Drug Coated Balloon Angioplasty in Failing AV Fistulas a Randomized Controlled Trial

Article

Drug Coated Balloon Angioplasty in Failing AV Fistulas A Randomized Controlled Trial

Scott O. Trerotola,1 Jeffrey Lawson,2,3 Prabir Roy-Chaudhury,4 and Theodore F. Saad,5 for the Lutonix AV Clinical Trial Investigators

Abstract Background Restenosis remains a problem in hemodialysis access interventions. Paclitaxel-coated balloons have 1Department of shown promise in reducing access-related restenosis in small trials. The primary hypotheses for our multicenter Radiology, Perelman trial were superior effectiveness at 180 days and noninferior safety at 30 days of a drug-coated balloon compared School of Medicine with conventional angioplasty for treatment of dysfunctional arteriovenous fistulas. of the University of Pennsylvania, Philadelphia, Design, setting, participants, & measurements This randomized trial enrolled 285 patients with dysfunctional Pennsylvania; arteriovenous fistulas at 23 centers. Grafts, central venous stenoses, thrombosed fistulas, and immature fistulas Departments of were excluded. All patients received angioplasty of the lesion responsible for access dysfunction. After successful 2Surgery and 3 angioplasty (#30% residual stenosis), lesions were treated with either a paclitaxel-coated balloon or an uncoated Pathology, Duke University, Durham, control balloon of similar design to the drug-coated balloon. Access function during follow-up was determined per North Carolina; ’ fi 4 centers usual protocols; reintervention was clinically driven. The primary ef cacy outcome assessment was done Division of at 6 months, and the safety assessment was done within 30 days of the procedure. Prespecified secondary end Nephrology, points includedassessmentofpostintervention targetlesion primary patency andaccesscircuitprimary patency at University of Arizona 6months. Health Sciences and Southern Arizona Veterans Affairs Results The 180-day end point was not met with target lesion primary patency (71%64% for the drug-coated Healthcare System, balloon and 63%64% for control; P=0.06), representing a difference of 8%66% (95% confidence interval, 23% to Tucson, Arizona; and 5 20%). Access circuit primary patency did not differ between groups. Interventions to maintain target lesion Section of Renal and Hypertensive patency were fewer for the drug-coated balloon at 6 months (0.31 versus 0.44 per patient; P=0.03). The primary Diseases, Christiana safety noninferiority end point was met and did not differ between groups (P=0.002). Care Health System, Newark, Delaware Conclusions Paclitaxel-coated balloon–assisted angioplasty did not meet the primary effectiveness end point at 180 days compared with conventional angioplasty. Both arms showed equivalent safety (ClinicalTrials.gov Correspondence: Dr. number NCT02440022). Scott O. Trerotola, – Department of Clin J Am Soc Nephrol 13: 1215 1224, 2018. doi: https://doi.org/10.2215/CJN.14231217 Radiology, Division of Interventional Radiology, Hospital Introduction used specialized angioplasty balloons that have the of the University of Pennsylvania, 1 It has been 51 years since the original description of ability to deliver the drug to the vessel wall, where it is fi fi Silverstein, 3400 hemodialysis stulas (1), and the superiority of stulas rapidly taken up and remains in the vessel wall. The Spruce Street, over other forms of hemodialysis access remains widely purpose of this study was to investigate the hypothesis Philadelphia, PA accepted (2). However, failure of fistulas remains as that paclitaxel-coated balloon treatment after successful 19104. Email: pervasive a problem today as it was half a century ago. fi [email protected]. angioplasty of stenosis in hemodialysis stulas would edu Access failure results in missed or inadequate treat- improve outcomes compared with angioplasty alone. ments, hospitalization, and catheter use, costing the United States health care system approximately $2 billion annually (3). In spite of substantial advances in Materials and Methods our understanding of the pathophysiology of access Study Design stenosis and restenosis, there have been no large-scale This multicenter (n=23), prospective, randomized, studies showing superiority of any intervention for controlled trial was carried out under an investiga- treating fistula-related stenosis. Among the most prom- tional device exemption from the US Food and Drug ising candidates for preventing de novo stenosis or Administration, and it was designed to test the safety restenosis after intervention is paclitaxel, which has and effectiveness of a drug-coated balloon in hemo- proven to be beneficial in preventing restenosis in large dialysis fistula–related venous stenosis. The study studies in other vascular beds (4–6) and several small, was carried out in full compliance with the Heath randomized, single-center studies in hemodialysis ac- Insurance Portability and Accountability Act and cess (both grafts and fistulas) (7–11). These studies have the Declaration of Helsinki, and each site obtained www.cjasn.org Vol 13 August, 2018 Copyright © 2018 by the American Society of Nephrology 1215 1216 Clinical Journal of the American Society of Nephrology

Institutional Review Board approval for participation in within the access circuit (defined as the anastomosis to the the study; all patients signed informed consent. Inclusion axillary vein inclusive) $50% with a matching clinical and exclusion criteria are shown in detail in Supplemental indicator (i.e., the lesion was such that the clinical indicator Appendix 1. The primary inclusion criterion was a dys- could be explained by the lesion) (2). After a patient was functional mature arteriovenous fistula (AVF) in the arm; determined to have a matching lesion, other fistulographic thrombosed or immature AVF, central venous stenosis, criteria had to be met. Specifically, there could be no more and grafts were excluded. The Consolidated Standards of than one additional (“nontarget”) lesion in the circuit that Reporting Trials (CONSORT) diagram (Figure 1) details had to be successfully treated (#30% residual stenosis) patient selection and enrollment. To be considered for before randomization. Overall lesion length and configu- the study, patients had to have a nonthrombosed native ration criteria also had to be met as shown in Supplemental AVF in the arm that had at least one clinical indicator Appendix 2. The target (study) lesion was then treated with of dysfunction as defined by the National Kidney Foun- high-pressure angioplasty with the requirements that the dation’s Kidney Disease Outcomes Quality Initiative balloon waist be fully effaced and that there be #30% (KDOQI) (2). residual stenosis after angioplasty. Provided that these criteria were met, the patient was then randomly Screening and Enrollment assigned to the drug-coated balloon group or control Patients who were candidates for the study underwent using sealed envelopes provided to the sites by study diagnostic fistulography to determine if they had a stenosis personnel.

Figure 1. | Consolidated Standards of Reporting Trials diagram for study accrual. *Not enrolled/not randomized. Clin J Am Soc Nephrol 13: 1215–1224, August, 2018 Drug-Coating Balloons for Failing AV Fistula, Trerotola et al. 1217

Table 1. Demographics

Variable Drug-Coated Balloon, n=141 Control, n=144 Total, n=285

Age, yr, mean6SD 64615 61613 62614 Men, n/total (%) 87/141 (62) 85/144 (59) 172/285 (60) Diabetes, n/total (%) 82/141 (58) 94/144 (65) 176/285 (62) Dyslipidemia, n/total (%) 85/141 (60) 84/144 (58) 169/285 (59) Hypertension, n/total (%) 133/141 (94) 142/144 (99) 275/285 (97) Current and former cigarette smoking, n/total (%) 64/141 (45) 66/144 (46) 130/285 (46) Cardiovascular disease, n/total (%) 83/141 (59) 96/144 (67) 179/285 (63) Congestive heart failure, n/total (%) 33/141 (23) 33/144 (23) 66/285 (23) Stroke, n/total (%) 18/141 (13) 13/144 (9) 31/285 (11) Coronary artery disease, n/total (%) 43/141 (31) 40/144 (28) 83/285 (29) Myocardial infarction, n/total (%) 16/141 (11) 23/144 (16) 39/285 (14) Peripheral arterial disease, n/total (%) 14/141 (10) 26/144 (18) 40/285 (14) Access age, d, mean6SD 11646973 105661122 110961050 Fistula location, n/total (%) Left arm 94/141 (67) 110/144 (76) 204/285 (72) Right arm 47/141 (33) 34/144 (24) 81/285 (28) Access site/type, n/total (%) Across antecubital fossa 6/141 (4) 8/144 (6) 14/285 (5) Transposed 0 2/8 (25) 2/14 (14) Nontransposed 6/6 (100) 6/8 (75) 12/14 (86) Forearm 47/141 (33) 31/144 (22) 78/285 (27) Transposed 5/47 (11) 4/31 (13) 9/78 (12) Nontransposed 42/47 (89) 27/31 (87) 69/78 (89) Upper arm 88/141 (62) 105/144 (73) 193/285 (68) Transposed 40/88 (46) 42/105 (40) 82/193 (43) Nontransposed 48/88 (55) 63/105 (60) 111/193 (58) Prior intervention in fistula, n/total (%) 120/141 (85) 125/144 (87) 245/285 (86) No. of prior interventions, mean6SD 4.063.3 3.662.8 3.863.1 No. of months on dialysis, mean6SD 41633 36631 38632 Clinical indicator of access dysfunction, n/total (%) 141/141 (100) 143/143 (100) 284/284 (100) Decreased blood flow 37/141 (26) 39/143 (27) 76/284 (27) High venous pressure 24/141 (17) 28/143 (20) 52/284 (18) Decreased Kt/V 5/141 (4) 6/143 (4) 11/284 (4) Diminished/abnormal thrill 39/141 (28) 25/143 (18) 64/284 (23) Pulsatility 87/141 (62) 84/143 (59) 171/284 (60) Flaccid access 1/141 (1) 0 1/284 (0.4) Abnormal bruit 9/141 (6) 10/143 (7) 19/284 (7) Arm or hand swelling 4/141 (3) 3/143 (2) 7/284 (3) Prolonged bleeding 28/141 (20) 27/143 (19) 55/284 (19) Difficult puncture 23/141 (16) 25/143 (17) 48/284 (17) Infiltration 3/141 (2) 4/143 (2) 7/284 (3) Recirculation 4/141 (3) 1/143 (1) 5/284 (2) Pulling clots 6/141 (4) 2/143 (1) 8/284 (3) Other 20/141 (14) 19/143 (13) 39/284 (14)

Note that clinical indicators total .100%, because patients may have had more than one. Exploratory analyses revealed no statistically signiﬁcant difference between groups for any of the variables shown.

Randomization and Masking the drug-coated and control balloons were visibly Randomization 1:1 was performed. Each site was pro- different). vided with one set of randomization envelopes by the sponsor numbered in sequential order, with each enve- Study End Points lope containing a treatment assignment card. The in- The study had both primary efficacy and primary safety vestigator was not allowed to open the randomization end points. The primary efficacy end point was target lesion envelope (i.e., remain blinded to the treatment assign- primary patency at 6 months defined as no need for clinically ment) until after all successful predilation inflations were driven reintervention on the target lesion or access throm- complete (when the last angioplasty balloon was deflated bosis (a complete list of definitions can be found in Supple- and the fistulogram confirmed success). To minimize bias, mental Appendix 2). Of note, a patient could undergo the subject, the nephrologist, and the dialysis staff were reintervention on the AVF without target lesion primary blinded to the treatment until after 1 year, the follow-up patency ending provided that the reintervention did not period required for all hypothesis-tested secondary end include retreatment of the target lesion. This was tested using points to be collected. The investigator and members of the hypothesis of superior efficacy of the drug-coated balloon the study team were of necessity aware of the randomiza- over control. The primary safety end point was freedom from tion group, and therefore, they were not blinded (because localized or systemic serious adverse events through 30 days 1218 Clinical Journal of the American Society of Nephrology

Table 2. Target lesion and procedural information

Drug-Coated Control, Total, Variable Balloon, P Valuea n=144 n=285 n=141

Target lesion location, n/total (%) Anastomotic 6/139 (4) 5/142 (4) 11/281 (4) Cephalic arch 26/139 (19) 32/142 (23) 58/281 (21) Cannulation zone 6/139 (4) 14/142 (10) 20/281 (7) Inflow 47/139 (34) 42/142 (30) 89/281 (32) Outflow 34/139 (24) 32/142 (23) 66/281 (23) Swing point 20/139 (14) 17/142 (12) 37/281 (13) Tandem lesion (#2cmapart),n/total (%) 4/141 (3) 10/143 (7) 14/284 (5) Vessel, n/total (%) Subclavianb 1/141 (1) 0 1/285 (0z4) Brachial 1/141 (1) 1/144 (1) 2/285 (1) Cephalic 97/141 (69) 97/144 (67) 194/285 (68) Basilic 36/141 (26) 41/144 (28) 77/285 (27) Median cubital 2/141 (1) 1/144 (1) 3/285 (1) Other 4/141 (3) 4/144 (3) 8/285 (3) De novo lesion, n/total (%) 43/141 (31) 39/144 (27) 82/285 (29) 0.60 Lesion length, mm, mean6SD 28z4615z129z5618z728z9617 0.77 Balloon fully effaced, n/total (%) 0.84 Yes 121/141 (86) 126/143 (88) 247/284 (87) No 10/141 (7) 8/143 (6) 18/284 (6) Unable to assess 10/141 (7) 9/143 (6) 19/284 (7) Reference vessel diameter, mm, mean6SD 7.8562 7.68627.77620.38 Range, mm 2.4–12.0 4.0–12.0 2.4–12.0 Nontarget lesion treated, n/total (%) 28/141 (20) 35/144 (24) 63/285 (22) 0.39 Preprocedure percentage stenosis (Core Lab), 646964610 64690.92 mean6SD Range, % 41–88 41.0–86 41–88 — Adjunctive procedure (e.g., angioplasty, 0.10 stent-graft, embolectomy) performed in the AV access circuit, n/total (%) Yes 7/141 (5) 2/144 (1) 9/285 (3) — No 134/141 (95) 142/144 (99) 276/285 (97) — Any device deficiencies or malfunctions, n/total (%) 0.06 Yes 4/141 (3) 0 4/285 (1) — No 137/141 (97) 144/144 (100) 281/285 (99) — Periprocedural complication, n/total (%) 2/141 (1) 2/144 (1) 4/285 (1) .0.99 Largest predilation balloon diameter, mm, mean6SD 8.361.8 8.161.8 8.261.8 0.42 Range, mm 4–12 5–12 4–12 — Predilation balloon length, mm, mean6SD 44613 44613 44613 0.99 Range, mm 20–80 20–80 20–80 — Maximum pressure of predilation, atm, mean6SD 2168226822680.56 Range, atm 8–40 6–40 6 40 — After predilation stenosis (Core Lab) percentage 6SD 22610 22611 22610 0.43 Range, % 22to50 29to50 29to50 — No. of study devices, n/total (%) 0.001 1 128/141 (91) 143/144 (99) 271/285 (95) — 2 13/141 (9) 1/144 (1) 14/285 (5) — Study balloon diameter, mm, mean6SD 8.361.8 8.161.8 8.261.8 0.36 Range, mm 4–12 5–12 4–12 — Study balloon length, mm, mean6SD 51612 46612 49613 ,0.001 Range, mm 40–100 40–100 40–100 — Transit time, s, mean6SD 18632 19639 19635 0.97 Range, s 0–320 1–420 0–420 Maximum pressure of study balloon, atm, mean6SD 9.762.1 12.16511.064.1 ,0.001 Range, atm 4–14 3–35 3–35 Total duration of inflation, s, mean6SD 115634 90641 101640 ,0.001 Range, s 30–180 5–240 5–240 — Operator able to deliver to intended site, inflate, 151/151 (100) 171/171 (100) 322/322 (100) NA and retrieve device, n/total (%) Final stenosis (Core Lab) percentage 6SD 17611 17610 17611 0.70 Range, % 228 to 47 216 to 44 228 to 47 — Adjunctive procedure performed after study 2/2 1/2 3/4 0.62 device (e.g., angioplasty, embolectomy), patients/no. of procedures Reason, n/total (%) .0.99 Residual stenosis 1/2 (50) 0 1/4 (25) — Other 1/2 (50) 2/2 (100) 3/4 (75) — Clin J Am Soc Nephrol 13: 1215–1224, August, 2018 Drug-Coating Balloons for Failing AV Fistula, Trerotola et al. 1219

Table 2. (Continued)

Drug-Coated Control, Total, Variable Balloon, P Valuea n=144 n=285 n=141

Treatment, n/total (%) 0.33 Angioplasty 2/2 (100) 0 2/4 (50) — Thrombectomy 0 2/2 (100) 2/4 (50) — Procedure success, n/total (%) 138/141 (98) 143/144 (99) 281/285 (99) 0.37 Clinical success, n/total (%) 140/141 (99) 144/144 (100) 284/285 (100) 0.50

All comparisons were exploratory in nature. —,noP value determined; AV, arteriovenous; NA, not applicable. aP value associated with the Wilcoxon rank sum test comparing drug-coated balloon and standard angioplasty groups or the Fisher exact test for categorical data. bProtocol deviation.

that reasonably suggest the involvement of the arteriovenous was at least 5 mm extra length on either end compared with access circuit as adjudicated by a blinded Clinical Events the predilation balloon; this was done to minimize “geo- Committee. These would include life-threatening events or graphic miss.” The control balloon was of similar design to those resulting in death, requiring hospitalization, resulting the study balloon. in permanent disability, or requiring intervention to prevent permanent impairment; the latter definition included throm- Follow-Up bosis. Secondary end points included hypothesis-driven No postoperative anticoagulation or antiplatelet regimen (access circuit primary patency at 6 months) and descrip- was prescribed as part of the study, and investigators were tive ones (target lesion primary patency and access circuit permitted to prescribe any medications as part of their normal primary patency at 3 months; device, procedural, and clinical clinical practice. Follow-up assessments were conducted by success; access abandonment and number of interventions telephone at intervals of 1, 3, and 6 months for the dialysis required to maintain access circuit primary patency at 3 and 6 unit as well as the patient. Similar telephone assessments are months; number of interventions required to maintain target planned at 9, 12, 18, and 24 months. In addition, an office visit lesion primary patency at 3 and 6 months; device- and with physical examination of the access was required at 6 procedure-related safety events at 1, 3, and 6 months; and months. All reinterventions were clinically driven; if the access circuit primary patency/target lesion primary patency patient developed a clinical indicator of access dysfunction using the 30-day window approach; i.e., at 210 days). Of note, prompting reintervention, the details of that intervention unlike target lesion primary patency, access circuit primary were captured in an identical fashion as the index procedure. patency ended with any reintervention on the AVF, regard- Investigators were asked to only treat the lesion(s) respon- less of lesion location. The end points chosen for this study sible for the abnormal clinical indicator during these visits in are consistent with the reporting standards of the Society of accordance with the National Kidney Foundation’sKDOQI Interventional Radiology (12). Because the study is ongoing Initiative guidelines (2). (enrollment was completed in March 2016; follow-up will end in March 2018), only end points and analyses up to 6 months are described herein. Similar end points and analyses Oversight will be applied at 9, 12, 18, and 24 months. In addition to blinded Core Lab monitoring of all data points, independent Clinical Events Committee (CEC) ad- judication and Data and Safety Monitoring Board (DSMB) Study Device and Procedure review were carried out. The CEC and DSMB reviews were Supplemental Appendices 1–4 have details of the device conducted in a blinded fashion. and procedure. In brief, the drug-coated balloon used in the study (Lutonix 035 DCB Catheter; Lutonix, Minneapolis, MN) is an angioplasty balloon with a 2-mg/mm2 paclitaxel- Statistical Analyses based coating. In contrast to the predilation balloon, the A detailed description of the sample size determination drug-coated balloon is a partially compliant low-pressure and statistical analysis can be found in Supplemental balloon. When determining percentage of stenosis for Appendix 3. In brief, intention-to-treat analyses were eligibility, stenosis was measured from the fistulogram performed to evaluate target lesion primary patency and images by determining the minimal luminal diameter of other end points. Kaplan–Meier analysis was used to the stenosis and the diameter of the adjacent normal vessel estimate the target lesion primary patency rate in the (reference) and then applying the formula percentage of drug-coated balloon and angioplasty groups. A log rank stenosis =[12(minimal luminal diameter/reference)]3100. test comparing the drug-coated balloon with angioplasty All fistulograms were analyzed by the Core Lab (Yale was used to test the primary hypothesis to determine if the Cardiovascular Research Group, New Haven, CT) in a drug-coated balloon was superior to angioplasty. The test blinded fashion. Study and control balloon diameters were was successful if the one-sided P value was ,0.03 and the identical to that of the high-pressure predilation balloon. result was in favor of the drug-coated balloon. The target Drug-coated balloon length was selected, such that there lesion primary patency at 6 months was primarily assessed 1220 Clinical Journal of the American Society of Nephrology

Drug Coated Balloon (N=141) Angioplasty (N=144) Difference % (95% CI) P-value

180 Day Event Free 0.06 Rate (SE) 71% (4%) 63% (4%) 8% (6%) 95% CI (63%-78%) (54%-70%) (-3%-20%) 210 Day Event Free 0.02 Rate (SE) 64% (4%) 53% (4%) 12% (6%) 95% CI (55%-72%) (44%-61%) (-0.2%-23%)

Time Cumulative Subjects with Events Cumulative Subjects Censored Cumulative Subjects Left

Drug Coated Balloon 30 days 3 5 133 90 days 10 11 120 180 days 37 16 88 210 days 46 16 79 Angioplasty 30 days 2 0 142 90 days 19 2 123 180 days 57 7 85 210 days 66 9 69

Figure 2. | Survival curves for target lesion primary patency show no significant difference at 180 days, but a significant improvement in patency for the DCB at 210 days (exploratory analysis). 95% CI, 95% confidence interval; SE, SEM. up to day 180, and it was assessed up to the end of the rank sum test. Subgroup analyses and covariate analyses 6-month window (day 210) in an exploratory analysis The were performed to evaluate the influence of potential study protocol and statistical analysis plan described the confounding factors. 6-month visit as occurring at 180 days with a 30-day visit window, meaning that subjects could return for the 6-month assessment within the protocol-specified window Results from day 150 to day 210. The primary efficacy analysis Study Participants involved Kaplan–Meier survival analysis of target lesion The study, performed under an approved US Food and primary patency at 6 months. Because Kaplan–Meier Drug Administration Investigative Device Exemption, requires assessment of survival as of a specifictime enrolled 285 (141 drug-coated balloon and 144 control) point, a single point in time had to be chosen (rather subjects at 23 centers between June 2015 and March 2016. than a window). Given that the analysis looked at 6-month Participating sites included hospital-based and free-stand- survival, 180 days was used as the analysis point. ing practices, with investigators representing interven- The sample size calculation was on the basis of the tional nephrology (n=10; 44%), interventional radiology primary efficacy end point of target lesion primary patency (n=9; 39%), and vascular surgery (n=4; 17%) backgrounds. at 6 months, which was estimated as 70% in the drug- Demographic data are shown in Table 1, and procedural coated balloon group and 50% in the balloon angioplasty details are shown in Table 2. As can be seen from the group. With a significance threshold of P=0.03 on a one- CONSORT diagram (Figure 1), of 581 patients screened, 489 fi tailed test and 90% statistical power, the total sample size met initial screening criteria and proceeded to stulography, fi fi was calculated as 284 after adjusting for 10% attrition, with 175 were excluded on the basis of stulographic ndings, 142 in each treatment group. and 29 (9% of those undergoing angioplasty) were excluded The 30-day safety end point was tested with a non- on the basis of unsuccessful predilation. inferiority approach using a binary approach (Farrington and Manning; 10% noninferiority margin). Efficacy Secondary end points include access circuit primary The primary effectiveness end point, superior target patency, which was analyzed similar to target lesion lesion primary patency evaluated at 6 months, was not met: primary patency, and the number of interventions required Kaplan–Meier survival analysis at day 180 was 71%64% to maintain target lesion primary patency or access circuit (95% confidence interval [95% CI], 63% to 78%) for the drug- primary patency, which was analyzed using a Wilcoxon coated balloon and 63%64% (95% CI, 54% to 70%) for Clin J Am Soc Nephrol 13: 1215–1224, August, 2018 Drug-Coating Balloons for Failing AV Fistula, Trerotola et al. 1221

control angioplasty, representing an absolute difference of associated with differential treatment group responses. 8%66% (95% CI, 23% to 20%; P=0.06) (Figure 2). An ex- Likewise, Cox regression analysis did not reveal any ev- ploratory analysis of effectiveness applying a 30-day win- idence of treatment-factor interactions, including the pro- dow to the 6-month end point to minimize missing data cedural variables shown in Table 2. points due to patients not completing their 6-month visit fi by 180 days (i.e., 210 days) showed statistical signi cance; Safety target lesion primary patency was superior for the drug- The primary safety end point for noninferiority for the 6 coated balloon (64% 4%; 95% CI, 55% to 72%) versus drug-coated balloon versus angioplasty was met with one- 6 control (53% 4%;95%CI,44%to61%;P=0.02). This sided P value of P=0.002 (Figure 4). The result of the 6 represented an absolute difference of 12% 6% (95% CI, 0% primary safety end point by binary analysis at day 30 was to 23%). Six-month access circuit primary patency did not 95% (130 of 137) for the drug-coated balloon and 96% (138 6 differ between the drug-coated balloon (62% 4%; 95% CI, of 144) for angioplasty. 53% to 70%) and control (58%64%; 95% CI, 49% to 66%; P=0.25) (Figure 3). Six-month access circuit primary patency also did not differ in the exploratory 210-day Discussion analysis. Interventions to maintain target lesion patency The consequences of fistula failure to a patient relying on were fewer for the drug-coated balloon at 6 months (0.31 hemodialysis are substantial: missed or inadequate dialysis versus 0.44 per patient; P=0.03). Most reintervention was treatments, avoidable hospitalization, and catheter use all performed for recurrence of access dysfunction; thrombo- adversely affect quality of life and contribute substantially sis was uncommon at 6 months (2.4% in the drug-coated to health care costs (13). A variety of treatments for stenosis balloon arm and 4.3% in the control arm). Anticoagulant in fistulas have been introduced in the half century since and antiplatelet medication use during the study is shown the original description of the fistula (1), including percu- in Table 3. taneous transluminal angioplasty, cutting balloons, stents, As shown in Table 2, there were no differences in and stent grafts. Although some of these therapies have demographics, access characteristics, or indications as shown promise, to date, no treatment modality for fistulas determined by exploratory analyses. Results were evalu- has been shown in a large, multicenter trial to be any bet- ated by site; a test for a site-treatment interaction was ter than conventional angioplasty. This study builds on performed, and it was not significant (P=0.71; Cox). Hence, smaller, randomized, single-center studies that showed there was no evidence that an investigator effect was benefit of the drug-coated balloon over angioplasty (7–11).

Drug Coated Balloon (N=141) Angioplasty (N=144) Difference % (95% CI) P-value

180 Day Event Free 0.25 Rate (SE) 62% (4%) 58% (4%) 4% (6%) 95% CI (53%-70%) (49%-66%) (-8%-16%) 210 Day Event Free 0.23 Rate (SE) 52% (4%) 48% (4%) 5% (6%) 95% CI (43%-61%) (39%-56%) (-7%-17%)

Time Cumulative Subjects with Events Cumulative Subjects Censored Cumulative Subjects Left

Drug Coated Balloon 30 days 5 5 131 90 days 18 11 112 180 days 49 16 76 210 days 61 16 64 Angioplasty 30 days 3 0 141 90 days 20 2 122 180 days 59 5 80 210 days 73 7 64

Figure 3. | Survival curves for access circuit primary patency show no difference between DCB and control. 95% CI, 95% conﬁdence interval; SE, SEM. 1222 Clinical Journal of the American Society of Nephrology

Table 3. Anticoagulant and antiplatelet medications

Time Point and Medication Drug-Coated Balloon, n=141 Control, n=144 Total, n=285

Preprocedure (within 72 h), n (%) Any medication 76 (54) 72 (50) 148 (52) Aspirin 61 (43) 56 (39) 117 (41) Clopidogrel 17 (12) 13 (9) 30 (11) Heparin 1 (1) 3 (2) 4 (1) Other antiplatelet/antithrombotic/anticoagulant 18 (13) 14(10) 32 (11) Intraprocedure, n (%) Any medication 11 (8) 10 (7) 21 (7) Aspirin 3 (2) 3 (2) 6 (2) Clopidogrel 0 1 (1) 1 (0.4) Heparin 7 (5) 7 (5) 14 (5) Other antiplatelet/antithrombotic/anticoagulant 1 (1) 0 1 (0.4) Postprocedure to discharge, n (%) Any medication 15 (11) 11 (8) 26 (9) Aspirin 12 (9) 7 (5) 19 (7) Clopidogrel 4 (3) 4 (3) 8 (3) Heparin 1 (1) 3 (2) 4 (1) Other antiplatelet/antithrombotic/anticoagulant 2 (1) 2 (1) 4 (1) Postdischarge to 30-d visit, n (%) Any medication 72 (51) 69 (48) 141 (50) Aspirin 54 (38) 55 (38) 109 (38) Clopidogrel 15 (11) 12 (8) 27 (10) Heparin 6 (4) 4 (3) 10 (4) Other antiplatelet/antithrombotic/anticoagulant 18 (13) 14 (10) 32 (11) After 30-d visit to 3-mo visit, n (%) Any medication 73 (52) 72 (50) 145 (51) Aspirin 57 (40) 55 (38) 112 (39) Clopidogrel 18 (13) 11 (8) 29 (10) Heparin 1 (1) 6 (4) 7 (3) Other antiplatelet/antithrombotic/anticoagulant 17 (12) 16 (11) 33 (12) After 3-mo visit to 6-mo visit, n (%) Any medication 72 (51) 73 (51) 145 (51) Aspirin 60 (43) 54 (38) 114 (40) Clopidogrel 20 (14) 11 (8) 31 (11)

According to the specified 180-day primary effectiveness effect size). In those studies, the drug-coated balloon end point, this study failed to show superiority of a (which is a lower-pressure device) was used first, with paclitaxel-coated balloon over standard balloon angio- further high-pressure angioplasty performed if the balloon plasty when applied after successful angioplasty of stenosis waist was not effaced. It is possible that this study protocol, in fistulas. However, applying a commonly used 6-month designed to maximize the pretreatment angioplasty result end point definition in an exploratory analysis, namely a by using high-pressure angioplasty for the pretreatment, is 30-day window surrounding 180 days, did show a statis- biased in favor of the control arm, because more angioplasty tically significant benefit. The failure to meet the primary was performed in that group than is usually done (two end point relates in large part to the decision to use a single separate treatments; to control the variable necessitated by time point (i.e., 180 days) instead of a 30-day window. the drug application) and because the investigators were Because subjects were permitted to return for the 6-month held to a very high standard of angioplasty result in both time point through day 210, this inadvertently eliminated arms of the study. This is but one of many possibilities that relevant 6-month data. In the case of this study, the choice of deserve further study as more research is done on the role of the 180-day cutoff resulted in an undesirable censoring of drug-coated balloons in AVF interventions. The study is subject data between 180 and 210 days. Per the protocol, ongoing, and it remains to be seen whether the clear all of these data were to be considered 6-month data. Thus, separation of the survival curves will continue through the after study completion, the planned Kaplan–Meier analysis 24-month study period. was updated in an exploratory analysis to adjust the Although the drug-coated balloon arm showed probable analysis time point to day 210, such that all 6-month benefit over angioplasty in terms of target lesion primary data would be included. Also worthy of note is that the patency, access circuit primary patency did not differ 6-month observed patency exceeded the KDOQI recom- between groups. This is not unexpected given the mul- mendations in both groups. Furthermore, although the tifocal nature of stenosis and restenosis in hemodialysis effect size is smaller than that seen in some prior trials, the fistulas. These data suggest that, to achieve improved 6-month target lesion primary patency in the drug-coated access circuit primary patency, the drug-coated balloon balloon group is very similar to that reported by Katsanos may need to be used to treat all lesions in a given circuit. et al. (8) and Kitrou et al. (10), whereas the control arm Subgroup analysis was unable to identify any specific patency was much lower in those studies (hence, the larger lesion location where the drug-coated balloon performed Clin J Am Soc Nephrol 13: 1215–1224, August, 2018 Drug-Coating Balloons for Failing AV Fistula, Trerotola et al. 1223

Drug Coated Balloon (N=141) Angioplasty (N=144) Difference % (95% CI) P-value

30 Day Event Free <0.001 Rate (SE) 95% (2%) 96% (2%) -1% (3%) 95% CI (90%-98%) (91%-98%) (-6%-4%)

Time Cumulative Subjects with Events Cumulative Subjects Censored Cumulative Subjects Left Drug Coated Balloon 30 days 7 4 130 Angioplasty 30 days 6 0 138

Figure 4. | Survival curves for safety show no difference between DCB and control, meeting the non-inferiority endpoint. Note that the P value is for the Kaplan–Meier analysis and is not for the primary safety analysis, which was binary and P=0.002. 95% CI, 95% confidence interval; SE, SEM. better than the overall results, likely due to small sample above. Having a single drug-coated balloon treatment size in such subset populations. instead of high-pressure angioplasty followed by the drug- The primary safety outcome was met in this study, coated balloon would be desirable; however, this drug- indicating that paclitaxel in the dosage and delivery coated balloon is not a high-pressure balloon, and high method used shows a lack of toxicity. The safety of the pressures are frequently needed to efface the balloon waist paclitaxel-coated balloon has been established in other in fistulas as shown in this study (30% of stenoses required applications as well as this one, and it is thus not .25 atm to efface the waist). Future trials will be needed to surprising, but nonetheless reassuring (in the setting of determine if the observed benefit with the drug-coated uremia, oxidative stress, and inflammation), that the safety balloon applies to these situations as well. Fistula flow goal of the study was met here as well. Although there is no measurement would have been the most objective and reason to believe that late complications will occur related standardized clinical monitoring parameter; however, be- to the drug-coated balloon, the completed 2-year data will cause flow measurement was not available in all centers, it hopefully confirm the 6-month data. Unlike some other was not required in this study. Although the use of the applications, such as arterial disease, it is possible that blinded Core Lab strengthens the observations and mea- relatively frequent use of the drug-coated balloon will surements, there is likely to have been variability in the occur in hemodialysis access, because restenosis is much qualityoftheangioplastyand/ordrugdeliveryamong more aggressive in this setting compared with arterial investigators. Ideally, the study would have been double disease. Nonetheless, in the small doses used, even re- blinded; however, as noted above, the trial device has a peated use of the drug-coated balloon is highly unlikely to different appearance from a standard angioplasty balloon, result in long-term adverse events. and thus, blinding the investigator performing the index Although this was a relatively all-inclusive study, certain procedure was not possible. However, it is believed that lesions and clinical scenarios seen in fistulas were ex- this was substantially mitigated by blinding the patient and cluded, including immature fistulas, thrombosed fistulas, dialysis center to the treatment performed. In this ran- and central venous stenosis. The approach taken in the con- domized trial, a paclitaxel-coated balloon used after trol group, namely two separate angioplasty treatments, successful angioplasty in AVF stenosis was not shown differs from conventional angioplasty, but it was neces- to be superior to a standard balloon using a strict 180-day sary to minimize variables between groups. This might definition of the 6-month end point. Fewer interventions have affected the results seen in the control group as noted were needed in the drug-coated balloon group to maintain 1224 Clinical Journal of the American Society of Nephrology

target lesion patency. The drug-coated balloon was as safe 4. Tepe G, Laird J, Schneider P, Brodmann M, Krishnan P, Micari A, as the standard balloon. Metzger C, Scheinert D, Zeller T, Cohen DJ, Snead DB, Alexander B, Landini M, Jaff MR; IN.PACT SFA Trial Investigators: Drug- coated balloon versus standard percutaneous transluminal an- Acknowledgments gioplasty for the treatment of superficial femoral and popliteal This study was sponsored by Lutonix. The investigators of the peripheral artery disease: 12-Month results from the IN.PACT SFA Lutonix AV Clinical Trial are Saravanan Balamuthusamy (Tarrant randomized trial. Circulation 131: 495–502, 2015 Vascular Clinic, Fort Worth, TX), Umar Waheed (Southwest Vascular 5. Jaff MR, Rosenfield K, Scheinert D, Rocha-Singh K, Benenati J, Nehler M, White CJ: Drug-coated balloons to improve Center, Phoenix, AZ), George Lipkowitz (Western New England Renal femoropopliteal artery patency: Rationale and design of the and Transplant Associates, Springfield, MA), Jeffrey Hoggard (Capital LEVANT 2 trial. Am Heart J 169: 479–485, 2015 Nephrology Associates, Raleigh, NC), David Peeler (University Vas- 6. Rosenfield K, Jaff MR, White CJ, Rocha-Singh K, Mena-Hurtado cular Access, Memphis, TN), Roxana Neyra (Arizona Kidney Disease & C, Metzger DC, Brodmann M, Pilger E, Zeller T, Krishnan P, Gammon R, Mu¨ller-Hu¨lsbeck S, Nehler MR, Benenati JF, Hypertension Centers, Phoenix, AZ), Michael Lawless (Life Access Scheinert D; LEVANT 2 Investigators: Trial of a paclitaxel-coated Center/St.John,Tulsa,OK),JonahLicht(ProvidenceInterventional balloonfor femoropopliteal artery disease. N Engl J Med 373: 145– Associates, Providence, RI), Angelo Makris (Chicago Access Care, 153, 2015 Chicago, IL), Matthew Schaefer (San Antonio Kidney Disease Center, 7. Lai CC, Fang HC, Tseng CJ, Liu CP, Mar GY: Percutaneous an- San Antonio, TX), Robert Molnar (Michigan Vascular Access, Flint, MI), gioplasty using a paclitaxel-coated balloon improves target lesion restenosis on inflow lesions of autogenous radiocephalic fistulas: AriKramer(SpartanburgRegionalHospital,Spartanburg,SC),Micah A pilot study. J Vasc Interv Radiol 25: 535–541, 2014 Chan (University of Wisconsin, Madison, WI), Greg Nadolski (Hospital 8. Katsanos K, Karnabatidis D, Kitrou P, Spiliopoulos S, Christeas N, of the University of Pennsylvania, Philadelphia, PA), Naveen Atray Siablis D: Paclitaxel-coated balloon angioplasty vs. plain balloon (Capital Nephrology Medical Group, Sacramento, CA), Charles Bratton dilation for the treatment of failing dialysis access: 6-Month interim results from a prospective randomized controlled trial. J (Medical University of South Carolina, Charleston, SC), Timothy Endovasc Ther 19: 263–272, 2012 fl P ederer (Renal Care Associates, Peoria, IL), Ahmed Kamel (University 9. Kitrou PM, Katsanos K, Spiliopoulos S, Karnabatidis D, Siablis D: ofAlabamaatBirmingham,Birmingham,AL),ScottSchultz(Minnesota Drug-eluting versus plain balloon angioplasty for the treatment of Vascular Surgery Center, Minneapolis, MN), Luke Wilkins (University failing dialysis access: Final results and cost-effectiveness analysis of Virginia, Charlottesville, VA), Zubin Irani (Massachusetts General from a prospective randomized controlled trial (NCT01174472). Eur J Radiol 84: 418–423, 2015 Hospital, Boston, MA), Jordan Tasse (Rush University, Chicago, IL), 10. Kitrou PM, Spiliopoulos S, Katsanos K, Papachristou E, Siablis D, William DaVanzo (Phoebe Putney Memorial Hospital, St. Simons Is- Karnabatidis D: Paclitaxel-coated versus plain balloon angio- land,GA),ScottResnick(NorthwesternUniversity,Chicago,IL),and plasty for dysfunctional arteriovenous fistulae: One-year results John Ross (Access Connections, Bamberg, SC). of a prospective randomized controlled trial. J Vasc Interv Radiol 26: 348–354, 2015 11. Kitrou PM, Papadimatos P, Spiliopoulos S, Katsanos K, Christeas Disclosures N, Brountzos E, Karnabatidis D: Paclitaxel-coated balloons for the S.O.T. is a consultant for Bard Peripheral Vascular, Lutonix, B. treatment of symptomatic central venous stenosis in dialysis ac- Braun, Cook, Medcomp, W.L. Gore, Orbimed, and Teleflex. He cess: Results from a randomized controlled trial. J Vasc Interv Radiol 28: 811–817, 2017 receives royalties from Teleflex andCook. J.L. is aconsultant for Bard 12. Gray RJ, Sacks D, Martin LG, Trerotola SO; Society of Interven- Peripheral Vascular; J.L. is employed by and has stock options in tional Radiology Technology Assessment Committee: Reporting Humacyte and has ownership and stock in InnAVasc. P.R.-C. is a standards for percutaneous interventions in dialysis access. J Vasc consultant for Lutonix, W.L. Gore, TVA, Medtronic, and Akebia and Interv Radiol 14: S433–S442, 2003 the founder and Chief ScientificOfficer for Inovasc. T.F.S. is a 13. McCarley P, Wingard RL, Shyr Y, Pettus W, Hakim RM, Ikizler TA: Vascular access blood flow monitoring reduces access morbidity consultant for Bard Peripheral Vascular. and costs. Kidney Int 60: 1164–1172, 2001

Received: December 22, 2017 Accepted: April 30, 2018 References 1. Brescia MJ, Cimino JE, Appel K, Hurwich BJ: Chronic hemodi- Published online ahead of print. Publication date available at www. alysis using venipuncture and a surgically created arteriovenous cjasn.org. ﬁstula. N Engl J Med 275: 1089–1092, 1966 2. Foundation NK; Vascular Access 2006 Work Group: Clinical See related editorial, “Drug-Coated Balloon Angioplasty for practice guidelines for vascular access. Am J Kidney Dis 48[Suppl ” – 1]: S176–S247, 2006 Hemodialysis Fistula Maintenance, on pages 1140 1141. 3. USRDS: USRDS 2010 Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal Disease in the United States, This article contains supplemental material online at http://cjasn. Bethesda, MD, National Institutes of Health, National Institute of asnjournals.org/lookup/suppl/doi:10.2215/CJN.14231217/-/ Diabetes and Digestive and Kidney Diseases, 2010 DCSupplemental. Statement of Clariﬁcation

Statement of Clarification: Drug Coated Balloon of additional follow-up data. The interim values were Angioplasty in Failing AV Fistulas correct at the time of publication, but with the Scott O. Trerotola, Jeffrey Lawson, Prabir Roy- additional data, the mean difference in target lesion Chaudhury and Theodore F. Saad; for the Lutonix primary patency at 210 days changed from 11.6% AV Clinical Trial Investigators. Drug coated balloon (95% confidence interval, –0.2% to 23.4%; P50.02) to angioplasty in failing AV fistulas: A randomized 11.5% (95% confidence interval, –0.3% to 23.3%; controlled trial. Clin J Am Soc Nephrol 13(8): 1215– P50.03). The primary end points remain un- 1224, 2018. changed. The authors wish to update readers of this article that the exploratory analysis on target lesion Published online ahead of print. Publication date available at primary patency at 210 days changed with availability www.cjasn.org.