon M, Jennings WC, Padberg FT Jr, et al. The Society for Vascular Surgery: clinical practice guidelines for the surgical placement and maintenance of arteriovenous hemodialysis access. J Vasc Surg 2008;48(5 Suppl):S2-25. 4. Collins AJ, Foley RN, Herzog C, Chavers BM, Gilbertson D, Ishani A, et al. US Renal Data System 2010 Annual Data Report. Am J Kidney Dis 2010;57(1 Suppl 1):A8.
Brownie et al 7
5. Allon M, Daugirdas J, Depner TA, Greene T, Ornt D, Schwab SJ. Effect of change in vascular access on patient mortality in hemodialysis patients. Am J Kidney Dis 2006;47:469-77. 6. Kensinger C, Brownie E, Bream P Jr, Moore D. Multidisciplinary team approach to end-stage dialysis access patients. J Surg Res 2015;199: 259-65. 7. Sidawy AN, Gray R, Besarab A, Henry M, Ascher E, Silva M Jr, et al. Recommended standards for reports dealing with arteriovenous hemodialysis accesses. J Vasc Surg 2002;35:603-10. 8. Gage SM, Katzman HE, Ross JR, Hohmann SE, Sharpe CA, Butterly DW, et al. Multi-center experience of 164 consecutive hemodialysis reliable outflow [HeRO] graft implants for hemodialysis treatment. Eur J Vasc Endovasc Surg 2012;44:93-9. 9. Cull JD, Cull DL, Taylow SM, Carsten CG 3rd, Snyder BA, Yourkey JR, et al. Prosthetic thigh arteriovenous access: outcome with SVS/AAVS reporting standards. J Vasc Surg 2004;39:381-6. 10. Steerman SN, Wagner J, Higgins JA, Kim C, Mirza A, Pavela J, et al. Outcomes com- parison of HeRO and lower extremity arteriovenous grafts in patients with long-standing renal failure. J Vasc Surg 2013;57:776e83. 11. Kudlaty EA, Pan J, Allemang MT, Kendrick DE, Kashyap VS, Wong VL, et al. The end stage of dialysis access: femoral graft or HeRO vascular access device. Ann Vasc Surg 2015;29:90-7. 12. Dageforde LA, Bream PR, Moore DE. Hemodialysis Reliable Outflow (HeRO) device in end-stage dialysis access: a decision analysis model. J Surg Res 2012;177:165e71. 13. Miller CD, Robbin ML, Barker J, Allon M. Comparison of arteriovenous grafts in the thigh and upper extremities in hemodialysis patients. J Am Soc Nephrol 2003;14:2942-7. Submitted Jan 14, 2016; accepted Apr 28, 2016.
DISCUSSION Dr Thomas Huber (Gainesville, Fla). The authors have retrospectively reviewed their experience with thigh prosthetic hemodialysis accesses and the Hemodialysis Reliable Outflow (HeRO; Hemosphere Inc, Minneapolis, Minn) device in the upper extremity. They reported that the patency rates were superior in the thigh group and that the infections and ischemic complications were similar, although there was an association with increased number of interventions for the HeRO group in their multivariate model. Notably, the secondary or cumulative patency rates at 1 year were 88% in the thigh group and 43% in the HeRO group. Based upon these findings, the authors concluded that the thigh accesses were a more durable option. Although I don’t disagree with the authors’ results or conclusion, the results must be interpreted with some caution given the retrospective study design. The patency rates reported for the HeRO device were comparable or better than those reported in the literature, while those for the thigh access were comparable or better. The infectious complication rates were not significantly different between the groups although they were quite “significant,” with 15% of the HeRO devices and 21% of the thigh accesses requiring explant. Furthermore, I am not certain that the thigh accesses and HeRO devices are really competing strategies, but perhaps should be viewed as complementary in the expanding groups of complex patients with central vein stenoses or occlusions. I have three questions or requests for the authors: First, please comment on the infectious risk of both the thigh accesses and HeRO devices. Our overall experience in terms of the infectious risk has been comparable or worse and we have really tried to limit both approaches. It seems that every HeRO device we have implanted has needed to be removed. Furthermore, other authors have proposed that the infectious complication rates of the thigh accesses are higher than tunneled dialysis catheters in the upper extremity and should be avoided.
Second, please detail your surveillance protocol for the thigh accesses given your excellent patency rates. Third, please outline your current treatment algorithm for patients with severe upper extremity central vein occlusions or stenoses that preclude the traditional autogenous or prosthetic options. I would like to congratulate the authors on their presentation and contribution to the literature and thank them for sending me a copy of the manuscript well in advance of the meeting. Thanks you. Dr Evan R. Brownie. Thank you, Dr Huber, for your comments and questions, as well as your contributions to the literature on this topic. First of all, I have to agree with you that these are complimentary strategies and not necessarily competing strategies, but the facts remain that hemodialysis centers, nephrologists, and surgeons are under a lot of pressure to minimize long-term catheter use. Dialysis units may now receive financial penalties for failing to meet benchmarks for tunneled catheter rates. So if a patient is not a peritoneal dialysis candidate, we are left with the two options of either a thigh or HeRO graft. One of the purposes in designing this study was to assess the outcomes as well as to guide the surgeon on an informed discussion and consent with the patient prior to proceeding. With regard to your first question, I agree that the infection rate is high, but it is similar to previously described studies comparing both thigh grafts and the HeRO catheter as well as thigh to upper extremity catheters. Again, we fall back to that pressure that nephrologists and dialysis centers are really pushing to avoid tunnel dialysis catheters. Anecdotally, our tunneled catheter infection rates are probably a little higher than both the lower extremity access and the HeRO groups, and this has been born out in two prior comparisons of thigh and HeRO groups. I do not have a specific timeline of the infections in terms of when the thigh grafts or when the HeRO grafts had an infectious issue, but on a cursory review of our study it does not appear to be any different.
8 Brownie et al
Secondly, in regard to the surveillance for our thigh grafts, we followed the same surveillance protocol that we do for our upper extremity access. We do monthly transonic measurements of inline flow. If it falls to <600 mL/min or <25% of the patient’s baseline, then they are referred to our interventional radiologist for a fistulogram. If there is a stenosis present at that time, an angioplasty is performed. If that stenosis is not amenable to any percutaneous intervention or it is recurrent and they are undergoing multiple angioplasties, then the patient is then evaluated and referred for surgical revision. With the thigh grafts, we try to discourage stents at the venous anastomosis as this makes revision more difficult in the future.
JOURNAL OF VASCULAR SURGERY --- 2016
Finally, to answer your question regarding central venous stenosis and the algorithm we used, the first thing we obtain is a computed tomography angiography/computed tomography venography of the chest and the bilateral upper extremities. If the patient has adequate arterial inflow, then our first option to the patient is really a HeRO graft so long as they are a candidate and interested in that procedure. We discuss the risks and benefits of both, but again, one of the points of the study was to really be able to educate the patient that maybe one procedure lasts longer but requires more interventions along the way. If the patient is not a candidate for either, then we just determine if they are going to have to rely on a catheter for the remainder of their time.
A comparison of patency and interventions in thigh versus Hemodialysis Reliable Outflow grafts for chronic hemodialysis vascular access Evan R. Brownie, MD,a Clark D. Kensinger, MD,a Irene D. Feurer, PhD,b Derek E. Moore, MD, MPH,a and David Shaffer, MD,a Nashville, Tenn Objective: With improvements in medical management and survival of patients with end-stage renal disease, maintaining durable vascular access is increasingly challenging. This study compared primary, assisted primary, and secondary patency, and procedure-specific complications, and evaluated whether the number of interventions to maintain or restore patency differed between prosthetic femoral-femoral looped inguinal access (thigh) grafts and Hemodialysis Reliable Outflow (HeRO; Hemosphere Inc, Minneapolis, Minn) grafts. Methods: A single-center, retrospective, intention-to-treat analysis was conducted of consecutive thigh and HeRO grafts placed between May 2004 and June 2015. Medical history, interventions to maintain or restore patency, and complications were abstracted from the electronic medical record. Data were analyzed using parametric and nonparametric statistical tests, Kaplan-Meier survival methods, and multivariable proportional hazards regression and logistic regression. Results: Seventy-six (43 thigh, 33 HeRO) grafts were placed in 61 patients (54% male; age 53 [standard deviation, 13] years). Median follow-up time in the intention-to-treat analysis was 21.2 months (min, 0.0; max, 85.3 months) for thigh grafts and 6.7 months (min, 0.0; max, 56.3 months) for HeRO grafts (P [ .02). The groups were comparable for sex, age, coronary artery disease, diabetes mellitus, peripheral vascular disease, and smoking history (all P $ .12). One thigh graft (2%) and five HeRO (15%) grafts failed primarily. In the intention-to-treat analysis, patency durations were significantly longer in the thigh grafts (all log-rank P # .01). Point estimates of primary patency at 6 months, 1 year, and 3 years were 61%, 46%, and 4% for the thigh grafts and 25%, 15%, and 6% for the HeRO grafts. Point estimates of assisted primary patency at 6 months, 1 year, and 3 years were 75%, 66%, and 54% for the thigh grafts and 41%, 30%, and 10% for the HeRO grafts. Point estimates of secondary patency at 6 months, 1 year, and 3 years were 88%, 88%, and 70% for the thigh grafts and 53%, 43%, and 12% for the HeRO grafts. There were no differences in ischemic (P [ .63) or infectious (P [ .79) complications between the groups. Multivariable logistic regression demonstrated that after adjusting for follow-up time, HeRO grafts were associated with an increased number of interventions (P [ .03). Conclusions: Thigh grafts have significantly better primary, assisted primary, and secondary patency compared with HeRO grafts. There is no significant difference between thigh grafts and HeRO grafts in ischemic or infectious complications. Our logistic regression model demonstrated an association between HeRO grafts and an increased number of interventions to maintain or restore patency. Although HeRO grafts may extend the use of the upper extremity, thigh grafts provide a more durable option for chronic hemodialysis. (J Vasc Surg 2016;-:1-8.)
Long-term, durable vascular access for hemodialysis (HD) remains a challenging problem for surgeons and interventionists caring for patients with end-stage renal disease (ESRD). The preferred surgical procedures for HD access, either autogenous vein arteriovenous (AV) fistulas or upper extremity prosthetic AV grafts, unfortunately, From the Department of Surgery,a and the Departments of Surgery and Biostatistics,b Vanderbilt University Medical Center. Author conflict of interest: none. Presented at the Fortieth Annual Meeting of the Southern Association for Vascular Surgery, Cancun, Mexico, January 20-23, 2016. Correspondence: Evan R. Brownie, MD, Department of Surgery, Vanderbilt University Medical Center, Medical Center North, Ste CCC-4312, 1161 21st Ave S, Nashville, TN 37232-2730 (e-mail: evan.r.brownie@ vanderbilt.edu). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest. 0741-5214 Copyright Ó 2016 by the Society for Vascular Surgery. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jvs.2016.04.055
have limited longevity due to venous stenosis, intimal hyperplasia, aneurysm formation, or infection. Central venous stenosis secondary to central venous catheters can preclude future AV access on the ipsilateral extremity due to inadequate flow rates, thrombosis, or arm swelling. Once upper extremity access sites have been exhausted, surgeons have historically resorted to lower extremity AV accesses, typically in the thigh, using prosthetic grafts or autogenous access with saphenous vein or femoral vein superficialization. A new graft, the Hemodialysis Reliable Outflow (HeRO) graft (Hemosphere Inc, Minneapolis, Minn), became available in 2008 after United States Food and Drug Administration approval for catheter-dependent patients or patients with failed or failing upper arm AV accesses due to central venous stenosis.1 The HeRO graft combines a standard prosthetic graft on the arterial end with a catheter on the venous end, implanted with the catheter extending into the right atrium and bypassing the central venous stenosis. Thus, the surgeon now has two options when faced with the chronic ESRD patient who is no longer a candidate for 1
2 Brownie et al
a traditional upper arm AV access, a thigh AV access, or an upper arm HeRO. Although the Kidney Disease Outcomes Quality Initiative and the Society for Vascular Surgery (SVS) guidelines provide an algorithm and consensus for establishing and placing initial permanent upper extremity HD access, there are no guidelines for the management of patients with central venous stenosis or loss of upper extremity access options who need HD.2,3 Tunneled dialysis catheters are associated with increased risk of infection, thrombosis, and increased death compared with AV fistulas or grafts4,5 and, thus, are not an ideal option for maintaining permanent HD access. At our institution, the two most common alternative HD access procedures performed in patients with central venous stenosis not amenable to percutaneous intervention or otherwise unsuitable upper extremity veins are prosthetic femoral-femoral looped inguinal access (thigh) grafts and HeRO grafts. This study compared patency durations and procedure-specific complication rates and evaluated whether the number of interventions to maintain or restore patency differed between thigh and HeRO grafts. METHODS After Institutional Review Board approval under a “Waiver of Consent and/or Authorization,” a singlecenter, retrospective, intention-to-treat analysis was performed of consecutive thigh and HeRO grafts placed between May 2004 and June 2015. To ensure the most potentially representative consecutive sample, grafts that were placed in a given patient after a previous thigh or HeRO graft were specifically included; whether a previous thigh or HeRO graft had been placed was addressed in the statistical analyses. All HeRO grafts were placed during or after 2010. Grafts were identified by a surgeon’s query for Current Procedural Terminology (American Medical Association, Chicago, Ill) codes during the study period. The access modality was selected by the operative surgeon in accordance with Kidney Disease Outcomes Quality Initiative guidelines. The electronic medical record was examined for patient demographics and comorbidities before the index procedure, postprocedural complications, and subsequent operative and endovascular interventions to restore or maintain access patency. All thigh grafts were placed in a loop (femoral arteryto-femoral vein) configuration after a preoperative evaluation to exclude significant lower extremity peripheral arterial disease that included a history to evaluate for claudication and a physical examination evaluating lower extremity pulses. Adjunctive imaging, at the surgeon’s discretion, included any or all of lower extremity arterial duplex ultrasound imaging, ankle-brachial indices, or lower extremity computed tomography angiography. Polytetrafluoroethylene (PTFE) grafts ranging from 6 to 8 mm, at the surgeon’s discretion and depending on the size of the recipient’s vessels, were used as a conduit. HeRO grafts were inserted as previously described.6 In brief, access to the right atrium is obtained in the
JOURNAL OF VASCULAR SURGERY --- 2016
interventional radiology suite by direct central access or chest wall access. The patient is then transported to the operating room with an occlusive dressing over the venous puncture site to secure the placeholder catheter. This catheter is then exchanged for the 5-mm nitinol-reinforced silicone outflow catheter of the HeRO device. An anastomosis is created from the ipsilateral brachial or axillary artery to the 6-mm expanded PTFE inflow component of the HeRO device. The inflow and outflow components of the HeRO device are then tunneled subcutaneously and coupled. Demographic data included age at first or only procedure and gender. Clinical data included the key comorbidities of diabetes mellitus, tobacco use, cerebrovascular disease, coronary artery disease, peripheral arterial disease, and documented central venous stenosis. Additional variables were whether the patient had undergone a prior attempt at peritoneal dialysis, whether listed for kidney transplant, time on dialysis, and the number of prior attempts at permanent HD access. Primary outcomes were primary, assisted primary, and secondary patency, as described by SVS recommended standards for reporting.7 Secondary outcomes were the number of interventions to maintain or restore access patency, ischemic, and infectious complications. Procedures to maintain or restore access patency included surgical thrombectomy, angiography with intended intervention, percutaneous thrombolysis, and balloon angioplasty, stenting, or operative revision. Ischemic complications were defined as steal syndrome, ischemic rest pain, tissue loss, or any of the aforementioned warranting abandonment of the access. Infectious complications included documented graft infection based on admission or discharge diagnosis, details of operative notes and postmortem analysis, persistent bacteremia attributed to graft infection, or any of the aforementioned warranting abandonment of the access. Parametric and nonparametric tests were used for between-group comparisons as appropriate for a particular measure. Kaplan-Meir survival analyses with log-rank tests were used to compare primary, assisted primary, and secondary patency between thigh and HeRO grafts in the intention-to treat (three analyses/end points) and in the sample that excluded grafts with initial failure (three additional analyses/end points). Data encoding for the survival analyses adhered to SVS recommendations. Specifically, primary patency reflected intervention-free graft survival. Events were the first/only intervention to maintain or restore patency, thrombosis (if no intervention was performed), and failure for any reason (if no intervention was performed). Observations without interventions were censored on the date of death, transplantation, or final event-free follow-up. Assisted primary patency reflected thrombosis-free graft survival. Events were the first/only thrombosis or graft failure (with or without thrombosis), and observations were censored as in the primary patency model. Secondary patency reflected graft survival to the time of abandonment. Events were graft failures, and observations were censored as in the primary and
JOURNAL OF VASCULAR SURGERY Volume -, Number -
Brownie et al 3
Table I. Demographic and clinical characteristics Variablea
Table II. Access outcomes
Thigh graft HeRO graft P (n ¼ 43) (n ¼ 33) value
Age at index graft, years 52.2 Male gender, No. (%) 21 Diabetes mellitus, No. (%) 10 Tobacco use, No. (%) 13 Coronary artery disease, 8 No. (%) Peripheral vascular disease, 11 No. (%) Central venous stenosis, No. (%) 14 Prior peritoneal dialysis attempt, 14 No. (%) Listed for kidney transplant, 6 No. (%) Time on dialysis before index 60 procedure, months Prior HD access attempts, No. 3
(13.9) (49) (23) (30) (19)
53.3 22 11 14 12
(13.0) (67) (33) (2) (36)
.74 .16 .44 .34 .12
(26)
8 (24)
1.00
(33) (33)
33 (100) 11 (33)
<.01 1.00
(14)
4 (12)
1.00
(0, 256) 95 (2, 255) .16 (0, 6)
3 (1, 6)
.91
HD, Hemodialysis; HeRO, Hemodialysis Reliable Outflow; thigh, prosthetic femoral-femoral looped inguinal access. a Continuous data are shown as mean (standard deviation) or as median (min, max), and categoric data are shown as indicated.
assisted primary patency models. Intention-to-treat analyses included all grafts. Because the sample included consecutive grafts, some in patients with a history of previous thigh or HeRO grafts, the presence of a previous thigh or HeRO graft was adjusted for in six parallel multivariable Cox proportional hazards regression models (three intention-to-treat and three that excluded initial failures). These analyses tested the likelihood of graft failure using the three patency end points in HeRO grafts (reference thigh) after adjusting for whether a previous thigh or HeRO graft had been placed (reference no previous thigh or HeRO graft). The number of interventions to maintain or restore patency was calculated as the number per graft per year and as the total number of interventions per graft. Between-group comparisons of numbers of interventions were performed for those grafts that did not fail initially. Two statistical approaches were used: (1) the median test compared the number of interventions per graft per year, and (2) multivariable logistic regression tested whether an increased number of interventions per graft, after adjusting for follow-up time in months, were associated with graft type (coded HeRO, reference thigh). Data were analyzed using SPSS 22 statistical software (BM Corp, Armonk, NY), and nondirectional Type I error probabilities of <.05 were considered statistically significant. RESULTS Seventy-six (43 thigh, 33 HeRO) grafts were placed in 61patients (54% male; age 53 [standard deviation] 13 years); of which, 22 were placed after a previous thigh (15 grafts) or HeRO (seven grafts) graft. All of the HeRO grafts were placed in the upper extremities, with
Variablea
Thigh graft (n ¼ 43)
HeRO graft (n ¼ 33)
P value
Follow-up time, 21.2 (0.0, 85.3) 6.7 (0.0, 56.3) months 22.9 (<1.0, 85.3) 10.3 (<1.0, 56.3) Follow-up time,b months Ischemic 3 (7) 1 (3) complications Infectious 11 (26) 7 (21) complications Interventions per 0.6 (0.0, 8.5) 1.5 (0.0, 16.0) year to maintain or restore patency, No.b
.02 .05 .63 .79 .09
HeRO, Hemodialysis Reliable Outflow; thigh, prosthetic femoral-femoral looped inguinal access. a Continuous data are shown as median (min, max) and categoric variables as frequency (%). b Includes 42 thigh and 28 HeRO grafts that did not fail initially.
right atrial access being successfully achieved in 33 of the HeRO grafts (100%) in the interventional radiology suite. Central access for 14 of the 33 HeRO grafts (43%) was gained through chest wall collaterals. Five HeRO grafts (15%) had initial failures. Two of these resulted from an inability to exchange and upsize a placeholder catheter located in a chest wall collateral vein for the 19F sheath introducer for the outflow component of the HeRO graft due to significant central venous thrombosis incompatible with stiff wire retrograde dilation of the access site. One HeRO graft failed to achieve significant flow and was eventually excised, one patient sustained an intraoperative air embolus and the graft was never used, and one patient was lost to follow-up without any documentation of the graft being accessed. One thigh graft (2%) was an initial failure because the patient experienced immediate foot ischemia after the procedure, warranting graft excision the same day. Demographic and clinical data are summarized in Table I. The groups were comparable for sex, age, coronary artery disease, diabetes mellitus, peripheral vascular disease, and smoking history (all P $ .12). Central venous stenosis was documented by venogram before the procedure in a significantly greater proportion of HeRO grafts compared with thigh grafts (100% vs 33%; P < .01). The median time on HD before the index graft placement did not differ between the groups (P ¼ 1.00). Follow-up time, complications, and the number of interventions are summarized in Table II. Grafts deemed initial failures were included in the intention-to-treat analyses and had time-to-event follow-up durations of 0 months. Median follow-up time for secondary patency was longer in the thigh grafts (21.2 vs 6.7 months; P ¼ .02). The three separate intention-to-treat Kaplan-Meier analyses demonstrated that primary (Fig 1), assisted primary (Fig 2), and secondary patency (Fig 3) were significantly longer in the thigh grafts (all P # .01 by log-rank).
4 Brownie et al
Fig 1. Kaplan-Meier primary patency of prosthetic femoralfemoral looped inguinal access (thigh) grafts (n ¼ 43) and Hemodialysis Reliable Outflow (HeRO) grafts (n ¼ 33). Primary patency was significantly longer in thigh grafts through 3 years, with all standard errors (SE) of <10.0%.
This is reflected by the percentage survival point estimates and their standard errors that are summarized in Table III and by the corresponding observed mean patency times. Primary patency at 6 months, 1 year, and 3 years was 61%, 46%, and 4% for the thigh grafts and 25%, 15%, and 6% for the HeRO grafts, and all standard errors of these estimates were <10%. In addition, point estimates of assisted primary and secondary patency at 6 months, 1 year, and 3 years were consistently higher in the thigh grafts, and all standard errors were <10% (Table III). Mean primary patency times (95% confidence interval [CI]) were 15.5 months (10.7-20.2 months) and 7.0 months (3.4-10.6 months), assisted primary patency averaged 39.0 months (27.3-50.7 months) and 10.5 months (6.0-15.0 months), and secondary patency averaged 53.3 months (41.5-65.1 months) and 16.1 months (9.3-22.8 months) for thigh and HeRO grafts, respectively. In the intention-to treat sample, multivariable Cox proportional hazards regression demonstrated that, after adjusting for whether a previous thigh or HeRO graft had been placed (hazard ratio [HR], 1.95; 95% CI, 1.07-3.59; P ¼ .03), primary patency failure was 2.6times more likely in HeRO grafts than in thigh grafts (HR, 2.57; 95% CI, 1.46-4.52; P < .01). Similarly, after adjusting for whether a previous thigh or HeRO graft had been placed (P $ .11), assisted primary patency failure was 4.0-times more likely (HR, 4.04; 95% CI, 2.10-7.78; P < .01), and secondary patency failure was 6.5-times more likely (HR, 6.47; 95% CI, 2.991-3.97; P < .01) in HeRO grafts than in thigh grafts.
JOURNAL OF VASCULAR SURGERY --- 2016
Fig 2. Kaplan-Meier assisted primary patency of prosthetic femoral-femoral looped inguinal access (thigh) grafts (n ¼ 43) and Hemodialysis Reliable Outflow (HeRO) grafts (n ¼ 33). Assisted primary patency was significantly longer in thigh grafts, with standard errors (SE) of <10.0% through 50 months. The vertical reference line indicates the point where the SEs transitioned to being #10.5% in thigh grafts.
The three additional Kaplan-Meier models that excluded the six grafts with initial failure were consistent with the intention-to-treat analyses in that patency durations were significantly longer in the thigh grafts (all P < .05 by log-rank). Primary patency at 6 months, 1 year, and 3 years was 62%, 47%, and 4% for thigh grafts and 30%, 18%, and 7% for HeRO grafts, respectively, and all standard errors for these estimates were #10% (Table III). Consistent with the intention-to-treat analyses, point estimates of assisted primary and secondary patency at 6 months, 1 year, and 3 years were, with the exception of the 3-year primary patency estimates, also generally higher in thigh grafts (Table III). Mean (95% CI) primary patency for thigh and HeRO grafts were 15.8 (11.0-20.6) and 8.2 (4.1-12.3) months, assisted primary patency averaged 39.9 (28.1-51.7) and 12.4 (7.4-17.4) months, and secondary patency averaged 54.5 (42.8-66.4) and 18.9 (11.4-26.4), respectively. In the sample that excluded primary graft failures, multivariable Cox proportional hazards regression demonstrated that after adjusting for whether a previous thigh or HeRO graft had been placed (HR, 1.94; 95% CI, 1.01-3.71; P ¼ .05), primary patency failure was 2.3times more likely in HeRO grafts than in thigh grafts (HR, 2.35; 95% CI, 1.29-4.28; P < .01). Similarly, after adjusting for whether a previous thigh or HeRO graft had been placed (P $ .14), assisted primary patency failure was 3.8-times more likely (HR, 3.78; 95% CI, 1.88-7.60;
JOURNAL OF VASCULAR SURGERY Volume -, Number -
Brownie et al 5
Table II). However, the multivariable logistic regression model, which adjusted for follow-up time and used the metric of total interventions per graft, demonstrated that HeRO grafts were associated with an increased number of interventions to maintain or restore patency. Specifically, the odds ratio for follow-up time in months (0.933; 95% CI, 0.892-0.977; P < .01) reflects the shorter observed total follow-up time for the HeRO grafts. Then, the odds ratio for the number of interventions per graft (1.224; 95% CI, 1.017-1.473; P ¼ .03) indicated that after adjusting for between-group differences in follow-up time, each additional intervention to maintain or restore patency for a particular graft was 1.2-times more likely to be associated with a HeRO graft than with a thigh graft. DISCUSSION
Fig 3. Kaplan-Meier secondary patency of prosthetic femoralfemoral looped inguinal access (thigh) grafts (n ¼ 43) and Hemodialysis Reliable Outflow (HeRO) grafts (n ¼ 33). Secondary patency was significantly longer in thigh grafts, with standard errors (SE) of <10.0% through 41 months. The vertical reference line indicates the point where SEs transitioned to being >10.0% in thigh grafts.
P < .01) and secondary patency failure was 6.4-times more likely (HR, 6.41; 95% CI, 2.79-14.71; P < .01) in HeRO grafts than in thigh grafts. Ischemic complication occurred in three thigh grafts (7%). As mentioned, one patient with steal syndrome in the recovery room underwent immediate resection. Steal syndrome in the second patient improved after graft revision from an 8-mm to 6-mm PTFE graft, and the third patient had a nonhealing ulcer of a pre-existing transtibial amputation stump, but the graft was left in place. Only one patient (3%) of the HeRO group had an ischemic complication, and his graft was removed for steal phenomenon. Overall, there was no difference in ischemic (P ¼ .63) complications between the groups. The number of infectious complications was similar (P ¼ .79) between the thigh and HeRO groups. An associated infection occurred in 11 of 43 thigh grafts (26%), nine of which were resected, one required operative revision, and the final graft was managed with antibiotics for persistent bacteremia attributed to the graft. An infectious complication occurred in seven of 33 grafts in the HeRO group (21%), and five required operative excision. In addition, one patient in the HeRO cohort required operative revision of a bleeding pseudoaneurysm, but the graft remained patent. Among grafts without initial failure, the median number of procedures per graft per year to maintain or restore patency was 0.6 in the thigh group and 1.5 in the HeRO group and did not differ between the groups (P ¼ .22;
With the increasing prevalence of ESRD and longer patient survival on HD, maintaining long-term dialysis access presents an increasing challenge to vascular surgeons. Limited data are available to guide the vascular surgeon on the choice of AV access procedures in patients who have exhausted traditional upper extremity AV fistula or graft options. We reviewed a 10-year experience at a single center to compare outcomes between thigh AV grafts and upper extremity HeRO grafts in this difficult patient population. Several reports have reviewed thigh and HeRO grafts independently. Katzman et al1 initially reported 36 HeRO devices averaging 8.4 months’ follow-up and noted primary and secondary patency at 8 months was 39% and 72%, and an average of 2.5 procedures per year were required to maintain patency. In a larger study of 164 patients from four centers, Gage et al8 reported 1-year primary and secondary patency rates of 41% and 91%, and 1.5 procedures per year were required to maintain patency. Cull et al9 reported similar results in 125 lower extremity AV grafts over a mean of 20 months, with 1-year primary and secondary patency of 34% and 68% and 1.68 interventions per patient per year. Two prior studies directly compared thigh and HeRO grafts. Steerman et al10 was the first and noted 1-year primary and secondary patency of 51% and 60% for thigh grafts compared with 15% and 57% for HeRO grafts, which was not statistically significant. The principal difference observed was significantly fewer interventions to maintain access patency of lower extremity grafts, 2.21 per year in the HeRO and 1.17 per year in the thigh group. The results of Kudlaty et al11 differed, with similar short-term patency rates for the two procedures and no statistically significant difference in the mean number of interventions to maintain patency, 1.1 vs 1.65 per year, in thigh and HeRO grafts. In contrast to previous reports, we found significantly longer primary patency, assisted primary patency, and secondary patency in thigh grafts. In addition, after adjusting for follow-up time, we observed an association between HeRO grafts and an increased number of interventions
JOURNAL OF VASCULAR SURGERY --- 2016
6 Brownie et al
Table III. Kaplan-Meier survival point estimates for Hemodialysis Reliable Outflow (HeRO) and prosthetic femoral-femoral looped inguinal access (thigh) grafts by sample and outcome 6 months Sample and outcome measurea Intention-to-treat (n ¼ 76) Primary patency Assisted primary patency Secondary patency Exclude initial failures (n ¼ 70) Primary patency Assisted primary patency Secondary patency
1 year
3 years
HeRO
Thigh
HeRO
Thigh
HeRO
Thigh
25 (8) 41 (9) 53 (9)
61 (8) 75 (7) 88 (5)
15 (6) 30 (8) 43 (9)
46 (8) 66 (8) 88 (5)
6 (5) 10 (7) 12 (7)
4 (4) 54 (9) 70 (9)
30 (9) 48 (10) 63 (9)
62 (8) 77 (7) 90 (5)
18 (8) 35 (10) 50 (10)
47 (9) 68 (8) 90 (5)
7 (6) 12 (8) 14 (9)
4 (4) 55 (9) 72 (9)
a
Survival estimates are presented as percentages (standard error of the estimate).
to maintain or restore patency. Our study does not address why the thigh grafts remained patent so much longer than the HeRO graft. One possibility is the larger size of the femoral vessels, particularly the venous outflow. It is noteworthy that 43% of the patients in the HeRO cohort obtained central venous access through a chest wall collateral and not direct cannulation of the central venous system, which may partially account for the lower HeRO patency rates we observed. This practice is suggested by Gage et al8 in 2.2% of patients but is not thoroughly described and reflects the comfort of our interventional radiologists with this technique, allowing patients who would otherwise be relegated to catheter dependence an attempt at alternative HD access. Our observation that, after adjusting for follow-up time, HeRO grafts were associated with an increased number of interventions to maintain or restore patency is consistent with Steerman et al10 but differs from the findings of Kudlaty et al.11 Although not statistically significant, our data demonstrated that a median of 0.6 interventions were performed per year to maintain or restore patency of thigh grafts compared with 1.5 interventions per year for HeRO grafts. Additional costs associated with the HeRO graft include the increased cost of the HeRO device itself, fluoroscopy equipment and personnel, and potentially increased operating room time. One decision analysis model that reviewed previously published data showed that the HeRO device is the least costly in patients at the end-stage of dialysis access.12 As health care costs become an increasing concern, the number of interventions to maintain patency over time may function as a surrogate for resource utilization or cost. Although our data show an association between the number of interventions to maintain patency and HeRO grafts, further study will be needed to evaluate the comparative costeffectiveness of these procedures. Historically, a key concern with using the groin as a site for HD access has been the risk of infection. Previous studies of thigh accesses reported infection rates from 9% to 41%,9,13 but the differences of thigh and HeRO grafts in our study and the two prior studies10,11 were not
statistically significant. In addition, we found no difference in ischemic complications. Limb ischemia or steal syndrome were identified in only three patients (7%) in our thigh graft cohort despite 23% of our patients having diabetes mellitus and the occasional use of 8-mm PTFE grafts. This is comparable to the 5% ischemic complication rate reported by the two prior studies comparing thigh and HeRO grafts,10,11 suggesting the utility of an appropriate preoperative screening for peripheral vascular disease. Therefore, our findings do not favor one procedure or the other based on the risk of infectious or ischemic complications alone. The strengths of our study include a relatively larger sample and longer-term follow-up that enabled statistically stable patency estimates through at least 3 years. Inclusion of all sequential HeRO and thigh grafts, regardless of whether a previous graft of either type had been placed, coupled with statistical adjustment for this consideration, allowed for greater statistical power and potentially enhances the generalizability of our findings. Limitations of our study include its retrospective, nonrandomized, single-center design. In addition, our patients reflect a heterogeneous population, including those with central venous stenosis and those with patent central veins who were otherwise unsuitable or in whom upper extremity vascular sites had been exhausted. CONCLUSIONS Maintaining long-term, durable, cost-effective vascular access for HD in patients with central venous stenosis or who have otherwise exhausted upper extremity sites remains a challenge. In a single-center, intention-to-treat analysis during a 10-year period, we found that thigh grafts have significantly longer primary, assisted primary, and secondary patency and that HeRO grafts are associated with an increased number of interventions to maintain or restore patency. We found no significant difference in ischemic or infectious complications between thigh and HeRO grafts. Although HeRO grafts may extend the short-term use of the upper extremity, thigh grafts remain a more durable option for chronic HD.
JOURNAL OF VASCULAR SURGERY Volume -, Number -
AUTHOR CONTRIBUTIONS Conception and design: EB, IF, DM, DS Analysis and interpretation: EB, IF, DM, DS Data collection: EB, CK Writing the article: EB, IF, DS Critical revision of the article: EB, CK, IF, DM, DS Final approval of the article: EB, CK, IF, DM, DS Statistical analysis: IF Obtained funding: Not applicable Overall responsibility: EB
REFERENCES 1. Katzman HE, McLafferty RB, Ross JR, Glickman MH, Peden EK, Lawson JH. Initial experience and outcome of a new hemodialysis access device for catheter-dependent patients. J Vasc Surg 2009;5: 600-7. 2. Vascular Access 2006 Work Group. Clinical practice guidelines for vascular access. Am J Kidney Dis 2006;48(Suppl 1):S176-247. 3. Sidawy AN, Spergel LM, Besarab A, Allon M, Jennings WC, Padberg FT Jr, et al. The Society for Vascular Surgery: clinical practice guidelines for the surgical placement and maintenance of arteriovenous hemodialysis access. J Vasc Surg 2008;48(5 Suppl):S2-25. 4. Collins AJ, Foley RN, Herzog C, Chavers BM, Gilbertson D, Ishani A, et al. US Renal Data System 2010 Annual Data Report. Am J Kidney Dis 2010;57(1 Suppl 1):A8.
Brownie et al 7
5. Allon M, Daugirdas J, Depner TA, Greene T, Ornt D, Schwab SJ. Effect of change in vascular access on patient mortality in hemodialysis patients. Am J Kidney Dis 2006;47:469-77. 6. Kensinger C, Brownie E, Bream P Jr, Moore D. Multidisciplinary team approach to end-stage dialysis access patients. J Surg Res 2015;199: 259-65. 7. Sidawy AN, Gray R, Besarab A, Henry M, Ascher E, Silva M Jr, et al. Recommended standards for reports dealing with arteriovenous hemodialysis accesses. J Vasc Surg 2002;35:603-10. 8. Gage SM, Katzman HE, Ross JR, Hohmann SE, Sharpe CA, Butterly DW, et al. Multi-center experience of 164 consecutive hemodialysis reliable outflow [HeRO] graft implants for hemodialysis treatment. Eur J Vasc Endovasc Surg 2012;44:93-9. 9. Cull JD, Cull DL, Taylow SM, Carsten CG 3rd, Snyder BA, Yourkey JR, et al. Prosthetic thigh arteriovenous access: outcome with SVS/AAVS reporting standards. J Vasc Surg 2004;39:381-6. 10. Steerman SN, Wagner J, Higgins JA, Kim C, Mirza A, Pavela J, et al. Outcomes com- parison of HeRO and lower extremity arteriovenous grafts in patients with long-standing renal failure. J Vasc Surg 2013;57:776e83. 11. Kudlaty EA, Pan J, Allemang MT, Kendrick DE, Kashyap VS, Wong VL, et al. The end stage of dialysis access: femoral graft or HeRO vascular access device. Ann Vasc Surg 2015;29:90-7. 12. Dageforde LA, Bream PR, Moore DE. Hemodialysis Reliable Outflow (HeRO) device in end-stage dialysis access: a decision analysis model. J Surg Res 2012;177:165e71. 13. Miller CD, Robbin ML, Barker J, Allon M. Comparison of arteriovenous grafts in the thigh and upper extremities in hemodialysis patients. J Am Soc Nephrol 2003;14:2942-7. Submitted Jan 14, 2016; accepted Apr 28, 2016.
DISCUSSION Dr Thomas Huber (Gainesville, Fla). The authors have retrospectively reviewed their experience with thigh prosthetic hemodialysis accesses and the Hemodialysis Reliable Outflow (HeRO; Hemosphere Inc, Minneapolis, Minn) device in the upper extremity. They reported that the patency rates were superior in the thigh group and that the infections and ischemic complications were similar, although there was an association with increased number of interventions for the HeRO group in their multivariate model. Notably, the secondary or cumulative patency rates at 1 year were 88% in the thigh group and 43% in the HeRO group. Based upon these findings, the authors concluded that the thigh accesses were a more durable option. Although I don’t disagree with the authors’ results or conclusion, the results must be interpreted with some caution given the retrospective study design. The patency rates reported for the HeRO device were comparable or better than those reported in the literature, while those for the thigh access were comparable or better. The infectious complication rates were not significantly different between the groups although they were quite “significant,” with 15% of the HeRO devices and 21% of the thigh accesses requiring explant. Furthermore, I am not certain that the thigh accesses and HeRO devices are really competing strategies, but perhaps should be viewed as complementary in the expanding groups of complex patients with central vein stenoses or occlusions. I have three questions or requests for the authors: First, please comment on the infectious risk of both the thigh accesses and HeRO devices. Our overall experience in terms of the infectious risk has been comparable or worse and we have really tried to limit both approaches. It seems that every HeRO device we have implanted has needed to be removed. Furthermore, other authors have proposed that the infectious complication rates of the thigh accesses are higher than tunneled dialysis catheters in the upper extremity and should be avoided.
Second, please detail your surveillance protocol for the thigh accesses given your excellent patency rates. Third, please outline your current treatment algorithm for patients with severe upper extremity central vein occlusions or stenoses that preclude the traditional autogenous or prosthetic options. I would like to congratulate the authors on their presentation and contribution to the literature and thank them for sending me a copy of the manuscript well in advance of the meeting. Thanks you. Dr Evan R. Brownie. Thank you, Dr Huber, for your comments and questions, as well as your contributions to the literature on this topic. First of all, I have to agree with you that these are complimentary strategies and not necessarily competing strategies, but the facts remain that hemodialysis centers, nephrologists, and surgeons are under a lot of pressure to minimize long-term catheter use. Dialysis units may now receive financial penalties for failing to meet benchmarks for tunneled catheter rates. So if a patient is not a peritoneal dialysis candidate, we are left with the two options of either a thigh or HeRO graft. One of the purposes in designing this study was to assess the outcomes as well as to guide the surgeon on an informed discussion and consent with the patient prior to proceeding. With regard to your first question, I agree that the infection rate is high, but it is similar to previously described studies comparing both thigh grafts and the HeRO catheter as well as thigh to upper extremity catheters. Again, we fall back to that pressure that nephrologists and dialysis centers are really pushing to avoid tunnel dialysis catheters. Anecdotally, our tunneled catheter infection rates are probably a little higher than both the lower extremity access and the HeRO groups, and this has been born out in two prior comparisons of thigh and HeRO groups. I do not have a specific timeline of the infections in terms of when the thigh grafts or when the HeRO grafts had an infectious issue, but on a cursory review of our study it does not appear to be any different.
8 Brownie et al
Secondly, in regard to the surveillance for our thigh grafts, we followed the same surveillance protocol that we do for our upper extremity access. We do monthly transonic measurements of inline flow. If it falls to <600 mL/min or <25% of the patient’s baseline, then they are referred to our interventional radiologist for a fistulogram. If there is a stenosis present at that time, an angioplasty is performed. If that stenosis is not amenable to any percutaneous intervention or it is recurrent and they are undergoing multiple angioplasties, then the patient is then evaluated and referred for surgical revision. With the thigh grafts, we try to discourage stents at the venous anastomosis as this makes revision more difficult in the future.
JOURNAL OF VASCULAR SURGERY --- 2016
Finally, to answer your question regarding central venous stenosis and the algorithm we used, the first thing we obtain is a computed tomography angiography/computed tomography venography of the chest and the bilateral upper extremities. If the patient has adequate arterial inflow, then our first option to the patient is really a HeRO graft so long as they are a candidate and interested in that procedure. We discuss the risks and benefits of both, but again, one of the points of the study was to really be able to educate the patient that maybe one procedure lasts longer but requires more interventions along the way. If the patient is not a candidate for either, then we just determine if they are going to have to rely on a catheter for the remainder of their time.