Robustness of Longitudinal Safety and Efficacy After Paclitaxel-Based Endovascular Therapy for Treatment of Femoro-Popliteal Artery Occlusive Disease: An Updated Systematic Review and Meta-Analysis of Randomized Controlled Trials

Background: The aims of this study were: i) to assess fragility indices (FIs) of individual randomized controlled trials (RCTs) that compared paclitaxel-based drug-coated balloons (DCBs) or drug-eluting stents (DESs) versus standard endovascular devices, and ii) to meta-analyze mid-term and long-term safety and efficacy outcomes from available RCT data while also estimating the FI of pooled results. Methods: This systematic review has been registered in the PROSPERO public database (CRD42022304326 http://www.crd.york.ac.uk/PROSPERO). A query of PubMed (Medline), EMBASE (Excerpta Medical Database), Scopus, and CENTRAL (Cochrane Central Register of Controlled Trials) databases was performed to identify eligible RCTs. Rates of primary patency (PP) and target lesion revascularization (TLR) were assessed as efficacy outcomes, while lower limb amputation (LLA) consisting of major amputation that is. below or above the knee and all-cause mortality were estimated as safety outcomes. All outcomes were pooled with a random effects model to account for any clinical and study design heterogeneity. The analyses were performed by dividing the RCTs according to their maximal follow-up length (mid-term was defined as results up to 2-3 years, while long-term was defined as results up to 4-5 years). For each individual outcome, the FI and reverse fragility index (RFI) were calculated according to whether the outcome results were statistically significant or not, respectively. The fragility quotient (FQ) and reverse fragility quotient (RFQ), which are the FI or RFI divided by the sample size, were also calculated. Results: A total of 2,337 patients were included in the systematic review and meta -analysis. There were 2 RCTs examining DES devices and 14 RCTs evaluating different DCBs. For efficacy outcomes, there was evidence that paclitaxel-based endovascular therapy increased the PP rate and reduced the TLR rate at mid-term, with a calculated pooled risk ratio (RR) of 1.66 for patency (95% CI, 1.55-1.86; P < 0.001), with a corresponding number needed -totreat (NNT) of 3 patients (95% CI, 2.9-3.8) and RR of 0.44 for TLR (95% CI, 0.35-0.54; P = 0.027), respectively. Similarly, there was evidence that paclitaxel-based endovascular therapy both increased PP and decreased TLR rates at long-term, with calculated pooled RR values of 1.73 (95% CI, 1.12-2.61; P = 0.004) and 0.53 (95% CI, 0.45-0.62; P = 0.82), respectively. For safety outcomes, there was evidence that paclitaxel-based endovascular therapy increased all -cause mortality at mid-term, with a calculated pooled RR of 2.05 (95% CI, 1.21-3.24). However, there was no difference between treatment arms in LLA at mid-term (95% CI, 0.1-2.7; P = 0.68). Similarly, neither all -cause mortality nor LLA at long-term differed between treatment arms, with a calculated pooled RR of 0.66, 1.02 (95% CI, 0.31-3.42) and 1.02 (95% CI, 0.305.21; P = 0.22), respectively. The pooled estimates of PP at mid-term were robust (FI = 28 and FQ = 1.9%) as were pooled rates of TLR (FI = 18 and FQ = 0.9%). However, when safety outcomes were analyzed, the robustness of the meta -analysis decreased significantly. In fact, the relationship between the use of paclitaxel-coated devices and all -cause mortality at mid-term showed very low robustness (FI = 4 and FQ = 0.2%). At 5 years, only the benefit of paclitaxel-based devices to reduce TLR remained robust, with an FI of 32 and an FQ of 3.1%. Conclusions: The data supporting clinical efficacy endpoints of RCTs that examined paclitaxelbased devices in the treatment of femoral-popliteal arterial occlusive disease were robust; however, the pooled safety endpoints were highly fragile and prone to bias due to loss of patient follow-up in the original studies. These findings should be considered in the ongoing debate concerning the safety of paclitaxel-based devices.