Comparison of Ellipsys Percutaneous and Proximal Forearm Gracz-Type Surgical Arteriovenous Fistulas


Rationale & Objective

Percutaneous arteriovenous fistulas (AVF) are created by establishing a proximal forearm anastomosis and offer a safe and reliable vascular access. This study compares the Ellipsys percutaneous AVF with a proximal forearm Gracz-type surgical AVF, chosen for comparison as it is constructed at the same anatomical site.

Study Design

Retrospective study of prospectively collected clinical data.

Setting & Participants

All vascular access procedures conducted during a 34-month period were reviewed. The study groups comprised 89 percutaneous AVFs and 69 surgical AVFs.


Percutaneous or surgical AVF placement.


AVF patency, function, and complications.

Analytical Approach

Patency rates for each AVF group were evaluated by competing risk survival analysis using a cumulative incidence function. Association of primary, primary assisted, and secondary patency with the AVF groups was examined by Cox proportional hazard models.


Technical success was 100% for both groups. Average procedure times were 14 minutes for percutaneous AVFs and 74 minutes for surgical AVFs (P < 0.001). Proximal radial artery (PRA) was used in all percutaneous AVF cases. Inflow for surgical AVFs included radial (30%), ulnar (12%), and brachial (58%) arteries. Outflow veins for both groups were the cephalic and/or basilic veins. Access flow volumes, times to maturation, and overall numbers of interventions per patient-year were not significantly different. Cumulative incidence of primary patency failure at 12 months was lower for surgical AVF (47% vs 64%, P = 0.1), but secondary patency failure was not different between groups (20% vs 12%, P = 0.3). PRA surgical AVFs had similar primary patency (65% vs 64%, P = 0.8) but higher secondary patency failure rates than percutaneous AVFs at 12 months (34% vs 12%, P = 0.04).


Retrospective study with a relatively short follow-up period, and not all patients required hemodialysis at the end of study.


Both percutaneous and surgical AVFs demonstrated high rates of technical success and secondary patency. Percutaneous AVFs required shorter procedure times. The rate of intervention was similar. When a distal radial artery AVF is not feasible, percutaneous AVF might offer an appropriate procedure for creating a safe and functional access, maintaining further proximal forearm surgical AVF creation options.

Plain-Language Summary

We compared Ellipsys percutaneous arteriovenous fistulas (AVF) with a surgical proximal forearm surgical AVF as they use the same anatomical site and vascular anatomy. We retrospectively reviewed our center’s vascular access procedures over a 34-month period. No demographic differences were observed. Both percutaneous and surgical AVF demonstrated similarly high technical success rates and secondary patency rates. The percutaneous AVFs required dramatically shorter procedure times. The rate of required interventions was similar. When both surgical and percutaneous AVF used proximal radial artery inflow, secondary patency was higher for percutaneous AVFs. If a distal radial artery AVF is not feasible, percutaneous AVF might offer an appropriate procedure for creating a safe and functional access.

Various vascular access options are available for the hemodialysis patient; however, almost 60 years after its introduction, the arteriovenous fistula (AVF) remains the preferred vascular access in most patients. Unfortunately, an AVF is at risk for loss of primary patency due to required interventions in addition to prolonged or failed maturation. A systematic patient-centered approach to hemodialysis vascular access becomes more challenging and increasingly important with the advancing age and increasing comorbidity burden of the kidney failure population. It is critically important that access planning not only considers the present needs of the patient but also includes preservation of vascular anatomy for the next vascular access.12

Recently, an innovation in vascular access has been introduced: the percutaneous AVF. Two devices have been approved for the creation of percutaneous AVF. Using distinctly different approaches, techniques, and mechanisms, both create successful proximal forearm AVFs.13 It is recommended that consideration for access creation starts distally at the wrist and progresses proximally to the upper arm.14,15 At our center, a standardized anatomical succession algorithm has been previously described 16 and was used in site selection and planning of vascular access for these patients throughout the study period. This autogenous access succession sequence included, in order of preference, a distal radial-cephalic AVF (including snuffbox fistula), percutaneous AVF options, a Gracz-type surgical AVF, upper arm AVF (brachial cephalic/brachial basilic), brachial vein AVF, and arteriovenous graft.

Our study compared the outcomes of the 2 most anatomically similar percutaneous and surgical AVFs: the Ellipsys percutaneous AVF (Avenu Medical, Inc)17 and the Gracz-type surgical AVF.

 This surgical AVF has shown excellent results in a wide range of patient groups, including the elderly, children, and individuals with peripheral arterial disease. In addition, it has been found to have higher patency rates than those reported for a distal radial-cephalic AVF and a low incidence of morbidity, such as hemodialysis access–induced hand ischemia, arm edema, and other complications associated with higher flow upper arm accesses.


From September 2017 through July 2020, 430 permanent vascular accesses were created in 387 individuals at a university-affiliated surgical center, including 280 surgical AVFs, 127 percutaneous AVFs, and 23 arteriovenous grafts (Fig 1). A prospectively maintained database of all cases was retrospectively reviewed for this study. The study follows the requirements of our institutional review board and is in accordance with the Declaration of Helsinki. The ethics committee of Hamburg ruled that the study was exempt from approval due to its retrospective anonymized nature.

Figure thumbnail gr1

Figure 1Vascular access procedures flow chart resulting in Ellipsys percutaneous arteriovenous fistula (AVF) and Gracz-type surgical AVF groups during the study period. VasQ is an external AVF anastomotic support device (Laminate Medical Inc).


Inclusion Criteria

The protocol for AVF planning and creation was previously described elsewhere. Briefly, the first choice for access placement was a distal forearm surgical AVF. If the distal anatomy was not suitable, the next consideration was a proximal forearm percutaneous AVF option. The next choice in the sequence plan was a surgical proximal forearm (cubital fossa) AVF using either the Gracz-type surgical AVF technique or a modified version. These surgical AVFs required some variation according to existing anatomy with the potential for confounding and associated bias in analyzing these data. The first 37 surgical AVFs were created at a time when Ellipsys was not available. After this period, surgical AVF was created only if a percutaneous AVF creation was not technically feasible or was abandoned (2 patients).

The proximal radial artery (PRA) was used for all percutaneous AVFs. The Gracz-type surgical AVF inflow involved either the PRA, proximal ulnar artery, or antecubital (distal) brachial artery, depending on vessel suitability. An inner diameter of ≥2 mm was required. Use of radial or ulnar artery inflow also required that the distal blood flow in the non-AVF forearm artery was intact and had triphasic flow. All patients had a deep communicating vein (DCV/perforator) and outflow vein with inner diameters of ≥2.0 mm with a tourniquet in place. The percutaneous AVFs also required a distance of ≤1.5 mm between the PRA and the DCV.

The only specific exclusion criteria for percutaneous AVF was failure to meet the anatomically suitable vessel requirements according to the inclusion criteria.


All patients provided written consent for each procedure. All procedures were performed under regional anesthesia except in 4 percutaneous AVF cases, which were performed under general anesthesia (patients’ preference). The patients received 2,000 units of heparin intraoperatively. Prophylactic antibiotics were not used. A standardized technique was used for the creation of both AVF types. The Ellipsys system was used to create percutaneous AVF anastomosis between the PRA and DCV. The procedure has been described in detail elsewhere.

The venous outflow used the cephalic, median cubital, or both veins, with both left undisturbed when available for outflow. Completion of the balloon angioplasty of the Ellipsys anastomosis was a routine part of the procedure and was always performed as recommended by the manufacturer. The arterial preference for creation of the surgical AVF anastomosis was the PRA (or ulnar artery if the radial artery was the dominant supply to the hand), followed by the brachial artery (if a PRA or ulnar artery was not feasible). When the brachial artery was used, the anastomosis was limited to a diameter of 4 mm or 75% of the diameter of the brachial artery (lesser of the 2).

The venous preference for the surgical AVF anastomosis creation was the DCV (Fig 2). If the DCV was not adequate, the proximal forearm cephalic vein was used. As in the percutaneous AVFs, our routine practice was to leave the cephalic and median cubital venous outflow branches undisturbed. All procedures were performed by a single vascular surgeon (R.S.).

Figure thumbnail gr2

Figure 2 Photograph of a typical Gracz-type proximal radial artery to deep communicating vein arteriovenous fistula. This configuration is the same as that seen in an Ellipsys percutaneous arteriovenous fistula.


Follow-Up Evaluation

Flow measurements were calculated during the follow-up examinations by duplex ultrasound, measuring volume flow in the distal brachial artery on postprocedure days 1 to 2, week 4, and every 3 to 6 months thereafter by the primary surgeon and his team. Outflow veins were evaluated with physical examination and ultrasound to determine progress toward maturation and the potential for cannulation. Indications for interventions were left to the judgment of the clinical team (surgeon/interventionalist and nephrologist/dialysis unit).

End Points/Definitions

Primary end points were procedural technical success, time to physiologic maturation, primary failure, time to first successful clinical use, and access patency. Technical success was defined as the presence of thrill/bruit and fistula flow in the outflow vein(s) by duplex ultrasound at completion of the procedure. Physiological maturity was defined as AVF blood flow of ≥500 mL/min and an outflow vein diameter of ≥5 mm. Primary patency was defined as the time from creation to reintervention, abandonment, or reaching certain events (eg, death, transfer to peritoneal dialysis, or kidney transplantation). Secondary patency was defined as the time from creation to either abandonment or reaching certain events. Patency status was evaluated after the procedure at 4 weeks and every 3 to 6 months thereafter, when feasible.

Primary failure was defined as abandonment/conversion before becoming physiologically mature or being used for dialysis. Time to successful clinical use was the time from creation to successful 2-needle cannulation for treatment to achieve the dialysis prescription. Any secondary unplanned procedure on the AVF (surgical or endovascular) was defined as an intervention. Interventions were reported as the number of interventions per patient-year. Definition for high-flow AVF was brachial artery flow > 2,000 mL/min (if asymptomatic) or > 1,500 mL/min (if symptomatic).

Statistical Analysis

Study variables were evaluated by obtaining descriptive statistics. Frequencies and percentages were calculated for categorical variables. Median and ranges were calculated for continuous variables. Data distribution was tested for normality for continuous variables using the Kolmogorov-Smirnov and Shapiro-Wilk tests. The Mann-Whitney U test was used to determine the difference between the percutaneous and surgical AVF groups for age, body mass index, and procedure-related continuous variables. Any association between categorical variables and the 2 AVF groups were assessed by χ2 test of independence or Fisher exact test, as appropriate.

Patency rates at different time points for each AVF group were evaluated by competing risk survival analysis using cumulative incidence function. Patency failure curves were created, and the Gray test of equality of cumulative incidence function was performed. The association of primary, primary assisted, and secondary patency with the AVF groups was examined by Cox proportional hazard models with subdistribution function. Hazard ratios and 95% confidence intervals were reported. Poisson regression analysis was performed to compare the number of interventions per patient-year. All statistical analyses were performed in SAS9.4 (SAS Institute).


Screening of 251 patients in whom a vascular access was planned found 152 patients (60.6%) who were eligible for the Ellipsys percutaneous AVF. The access sequence algorithm used during the study period for vascular access site selection and planning16 resulted in a radial-cephalic AVF created in 62 patients. As the second option when adequate vessels were present, an Ellipsys percutaneous AVF was created in 90 cases (20.9% of the total vascular access cases). One case was disqualified from the study because a brachial vein-artery anastomosis was used. This left 89 cases in the percutaneous AVF study group with 100% technical success.

Gracz-type surgical AVF was created in 71 patients (16.5% of the total vascular access cases) with 100% success. Two surgical AVF patients in whom an external anastomotic support device was used were excluded from the study, leaving 69 surgical AVF cases for comparison. One patient contributed to both percutaneous and surgical AVF study groups when a surgical AVF was created after a technically successful percutaneous AVF creation that later failed.

No significant patient demographic differences were observed between the groups, although the differences may not have reached statistical significance due to sample sizes (Table 1). Nominally more predialysis patients were present in the percutaneous AVF group (44% vs 28%, P = 0.07), and a significantly higher percentage of patients with a central venous catheter was observed in the surgical group (73% vs 54%, P = 0.02). There was a history of a previous ipsilateral access in 38.2% of percutaneous AVF and 42% of surgical AVF cases.

Table 1 Patient Characteristics by Procedure Type



               AVF Type

P a

Gracz sAVF (n = 69)

Ellipsys pAVF (n = 89)

Male sex

93 (58.9%)

35 (50.7%)

58 (65.2%)


Age, y

67.8 (28.0-87.7)

67.9 (33.2-87.7)

66.0 (28.0-86.2)


BMI, kg/m2

26.6 (16.5-50.2)

28.7 (16.5-50.2)

26.2 (16.5-45.1)



65 (41.1%)

33 (47.8%)

32 (36%)


CKD status




58 (36.7%)

19 (27.5%)

39 (43.8%)

 Kidney failure

98 (62%)

49 (71%)

49 (55%)


2 (1.3%)

1 (1.4%)

1 (1.1%)


Previous ipsilateral AVF

63 (39.9%)

29 (42%)

34 (38.2%)


Dialysis with a CVC at time of index procedure

98 (62%)

50 (72.5%)

48 (53.9%)


Values for categorical variables given as count (%) and for continuous variables as median (range). Abbreviations: AVF, arteriovenous fistula; BMI, body mass index; CKD, chronic kidney disease; CVC, central venous catheter; pAVF, percutaneous arteriovenous fistula; sAVF, surgical arteriovenous fistula.

a For CKD status Fisher exact test was used whereas, for all other categorical variables, χ2 test of independence was used. Mann-Whitney U test was used for age and BMI.

In the percutaneous AVF group, the PRA was used for inflow in 100%. The inflow artery in the surgical AVF cases was radial (30.4%), ulnar (11.6%), and brachial (58%). In the percutaneous AVF cases, outflow was cephalic (23.6%), basilic (10.1%), or both (66.3%). Outflow for the surgical AVF procedures was cephalic (36.2%), basilic (29%), or both (34.8%). “Dual outflow” (cephalic and basilic) was more common in the percutaneous AVF group (66.3%) than that in the surgical AVF group (34.8%). More patients in the surgical AVF group had only basilic vein outflow (29%). The percutaneous AVF procedures required an average time of 14 minutes, and the surgical AVFs averaged 74 minutes (P < 0.001) (Table 2). A single percutaneous AVF patient developed a hematoma at the anastomotic site, which required later conversion to surgical AVF.


Table 2 Procedure Details



           AVF Type

P a

Gracz sAVF

Ellipsys pAVF

Technical success

158 (100%)

69 (100%)

89 (100%)






40 (25.3%)

40 (58%)

0 (0)


110 (69.6%)

21 (30%)

89 (100%)


8 (5%)

8 (12%)

0 (0)

Outflow vein



 Cephalic vein

46 (29.1%)

25 (36%)

21 (24%)

 Basilic vein

29 (18.3%)

20 (29%)

9 (10%)

 Cephalic and basilic vein

83 (52.5%)

24 (34.8%)

59 (66%)

Procedure time, min

20 (8-128)

70 (45-128)

14 (8-31)


Values given as count (%), except for procedure time, given as median (range). Abbreviations: AVF, arteriovenous fistula; pAVF, percutaneous arteriovenous fistula; sAVF, surgical arteriovenous fistula.

a Fisher exact test was used for anastomosis, whereas χ2 test of independence was used for all other categorical variables. Mann-Whitney U test was used for continuous variables.


The mean follow-up time for the percutaneous AVF group was 266 (range, 1-674) days and for surgical AVFs 472 (range, 2-1,016) days. Two patients from the surgical AVF group died within 4 weeks after the procedure of pneumonia. The access blood flow volumes at day 1, week 4, month 3, and month 6 for percutaneous AVF versus surgical AVF were not significantly different. Flow measurements were not available for some patients at specific time points (missed follow-up visit, censored and did not reach the study-specific time point, or not documented). High-flow AVF developed in 2 patients with percutaneous AVF (0.3 per 100 person-months) and in 7 patients with surgical AVF (0.7 per 100 person-months) (P = 0.4). All high-flow surgical AVFs had brachial artery inflow, and 3 of these patients developed symptoms of congestive heart failure and underwent flow reduction procedures. At the time of writing, these individuals remain in 6- to 12-month follow-up status to ensure flow reduction is maintained. Physiological maturity within the 2 groups at 4 weeks, 3 months, and 6 months was achieved in the percutaneous AVF group in 68 of 89 (76%), 72 of 89 (80.9%), and 76 of 89 (85%) cases, and in the surgical AVF group in 51 of 67 (76%), 53 of 67 (79%), and 53 of 67 (79%) cases, respectively. These differences were not statistically significant.

Time to Cannulation

Dialysis was required in 58 percutaneous AVF and 63 surgical AVF patients. Of these, 49 (85%) percutaneous AVFs and 50 (79%) surgical AVFs were successfully used for 2-needle dialysis for 6 consecutive sessions. Time from access creation to cannulation was 57 (range, 1-426) days for percutaneous AVF versus 68 (range, 1-403) days for surgical AVF, including the patients with preemptive AVF creation and first-day cannulations in preexisting AVF. These differences were not statistically significant (Table 3).

Table 3. Postoperative Follow-Up Results



Gracz sAVF

Ellipsys pAVF

P a





Follow-up, d

355.7 (1-1,016)


472 (2-1,016)


266 (1-674)


AVF blood flow, mL/min b


 Day 1

472.5 (150-1,800)


540 (190-1,000)


470 (150-1,800)


 Week 4

820 (150-2,000)


1,000 (150-2,000)


810 (290-1,400)


 Month 3

895 (80-3,000)


960 (80-3,000)


842 (280-1,500)


 Month 6

820 (70-1,900)


970 (320-1,900)


790 (70-1,600)


Maturation by ultrasound criteria

137 (7.37)


57 (5.34)


79 (10.15)


High-flow AVF

9 (0.49)


7 (0.66)


2 (0.26)



0 (0)


0 (0)


0 (0)


AVF abandoned

19 (0.1)


12 (1.12)


7 (0.9)


No. of dialysis patients at study end






AVF use in dialysis patients

99 (5.36)


50 (4.68)


49 (6.3)


Time to 1st use, d c

53 (1-185)


52 (1-185)


54 (1-183)


Unless otherwise indicated, values reported as median (range) or count and event rate (per 100 person-months). Abbreviations: AVF, arteriovenous fistula; HAIDI, hemodialysis access–induced hand ischemia; pAVF, percutaneous arteriovenous fistula; sAVF, surgical arteriovenous fistula.

a Fisher exact test was used for high flow, whereas χ2 test of independence was used for all other categorical variables. Mann-Whitney U test was used for continuous variables.

b AVF blood flow information was not available from all the study participants.

c In patients with kidney failure at the time of the procedure.



Surgical or endovascular interventions were required in both groups (Table 4). The number of patients requiring 0, 1, or 2 or more interventions in the percutaneous AVF and surgical AVF groups were 22 (32%) versus 34 (49%), 52 (58%) versus 13 (19%), and 13 (15%) versus 24 (27%), respectively. For some encounters, several simultaneous interventions were performed. Inflow angioplasty was required in 40 of the percutaneous AVF patients (45%) versus 6 of surgical AVF patients (9%). Ultrasound-guided sharp recanalization (needle re-entry through the DCV into the PRA) was performed successfully in 6 local occlusions (7%) of the percutaneous AVF patients before angioplasty. Outflow angioplasty was performed in 9 percutaneous AVF patients (10%) and 13 surgical AVF patients (19%). This included placing a stent graft in 2 surgical AVF patients. In addition, the outflow vein in 5 of the surgical AVF and 1 of the percutaneous AVF patients required a patchplasty. Outflow thrombosis occurred in 6 of the surgical AVF patients and was treated using a mechanical thrombectomy. No outflow thrombotic events occurred in the percutaneous AVF group. Vein transposition, either basilic or cephalic, was performed in 3 percutaneous AVF patients (3%) and 20 surgical AVF patients (29%). Vein superficialization by liposuction was performed in 2 surgical AVF patients (3%). Banding of the median cubital vein or ligation/embolization of a secondary draining brachial vein was necessary to direct blood flow to the targeted primary cannulation vein in 15 percutaneous AVF patients (17%) and 2 surgical AVF patients (3%). In total, 56 intervention procedures were performed in 37 of 89 percutaneous AVF cases (42%) versus 58 in 35 of 69 surgical AVF cases (51%). The number of interventions per patient-year was 0.86 for percutaneous AVF versus 0.66 for surgical AVF (P = 0.2).

Table 4. Postoperative Interventions per Patient to Maintain the AVF (Excluding Conversion to Another Access)


Overall (N = 158)

Gracz sAVF (n = 69)

Ellipsys pAVF (n = 89)


No. of interventions per patient




86 (54.4%)

34 (49%)

52 (58%)


46 (29.1%)

22 (32%)

24 (27%)

 2 or more

26 (16.5%)

13 (19%)

13 (15%)

No. of interventions per patient-year





Specific Interventions


 Balloon angioplasty


Arterial inflow (proximal radial/ulnar distal brachial)

5 (3.2%)

0 (0)

5 (6%)



8 (5.1%)

0 (0)

8 (9%)


Juxtaanastomotic (perforator) vein

33 (20.9%)

6 (9%)

27 (30%)


With sharp recanalization

6 (3.8%)

0 (0)

6 (7%)


Outflow vein (including cephalic arch/basilic swing points)

20 (12.7%)

12 (17%)

8 (9%)


Outflow central vein obstruction

2 (1.3%)

1 (1%)

1 (1%)


Placing stent/stent graft

2 (1.3%)

2 (3%)

0 (0)


Patchplasty of outflow vein

6 (3.8%)

5 (7%)

1 (1%)





6 (3.8%)

6 (9%)

0 (0)



0 (0)

0 (0)

0 (0)



23 (14.6%)

20 (29%)

3 (3%)


Basilic vein

19 (12.0%)

17 (25%)

2 (2%)


Cephalic vein

4 (2.5%)

3 (4%)

1 (1%)



2 (1.3%)

2 (3%)

0 (0)


 Banding procedure

14 (8.9%)

5 (7%)

9 (10%)


For high-flow AVF

3 (1.9%)

3 (4%)

0 (0)


Of median cubital vein to direct flow to cephalic vein

11 (7.0%)

2 (3%)

9 (10%)


 Ligation/coiling of brachial vein

6 (3.8%)

0 (0)

6 (7%)


Serious adverse events

2 (1.3%)

1 (1%)

1 (1%)


Abbreviations: AVF, arteriovenous fistula; pAVF, percutaneous arteriovenous fistula; sAVF, surgical arteriovenous fistula.



Cumulative incidence of primary patency failure at 3, 6, and 12 months for the percutaneous AVFs and surgical AVFs was 35% versus 25%, 53% versus 32%, and 64% versus 47%, respectively. Secondary patency failure was 9% versus 16%, 11% versus 16%, and 12% versus 20% (Fig 3A and B ). Using surgical AVF as the reference group, a comparison of patency data for the 2 groups showed an adjusted hazard ratio of 1.47 (95% CI, 0.93-2.31) for primary patency and 0.66 (95% CI, 0.29-1.52) for secondary patency.

Figure thumbnail gr3

Figure 3 Cumulative incidence plots of vascular access patency failure show (A) primary and (B) secondary patency rates by procedure type. Surgical Gracz-type arteriovenous fistulas had artery inflow supplied by the radial, ulnar, and brachial arteries (30%, 12%, and 58%, respectively). All Ellipsys percutaneous arteriovenous fistulas had radial artery inflow. (A) Gray test for equality of cumulative incidence functions, P = 0.1 and 0.3 for panels A and B, respectively.

To more closely match the procedural anatomy of the percutaneous AVF and surgical AVF and minimize confounding variables, comparisons of cumulative incidence of patency failure were made between percutaneous AVF and the subgroup of surgical AVFs where the PRA was used in the creation of the surgical AVF anastomosis. This showed a primary patency failure at 3, 6, and 12 months for the percutaneous AVFs and surgical AVFs of 35% versus 42%, 53% versus 42%, and 64% versus 65%, respectively. At the same time intervals, assisted primary patency failure was 14% versus 31%, 17% versus 40%, and 22% versus 40%, respectively; and secondary patency failure was 9% versus 26%, 11% versus 26%, and 12% versus 34%, respectively (Fig 4A and B ). Using the surgical AVF subgroup as the reference, a comparison of patency data for the 2 groups gave a hazard ratio of 0.37 (95% CI, 0.11-1.19) when adjusted for age, sex, body mass index, diabetes mellitus, previous ipsilateral AVF, and dialysis with a central catheter at time of index procedure.

Figure thumbnail gr4

Figure 4 Cumulative incidence plots of vascular access patency failure show (A) primary and (B) secondary patency rates by procedure type. Surgical Gracz-type arteriovenous fistulas (AVFs) in this subset analysis had arterial inflow supplied by the radial artery. All Ellipsys percutaneous AVFs had radial artery inflow. There was no statistically significant difference in cumulative incidence of primary patency failure between the 2 groups (Grays test for equality of cumulative incidence functions, P = 0.8). Ellipsys percutaneous AVFs had significantly lower secondary patency failure than those with radial artery inflow Gracz-type surgical AVFs (Gray test for equality of cumulative incidence functions, P = 0.04).

Overall, there were 19 primary failures as defined. Within the percutaneous AVF group the incidence was 8% (7 of 89) compared with 17% (12 of 69) in the surgical AVF group. The number of failures per patient-month was 0.009 for the percutaneous AVF group and 0.011 for the surgical AVF group, respectively. All the surgical AVF cases had either an attempt at salvage or imaging of the surgical AVF that found no possible resolution available. Six of the 7 percutaneous AVF patients elected to forgo an attempt at percutaneous AVF salvage, opting for conversion to a surgical AVF.


Before innovations in surgical technique can be introduced into clinical practice, several questions must be answered to understand the innovations’ clinical utility and where to introduce the technology in current clinical practice. For vascular access procedures, the specific questions that need to be addressed should examine the 5 clinical phases of the lifecycle of an AV access: (1) creation, (2) maturation, (3) initial clinical use, (4) sustained clinical use, and (5) dysfunction.31 In this study, we evaluated the newest innovation in dialysis access surgery, the percutaneous AVF created with the Ellipsys device, on all these domains. In effort to make a direct comparison with surgical AVF, we used the Gracz-type surgical AVF as the control group because it uses similar antecubital vascular anatomy.

Both procedures had a 100% technical success rate, but procedure time favored percutaneous AVF (14 minutes) over surgical AVF (74 minutes). These values compare favorably with those that have been previously reported, which have been in the range of 88% to 100% technical success and 18 to 24 minutes for percutaneous AVF procedure times. Creation of the Ellipsys percutaneous AVF avoids some issues with surgical AVFs that result from skin incision, inflammation associated with vessel dissection and manipulation, and a high rate of primary failure reported in many studies.32 In addition, the procedure is minimally invasive, does not leave a surgical scar, involves less postoperative pain, and can be performed as an office-based procedure. These attributes have been reflected in reports of a high level of patient satisfaction with this procedure.34

Primary failure, or failure to develop physiological maturity capable of supporting hemodialysis, was observed in 7 patients (8%) in the percutaneous AVF group, which was half that of the surgical AVF group (17%). Our experience compares favorably against the 20% to 60% reported by others for newly created antecubital AVFs and significantly better than what has been observed with the creation of other types of AVFs.38

Time to fistula maturation impacts whether patients initiate hemodialysis with a catheter and how long they are catheter dependent. In this study, the ultrasound criteria to assess maturation included vessel size and access blood flow volumes that are necessary for clinical use of the AVF. The differences in time required to achieve physiological maturity did not differ significantly between the 2 groups. In the percutaneous AVF group, physiological maturity was achieved in 76% at 4 weeks and 85% at 6 months. Similar rates were observed in the surgical AVF group (76% at 4 weeks and 79% at 6 months). This stands in marked contrast to data presented in a multicenter trial4 involving 877 hemodialysis patients in which only 40% of newly created AVFs matured within 6 months and a single-center study35  involving 1,206 AVFs that had a median maturation time of 10.3 weeks.

Only 58 patients in the percutaneous AVF group progressed to the point of requiring kidney replacement therapy. The access was used to provide dialysis in 85%. The average time from creation to initial clinical use was 8 weeks (57 days). This did not differ significantly from the corresponding time period observed in the surgical AVF group. It is important to note that in some instances, the percutaneous AVF was used within 14 days to avoid catheter placement, as has been noted in other studies.28

The ability to use a newly created access early is a valuable asset in the overall strategy to decrease catheter use. The lack of surgical inflammation and edema in the cubital fossa plays an important role in allowing early cannulation with these readily accessible sites for selected patients. In addition, percutaneous AVFs distribute flow through both median cephalic and median cubital veins, resulting in lower pressure in these vessels that are often larger and offer easy cannulation. The availability and increased usage of plastic cannulas and ultrasound guidance in dialysis units adds to the potential for early cannulation.

All dialysis access is subject to the risk of failure of maturation and the development of dysfunction. As a result, many will require some form of intervention to become usable or to maintain functional patency. Within the percutaneous AVF group, the majority of patients (58%) did not require any intervention over the course of the study. Although nominally fewer interventions were required than in the surgical AVF group, the difference was not statistically significant. Significant variations were evident in the types of interventional procedures required for each of the 2 groups (Table 4). Outflow vein stenosis with thrombosis occurred only in the surgical AVF group. We hypothesize that the broader outflow distribution of a percutaneous AVF may result in lower pressure and provide protection against excessive turbulence, minimizing the risk of neointimal hyperplasia. Alternatively, surgical inflammation that contributes to swing segment stenosis does not occur with percutaneous techniques. Conversely, more anastomotic angioplasties were required in the percutaneous AVF group, possibly as a surgical AVF anastomosis may be somewhat larger than that of the percutaneous AVF.

The limited number of transpositions required in the percutaneous AVF group has been seen in other studies and is likely related to the extended antecubital cannulation zone that characterizes this access. Both median cubital and median cephalic veins become reliable cannulation segments within the cubital fossa, in addition to the outflow segment adjacent to the anastomosis that would likely be unusable with a surgically created AVF because of the surgical scar. These cannulation sites become even more important with obese individuals.

High access blood flow can be problematic, particularly with the increasing age of the dialysis population and associated comorbidities, creating a risk for cardiac complications or hand ischemia.29, 30 This was seen in 2 of 89 percutaneous AVF patients (2%) and 7 of 69 surgical AVF patients (10%) (all with a brachial artery anastomosis). Cardiac symptoms occurred in 3 surgical AVF patients, and a banding procedure was performed in these cases for flow reduction.

Primary patency was similar between the groups; though it appeared to favor the surgical AVF group, the proportional hazards modeling of survival data failed to meet statistical significance. Similar observations were made for secondary patency. The most direct comparison of the percutaneous and surgical AVFs is possible by restricting the surgical AVF group to only those cases in which inflow is provided only by the PRA. In this subgroup, we found assisted primary patency and cumulative patency to be higher in the percutaneous AVF group. Our results were similar to a study involving 105 Ellipsys percutaneous AVFs, where the cumulative patency rate at 6, 12, 18, and 24 months was 97.1%, 93.9%, 93.9%, and 92.7%, respectively.34 These values are somewhat higher than those in this study; however, the specifics of patient selection involved were not reported.

The current study reflects the results obtained using an access succession algorithm strategy based on physical and ultrasound examinations to determine the appropriate anatomy while conserving forearm sites whenever possible. Patency values from both groups were superior to the values reported for AVFs in general. In a meta-analysis of AVF patency rates involving several thousand patients, the cumulative patency was only 64%, significantly less than observed in either group in this study.

The results presented in this study demonstrate that the percutaneous AVF is comparable to the surgical AVF in terms of creation, maturation, and clinical use. In our experience, both exhibited traits that are superior to other types of vascular access as reported in the literature. It has been recommended that every dialysis patient have a “life plan” to direct both planning and creation of dialysis vascular access.39 Part of this plan should be vessel preservation. Based on our successful experience, we propose that the percutaneous AVF be incorporated into this strategy. 16 The comparison between the percutaneous and surgical AVF groups in this study confirmed equipoise on the major outcome domains of creation, maturation, initial and sustained clinical use, and dysfunction, and that this inclusion is both logical and appropriate.

This study has limitations associated with all retrospective reviews in addition to a relatively short follow-up period and not all patients requiring hemodialysis by the end of the study period. The Ellipsys cases included an initial learning curve, so it is reasonable to expect future results for both percutaneous AVF creation and interventions may show higher success rates. Further studies contrasting percutaneous and surgical vascular access procedures as well as randomized studies comparing percutaneous and surgical AVF are needed.

In conclusion, both Ellipsys percutaneous AVF and Gracz-type surgical AVF demonstrate 100% technical success with similarly high cumulative patency rates. Ellipsys percutaneous AVFs required shorter procedure times, had no thrombotic occlusions of outflow veins, had a low rate of high-flow AVFs, and had easily accomplished percutaneous ultrasound-guided salvage procedures. Creating a Gracz-type surgical AVF is possible after a failed Ellipsys percutaneous AVF. When a distal radial-cephalic AVF is not feasible, an Ellipsys percutaneous AVF might be an appropriate and logical next-step procedure for creating a safe and functional access while maintaining future proximal forearm surgical AVF creation options.

Fuente:  Am J Kidney Dis. 78(4):520-529      DOI:


Leave A Comment