Paclitaxel coated balloon versus conventional balloon angioplasty in dysfunctional dialysis arteriovenous fistula: a systematic review and meta-analysis of randomized controlled trials

Abstract Purpose To compare the efficacy and safety between paclitaxel coated balloon (PCB) angioplasty and conventional balloon (CB) angioplasty in the treatment of dysfunctional arteriovenous fistula (AVF). Methods We searched four major electronic databases (PubMed, EMBASE, Web of Science and the Cochrane Library) for randomized controlled trials (RCTs) published from inception through November 28, 2021. Outcomes of interest included target lesion primary patency (TLPP), technical success and all-cause mortality. The STATA package version 15.1 was utilized to undertake meta-analyses. Results Fourteen RCTs totaling 1535 patients were analyzed. The available data showed that there were no significant differences of TLPP rates at 3, 6, 9 and 12 months between the PCB group and the CB group (risk ratio (RR) 1.00, 95% confidence interval (CI) 0.93–1.07, p = 1.000, I2 = 33.5%, Cochrane Q test p = 0.185, fixed-effect model; RR 1.17, 95% CI 0.99–1.39, p = 0.065, I2 = 75.4%, Cochrane Q test p = 0.000, random-effect model; RR 0.81, 95% CI 0.35–1.89, p = 0.625, I2 = 62.8%, Cochrane Q test p = 0.045, random-effect model; RR 1.19, 95% CI 0.97–1.47, p = 0.096, I2 = 40.5%, Cochrane Q test p = 0.071, random-effect model). In addition, two groups had similar technical success rates (RR 1.00, 95% CI 0.97–1.03, p = 1.000, I2 = 0.0%, Cochrane Q test p = 0.596, fixed-effect model) and all-cause mortality rates (RR 1.00, 95% CI 0.54–1.84, p = 1.000, I2 = 0.0%, Cochrane Q test p = 0.599, fixed-effect model). Conclusions PCB angioplasty did not appear to convey any obvious advantage over CB angioplasty in the treatment of dysfunctional AVF. However, further multi-center, large-scale and well-designed RCTs are needed to prove outcomes.


Introduction
Reliable vascular access is known as the lifeline of maintenance hemodialysis patients. There are several commonly used permanent hemodialysis vascular access types such as autologous arteriovenous fistula (AVF), tunnel-cuffed catheter (TCC) and arteriovenous graft (AVG). However, the extensive application of AVG in clinical practice has not yet been realized due to its high price and technological problems. Note that TCC was chosen only when AVF could not be established or patients were expected to have a relatively short survival time. It has been found that the patients with AVF had a better survival rate compared with patients with other access types [1]. As a consequence, AVF is currently the preferred choice for vascular access. And its functional status directly affects the dialysis efficiency and quality of life of patients undergoing maintenance hemodialysis. Nevertheless, the persistence of AVF was not satisfactory enough and the most prevalent causes of dysfunctional AVF were thrombosis and vascular stenosis [2]. Therefore, long-term patency preservation of the fistula tract presented an urgent clinical problem to be solved. In the past era, dysfunctional AVF was generally treated with surgical methods. With the rapid advancement of intraluminal interventional techniques, it has emerged as a primary therapeutic approach in the treatment of this disease. Conventional balloon (CB) angioplasty was thought to be the gold standard for the treatment of dysfunctional dialysis access, either AVF or AVG. But Haskal's study showed that the incidence of patency of the treatment area and the access circuit in the CB group was only 23% and 20%, respectively [3]. Compared with CB, high-pressure balloons and cutting balloons are able to improve the patients' prognosis, but the stenosis rate is still high in the short term [4,5]. Accordingly, the emergence of paclitaxel coated balloon (PCB) is expected to be useful for solving the foregoing issues. The role of PCB in coronary artery diseases and peripheral arterial diseases has been widely recognized [6,7]. However, whether PCB angioplasty outperforms CB angioplasty in the treatment of dysfunctional AVF is still in controversy.

Materials and methods
The present meta-analysis was reported referring to the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) statement [22].

Search strategy
A systematic search of relevant literature available on PubMed, EMBASE, Web of Science and the Cochrane Library containing several keywords "arteriovenous fistula," "dialysis fistula," "drug-coated balloon," "drugeluting balloon" and "paclitaxel" published from their date of inception to November 28, 2021, was carried out (Appendix 1). We did not use any language or data restrictions, although we used only English search terms. References of these articles were also searched to find potential relevant articles.

Inclusion and exclusion criteria
Titles, abstracts and the full texts of all retrieved studies were preliminarily filtrated by a pair of authors to determine the inclusion (LC and LM). Disagreements from the two authors were solved by consensus or by appeal to a third review author (JJ). Inclusion criteria: (1) RCTs with two parallel arms; (2) Hemodialysis patients with dysfunctional AVFs; (3) Patients were treated with PCB angioplasties or CB angioplasties; (4) TLPP rates, technical success rates or all-cause mortality rates of both methods were provided in the literature; (5) Clinical follow-up of at least 6 months. Exclusion criteria: (1) Observational studies, animal studies, in vitro tests, reviews, comments, editorials, case reports and series, protocols, letters, conference abstracts, crossover trials and single-arm tests; (2) Repeated reporting; (3) Full text not available; (4) AVF and AVG data reported together; (5) Use of a stent.

Outcomes of interest and data extraction
The endpoint events were defined in accordance with the Society of Interventional Radiology (SIR) criteria for percutaneous interventional procedures in dialysis access [23] and the previous literature [12,15,19]. TLPP was adjudicated as freedom from clinically-driven target lesion revascularization (CD-TLR) or access circuit thrombosis during the follow-up period. TLPP ended when any one of the followings occurred: (1) decreased access blood flow (<500mL/min, 25% decrease in flow); (2) elevated venous pressures; (3) decreased dialysis dose (Kt/V); (4) abnormal physical exam included: i. diminished or abnormal thrill (focal, systolic only, etc); ii. pulsatility; iii. flaccid access; iv. abnormal bruit; v. arm or hand swelling; (5) prolonged bleeding; (6) difficult puncture; (7) infiltration; (8) recirculation; (9) pulling clots. Technical success was defined as successful completion of the angioplasty procedure with <30% residual stenosis by visual estimate and a palpable thrill. All-cause mortality was reported through 12 months. Data were separately extracted by two review authors (LC and LM).

Risk of bias and quality assessment
Methodological quality appraisal was conducted by two independent reviewers (LC and LM). In the event of a discrepancy between the two authors, a third author will decide (JJ). The quality evaluation of selected studies was performed using 7 elements from the recommended Cochrane Collaborations tool: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and other bias [24]. Publication bias was assessed through visual inspection of funnel charts, whereby an asymmetric funnel diagram indicated the presence of publication bias.

Statistical analysis
Statistical heterogeneity was assessed via the I 2 statistic and Cochrane's Q test. If evident heterogeneity (I 2 > 50% or p < 0.1), the random-effect model was employed for analysis; if not, the fixed-effect model. We conducted sensitivity analyses and subgroup analyses to search for the potential sources of heterogeneity. A value of p < 0.05 was accepted as a statistically significant difference. Meta-analyses were processed with STATA software version 15.1 (StataCorp, College Station, TX, USA) to calculate risk ratios (RRs) and their 95% confidence intervals (CIs). The kappa coefficients were computed by use of SPSS 25.0 (SPSS Inc., Chicago, IL, USA) to assess the degree of concordance between the two investigators.

Results
We identified 871 potentially eligible studies. After deduplication, 438 documents remained. Of these abstracts, 409 were excluded based upon the inclusion and exclusion criteria. Then, a total of 29 potentially relevant full-texts were retrieved and subjected to further review. Studies excluded and reasons for exclusion after full-text screening were provided in Table 1. Eventually, fourteen RCTs [8][9][10][11][12][13][14][15][16][17][18][19][20][21], including eight multicenter trials [8,[12][13][14][15][16]18,20], with 1535 patients fulfilled the criteria for inclusion. A moderately high level of agreement between two independent investigators was observed at the title and abstract review (kappa ¼ 0.695) and full-text evaluation (kappa ¼ 0.861) stages. The detailed steps of the study search and selection process were outlined in Figure 1 and the baseline characteristics of the selected trials were summarized in Table 2. Two of the fourteen studies included patients with AVF and AVG, but we extracted data only for   patients with AVF [9,14]. As for the methodological quality assessment, all studies scored three to seven points, among which four articles scored seven points [14,15,17,21]. The kappa values of agreement during the quality appraisal and data extraction were 0.863 and 1.000, respectively. The detailed score of each item for each article was described in Figure 2 and the proportion of each item in the methodological evaluation was shown in Figure 3.

6-Month TLPP
Thirteen studies [8][9][10][11][12][13][15][16][17][18][19][20][21] investigated the 6-month TLPP. The pooled event rates at 6 months in the PCB group and the CB group were 70.9% (482/680) and 59.5% (402/676), respectively. While the PCB group did improve the 6-month TLPP rates, the difference did not reach statistical significance (RR 1.17, 95% CI 0.99-1.39, p ¼ 0.065, I 2 ¼ 75.4%, Cochrane Q test p ¼ 0.000, random-effect model, Figure 4(B)). In order to estimate the possible sources of heterogeneity, sensitivity analysis and subgroup analyses based on PCB type, paclitaxel dose and inflation time were performed for this outcome. It is a pity that subgroup analyses of study characteristics did not find any factors that accounted for the heterogeneity ( Table 3). The results of a subgroup analysis according to PCB type indicated that the APERTO (Cardionovum) balloons significantly    We also did not find the source of heterogeneity through the sensitivity analysis. After removing one study, the estimates did not change significantly, which means these results were relatively robust in this meta-analysis (Appendix 2). The funnel graph was symmetrically distributed, and publication bias was not evident (Egger's test p ¼ 0.443, Appendix 2).

Discussion
In this meta-analysis, we conducted a comprehensive search for all studies regarding PCB angioplasty versus CB angioplasty in dysfunctional AVF. A total of fourteen articles containing 1535 subjects were included, and results showed no significant differences between the two groups in TLPP rates after 3, 6, 9 and 12 months of treatment (RR 1.00, 95% CI 0.93-1.07, p ¼ 1.000,  [40]. This is because while an uncoated balloon was used in the CB angioplasty, both an uncoated balloon for predilation and a PCB were required in the  PCB angioplasty. From the above results, it appears that PCB angioplasty is neither more effective nor much safer and is more costly. Thereby, PCB angioplasty seems to be not cost-effective compared to CB angioplasty from an economic point of view. Since the first RCT using PCB in AVF was conducted by Lai et al. [11], the debates on the benefits of PCB angioplasty for the treatment of dysfunctional AVF have never ceased. A number of meta-analyses evaluating the efficacy and safety of PCB angioplasty versus CB angioplasty for the treatment of dysfunctional AVF have been published [41][42][43]. In line with our findings, a recent meta-analysis by Liao et al demonstrated no significant improvement of TLPP rates in the PCB group, either for that at 6 months (RR 0.75, 95% CI 0.56-1.01, p ¼ 0.06) or 12 months (RR 0.89, 95% CI 0.79-1.00, p ¼ 0.06) [41]. Similar results have also been reported by Abdul Salim et al and Lazarides et al. [42,43]. In comparison to the meta-analyses described above, the strength of this paper lies in the incorporation of the most recent studies with large sample sizes into the final meta-analysis, making the results more persuasive.
Obvious heterogeneity existed in the included literature. We noticed that there was one study that obviously deviated from the axis of symmetry in the forest graph, which might have a great influence on heterogeneity [21]. Bjorkman et al suggested that the target lesion revascularization-free survival after the PCB angioplasty was clearly worse with 1-year follow-up, which was opposite to other studies [21]. In order to find out the source of heterogeneity, sensitivity analyses were performed by omitting one study at a time. The results of sensitivity analyses showed that there were no significant changes in the overall effect measures, indicating the results were relatively reliable.
In extension, we performed subgroup analyses based on paclitaxel dose, PCB type and inflation time. Of interest, the 6-month outcomes of the endovascular invention utilizing PCBs for dysfunctional AVFs were linked with the doses of paclitaxel delivered to vessels. We found that standard-dose PCBs (3.0 lg/mm 2 and 3.5 lg/ mm 2 ) were significantly more effective compared with CBs in improving 6-month TLPP rates, while there were no significant differences between low-dose PCBs (2.0 lg/mm 2 ) and CBs at 6-month TLPP rates. Katsanos and colleagues, pooling data from eleven RCTs, showed that standard-dose PCBs (3.0 lg/mm 2 and 3.5 lg/mm 2 ) were superior to low-dose PCBs (2.0 lg/mm 2 ) in reducing the rates of restenosis and target lesion revascularization (TLR) in the femoropopliteal artery, a finding that was generally consistent with the conclusions of our meta-analysis [44]. For these, the current study recommended that priority should be given to standard-dose PCBs (3.0 lg/mm 2 and 3.5 lg/mm 2 ) in the treatment of dysfunctional AVFs.
Meanwhile, we also found that APERTO (Cardionovum) balloons significantly outperformed CBs in terms of 6-month TLPP, whereas the Lutonix 035 (Bard), IN.PACT Admiral (Medtronic) and SeQuent Please (B.Braun) balloons did not improve 6-month TLPP rates compared with CBs. And Passeo-18 Lux (Biotronik) balloons were favored over CBs in terms of 12-month TLPP, whereas the Lutonix 035 (Bard), IN.PACT Admiral (Medtronic) and SeQuent Please (B.Braun) balloons did not improve 12-month TLPP rates compared with CBs. We speculated that this may be explained by the different doses of drug delivered to vessels.
Another interesting finding was that 6-and 12month TLPP rates were significantly higher in the group with the PCB inflation time !120 s compared with the controls. However, no significant differences were detected between the group with the PCB inflation time <120 s and the control group. In agreement with this study, Rhee and colleagues also reported that fully optimized PCB angioplasty with prolonged inflation time plays an important role in reducing target lesion failure after PCB angioplasty [45].
Although all of the enrolled studies were RCTs, allowing our findings to be reliable, several limitations should be acknowledged. Firstly, because of the macroscopic differences between PCBs and CBs, investigators cannot be unaware of the treatment assignment. None of the included RCTs was double-blinded, ineluctably increasing the risk of bias. Secondly, heterogeneity existed in the included literature. Sensitivity analyses and subgroup analyses according to PCB type, paclitaxel dose and inflation time were conducted to find the source of heterogeneity, compensating for this deficiency to some extent. However, the subgroup analysis based on AVF age was not performed in this meta-analysis. TLPP has been shown to have a positive correlation with AVF age in a previous study by Irani et al [28]. Thirdly, not all included articles reported the six outcomes and the data used for meta-analysis were incomplete, despite our efforts to contact authors of the included studies. Fourthly, all follow-ups were clinically driven. That is to say, not all patients have undergone ultrasound examinations before endovascular interventions, therefore causing a potential bias. Fifthly, unpublished results were not available, which inevitably produced publication bias. Finally, it should be acknowledged that the present study was not registered, with the possibility of a small offset. But it has to be pointed out that our meta-analysis was conducted strictly in compliance with the process of a systematic review.
In conclusion, this meta-analysis showed that there was insufficient evidence to support the distinct superiority of PCB angioplasty over CB angioplasty in the treatment of dysfunctional AVF. Due to the heterogeneity across studies, the findings from our study should be dealt with with some caution, although sensitivity analyses and subgroup analyses were performed to compensate for this deficiency to some extent. Thus, more multi-center, large-scale and well-designed RCTs are required to confirm our conclusions in the future.