Comparison of drug‐coated balloon angioplasty versus common balloon angioplasty for arteriovenous fistula stenosis: A systematic review and meta‐analysis

Abstract Drug‐coated balloons (DCBs) have been used in dialysis patients with arteriovenous fistula (AVF) stenosis, but whether DCBs have advantages over ordinary balloons is still controversial. A meta‐analysis was designed to investigate the safety and efficacy of DCBs and common balloons (CBs) in the treatment of AVF stenosis. We searched the PubMed, EMBASE, and China National Knowledge Internet (CNKI) databases for randomized controlled trials that evaluated the comparison of DCB angioplasty versus CB angioplasty for AVF stenosis in dialysis patients and reported at least one outcome of interest. The results showed that the DCB group had a higher first‐stage patency rate of the target lesion 6 months [odds ratio, OR = 2.31, 95% confidence interval, CI: (1.69, 3.15), p < .01] and 12 months [OR = 2.09, 95% CI: (1.50, 2.91), p < .01] after surgery. There was no statistically significant difference in all‐cause mortality between the two groups at 6 months [OR = 0.85, 95% CI: (0.47, 1.52), p = .58] and 12 months [OR = 0.99, 95% CI: (0.60, 1.64), p = .97]. Compared with CB, DCBs as a new endovascular treatment for AVF stenosis have a higher primary patency rate of target lesions and can delay the occurrence of restenosis. There is no evidence that DCB can increase the mortality of patients.


| INTRODUCTION
Worldwide, approximately 2 million people undergo dialysis for endstage renal disease, 1 which is usually treated with hemodialysis. 2 Autogenous arteriovenous fistula (AVF) is the preferred vascular access for hemodialysis in patients 3 ; however, with gradual increases in the application time of AVF, the incidence of vascular accessrelated complications such as AVF stenosis or even occlusion increases significantly. Several studies have shown that the AVF patency rate is usually only 60%-65% at 1 year after operation. 4 An effective solution for AVF stenosis is to perform a second operation to re-establish the internal fistula, but this will reduce and eventually The following keywords were used for the search strategy:

| Eligibility criteria
Two authors (Yong Zhang and Xiang-Yang Hu) independently carried out the primary review to search for trials that met the inclusion criteria (Supporting Information: Table S3). Any discrepancy was resolved by discussion and consensus (Supporting Information: Figure S1). The following criteria were used: (1) Adult participants (≥18 years) on dialysis for at least 6 months, irrespective of age, sex, and race; (2) comparison of hemodialysis patients with stenotic AVF or arteriovenous graft (AVG) who underwent DCB angioplasty and those who underwent CB angioplasty at 6 months or 12 months; (3) inclusion of one of the following outcomes: primary patency and all-cause mortality at 6 or 12 months; (4) only RCTs were included in the meta-analysis. The main characteristics of the included studies are listed in Table 1 and Supporting Information: Table S1.

| Data extraction
Two reviewers (Xiang-Yang Hu and Yong Zhang) independently extracted data from the same set of publications. The following information was extracted: author, year, country, sample size, average age, paclitaxel dose, follow-up, and main results.

| Summary of effect size
Odds ratios (ORs) with 95% confidence intervals (CIs) were used as the effect size measures of dichotomous data. The weight of enrolled studies was determined by accounting for the size of the treatment group, control group, and total sample size. A Z-test was calculated, and therapeutic efficacy was deemed significant with a p < .05 cut-off. 12

| Risk of bias
The quality of all trials was independently evaluated by two authors (Xiang-Yang Hu and Yong Zhang) according to the Cochrane quality criteria (Supporting Information: Figure S2). Any disagreement between the authors was settled by a discussion with a third author (Zhen-Wu Zou). A weighted kappa value was calculated to examine agreement between reviewers for the overall study risk of bias assessment (Supporting Information: Table S5).

| GRADE quality assessment
The overall quality of evidence was evaluated by two authors (Fan-Li Yuan and Yong Zhang) according to The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria.
The results and the overall quality of evidence are presented in Supporting Information: Table S4.

| Study selection
A total of 484 studies were identified during the initial search after excluding duplicate records (n = 35). Eighty-five articles were retained after title/abstract curation (excluding 433 records). Thereafter, we read the full text and analyzed 15 RCTs 9,10,13-25 involving a total of 1525 patients for quantitative synthesis (Supporting Information: Figure S1). The main characteristics of the included RCTs (country, design, sample size, age, intervention, follow-up, and main results) are described in Table 1.   Figure S4). The choropleth map revealed regional differences in the 6-month primary patency rate between Belgium, the Netherlands, Sweden, the United Kingdom, and the United States of America (Figure 1). Overall, subgroup analysis showed that in Asia and Oceania, DCBs had a higher primary patency rate at 6 months after balloon expansion, but in Europe and North

| Primary patency rate of the target lesion
The choropleth map of the 6-month primary patency rate. CI, confidence interval.
America, there was no significant difference between DCBs and CBs (Supporting Information: Figure S5). The meta-regression by bubble plot revealed no significant heterogeneity in the study size (p = .557, Supporting Information: Figure S6) OR publication year (p = .816, Supporting Information: Figure S7). Sensitivity analysis (Supporting Information: Figure S8) was performed, and Galbraith plots (Supporting Information: Figure S9) were generated to evaluate the stability of our results. The analysis suggested that no individual studies significantly affected the pooled OR, indicating that the results were statistically robust. No significant publication bias was found by Begg's plots (p = .228, Supporting Information: Figure S10), Egger's test (p = .271, Supporting Information: Figure S11) OR funnel plots (Supporting Information: Figure S12).  Figure S13). No significant heterogeneity was observed (I 2 = 9.28%). The L'Abbe plot showed that all 13 included studies fell to the upper left of the diagonal line, also indicating a higher primary patency rate at 12 months in the DCB group (Supporting Information: Figure S14). The choropleth map revealed regional differences in the 12-month primary patency rate in Australia, Belgium, the Netherlands, Sweden, Singapore, Spain, and the United Kingdom ( Figure 2). Overall, subgroup analysis showed that in Asia and Europe, DCBs had a higher primary patency rate at 12 months after balloon expansion, but in Oceania, there was no The choropleth map of the 12-month primary patency rate. CI, confidence interval. ZHANG ET AL. | 881 significant difference between DCBs and CBs (Supporting Information: Figure S15). The meta-regression by bubble plot analysis revealed no significant heterogeneity in the study size (p = .067, Supporting Information: Figure S16) OR publication year (p = .597, Supporting Information: Figure S17). Sensitivity analysis (Supporting Information: Figure S18) was performed, and Galbraith plots (Supporting Information: Figure S19) were generated to evaluate the stability of our results. The analysis suggested that no individual studies significantly affected the pooled OR, indicating that the results were statistically robust. Moreover, no significant publication bias was found by Begg's plots (p = .161, Supporting Information: Figure S20), Egger's test (p = .942, Supporting Information: Figure S21) OR funnel plots (Supporting Information: Figure S22).

| All-cause mortality
Five studies 9,13,16,22,23 assessed the incidence of all-cause mortality at 6 months after balloon expansion between the two groups in a total of 757 patients, with 382 assigned to the DCB group and 375 assigned to CB group (OR = 0.85, 95% CI: 0.47, 1.52, p = .58, Figure 3). No significant heterogeneity was observed (I 2 = 11.64%).
The meta-regression by bubble plot revealed no significant heterogeneity of the sample size (p = .090, Supporting Information: Figure S23) OR publication year (p = .476, Supporting Information: Figure S24). Sensitivity analysis (Supporting Information: Figure S25) was performed, and Galbraith plots (Supporting Information: Figure S26) were generated to evaluate the stability of our results and we found that no individual studies significantly affected the pooled OR, confirming that the results were statistically robust. No significant publication bias was found by Begg's plots (p = .327, Supporting Information: Figure S27), Egger's test (p = .318, Supporting Information: Figure S28) OR funnel plot (Supporting Information: Figure S29).
Seven studies 9,13,[16][17][18]23,24 assessed the incidence of all-cause mortality at 12 months after balloon expansion between the two groups in a total of 774 patients, with 382 assigned to DCB groups and 392 assigned to CB groups (OR = 0.99, 95% CI: 0.60, 1.64, p = .97, Figure 4). No significant heterogeneity was observed (I 2 = 0%). The meta-regression by bubble plot revealed no significant heterogeneity of the sample size (p = .241, Supporting Information: Figure S30) OR publication year (p = .912, Supporting Information: Figure S31). Sensitivity analysis (Supporting Information: Figure S32) and Galbraith plots (Supporting Information: Figure S33) were performed to evaluate the stability of our results. The analysis results suggested that no individual studies significantly affected the pooled OR, indicating that the results were statistically robust. No significant publication bias was found in the results of Begg's plots (p = .453, Supporting Information: Figure S34), Egger's test (p = .793, Supporting Information: Figure S35) OR the funnel plot (Supporting Information: Figure S36).

| Main findings
Contrary to previous studies, 26 our meta-analysis demonstrated that DCBs had a significantly higher primary patency rate of target lesions than CBs at 6 and 12 months after angioplasty. Meanwhile, we observed that there were some regional differences in the above conclusions. Overall, DCB has a higher primary patency rate at 6 months after balloon expansion in Asia and Oceania, but in Europe and North America, there is no significant difference between DCB and CB treatment. DCB has a higher primary patency rate at 12 months after balloon expansion in Asia and Europe, but in Oceania, there is no significant difference between the two groups. There could be many reasons for these regional differences, such as degree of stenosis, primary stenosis, central vein stenosis, diabetes mellitus, Forest plot of all-cause mortality at 6 months after balloon expansion. CI, confidence interval.
hypertension, antiplatelet aggregation drug use, drug dosage, and so on. We did not perform subgroup analyses sequentially, which may have affected our results. There was no statistically significant difference in all-cause mortality between the two groups at 6 and 12 months.

| Interpretation
Most patients with end-stage renal disease require long-term hemodialysis treatment, and vascular access is essential for effective hemodialysis. At present, long-term hemodialysis treatment mainly relies on AVF and AVG. Among them, AVF has become the first choice for dialysis access due to its advantages of long application time and fewer intervention times. 27 However, with the gradual prolongation of the application time of AVF, the incidence of vascular access-related complications such as AVF stenosis or even occlusion has increased significantly. Several studies have shown that the AVF patency rate after 1 year is usually only 60%-65%. 4 A large number of studies have shown that the main mechanism of AVF stenosis is that under the pro-inflammatory state of chronic kidney disease, hemodynamic changes caused by different anatomical morphologies of dialysis access and repeated puncture promote the proliferation of vascular intima tissue, make the vascular lumen smaller, and eventually lead to the failure of dialysis access. 28 An effective solution to AVF stenosis is to use secondary surgery to re-establish the internal fistula, but this will reduce the vascular resources of dialysis patients and eventually lead to the depletion of vascular resources.
With the advancement of medicine, PTA has gradually become one of the main methods for the treatment of AVF stenosis. In recent years, high-pressure balloons, cutting balloons, and other instruments have been gradually used in clinical practice, but there is still a high incidence of restenosis, as high as 55%-75% within 1 year. 24 It has been suggested that PTA is associated with a high restenosis rate of AVF, and intimal hyperplasia is the main pathophysiological mechanism of restenosis after PTA. 6  In recent years, there have been a few studies on the use of DCBs in the treatment of AVF stenosis. However, the sample size of these studies was small, which may lead to bias in the results. 10 In addition, some RCT findings are also controversial. This meta-analysis showed that compared with CB angioplasty, DCB angioplasty had a higher primary patency rate of target lesions at 6 and 12 months after the operation, and the difference was statistically significant (p < .05).
This confirmed that DCB could delay the occurrence of restenosis of AVF, improve the primary patency rate of target lesions, and thus prolong the vascular access time of patients with end-stage renal disease.
In addition, studies have shown that the use of paclitaxel DCB can increase the risk of death in patients with peripheral artery disease, which has aroused widespread concern about the safety of paclitaxel-related endovascular devices. 29 However, some scholars questioned whether the above analysis results were affected by

| CONCLUSIONS
In conclusion, compared with CB, DCB as a new endovascular treatment for AVF stenosis has a higher primary patency rate of target lesions and can delay the occurrence of restenosis. At present, there is no evidence that DCB can increase the mortality of patients, but further clinical studies are needed to determine its mid-and long-term efficacy and safety in the treatment of AVF stenosis. We believe that future studies should optimize the application technology according to specific vascular access types or lesion characteristics and clarify the role of DCBs in hemodialysis vascular access management.