- Email: [email protected]
Factors influencing technical success and outcome of percutaneous balloon angioplasty in de novo native hemodialysis arteriovenous fistulas
European Journal of Radiology 81 (2012) 2298–2303
Contents lists available at SciVerse ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Factors influencing technical success and outcome of percutaneous balloon angioplasty in de novo native hemodialysis arteriovenous fistulas Sam Heye a,∗ , Geert Maleux a,1 , Johan Vaninbroukx a,1 , Kathleen Claes b,2 , Dirk Kuypers b,2 , Raymond Oyen a,1 a b
University Hospitals Leuven, Department of Radiology, Herestraat 49, 3000 Leuven, Belgium University Hospitals Leuven, Department of Nephrology, Herestraat 49, 3000 Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 20 July 2011 Received in revised form 2 September 2011 Accepted 6 September 2011 Keywords: Dialysis Arteriovenous fistula Angioplasty Balloon Treatment outcome
a b s t r a c t Objective: To determine predictors of technical success, dysfunction recurrence and patency after percutaneous transluminal angioplasty (PTA) of de novo dysfunctional hemodialysis arteriovenous fistulas (AVFs). Methods: Retrospective analysis of first time PTA of 167 AVF in 162 patients (100 men, 66 ± 13 years). Anatomical (location, length, grade and number of stenoses) and clinical variables (sex, age, prior AVF, diabetes mellitus and AVF age, side and location) were reviewed. Results: 217 stenoses or segmental occlusions were treated. Technical success rate (84.4%) was higher in radiocephalic AVF compared to brachial artery–median vein AVF (p = 0.030) and was negatively correlated with initial stenosis (p = 0.049). Dysfunction recurrence was seen in 52.7% and correlated negatively with technical success (p = 0.013) and AVF age (p = 0.008). Early dysfunction (within 6 months) was negatively correlated with AVF age (p = 0.016) and positively with diabetes (p = 0.003). Higher AVF age resulted in higher primary (p = 0.005) and secondary patency rates (p = 0.037–0.005). Conclusions: Technical success of PTA in hemodialysis AVF is affected by AVF type and initial stenosis and has significant effect on dysfunction recurrence, but not on AVF longevity. Younger AVF has increased risk for (early) recurrent dysfunction and lower patency rates. Patients with diabetes mellitus have higher risk for early dysfunction. © 2011 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Arteriovenous fistulas (AVFs) are the preferred type of access in patients who undergo hemodialysis [1]. Hemodynamic significant stenosis causing access dysfunction is a frequent complication and requires (repeated) percutaneous transluminal balloon angioplasty (PTA) to maintain patency [2–5]. Durability of PTA is however limited, with 1-year primary patency rates ranging between 26% and 62% [2–4]. Variables influencing outcome were studied in previously reported series [3,4]. In a series of 65 dysfunctional or occluded AVF Clark et al. showed that AVF with a stenosis longer than 2 cm in length and patients with at least one comorbid factor such as diabetes, coronary artery disease or peripheral vascular
disease had a significantly increased risk for access patency loss [3]. No variables affected the outcome of PTA in 151 dysfunctional AVF in a study of Rajan et al. [4]. Not only there were some conflicting results between these 2 studies, but also it was not clear whether previous interventions were performed on the studied AVF or not. Moreover these studies reported little on the impact of technical success on the outcome or on the location of the stenosis in case of recurrent dysfunction. The objective of our study was to determine anatomical and clinical variables that were predictors of technical success, dysfunction recurrence and patency after PTA on hemodialysis AVF that have not yet been treated for dysfunction or thrombosis.
2. Materials and methods ∗ Corresponding author. Tel.: +32 16 343782; fax: +32 16 343765. E-mail addresses: [email protected] (S. Heye), [email protected] (G. Maleux), [email protected] (J. Vaninbroukx), [email protected] (K. Claes), [email protected] (D. Kuypers), [email protected] (R. Oyen). 1 Tel.: +32 16 343782; fax: +32 16 343765. 2 Tel.: +32 16 34 45 80; fax: +32 16 34 45 99. 0720-048X/$ – see front matter © 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2011.09.004
This study was compliant to International Conference on Harmonization Guidelines on Good Clinical Practice (ICH-GCP) principles and was approved by the institutional Ethics committee (S52134). The requirement for informed patient consent was waived given the retrospective nature of the study.
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303 Table 1 Patient characteristics.
Patient age (year) AVF age (month)
Sex Male Female Prior AVF No 1 2 3 Cause end-stage renal disease Diabetes Nefroangiosclerosis/renal vascular disease Glomerular disease ADPKDa Tubulointerstitial Other/unknown AVF type Radiocephalic Brachiocephalic Brachiobasilic Transposed brachiobasilic Brachial artery–median vein AVF side Left Right Diabetes mellitus Present Absent a
Mean ± SD
Range
66.03 ± 13.41 18.95 ± 20.45
19–88 1.33–128.07
N = 162
%
100 62
61.73 38.27
139 20 5 1
84.24 12.12 3.03 0.61
40 35 20 17 20 30
24.69 21.60 12.35 10.49 12.35 18.52
70 48 15 11 23
41.92 28.74 8.98 6.59 13.77
108 59
64.67 35.33
57 105
35.19 64.81
Autosomal dominant polycystic kidney disease.
2.1. Patient population The records of 162 patients who underwent first time PTA of a dysfunctional native AVF between January 1993 and January 2009 were retrospectively reviewed. A total of 100 men and 62 women were treated. Mean patient age ± standard deviation was 66 ± 13 years. Five patients underwent angioplasty of two different dysfunctional AVF in time, resulting in 167 first time angioplasty procedures. Patient demographics are shown in Table 1. Patients with dysfunctional native AVF referred for fistulography and treatment but without previous history of stenosis or thrombosis were included. Thrombosed AVF requiring first thrombectomy, using thrombus aspiration, thrombolytics or mechanical thrombectomy-devices before PTA were excluded as well as arteriovenous grafts (AVG) and thrombosed nonmature AVF treated by surgical revision. Short segment occlusion/thrombosis treated by PTA alone were included. When AVF was already in use for hemodialysis (n = 159, 95.2%) indications for referral were decreasing flow rates, difficult cannulation, prolonged bleeding from puncture sites, elevated venous pressure/recirculation or upper limb edema. When the access was not yet used (n = 8, 4.8%), indication for fistulography with additional endovascular treatment was either non- or delayed maturation of the AVF or a change in the clinical presentation of the AVF in patients with a mature AVF but not yet in hemodialysis. 2.2. Pretreatment fistulography Although the method of access for fistulography was at the discretion of the interventional radiologist, until 1997 imaging of the dysfunctional AVF was in most of the cases initiated with puncture of the brachial artery with an 18-GA needle after local anesthesia. After digital subtraction angiography (DSA) of the feeding artery, the arteriovenous anastomosis, the outflow vein(s) and the central
2299
veins up to the right atrium, a second puncture of the outflow vein with an 18-GA needle was performed when a >50% stenosis or short segment occlusion was detected. Angioplasty of all stenoses was performed via access in the outflow vein. After 1997 fistulography was usually performed after puncture of the outflow vein close to the arteriovenous anastomosis. In order to obtain opacification of the arteriovenous anastomosis and the distal part of the feeding artery, reflux of contrast-material across the anastomosis was achieved during compression of the outflow vein either with a tourniquet or by manual compression. A second puncture of the outflow vein in the direction of the stenosis was performed when needed. Ultrasound guidance for creating access in the outflow vein was at the discretion of the operator. 2.3. Treatment After identification of a >50% stenosis or a (relatively short) thrombosis of the AVF, the stenosis/occlusion was treated by PTA with conventional or high-pressure non-compliant balloons. In all cases PTA was performed via transvenous approach, after replacing the needle for fistulography by an appropriately sized sheath over a guide wire. Balloon size (median diameter: 6 mm, range: 4–14 mm) was chosen by visual estimation of the diameter of a normal segment of the vessel adjacent to the stenosis. Balloon inflation was maintained for a minimum of 2 min. In 3 cases additional PTA with a peripheral cutting balloon (Boston Scientific, Natick, MA) was performed because of a waist in the conventional balloon during inflation. After ultra-high pressure balloons with a rated burst pressure of 30 ATM (Conquest, Bard, Oakville, Canada) became available, these balloons were used when persistent conventional balloon waist was seen. In one patient minimal waist in the balloon was accepted without additional angioplasty with ultra high-pressure balloons or cutting balloons. In 11 cases additional stenting after PTA with self-expanding stents was performed because of elastic recoil (n = 9) after 3 dilatations or because of (short) central vein occlusion (n = 2). Stents were placed in the juxta-anastomotic region (n = 1, 9.1%), in the draining vein (n = 2, 18.2%), in the cephalic arch (n = 3, 27.3%), in the subclavian vein (n = 1, 9.1%) and in the brachiocephalic vein (n = 4, 36.4%). Heparin was administered at the operator discretion. Doses ranged between 2000 and 5000 IU. After the procedure the sheath(s) remained in place if subsequent hemodialysis was possible or needed; otherwise the sheaths were removed and hemostasis was performed by manual compression. 2.4. Study endpoints and definitions Anatomical variables were location, length and grade of stenosis before and after PTA and the presence of more than one stenosis. For the location of the stenosis, the AVF was divided in 6 different segments: feeding artery, arteriovenous anastomosis, juxta-anastomotic segment of the fistula vein defined as the first 2 cm of the vein distally to the anastomosis, proximal venous outflow (including the cannulation sites), distal venous outflow and central veins. Clinical variables were patient gender, patient age, age of the AVF, AVF type, side of access (left or right arm), diabetes mellitus and prior AVF. Patient age was divided in 3 groups: <65 years, 65–75 years and >75 years. AVF age was also divided in 3 groups: <6 months old, 6–12 months old and >12 months old. Technical success was defined as a residual stenosis ≤ 30% after the procedure. A procedure was considered clinically successful when the AVF was patent and usable for hemodialysis. Complications of the procedure were categorized as minor and major
2300
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
following the Society of Interventional Radiology (SIR) guidelines [6]. Clinical success was defined as a resolution of pretreatment clinical indicators of access malfunction and the ability to provide adequate hemodialysis for at least one session. AVF dysfunction due to a stenosis or occlusion necessitating a new intervention at follow-up was recorded (recurrent dysfunction) as well as the time interval between initial intervention and dysfunction recurrence. The definitions of primary, assisted primary and secondary patency as reported by Gray et al. were used [7]. Primary patency was defined as the interval after PTA until the next AVF thrombosis or repeated intervention [7]. Assisted primary patency was defined as the interval after PTA until the next AVF thrombosis or surgical intervention excluding the treated lesion from the access circuit; percutaneous treatments of recurrent or new access stenosis/occlusion were considered compatible to assisted primary patency [7]. Secondary patency was defined as the interval after PTA until the AVF is surgically declotted, revised or abandoned [7]. 2.5. Statistical analysis Kaplan–Meier estimator was used to estimate recurrence and patency rates. The log rank test was used to compare two groups and the Cox proportional hazard model and was used to evaluate the effect of other explanatory variables. Patients were censored in case of lost to follow-up, renal transplantation, death with a functional AVF or switch to peritoneal dialysis. Cumulative incidence function (CIF) was also used to estimate dysfunction recurrence and patency rates. CIF takes the competing events into account in the statistical analysis. Here the test introduced by Pepe and Mori [8] was used to compare two groups, and the competing risk regression model proposed by Fine and Gray [9] for modeling the effect of other explanatory variable. For analysis of the relationship between stenosis location and recurrence of AVF dysfunction or patency rates, rarely observed stenosis locations (≤2 times) were not included. All analyses have been performed using the statistical package SAS version 9.2 (SAS, Cary, NC, USA). 3. Results A total of 217 stenoses or segmental occlusions were found: angiography demonstrated one stenosis/occlusion in 122 AVF (73.05%); two stenoses or one stenosis and an occlusion in 40 AVF (23.95%) and in 5 AVF (3%) 3 stenoses were found at the draining fistula vein. One stenosis was located at the feeding artery (0.5%); 75 (34.6%) were located at the juxta-anastomotic segment. In 119 (54.8%) cases the stenosis was at the proximal venous outflow segment and in 11 cases (5.1%) at the distal venous outflow (all of them located at the cephalic arch). Eleven (5.1%) central venous stenoses were seen, 9 at the innominate vein and 2 at the subclavian vein. Median and mean length of the stenosis was 16 mm and 21.8 mm, respectively (range: 1 mm–133 mm). Median and mean degree of stenosis was 71% and 72.2%, respectively (range: 51–100%). 3.1. Technical success, clinical success and complication rate Per stenosis, PTA was technically successful in 87.1% of cases (n = 189) after the procedure. In 28 cases (12.9%) residual stenosis was greater than 30%. Median and mean degree of residual stenosis was 38% and 40.9%, respectively (range: 31–60%). Per AVF, technical success, defined as residual stenosis ≤ 30%, was
84.4% (n = 141). None of the anatomical or clinical variables were significantly related to the technical success at AVF level. At stenosis level, there was a linear effect of the initial stenosis grade on the technical success: the higher (R#2, 12) the stenosis grade, the smaller the success probability (p = 0.044). There was also a significant difference in technical success between the different types of AVF (p = 0.042). Technical success was highest in radiocephalic AVF (94.3%) and in brachiobasilic AVF (94.1%) and lowest in brachial artery–median vein AVF (72.4%). Technical success rate in brachiocephalic AVF and transposed brachiobasilic AVF was 81.8% and 87.5%, respectively. The difference in technical success between radiocephalic AVF and brachial artery–median vein AVF was statistically significant (p = 0.030). Clinical success was found in 95.2% of cases (n = 159). Eight patients were not in hemodialysis at the time of PTA because hemodialysis was anticipated but not yet required. Complications during PTA occurred in 5 procedures (3%): 4 (2.4%) minor complications (3 minor hematomas at the puncture site and one contrast extravasation treated by balloon tamponade) and one (0.6%) major complication (stent migration to the right atrium).
3.2. Dysfunction (stenosis and/or occlusion) recurrence During follow-up, no recurrence of stenosis or occlusion was seen in 79 cases (47.3%). In 66 cases (39.5%) the next event at the level of the AVF was a stenosis and in 22 cases (13.2%) an occlusion occurred. Mean and median time until recurrent dysfunction after PTA was 13.1 months and 8.5 months, respectively (range 0.1–80 months). Based on Kaplan–Meier analysis, recurrence rates were 44.7%, 63.1% and 72.2% at 1 year, 2 years and 3 years, respectively. Using the cumulative index function (CIF), recurrence rates were 38.1%, 49.8% and 54.8% at 1 year, 2 years and 3 years, respectively. At stenosis level, 61 stenoses (79.2%) were at the same location, 12 stenoses (15.6%) were both at the same location as before and at a different location and 4 stenoses (5.2%) were located in a different location compared to the initial PTA procedure. At the end of follow-up, 48 AVF (28.7%) were patent and functioning, 27 patients (16.2%) underwent renal transplantation with a functional AVF and 43 patients (25.8%) died with a functional AVF. In 49 cases (29.3%) the AVF was abandoned because of dysfunction/thrombosis or surgery was performed on the AVF. Follow-up ranged from 6 months to 6 years. None of the variables was significantly related to recurrence (stenosis and occlusion) except age of the AVF, where a linear effect of the AVF age was seen: the older the AVF, the smaller the probability of recurrence (Cox model with linear effect of age, p = 0.008). When AVF age was divided in 3 groups (<6 months, 6–12 months and older than 12 months), the difference between the groups regarding recurrence was not significant. This significant linear effect of the age of the AVF was also present for recurrent stenosis alone (not occlusion) (Cox model with linear effect of age, p = 0.0352). For recurrent stenosis (not occlusion) technical successful procedures as defined earlier resulted in a significantly lower recurrence rate (log rank, p = 0.013) (Table 2). Patients with a history of a prior AVF had a higher occlusion occurrence that was significant with the log-rank test (p = 0.0002), but not significant with CIF (p = 0.170). Patients with a right-sided access had a higher occlusion rate during follow-up (20.7% versus 9.3% for left-sided access interventions) that was not significant with the log-rank test (p = 0.064), but marginally significant with CIF (p = 0.043).
1.0818
c
b
1.2455 1.7817
0.1072 0.2503 0.5077a 0.0795a 0.6703 2.5953 4.1051 0.4388a 3.0758a 0.1813
1.5419
0.1653 0.0078a 0.0965 0.1491 0.6804 0.2744 3.5999 7.09a 2.7617 2.0818 2.3018 1.1944
1.0608
0.2799 0.6580 0.4233 0.2753a 1.7195 1.1190a
Log rank to compare two groups, Cox proportional hazard model to evaluate the effect of other explanatory variables. Cumulative incidence function (CIF) to compare two groups, competing risk regression (CRR) for modeling the effect of other explanatory variables. Two locations of stenosis (feeding artery and arteriovenous anastomosis, 0.5% of all stenoses) were excluded for analysis; the remaining locations (99.5% of all stenoses) were used for analysis. a
3.5999 4.4340a 0.0393 1.9444 6.0076 0.0328 0.1007b 0.0074b 0.5968 0.4173 0.7629b 0.3030
6.1510 3.6532 1.4699a 2.9321a 0.6524
2.8306 1.5633a 0.0990b 0.2594b
0.2143 0.2575b 0.8095b 0.2734b 0.1882
0.0001 0.6888 0.1609 0.6436 0.2141
Clinical variables Sex Patient age Categorical Linear AVF age Categorical Linear Prior AVF DM AVF type AVF side (L/R) Anatomical variables Technical success Location stenosisc Degree of stenosis (linear) Length of stenosis (linear) Multiple stenoses
2301
3.3. Early dysfunction (stenosis and/or occlusion) recurrence within 6 months
0.2123b 0.6653b 0.2983
0.18878 1.7271
0.9425 1.8386 0.5712a 0.2567a 0.3034 0.0131 0.3014 0.2254a 0.868a 0.4193
12.544
0.0004 0.4795b 0.3526b 0.2022b 0.2644
0.3316 0.6066 0.4498a 0.6124a 0.5817
4.1048 0.0148
0.1653 0.0352a 0.8428 0.1632 0.1986 0.8564
0.2429 0.2112a
0.0984 0.4707
3.3043 2.7365a 14.05 0.2125 4.4244 3.4317 0.2746b 0.0589b 0.7538 0.4926 0.1804b 0.9032
0.9468 0.9763a 0.1093 0.0009a 0.0538b 0.1193b
0.1916 0.0981a 0.0002 0.6448 0.3516 0.0640
1.8796 0.5839
0.3088b 0.2501b 0.1704 0.4448 0.0763b 0.0428
0.7737b 0.6617b
0.2928 1.1069 0.3280 0.9567 0.0193 0.9943
0.8895
Log-rank/Cox-modela CIF/CRR Log-rank/Cox-model
Log-rank/Cox-model
a
Stenosis + occlusion Dysfunction recurrence
Table 2 Recurrence of dysfunction (stenosis and/or occlusion).
p-Value
CIF/CRR
b
p-Value
Stenosis
a
p-Value
b
p-Value
Occlusion
p-Value
CIF/CRRb
p-Value
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
Dysfunction recurrence within 6 months was seen in 38 AVF (27.3%), 32 recurrent stenoses (23.%) and 6 occlusions (4.3%). Twenty-eight AVF were censored (end of study, death or other competing event) and 101 (72.7%) AVF remained patent without problems during the first 6 months. The only variables with a significant effect on early dysfunction were age of the AVF and diabetes mellitus. A linear effect of AVF age was seen, with a higher recurrence probability for patients with younger AVF (p = 0.016). Patients with diabetes mellitus had also a higher probability of recurrence within 6 months (p = 0.003). The same negative relationship between AVF age and recurrence probability (p = 0.033) was seen at stenosis level as well as the higher probability of recurrence in patients with diabetes mellitus (p = 0.005). Since only 6 occlusions within 6 months of follow-up were observed, no further analyses were performed at occlusion level. 3.4. Primary patency Primary patency at 1 year, 2 years and 3 years was 48.5%, 31.4% and 22.5%, respectively based on Kaplan–Meier analysis. With CIF, primary patency at 1 year, 2 years and 3 years was 52.1%, 39.4% and 33.5%, respectively. The only variable with a significant effect on the primary patency was the age of the AVF for both the Cox model with linear effect (p = 0.009) and the competition risk regression test for linear effect of AVF age (p = 0.005): the older the AVF, the higher primary patency. 3.5. Assisted primary patency Assisted primary patency at 1 year, 2 years and 3 years was 77.6%, 57.3% and 46.5%, respectively based on Kaplan–Meier analysis. With CIF, assisted primary patency at 1 year, 2 years and 3 years was 79.2%, 64.6% and 58%, respectively. There was a significant difference in assisted primary patency between the different age groups: patients older than 75 years of age had a shorter assisted primary patency compared to patients between 65 and 75 years of age (log-rank, p = 0.011). This difference was not significant using competing risk regression as statistical model. 3.6. Secondary patency Secondary patency at 1 year, 2 years and 3 years was 83.6%, 68.4% and 60.8%, respectively (Kaplan–Meier analysis); with CIF this was 84.7%, 74.3% and 70.1%, respectively. There was a significant difference in secondary patency between different age groups: patients older than 75 years of age had a shorter secondary patency compared to patients between 65 and 75 years of age (log-rank, p = 0.014). This difference was not significant using competing risk regression as statistical model. Patients with an AVF created less than 6 months before had a significantly shorter secondary patency than patients with an AVF older than 6 months (p = 0.037 for AVF between 6 and 12 months old and p = 0.005 for AVF older than 1 year). 4. Discussion Both North American and European guidelines on hemodialysis access state that an AVF is the preferred type of access for hemodialysis [1,10]. However, in a majority of AVF stenosis will develop over time [1]. A hemodynamically relevant stenosis (>50%) may reduce dialysis efficacy or may cause access thrombosis when left untreated [1].
2302
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
In our series of native AVF with first-time dysfunction (no interventions before), the vast majority of the patients presented with one or two significant stenosis/stenoses, with nearly two-thirds of the patients having one stenosis. This is comparable to other reports [3,4]. The higher percentage of stenoses in the proximal venous outflow segment (54.8%) compared to the juxta-anastomotic region (34.6%) is different to other reports [2–4], where 47–64% of stenoses were reported to be in the juxta-anastomotic region. This difference is mainly related to the strict use of our definition of juxtaanastomotic segment within 2 cm of the anastomosis; a stenosis at 2.2 cm of the anastomosis was therefore considered a proximal venous outflow segment stenosis. As already shown by Clark et al. and Turmel-Rodrigues et al., most stenoses in AVF are short, with respectively, 42.6% and 86% of stenoses less than 2 cm long in their series [2,3]. Our results of a median and mean length of stenosis of 16 mm and 21.8 mm, respectively, reflect this. Technical and clinical success rates were comparable to other studies, where often clinical success rates exceed technical success rates and where it is suggested that an autologous vein is more subject to spasm, elastic recoil or other factors resulting in a residual stenosis of 30% or more [3,4,11–13]. As in a study of Trerotola et al. [14] we found a linear relationship between severity of initial stenosis and residual stenosis. Recurrent stenosis was significantly more seen in AVF with a technically unsuccessful outcome after PTA compared to technically successful procedures; early recurrence was however not influenced by technically successful or unsuccessful procedures. In addition patency rates were not significantly different between technically successful and unsuccessful procedures. This means that even though there is a higher risk for recurrent stenosis when residual stenosis after PTA is 30% or more, this will not affect AVF durability. In a study of Clark et al. on 65 dysfunctional or occluded AVF no difference in long-term patency was found between AVF with a >30% residual stenosis compared to <30% residual stenosis after treatment [3]. Similar findings were demonstrated by Forauer et al. who showed significantly higher technical success rate with balloon angioplasty inflation times of 3 min versus 1 min, but no significant post-intervention patency differences [15]. A higher technical success rate for radiocephalic AVF compared to upper arm AVF (and especially brachial artery–median vein AVF) was seen in our study. Trerotola et al. also mentioned access type as a significant variable for residual stenosis, but in this study access type was stratified in fistula, graft or hybrid, without review of the different types of AVF [14]. Turmel-Rodrigues et al. did not see a difference in success rate between forearm and upper arm AVF, with a success rate of 95 ± 2% and 97 ± 3%, respectively [2]. The reason for this lower technical success rate in brachial artery–median vein AVF is not clear, but it may have been because of a less aggressive approach (no overdilatation of stenoses), to elastic recoil or to an underestimation of residual stenosis in AVF with a larger diameter of fistula veins by visual estimate. Over 50% of dysfunctional AVF treated with PTA in our study showed a functional stenosis (in 75% of cases) or occlusion (in 25% of cases) during follow-up. Moreover almost 95% of the stenoses found during follow-up were at the same location as the first-time stenosis. These findings suggest that durability of PTA is only moderate. The mean recurrence time of 13.1 months is within range of the results in the study of Turmel-Rodrigues et al., where the mean interval between reintervention and initial dilatation was 18.1 months and 10.9 months for forearm and upper arm AVF, respectively [2]. Tessitori et al. found a median time to restenosis of 8 months [16]. Similar to the study of Turmel-Rodrigues et al., where a significantly shorter interval between reinterventions was seen in forearm AVF created less than 1 year before compared to AVF older than 1 year, we found a significant effect of AVF age on dysfunction recurrence [2]. This effect was only linear (the older the AVF, the smaller the probability of recurrence) and could not
be shown when we tried to stratify AVF age in groups younger than 6 months, older than 6 months but younger than 1 year and older than 1 year. The effect of prior AVF on dysfunction recurrence was only significant using the log-rank statistical test, but not significant with the cumulative incidence function model. Inversely the effect of AVF side was significant when calculated with the cumulative incidence function model, but not with the log-rank model (albeit a tendency towards significance was seen). Both analyses are valid but differ in interpretation of the results, because in the log-rank model, the competing events are considered as censored observations, while in the cumulative incidence function model the competing events are taken into account. This may result in conflicting outcomes and these variables were therefore not considered as variables with an absolute clinical effect on dysfunction recurrence. Early recurrence, defined as recurrent dysfunction (stenosis or occlusion) within 6 months after initial PTA was seen in more than 25% of cases and the majority were recurrent stenoses. Both AVF age and diabetes mellitus were variables with a significant effect on early recurrent dysfunction and may have implications for surveillance of AVF after initial PTA. The patency rates observed in our study were within range with the patency rates reported [2–4,11]. The linear effect of AVF age on primary patency was to be expected given the effect of AVF age on dysfunction recurrence. This effect was also demonstrated by Turmel-Rodrigues et al. [2] as mentioned before and by Clark et al. [3], who showed a substantial (but not significant) association between lower primary patency rates and AVF younger than 6 months. Secondary patency rates were also influenced by AVF age, but in contrast to primary patency, this effect was not linear but categorical. AVF younger than 6 months showed lower secondary patency rates compared to AVF older than 6 months. This cannot solely be explained by the 20% non-maturing AVF in this group of young AVF, because non-maturating AVF treated by PTA have similar patency rates compared to mature AVF in most studies [17,18]. Only Manninen et al. reported a significant lower secondary patency rate in percutaneously treated non-maturing AVF compared to mature AVF and suggested that this might have been related to poorer quality veins in non-mature AVF [13]. The effect of patient age on patency was only significant using the log-rank statistical test, but not significant with the competing risk regression model. Similar to the variables that may have an effect on dysfunction recurrence this resulted in conflicting outcomes. In the group of patients older than 75 years of age, the patency rate may have been interpreted lower in the log-rank model compared to the competing risk regression model, because in this group more patients will have died during follow-up and thus more patients will have been censored. As a result patient age was not considered as a variable with an absolute significant effect on assisted primary or secondary patency. Major limitations of the study were that this was a retrospective study from a single institution, without a control group and the relative small number of patients. In conclusion, technical success of PTA in hemodialysis AVF is influenced by the initial grade of stenosis and by AVF type, with highest success rate seen in radiocephalic AVF. Technical success itself has a significant effect on dysfunction recurrence, but not on AVF longevity. AVF age has a significant effect on dysfunction recurrence and on patency; younger AVF have an increased risk for (early) recurrent dysfunction and for lower patency rates. The only other variable with a significant effect on recurrent dysfunction within 6 months after PTA is diabetes mellitus. These results may help in tailoring surveillance programs, but large prospective trials are still warranted.
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
References [1] National Kidney Foundation. NKF-K/DOQI Clinical practice guidelines and clinical practice recommendations for 2006 updates: hemodialysis adequacy, peritoneal dialysis adequacy and vascular access. Am J Kidney Dis 2006;48(suppl. 1):S188–91. [2] Turmel-Rodrigues L, Pengloan J, Baudin S, et al. Treatment of stenosis and thrombosis in haemodialysis fistulas and grafts by interventional radiology. Nephrol Dial Transplant 2000;15:2029–36. [3] Clark TWI, Hirsch DA, Jindal KJ, Veugelers PJ, LeBlanc J. Outcome and prognostic factors of restenosis after percutaneous treatment of native hemodialysis fistulas. J Vasc Interv Radiol 2002;13:51–9. [4] Rajan DK, Bunston S, Misra S, Pinto R, Lok CE. Dysfunctional autogenous hemodialysis fistulas: outcomes after angioplasty – are there clinical predictors of patency? Radiology 2004;232:508–15. [5] Falk A. Maintenance and salvage of arteriovenous fistulas. J Vasc Interv Radiol 2006;17:807–13. [6] Sacks D, McClenny TE, Cardella JF, Lewis CA. Society of interventional radiology clinical practice guidelines. J Vasc Interv Radiol 2003;14(suppl):S199–202. [7] Gray RJ, Sacks D, Martin LG, Trerotola SO. Society of Interventional Radiology Technology Assessment Committee. Reporting standards for percutaneous interventions in dialysis access. J Vasc Interv Radiol 2003;14:S433–42. [8] Pepe MS, Mori M. Kaplan–Meier, marginal or conditional probability curves in summarizing competing risks failure time data? Stat Med 1993;12:737–51. [9] Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509.
2303
[10] Tordoir J, Canaud B, Haage P, et al. EBPG on vascular access. Nephrol Dial Transplant 2007;22(suppl 2):ii88–117. [11] Manninen HI, Kaukanen ET, Ikäheimo R, et al. Brachial artery access: endovascular treatment of failing Brescia–Cimino hemodialysis fistulas – initial success and long-term results. Radiology 2001;218:711–8. [12] Maeda K, Furukawa A, Yamasaki M, Murata K. Percutaneous transluminal angioplasty for Brescia–Cimino hemodialysis fistula dysfunction: technical success rate, patency rate and factors that influence the results. Eur J Radiol 2005;54:426–30. [13] Manninen HI, Kaukanen ET, Mäkinen K, Karhapää P. Endovascular salvage of nonmaturing autogenous hemodialysis fistulas: comparison with endovascular therapy of failing mature fistulas. J Vasc Interv Radiol 2008;19:870–6. [14] Trerotola SO, Kwak A, Clark TWI, et al. Prospective study of balloon inflation pressures and other technical aspects of hemodialysis access angioplasty. J Vasc Interv Radiol 2005;16:1613–8. [15] Forauer AR, Hoffer EK, Homa K. Dialysis access venous stenoses: treatment with balloon angioplasty – 1-min versus 3-min inflation times. Radiology 2008;249:375–81. [16] Tessitori N, Mansueto G, Bedogna V, et al. A prospective controlled trial on effect of percutaneous transluminal angioplasty on functioning arteriovenous fistulae survival. J Am Soc Nephrol 2003;14:1623–7. [17] Turmel-Rodrigues L, Mouton A, Birmelé B, et al. Salvage of immature forearm fistulas for haemodialysis by interventional radiology. Nephrol Dial Transplant 2001;16:2365–71. [18] Clark TWI, Cohen RA, KWak A, et al. Salvage of nonmaturing native fistulas by using angioplasty. Radiology 2007;242:286–92.
Contents lists available at SciVerse ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Factors influencing technical success and outcome of percutaneous balloon angioplasty in de novo native hemodialysis arteriovenous fistulas Sam Heye a,∗ , Geert Maleux a,1 , Johan Vaninbroukx a,1 , Kathleen Claes b,2 , Dirk Kuypers b,2 , Raymond Oyen a,1 a b
University Hospitals Leuven, Department of Radiology, Herestraat 49, 3000 Leuven, Belgium University Hospitals Leuven, Department of Nephrology, Herestraat 49, 3000 Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 20 July 2011 Received in revised form 2 September 2011 Accepted 6 September 2011 Keywords: Dialysis Arteriovenous fistula Angioplasty Balloon Treatment outcome
a b s t r a c t Objective: To determine predictors of technical success, dysfunction recurrence and patency after percutaneous transluminal angioplasty (PTA) of de novo dysfunctional hemodialysis arteriovenous fistulas (AVFs). Methods: Retrospective analysis of first time PTA of 167 AVF in 162 patients (100 men, 66 ± 13 years). Anatomical (location, length, grade and number of stenoses) and clinical variables (sex, age, prior AVF, diabetes mellitus and AVF age, side and location) were reviewed. Results: 217 stenoses or segmental occlusions were treated. Technical success rate (84.4%) was higher in radiocephalic AVF compared to brachial artery–median vein AVF (p = 0.030) and was negatively correlated with initial stenosis (p = 0.049). Dysfunction recurrence was seen in 52.7% and correlated negatively with technical success (p = 0.013) and AVF age (p = 0.008). Early dysfunction (within 6 months) was negatively correlated with AVF age (p = 0.016) and positively with diabetes (p = 0.003). Higher AVF age resulted in higher primary (p = 0.005) and secondary patency rates (p = 0.037–0.005). Conclusions: Technical success of PTA in hemodialysis AVF is affected by AVF type and initial stenosis and has significant effect on dysfunction recurrence, but not on AVF longevity. Younger AVF has increased risk for (early) recurrent dysfunction and lower patency rates. Patients with diabetes mellitus have higher risk for early dysfunction. © 2011 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Arteriovenous fistulas (AVFs) are the preferred type of access in patients who undergo hemodialysis [1]. Hemodynamic significant stenosis causing access dysfunction is a frequent complication and requires (repeated) percutaneous transluminal balloon angioplasty (PTA) to maintain patency [2–5]. Durability of PTA is however limited, with 1-year primary patency rates ranging between 26% and 62% [2–4]. Variables influencing outcome were studied in previously reported series [3,4]. In a series of 65 dysfunctional or occluded AVF Clark et al. showed that AVF with a stenosis longer than 2 cm in length and patients with at least one comorbid factor such as diabetes, coronary artery disease or peripheral vascular
disease had a significantly increased risk for access patency loss [3]. No variables affected the outcome of PTA in 151 dysfunctional AVF in a study of Rajan et al. [4]. Not only there were some conflicting results between these 2 studies, but also it was not clear whether previous interventions were performed on the studied AVF or not. Moreover these studies reported little on the impact of technical success on the outcome or on the location of the stenosis in case of recurrent dysfunction. The objective of our study was to determine anatomical and clinical variables that were predictors of technical success, dysfunction recurrence and patency after PTA on hemodialysis AVF that have not yet been treated for dysfunction or thrombosis.
2. Materials and methods ∗ Corresponding author. Tel.: +32 16 343782; fax: +32 16 343765. E-mail addresses: [email protected] (S. Heye), [email protected] (G. Maleux), [email protected] (J. Vaninbroukx), [email protected] (K. Claes), [email protected] (D. Kuypers), [email protected] (R. Oyen). 1 Tel.: +32 16 343782; fax: +32 16 343765. 2 Tel.: +32 16 34 45 80; fax: +32 16 34 45 99. 0720-048X/$ – see front matter © 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2011.09.004
This study was compliant to International Conference on Harmonization Guidelines on Good Clinical Practice (ICH-GCP) principles and was approved by the institutional Ethics committee (S52134). The requirement for informed patient consent was waived given the retrospective nature of the study.
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303 Table 1 Patient characteristics.
Patient age (year) AVF age (month)
Sex Male Female Prior AVF No 1 2 3 Cause end-stage renal disease Diabetes Nefroangiosclerosis/renal vascular disease Glomerular disease ADPKDa Tubulointerstitial Other/unknown AVF type Radiocephalic Brachiocephalic Brachiobasilic Transposed brachiobasilic Brachial artery–median vein AVF side Left Right Diabetes mellitus Present Absent a
Mean ± SD
Range
66.03 ± 13.41 18.95 ± 20.45
19–88 1.33–128.07
N = 162
%
100 62
61.73 38.27
139 20 5 1
84.24 12.12 3.03 0.61
40 35 20 17 20 30
24.69 21.60 12.35 10.49 12.35 18.52
70 48 15 11 23
41.92 28.74 8.98 6.59 13.77
108 59
64.67 35.33
57 105
35.19 64.81
Autosomal dominant polycystic kidney disease.
2.1. Patient population The records of 162 patients who underwent first time PTA of a dysfunctional native AVF between January 1993 and January 2009 were retrospectively reviewed. A total of 100 men and 62 women were treated. Mean patient age ± standard deviation was 66 ± 13 years. Five patients underwent angioplasty of two different dysfunctional AVF in time, resulting in 167 first time angioplasty procedures. Patient demographics are shown in Table 1. Patients with dysfunctional native AVF referred for fistulography and treatment but without previous history of stenosis or thrombosis were included. Thrombosed AVF requiring first thrombectomy, using thrombus aspiration, thrombolytics or mechanical thrombectomy-devices before PTA were excluded as well as arteriovenous grafts (AVG) and thrombosed nonmature AVF treated by surgical revision. Short segment occlusion/thrombosis treated by PTA alone were included. When AVF was already in use for hemodialysis (n = 159, 95.2%) indications for referral were decreasing flow rates, difficult cannulation, prolonged bleeding from puncture sites, elevated venous pressure/recirculation or upper limb edema. When the access was not yet used (n = 8, 4.8%), indication for fistulography with additional endovascular treatment was either non- or delayed maturation of the AVF or a change in the clinical presentation of the AVF in patients with a mature AVF but not yet in hemodialysis. 2.2. Pretreatment fistulography Although the method of access for fistulography was at the discretion of the interventional radiologist, until 1997 imaging of the dysfunctional AVF was in most of the cases initiated with puncture of the brachial artery with an 18-GA needle after local anesthesia. After digital subtraction angiography (DSA) of the feeding artery, the arteriovenous anastomosis, the outflow vein(s) and the central
2299
veins up to the right atrium, a second puncture of the outflow vein with an 18-GA needle was performed when a >50% stenosis or short segment occlusion was detected. Angioplasty of all stenoses was performed via access in the outflow vein. After 1997 fistulography was usually performed after puncture of the outflow vein close to the arteriovenous anastomosis. In order to obtain opacification of the arteriovenous anastomosis and the distal part of the feeding artery, reflux of contrast-material across the anastomosis was achieved during compression of the outflow vein either with a tourniquet or by manual compression. A second puncture of the outflow vein in the direction of the stenosis was performed when needed. Ultrasound guidance for creating access in the outflow vein was at the discretion of the operator. 2.3. Treatment After identification of a >50% stenosis or a (relatively short) thrombosis of the AVF, the stenosis/occlusion was treated by PTA with conventional or high-pressure non-compliant balloons. In all cases PTA was performed via transvenous approach, after replacing the needle for fistulography by an appropriately sized sheath over a guide wire. Balloon size (median diameter: 6 mm, range: 4–14 mm) was chosen by visual estimation of the diameter of a normal segment of the vessel adjacent to the stenosis. Balloon inflation was maintained for a minimum of 2 min. In 3 cases additional PTA with a peripheral cutting balloon (Boston Scientific, Natick, MA) was performed because of a waist in the conventional balloon during inflation. After ultra-high pressure balloons with a rated burst pressure of 30 ATM (Conquest, Bard, Oakville, Canada) became available, these balloons were used when persistent conventional balloon waist was seen. In one patient minimal waist in the balloon was accepted without additional angioplasty with ultra high-pressure balloons or cutting balloons. In 11 cases additional stenting after PTA with self-expanding stents was performed because of elastic recoil (n = 9) after 3 dilatations or because of (short) central vein occlusion (n = 2). Stents were placed in the juxta-anastomotic region (n = 1, 9.1%), in the draining vein (n = 2, 18.2%), in the cephalic arch (n = 3, 27.3%), in the subclavian vein (n = 1, 9.1%) and in the brachiocephalic vein (n = 4, 36.4%). Heparin was administered at the operator discretion. Doses ranged between 2000 and 5000 IU. After the procedure the sheath(s) remained in place if subsequent hemodialysis was possible or needed; otherwise the sheaths were removed and hemostasis was performed by manual compression. 2.4. Study endpoints and definitions Anatomical variables were location, length and grade of stenosis before and after PTA and the presence of more than one stenosis. For the location of the stenosis, the AVF was divided in 6 different segments: feeding artery, arteriovenous anastomosis, juxta-anastomotic segment of the fistula vein defined as the first 2 cm of the vein distally to the anastomosis, proximal venous outflow (including the cannulation sites), distal venous outflow and central veins. Clinical variables were patient gender, patient age, age of the AVF, AVF type, side of access (left or right arm), diabetes mellitus and prior AVF. Patient age was divided in 3 groups: <65 years, 65–75 years and >75 years. AVF age was also divided in 3 groups: <6 months old, 6–12 months old and >12 months old. Technical success was defined as a residual stenosis ≤ 30% after the procedure. A procedure was considered clinically successful when the AVF was patent and usable for hemodialysis. Complications of the procedure were categorized as minor and major
2300
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
following the Society of Interventional Radiology (SIR) guidelines [6]. Clinical success was defined as a resolution of pretreatment clinical indicators of access malfunction and the ability to provide adequate hemodialysis for at least one session. AVF dysfunction due to a stenosis or occlusion necessitating a new intervention at follow-up was recorded (recurrent dysfunction) as well as the time interval between initial intervention and dysfunction recurrence. The definitions of primary, assisted primary and secondary patency as reported by Gray et al. were used [7]. Primary patency was defined as the interval after PTA until the next AVF thrombosis or repeated intervention [7]. Assisted primary patency was defined as the interval after PTA until the next AVF thrombosis or surgical intervention excluding the treated lesion from the access circuit; percutaneous treatments of recurrent or new access stenosis/occlusion were considered compatible to assisted primary patency [7]. Secondary patency was defined as the interval after PTA until the AVF is surgically declotted, revised or abandoned [7]. 2.5. Statistical analysis Kaplan–Meier estimator was used to estimate recurrence and patency rates. The log rank test was used to compare two groups and the Cox proportional hazard model and was used to evaluate the effect of other explanatory variables. Patients were censored in case of lost to follow-up, renal transplantation, death with a functional AVF or switch to peritoneal dialysis. Cumulative incidence function (CIF) was also used to estimate dysfunction recurrence and patency rates. CIF takes the competing events into account in the statistical analysis. Here the test introduced by Pepe and Mori [8] was used to compare two groups, and the competing risk regression model proposed by Fine and Gray [9] for modeling the effect of other explanatory variable. For analysis of the relationship between stenosis location and recurrence of AVF dysfunction or patency rates, rarely observed stenosis locations (≤2 times) were not included. All analyses have been performed using the statistical package SAS version 9.2 (SAS, Cary, NC, USA). 3. Results A total of 217 stenoses or segmental occlusions were found: angiography demonstrated one stenosis/occlusion in 122 AVF (73.05%); two stenoses or one stenosis and an occlusion in 40 AVF (23.95%) and in 5 AVF (3%) 3 stenoses were found at the draining fistula vein. One stenosis was located at the feeding artery (0.5%); 75 (34.6%) were located at the juxta-anastomotic segment. In 119 (54.8%) cases the stenosis was at the proximal venous outflow segment and in 11 cases (5.1%) at the distal venous outflow (all of them located at the cephalic arch). Eleven (5.1%) central venous stenoses were seen, 9 at the innominate vein and 2 at the subclavian vein. Median and mean length of the stenosis was 16 mm and 21.8 mm, respectively (range: 1 mm–133 mm). Median and mean degree of stenosis was 71% and 72.2%, respectively (range: 51–100%). 3.1. Technical success, clinical success and complication rate Per stenosis, PTA was technically successful in 87.1% of cases (n = 189) after the procedure. In 28 cases (12.9%) residual stenosis was greater than 30%. Median and mean degree of residual stenosis was 38% and 40.9%, respectively (range: 31–60%). Per AVF, technical success, defined as residual stenosis ≤ 30%, was
84.4% (n = 141). None of the anatomical or clinical variables were significantly related to the technical success at AVF level. At stenosis level, there was a linear effect of the initial stenosis grade on the technical success: the higher (R#2, 12) the stenosis grade, the smaller the success probability (p = 0.044). There was also a significant difference in technical success between the different types of AVF (p = 0.042). Technical success was highest in radiocephalic AVF (94.3%) and in brachiobasilic AVF (94.1%) and lowest in brachial artery–median vein AVF (72.4%). Technical success rate in brachiocephalic AVF and transposed brachiobasilic AVF was 81.8% and 87.5%, respectively. The difference in technical success between radiocephalic AVF and brachial artery–median vein AVF was statistically significant (p = 0.030). Clinical success was found in 95.2% of cases (n = 159). Eight patients were not in hemodialysis at the time of PTA because hemodialysis was anticipated but not yet required. Complications during PTA occurred in 5 procedures (3%): 4 (2.4%) minor complications (3 minor hematomas at the puncture site and one contrast extravasation treated by balloon tamponade) and one (0.6%) major complication (stent migration to the right atrium).
3.2. Dysfunction (stenosis and/or occlusion) recurrence During follow-up, no recurrence of stenosis or occlusion was seen in 79 cases (47.3%). In 66 cases (39.5%) the next event at the level of the AVF was a stenosis and in 22 cases (13.2%) an occlusion occurred. Mean and median time until recurrent dysfunction after PTA was 13.1 months and 8.5 months, respectively (range 0.1–80 months). Based on Kaplan–Meier analysis, recurrence rates were 44.7%, 63.1% and 72.2% at 1 year, 2 years and 3 years, respectively. Using the cumulative index function (CIF), recurrence rates were 38.1%, 49.8% and 54.8% at 1 year, 2 years and 3 years, respectively. At stenosis level, 61 stenoses (79.2%) were at the same location, 12 stenoses (15.6%) were both at the same location as before and at a different location and 4 stenoses (5.2%) were located in a different location compared to the initial PTA procedure. At the end of follow-up, 48 AVF (28.7%) were patent and functioning, 27 patients (16.2%) underwent renal transplantation with a functional AVF and 43 patients (25.8%) died with a functional AVF. In 49 cases (29.3%) the AVF was abandoned because of dysfunction/thrombosis or surgery was performed on the AVF. Follow-up ranged from 6 months to 6 years. None of the variables was significantly related to recurrence (stenosis and occlusion) except age of the AVF, where a linear effect of the AVF age was seen: the older the AVF, the smaller the probability of recurrence (Cox model with linear effect of age, p = 0.008). When AVF age was divided in 3 groups (<6 months, 6–12 months and older than 12 months), the difference between the groups regarding recurrence was not significant. This significant linear effect of the age of the AVF was also present for recurrent stenosis alone (not occlusion) (Cox model with linear effect of age, p = 0.0352). For recurrent stenosis (not occlusion) technical successful procedures as defined earlier resulted in a significantly lower recurrence rate (log rank, p = 0.013) (Table 2). Patients with a history of a prior AVF had a higher occlusion occurrence that was significant with the log-rank test (p = 0.0002), but not significant with CIF (p = 0.170). Patients with a right-sided access had a higher occlusion rate during follow-up (20.7% versus 9.3% for left-sided access interventions) that was not significant with the log-rank test (p = 0.064), but marginally significant with CIF (p = 0.043).
1.0818
c
b
1.2455 1.7817
0.1072 0.2503 0.5077a 0.0795a 0.6703 2.5953 4.1051 0.4388a 3.0758a 0.1813
1.5419
0.1653 0.0078a 0.0965 0.1491 0.6804 0.2744 3.5999 7.09a 2.7617 2.0818 2.3018 1.1944
1.0608
0.2799 0.6580 0.4233 0.2753a 1.7195 1.1190a
Log rank to compare two groups, Cox proportional hazard model to evaluate the effect of other explanatory variables. Cumulative incidence function (CIF) to compare two groups, competing risk regression (CRR) for modeling the effect of other explanatory variables. Two locations of stenosis (feeding artery and arteriovenous anastomosis, 0.5% of all stenoses) were excluded for analysis; the remaining locations (99.5% of all stenoses) were used for analysis. a
3.5999 4.4340a 0.0393 1.9444 6.0076 0.0328 0.1007b 0.0074b 0.5968 0.4173 0.7629b 0.3030
6.1510 3.6532 1.4699a 2.9321a 0.6524
2.8306 1.5633a 0.0990b 0.2594b
0.2143 0.2575b 0.8095b 0.2734b 0.1882
0.0001 0.6888 0.1609 0.6436 0.2141
Clinical variables Sex Patient age Categorical Linear AVF age Categorical Linear Prior AVF DM AVF type AVF side (L/R) Anatomical variables Technical success Location stenosisc Degree of stenosis (linear) Length of stenosis (linear) Multiple stenoses
2301
3.3. Early dysfunction (stenosis and/or occlusion) recurrence within 6 months
0.2123b 0.6653b 0.2983
0.18878 1.7271
0.9425 1.8386 0.5712a 0.2567a 0.3034 0.0131 0.3014 0.2254a 0.868a 0.4193
12.544
0.0004 0.4795b 0.3526b 0.2022b 0.2644
0.3316 0.6066 0.4498a 0.6124a 0.5817
4.1048 0.0148
0.1653 0.0352a 0.8428 0.1632 0.1986 0.8564
0.2429 0.2112a
0.0984 0.4707
3.3043 2.7365a 14.05 0.2125 4.4244 3.4317 0.2746b 0.0589b 0.7538 0.4926 0.1804b 0.9032
0.9468 0.9763a 0.1093 0.0009a 0.0538b 0.1193b
0.1916 0.0981a 0.0002 0.6448 0.3516 0.0640
1.8796 0.5839
0.3088b 0.2501b 0.1704 0.4448 0.0763b 0.0428
0.7737b 0.6617b
0.2928 1.1069 0.3280 0.9567 0.0193 0.9943
0.8895
Log-rank/Cox-modela CIF/CRR Log-rank/Cox-model
Log-rank/Cox-model
a
Stenosis + occlusion Dysfunction recurrence
Table 2 Recurrence of dysfunction (stenosis and/or occlusion).
p-Value
CIF/CRR
b
p-Value
Stenosis
a
p-Value
b
p-Value
Occlusion
p-Value
CIF/CRRb
p-Value
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
Dysfunction recurrence within 6 months was seen in 38 AVF (27.3%), 32 recurrent stenoses (23.%) and 6 occlusions (4.3%). Twenty-eight AVF were censored (end of study, death or other competing event) and 101 (72.7%) AVF remained patent without problems during the first 6 months. The only variables with a significant effect on early dysfunction were age of the AVF and diabetes mellitus. A linear effect of AVF age was seen, with a higher recurrence probability for patients with younger AVF (p = 0.016). Patients with diabetes mellitus had also a higher probability of recurrence within 6 months (p = 0.003). The same negative relationship between AVF age and recurrence probability (p = 0.033) was seen at stenosis level as well as the higher probability of recurrence in patients with diabetes mellitus (p = 0.005). Since only 6 occlusions within 6 months of follow-up were observed, no further analyses were performed at occlusion level. 3.4. Primary patency Primary patency at 1 year, 2 years and 3 years was 48.5%, 31.4% and 22.5%, respectively based on Kaplan–Meier analysis. With CIF, primary patency at 1 year, 2 years and 3 years was 52.1%, 39.4% and 33.5%, respectively. The only variable with a significant effect on the primary patency was the age of the AVF for both the Cox model with linear effect (p = 0.009) and the competition risk regression test for linear effect of AVF age (p = 0.005): the older the AVF, the higher primary patency. 3.5. Assisted primary patency Assisted primary patency at 1 year, 2 years and 3 years was 77.6%, 57.3% and 46.5%, respectively based on Kaplan–Meier analysis. With CIF, assisted primary patency at 1 year, 2 years and 3 years was 79.2%, 64.6% and 58%, respectively. There was a significant difference in assisted primary patency between the different age groups: patients older than 75 years of age had a shorter assisted primary patency compared to patients between 65 and 75 years of age (log-rank, p = 0.011). This difference was not significant using competing risk regression as statistical model. 3.6. Secondary patency Secondary patency at 1 year, 2 years and 3 years was 83.6%, 68.4% and 60.8%, respectively (Kaplan–Meier analysis); with CIF this was 84.7%, 74.3% and 70.1%, respectively. There was a significant difference in secondary patency between different age groups: patients older than 75 years of age had a shorter secondary patency compared to patients between 65 and 75 years of age (log-rank, p = 0.014). This difference was not significant using competing risk regression as statistical model. Patients with an AVF created less than 6 months before had a significantly shorter secondary patency than patients with an AVF older than 6 months (p = 0.037 for AVF between 6 and 12 months old and p = 0.005 for AVF older than 1 year). 4. Discussion Both North American and European guidelines on hemodialysis access state that an AVF is the preferred type of access for hemodialysis [1,10]. However, in a majority of AVF stenosis will develop over time [1]. A hemodynamically relevant stenosis (>50%) may reduce dialysis efficacy or may cause access thrombosis when left untreated [1].
2302
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
In our series of native AVF with first-time dysfunction (no interventions before), the vast majority of the patients presented with one or two significant stenosis/stenoses, with nearly two-thirds of the patients having one stenosis. This is comparable to other reports [3,4]. The higher percentage of stenoses in the proximal venous outflow segment (54.8%) compared to the juxta-anastomotic region (34.6%) is different to other reports [2–4], where 47–64% of stenoses were reported to be in the juxta-anastomotic region. This difference is mainly related to the strict use of our definition of juxtaanastomotic segment within 2 cm of the anastomosis; a stenosis at 2.2 cm of the anastomosis was therefore considered a proximal venous outflow segment stenosis. As already shown by Clark et al. and Turmel-Rodrigues et al., most stenoses in AVF are short, with respectively, 42.6% and 86% of stenoses less than 2 cm long in their series [2,3]. Our results of a median and mean length of stenosis of 16 mm and 21.8 mm, respectively, reflect this. Technical and clinical success rates were comparable to other studies, where often clinical success rates exceed technical success rates and where it is suggested that an autologous vein is more subject to spasm, elastic recoil or other factors resulting in a residual stenosis of 30% or more [3,4,11–13]. As in a study of Trerotola et al. [14] we found a linear relationship between severity of initial stenosis and residual stenosis. Recurrent stenosis was significantly more seen in AVF with a technically unsuccessful outcome after PTA compared to technically successful procedures; early recurrence was however not influenced by technically successful or unsuccessful procedures. In addition patency rates were not significantly different between technically successful and unsuccessful procedures. This means that even though there is a higher risk for recurrent stenosis when residual stenosis after PTA is 30% or more, this will not affect AVF durability. In a study of Clark et al. on 65 dysfunctional or occluded AVF no difference in long-term patency was found between AVF with a >30% residual stenosis compared to <30% residual stenosis after treatment [3]. Similar findings were demonstrated by Forauer et al. who showed significantly higher technical success rate with balloon angioplasty inflation times of 3 min versus 1 min, but no significant post-intervention patency differences [15]. A higher technical success rate for radiocephalic AVF compared to upper arm AVF (and especially brachial artery–median vein AVF) was seen in our study. Trerotola et al. also mentioned access type as a significant variable for residual stenosis, but in this study access type was stratified in fistula, graft or hybrid, without review of the different types of AVF [14]. Turmel-Rodrigues et al. did not see a difference in success rate between forearm and upper arm AVF, with a success rate of 95 ± 2% and 97 ± 3%, respectively [2]. The reason for this lower technical success rate in brachial artery–median vein AVF is not clear, but it may have been because of a less aggressive approach (no overdilatation of stenoses), to elastic recoil or to an underestimation of residual stenosis in AVF with a larger diameter of fistula veins by visual estimate. Over 50% of dysfunctional AVF treated with PTA in our study showed a functional stenosis (in 75% of cases) or occlusion (in 25% of cases) during follow-up. Moreover almost 95% of the stenoses found during follow-up were at the same location as the first-time stenosis. These findings suggest that durability of PTA is only moderate. The mean recurrence time of 13.1 months is within range of the results in the study of Turmel-Rodrigues et al., where the mean interval between reintervention and initial dilatation was 18.1 months and 10.9 months for forearm and upper arm AVF, respectively [2]. Tessitori et al. found a median time to restenosis of 8 months [16]. Similar to the study of Turmel-Rodrigues et al., where a significantly shorter interval between reinterventions was seen in forearm AVF created less than 1 year before compared to AVF older than 1 year, we found a significant effect of AVF age on dysfunction recurrence [2]. This effect was only linear (the older the AVF, the smaller the probability of recurrence) and could not
be shown when we tried to stratify AVF age in groups younger than 6 months, older than 6 months but younger than 1 year and older than 1 year. The effect of prior AVF on dysfunction recurrence was only significant using the log-rank statistical test, but not significant with the cumulative incidence function model. Inversely the effect of AVF side was significant when calculated with the cumulative incidence function model, but not with the log-rank model (albeit a tendency towards significance was seen). Both analyses are valid but differ in interpretation of the results, because in the log-rank model, the competing events are considered as censored observations, while in the cumulative incidence function model the competing events are taken into account. This may result in conflicting outcomes and these variables were therefore not considered as variables with an absolute clinical effect on dysfunction recurrence. Early recurrence, defined as recurrent dysfunction (stenosis or occlusion) within 6 months after initial PTA was seen in more than 25% of cases and the majority were recurrent stenoses. Both AVF age and diabetes mellitus were variables with a significant effect on early recurrent dysfunction and may have implications for surveillance of AVF after initial PTA. The patency rates observed in our study were within range with the patency rates reported [2–4,11]. The linear effect of AVF age on primary patency was to be expected given the effect of AVF age on dysfunction recurrence. This effect was also demonstrated by Turmel-Rodrigues et al. [2] as mentioned before and by Clark et al. [3], who showed a substantial (but not significant) association between lower primary patency rates and AVF younger than 6 months. Secondary patency rates were also influenced by AVF age, but in contrast to primary patency, this effect was not linear but categorical. AVF younger than 6 months showed lower secondary patency rates compared to AVF older than 6 months. This cannot solely be explained by the 20% non-maturing AVF in this group of young AVF, because non-maturating AVF treated by PTA have similar patency rates compared to mature AVF in most studies [17,18]. Only Manninen et al. reported a significant lower secondary patency rate in percutaneously treated non-maturing AVF compared to mature AVF and suggested that this might have been related to poorer quality veins in non-mature AVF [13]. The effect of patient age on patency was only significant using the log-rank statistical test, but not significant with the competing risk regression model. Similar to the variables that may have an effect on dysfunction recurrence this resulted in conflicting outcomes. In the group of patients older than 75 years of age, the patency rate may have been interpreted lower in the log-rank model compared to the competing risk regression model, because in this group more patients will have died during follow-up and thus more patients will have been censored. As a result patient age was not considered as a variable with an absolute significant effect on assisted primary or secondary patency. Major limitations of the study were that this was a retrospective study from a single institution, without a control group and the relative small number of patients. In conclusion, technical success of PTA in hemodialysis AVF is influenced by the initial grade of stenosis and by AVF type, with highest success rate seen in radiocephalic AVF. Technical success itself has a significant effect on dysfunction recurrence, but not on AVF longevity. AVF age has a significant effect on dysfunction recurrence and on patency; younger AVF have an increased risk for (early) recurrent dysfunction and for lower patency rates. The only other variable with a significant effect on recurrent dysfunction within 6 months after PTA is diabetes mellitus. These results may help in tailoring surveillance programs, but large prospective trials are still warranted.
S. Heye et al. / European Journal of Radiology 81 (2012) 2298–2303
References [1] National Kidney Foundation. NKF-K/DOQI Clinical practice guidelines and clinical practice recommendations for 2006 updates: hemodialysis adequacy, peritoneal dialysis adequacy and vascular access. Am J Kidney Dis 2006;48(suppl. 1):S188–91. [2] Turmel-Rodrigues L, Pengloan J, Baudin S, et al. Treatment of stenosis and thrombosis in haemodialysis fistulas and grafts by interventional radiology. Nephrol Dial Transplant 2000;15:2029–36. [3] Clark TWI, Hirsch DA, Jindal KJ, Veugelers PJ, LeBlanc J. Outcome and prognostic factors of restenosis after percutaneous treatment of native hemodialysis fistulas. J Vasc Interv Radiol 2002;13:51–9. [4] Rajan DK, Bunston S, Misra S, Pinto R, Lok CE. Dysfunctional autogenous hemodialysis fistulas: outcomes after angioplasty – are there clinical predictors of patency? Radiology 2004;232:508–15. [5] Falk A. Maintenance and salvage of arteriovenous fistulas. J Vasc Interv Radiol 2006;17:807–13. [6] Sacks D, McClenny TE, Cardella JF, Lewis CA. Society of interventional radiology clinical practice guidelines. J Vasc Interv Radiol 2003;14(suppl):S199–202. [7] Gray RJ, Sacks D, Martin LG, Trerotola SO. Society of Interventional Radiology Technology Assessment Committee. Reporting standards for percutaneous interventions in dialysis access. J Vasc Interv Radiol 2003;14:S433–42. [8] Pepe MS, Mori M. Kaplan–Meier, marginal or conditional probability curves in summarizing competing risks failure time data? Stat Med 1993;12:737–51. [9] Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509.
2303
[10] Tordoir J, Canaud B, Haage P, et al. EBPG on vascular access. Nephrol Dial Transplant 2007;22(suppl 2):ii88–117. [11] Manninen HI, Kaukanen ET, Ikäheimo R, et al. Brachial artery access: endovascular treatment of failing Brescia–Cimino hemodialysis fistulas – initial success and long-term results. Radiology 2001;218:711–8. [12] Maeda K, Furukawa A, Yamasaki M, Murata K. Percutaneous transluminal angioplasty for Brescia–Cimino hemodialysis fistula dysfunction: technical success rate, patency rate and factors that influence the results. Eur J Radiol 2005;54:426–30. [13] Manninen HI, Kaukanen ET, Mäkinen K, Karhapää P. Endovascular salvage of nonmaturing autogenous hemodialysis fistulas: comparison with endovascular therapy of failing mature fistulas. J Vasc Interv Radiol 2008;19:870–6. [14] Trerotola SO, Kwak A, Clark TWI, et al. Prospective study of balloon inflation pressures and other technical aspects of hemodialysis access angioplasty. J Vasc Interv Radiol 2005;16:1613–8. [15] Forauer AR, Hoffer EK, Homa K. Dialysis access venous stenoses: treatment with balloon angioplasty – 1-min versus 3-min inflation times. Radiology 2008;249:375–81. [16] Tessitori N, Mansueto G, Bedogna V, et al. A prospective controlled trial on effect of percutaneous transluminal angioplasty on functioning arteriovenous fistulae survival. J Am Soc Nephrol 2003;14:1623–7. [17] Turmel-Rodrigues L, Mouton A, Birmelé B, et al. Salvage of immature forearm fistulas for haemodialysis by interventional radiology. Nephrol Dial Transplant 2001;16:2365–71. [18] Clark TWI, Cohen RA, KWak A, et al. Salvage of nonmaturing native fistulas by using angioplasty. Radiology 2007;242:286–92.