Approximately 202 million people worldwide live with lower extremity artery disease (LEAD) [1], a common cause of which is chronic obstruction of the femoropopliteal artery (FPA) [2]. Modern bare metal nitinol stents (BMSs) have been widely used to manage FP lesions; however, their long-term patency and durability in the FP region are suboptimal [3], while in-stent restenosis (ISR) is the main challenge related to their durability [4]. The 12-month ISR rates after BMS implantation in the superficial femoral artery (SFA) and proximal popliteal artery are 18–37%.

Several methods have been applied to manage ISR, including percutaneous transluminal angioplasty (PTA), drug-coated balloon (DCB) implantation, and repeat stenting [5]. The advent of drug-eluting stent (DES) technology has reduced ISR rates but eventually leads to disappointment [6]. In recent years, paclitaxel (PTX) has been applied in DES and DCB because of its potent inhibitory effect on smooth muscle cell migration and proliferation at low concentrations. Furthermore, owing to the efficacious delivery of PTX from balloons, PTX-coated balloons (PCBs) have been developed as an alternative to DES and have been particularly successful in the treatment of peripheral artery disease (PAD) [7].

Last year, the DAEDALUS study compared PCB and DES for coronary ISR [8] and concluded that PCB for FP ISR is promising. To date, many trials have assessed the effectiveness and safety of PCBs for FP ISR, while horizontal comparisons of PCBs and uncoated balloon (UCB) are rare.

In the last five years, several new trials have been published; therefore, this meta-analysis aimed to assess the overall outcomes of prospective controlled trials comparing PCB with UCB for FP ISR.

Our results showed that, compared with UCBA, participants treated with PCBA experienced decreased recurrent restenosis, increased primary patency, freedom from TLR, and clinical improvement over a 1-year duration. However, the ABI and MAEs of patients did not differ significantly between the two interventions.

Guidelines recommend primary stenting as a class I intervention for FP lesions, but controversy persists regarding ISR. Although several endovascular therapies are available, they provide suboptimal long-term patency rates. Therefore, a few treatment methods are strongly recommended by these guidelines. In the 2018 Society for Cardiovascular Angiography and Interventions consensus guidelines for device selection in femoropopliteal arterial interventions [18], DCB was recommended as a class I treatment for managing FP ISR. In contrast, the 2017 ESC guidelines [3] provided a class IIb recommendation for DCB angioplasty for FP ISR. The evidence level of both guidelines was B, which partly explains this controversy. The quality of the evidence required improvement. Thus, it is important to confirm the efficacy and safety of PCBA versus UCBA.

Paclitaxel is currently a widely used drug in DES and DCB for its antiproliferative effect in vivo. To date, many types of DCB are available on the market [19]. Studies included in our analysis applied 5 different balloon products (IN.PACT Admiral, FREEWAY, Orchid, Cotavance and Lutonix), and the PTX dose of them is 3 and 3.5 $\mu$g/mm${}^2$. For the time being, the paclitaxel dose varies between 2 and 3.5 $\mu$g/mm${}^2$ [20], which contains the dose used in studies included. Another difference is the technology used in different products. However, only a few studies have compared different PCB directly, and made opposite conclusions [21, 22]. Thus, the present data are underpowered to discern outcome differences between the different PTX devices. In our analysis, there is no heterogeneity across trials in most outcomes, and the outcomes observe a similar trend, indicating that the devices may not influence the overall effects.

As shown in Figs. 2,3,4, our trial yielded reliable and consistent findings that support the benefits of PCBA. In brief, the recurrent restenosis and primary patency results revealed a better angiographic endpoint for PCBA versus UCBA. The outcomes reflected the vessel patency after the interventions measured on DUS and CTA as recommended by the guidelines [3]. Angiographic success laid the foundation for PCB preference in the management of FP ISR.

In our analysis, we chose freedom from TLR and clinical improvement as the clinical success parameters. Freedom from TLR was clinically driven and not based on imaging features. Clinical improvement was evaluated as an increase in Rutherford classification, a method related to clinical symptoms. As reported in Figs. 5,6,7,8, both outcomes supported that the clinical results of PCBA were superior to those of UCBA at one year post-intervention.

In our analysis, ABI did not differ significantly between PCBA and UCBA, but there was high heterogeneity across the three trials [12, 13, 15]. We confirmed the Liao trial as the source of the heterogeneity in the sensitivity analysis but failed to identify the variable causing the heterogeneity. The regression results reflected that race, PTX dose, and balloon device were not sources of heterogeneity. ABI was defined as the ratio of systolic blood pressure (SBP) measured at the ankle to that measured at the brachial artery, and it has become a good non-invasive test for diagnosing LEAD because of its good sensitivity and specificity [3]. In addition to its frequently used diagnostic function, ABI can be used as a follow-up parameter when combined with angiographic methods such as DUS for revascularized patients with PAD [23]. ABI was stable in most situations, but its sensitivity was poorer in patients with DM or end-stage CKD due to medial artery calcification (MAC) [24, 25]. As shown in Table 1, the FAIR and PACUBA trials reported a similar baseline ABI of approximately 0.65, while the Liao trial reported an ABI of approximately 0.50, and there was no significant difference between the PCB and UCB groups in terms of ABI. However, the proportion of patients with DM differed among the three trials. In the FAIR trial, the ratio was 45.2% in the PCB group and 29.8% in the UCB group. In the PACUBA trial, the rates were 52% and 38%, respectively. Only in the Liao trial were the ratios close between the two groups (50% versus 47.2%). We hypothesized that the existence of DM might explain the heterogeneity across the three trials, especially between the Liao trial and the other two trials. On the other hand, compared with the baseline values, the summarized results of the three trials [12, 13, 15] showed an approximate 0.20 increase in ABI at 12 months post-intervention, which was close to the threshold of 0.90 for diagnosing PAD [26]. The SFA ISR trial reported a similar increase in ABI. To make the calculation of ABI more precise, future researchers should maintain the consistency of baseline characteristics (DM, hypertension, smoking, CKD) between patients in the experimental and control groups. The toe-brachial index (TBI), another measurement when ABI is unsuitable, is generally unaffected by MAC with better sensitivity but lower specificity than ABI [23, 27].

As shown in Figs. 10,11, we can conclude that PCBA has a safety profile similar to that of UCBA. PCBA did not significantly increase the incidence of MAEs compared to UCBA. In contrast, PCBA significantly decreased the incidence of MAEs compared with UCBA 12 months post-intervention. UCBA has been the most commonly used strategy for PAD for a long time, and its side effects are relatively low [28]. Therefore, with a low side effect ratio, PCBA can also be safely applied to manage FP ISR. Besides, it is noteworthy that, the results of a meta-analysis published in 2018 aroused concern about an increased risk of death associated with the use of PCB to manage PAD [29]. Right after that, several researches [30, 31, 32] made a different conclusion that PCB was safe. Although the researchers did not reach an agreement, they all admitted that more data were needed, which made the safety of PCB still under controversy. In our analysis, there is no difference between PCBA and UCBA in the incidence of MAEs.

As illustrated in the sensitivity analysis, some outcomes were insufficiently stable, as characterized by a change in conclusion when one trial was omitted. Theoretically, a larger number of participants and a shorter 95% CI line in the forest plot represented a more reliable conclusion. In terms of the outcomes “clinical improvement at 6-month follow-up” and “MAEs at 12-month follow-up”, a relatively reliable trial was omitted, and the left trials reported a wide 95% CI, leading to a wide 95% CI of the overall effect. In terms of the outcomes “freedom from TLR at the 6-month follow-up” and “primary patency”, there were more reasons. Although both overall effects were positive, most trials in the analysis were not completely supportive, with a 95% CI of the OR of 1. The FAIR and Liao trials were the only ones to report a positive effect of the outcome, and the conclusion changed when they were omitted (Supplementary Fig. 13B,D).

Upon summarizing the four outcomes, the bounds of the 95% CI of the OR for all four outcomes were very close to 1. Although we could explain the unstable effect of every outcome, the primary cause was limited data. In this context, a slight difference was observed. Particularly, in the outcome “primary patency”, which was reported by only three trials [13, 15, 17], once the Liao trial was omitted, the conclusion could turn to unsure from positive, although there was no heterogeneity across the three trials. Almost every trial mentioned the limitation of a small amount of participants, and we continued to urge the importance of a future multicenter randomized controlled trial with a large number of participants to ensure a powerful conclusion.

The majority of trials set endpoints 12 months after interventions. The pooled effects of the outcomes reflected an evident trend. The OR of the 6- and 12-month follow-ups was 0.22 versus 0.18 for the outcome “recurrent restenosis”, 2.70 versus 4.02 for the outcome “freedom from TLR”, 1.87 versus 2.38 for the outcome “clinical improvement”, and 0.71 versus 0.50 for the outcome “MAEs”. A similar trend is shown in Table 2, with a smaller change in the percentage of the PCBA group than that of the UCBA group. In the ISAR-PEBIS and COPA CABANA trials, researchers reported the outcome “freedom from TLR” at the 24-month follow-up, and the ratios of the PCBA and UCBA groups were 64.3% versus 44.8% and 48.1% versus 25.0%, respectively. The DEBATE-ISR trial reported a 36-month follow-up outcome of “freedom from TLR”, with a ratio of 59.1% in the PCBA group and 57.1% in the UCBA group [33]. Therefore, based on the above data, we concluded that PCBA might achieve better outcomes than UCBA and that the gap could be increased over time to up to 24 months. Long-time effectiveness is an important advantage of PCBA over other endovascular treatment methods, and more evidence is needed to extend the follow-up period.

This evidence report has several limitations. First, there was little relevant data. There have been only seven prospective controlled trials of PCBA versus UCBA in the management of FP ISR since 2010. Among the chosen outcomes, only “freedom from TLR (12 months)” was reported by all seven trials, while the majority were referred to by fewer than five trials. The limited data was a key limitation of this meta-analysis.

Second, some trials reported a high missing follow-up rate. Almost half of the participants in the PACUBA trial were lost at the 12-month follow-up, which decreased the value of our analysis. In particular, in terms of the outcome “primary patency”, data from the PACUBA trial hardly impacted the overall effect. A high missing follow-up rate would sharply reduce the amount of data, leading to questionable conclusions.

Third, every study but the SFA ISR trial conducted dual-antiplatelet therapy for different durations, but only the DEBATE-ISR and FAIR trials reported some adverse events that might be related to the therapy. Relevant adverse events, such as ischemic and hemorrhagic accidents, should be completely reported.

Fourth, most trials focused on efficacy and safety outcomes; only the Liao and SFA ISR trials examined functional outcomes such as the Walking Impairment Questionnaire, EuroQol 5 dimensions quality-of-life measure, and 6-minute walking test. Intermittent claudication is always the primary symptom in patients with LEAD; thus, functional improvements should be assessed, especially in patients with walking impairments.

Finally, as discussed in the previous section, PCBA might achieve increasingly better outcomes than UCBA over time, but studies presenting such outcomes are scarce. However, the long-term durability and effects of PCBA are unknown. Although a cohort study assessed the clinical efficacy of DCB for FP lesions over 3 years [34], few studies have focused on the long-term durability and efficacy of PCBA versus UCBA for FP ISR.

Findings from our meta-analysis showed a reliable beneficial effect in terms of both angiographic and clinical success and a similar effect on the safety outcomes of PCBA versus UCBA. These data support clinician decisions regarding the management of FP ISR. Specifically, PCBA as a treatment strategy could achieve better short-term outcomes than UCBA for FP ISR management, including potent recurrent restenosis-lowering and symptom-improving capacity without increased MAEs.

QL and HL had the idea for the study. QL, YW and HL selected studies for inclusion and abstracted data. LW did the statistical analyses. QL, LW, LZ, YW and HL interpreted the data. QL, LZ and HL wrote the first draft. QL, LW, LZ, YW, LMW and HL critically revised the paper for important intellectual content. All authors approved the final draft.

Not applicable.

Not applicable.

This work is supported by Sichuan Science and Technology Program [grant number 2019YFS0037].

The authors declare no conflict of interest.