Impact of Bypass Conduit and Early Technical Failure on Revascularization for Chronic Limb-Threatening Ischemia.

The design and primary outcomes of BEST-CLI are reported elsewhere. Trial oversight committees were selected to represent multiple specialties and stakeholders, and included attention to diversity in recruitment and retention of study participants. 7 , 8 De-identified BEST-CLI data and materials have been made publicly available by the National Heart, Lung and Blood Institutes (NHLBI) and is available under accession HLB02932424a through this NHLBI BIOLINCC website: https://biolincc.nhlbi.nih.gov/studies/best_cli/ BEST-CLI was conducted at 150 sites in the United States, Canada, Finland, Italy and New Zealand. The study protocol was approved by the appropriate ethics committee at each site, and all patients provided written informed consent. The disposition of patients enrolled in the trial, and the treatments received, is summarized in the modified CONSORT diagram ( Figure S1 and Table S1 ). Enrolled patients were placed into either Cohort 1 or 2 based on assessment of the GSV; they were further stratified by the limb status at presentation (rest pain or tissue loss) and the presence of significant infra-popliteal (IP) artery disease to ensure balance in these factors. Following randomization (N=1830) a total of 50 patients did not undergo any revascularization procedure; the remaining 1780 are included in the as-treated analysis and received ENDO (N=923), SSGSV bypass (N=621) or AC bypass (N=236). This includes 34 patients who crossed over from their randomized (OPEN vs ENDO) treatment assignment. Of the 718 patients randomized to OPEN in Cohort 1, 662 received an open operation as the first procedure per-protocol and 581 (88%) of these received a SSGSV bypass; 81 (12%) received an AC bypass. Of the 197 patients randomized to OPEN in Cohort 2, 188 received an open operation as their first procedure per-protocol and 151 (80%) of these received an AC bypass; 37 (20%) had their bypass completed using SSGSV. Technical failure in the trial was defined as failure to achieve a patent bypass graft or abandonment of the surgical bypass procedure for any cause in the OPEN arm; for ENDO it was defined as inability to cross a target lesion, occlusion or persistent obstruction (>50%) of the target artery path such that in-line flow to the foot was not achieved, or a complication (vessel rupture, dissection or embolization) that could not be managed via an endovascular approach. Of note, investigators were permitted to execute a planned second stage ENDO intervention within 4 days of the initial procedure without penalty if this was deemed clinically appropriate. Further details of the BEST-CLI trial design and procedures are reported elsewhere. 7 , 8 All endpoints were measured as time to event, except for total number of target limb reinterventions over time. Major adverse limb events (MALE) were defined in accordance with the Society for Vascular Surgery Objective Performance Goals 9 as either a major limb amputation or a major re-intervention (new open bypass graft, major open surgical revision (jump or interposition graft), or thrombectomy/thrombolysis of the target limb). Minor reinterventions included repeat percutaneous angioplasty, stenting, or atherectomy or a surgical patch angioplasty. The primary composite endpoint was MALE or all-cause mortality at any time. Secondary endpoints included MALE or perioperative (30 day) death (MALE-POD), major amputation, all cause mortality, any target limb reintervention, and the composite endpoint of reintervention, amputation, or death (RAD). All BEST-CLI patients who underwent a revascularization procedure (N = 1780) were included in this analysis. The outcome analyses were performed according to the as-treated principle. Time-to-event outcomes were described with the use of Kaplan–Meier (KM) methods, and treatment groups were compared with the use of log-rank test statistics; KM event rates are reported at 1 and 3 years. Multivariable Cox proportional hazards models were created for each of the endpoints, adjusting for pre-specified baseline variables including randomization strata, prior infrainguinal revascularization in the target limb, end stage renal disease (ESRD), diabetes, and smoking. Additional baseline covariates identified as being associated with study endpoints on univariate screen (P<0.2) were also included in these models. Hazard ratios (HR) and 95% confidence intervals (CI) are reported with ENDO as the referent group. The outcome of major limb amputation was also analyzed with death as a competing risk using the Fine–Gray subdistribution hazard model. 10 We used the sensitivity index of the KM estimate, defined as D(t) = 100 *(1-KM(t))/n(t) where KM(t) is the incidence rate at time t and n(t) is the number of subjects at risk, to determine the time beyond which the estimates are less meaningful. Based on Gebski et al 11 this threshold was determined to be 5 years, thus in our presentation the KM curves were truncated at 5 years. In computing the number of target limb reinterventions/person-year the follow-up time was determined by either major limb amputation or study termination for each subject. The widths of the confidence intervals have not been adjusted for multiplicity, so confidence intervals should not be used for hypothesis testing. All the analyses were performed with the use of SAS version 9.4 (SAS Institute), and R software, version 4.30.

Baseline patient characteristics, limb status, and procedural details are summarized in Tables 1 and 2 . In general, patient characteristics were well balanced across the three as-treated groups with the exceptions of prior coronary artery bypass grafting (CABG: 37.5% SSGSV, 68.8% AC, 49.6% ENDO: p<.001), prior CABG or percutaneous coronary intervention [PCI] (66.8% SSGSV, 85.7% AC, 76.2% ENDO; p<.001), prior transient ischemic attack or stroke (13.9% SSGSV, 20.8% AC, 17.3% ENDO; p=.037) and baseline antiplatelet use (dual therapy: 13.4% SSGSV, 20.3% AC, 20.6% ENDO; p<.001). There were no significant differences in prior index limb revascularization, WIFI stage, or baseline hemodynamics in the affected limb. Bypass targets differed across the two OPEN subgroups with SSGSV targets more commonly being infra-popliteal (IP: 56%, vs 44% for AC; p<.001). Levels of disease treated (femoropopliteal [FP] only, IP only, or combined FP+IP) were significantly different (p=.003) with notably few (9.5%) isolated IP bypasses using AC conduits. Among the ENDO subgroup, 41% underwent multi-level interventions, 46% were treated with a drug-eluting device (drug coated balloon [DCB] or drug eluting stent [DES]), stents were used in 39.8% and atherectomy was infrequently used (14.3%). In the SSGSV subgroup, 39% of bypasses were multi-level (FP+IP), 42.5% were FP only and 18.5% IP only. Overall immediate technical failure occurred in 157 subjects, the majority in the ENDO subgroup (16.3% ENDO vs 0.3% SSGSV, 3.9% AC; p<.001). Examination of the baseline patient and limb characteristics across the 3 subgroups with technical success (not shown) demonstrated similar findings to the overall population. Peri-procedural (30 day) outcomes across the three as-treated groups are summarized in Table 3 . Mortality (1.3% SSGSV, 2.5% AC, 0.9% ENDO; p=.114) was not different by treatment received. There were a greater number of MACE events in both OPEN subgroups, but the difference was not significant (4.5% SSGSV, 4.7% AC, 2.6% ENDO; p=.084). There was no difference in the proportion of patients experiencing at least one serious adverse event (SAE). MALE was significantly greater in the ENDO subgroup (2.7% SSGSV, 3% AC, 13.1% ENDO; p<.001). In comparison to ENDO treated patients, those treated with SSGSV bypass experienced significantly reduced rates of the primary endpoint MALE or all-cause death (HR 0.65; 95% CI: 0.56, 0.76; p<.001), as well as major amputation (HR 0.7; 95% CI:0.52, 0.94; p=.016), MALE-POD (HR 0.51; 95% CI:0.41, 0.62; p<.001), RAD (HR 0.69; 95% CI: 0.61, 0.79; p<.001), and any target limb reintervention (HR 0.64; 95% CI: 0.55, 0.76; p<.001). In comparison to ENDO, those treated with AC bypass experienced reduced rates of MALE-POD (HR 0.69; 95% CI: 0.52, 0.92; p=.01), RAD (HR 0.79; 95%CI: 0.66, 0.95; p=.014) and any reintervention (HR 0.73; 95% CI: 0.57, 0.93; p=.011). Unadjusted time-to-event curves are illustrated in Figure 1 (A - F ). The estimated KM outcome rates at 3 years for each of the as treated subgroups are summarized in Table 4 (top section). Table S2 provides 1-year KM estimates for all as-treated patients. Table S3 provides 3-year KM estimates separately for the two ITT cohort subpopulations. To examine the impact of anatomic levels of disease treated, we repeated the Cox regression models separately within the three subgroups of FP only, IP only, FP + IP. These results ( Figure S2 ) demonstrate that the improved outcomes associated with SSGSV bypass versus ENDO were greatest in the patients who underwent FP-only procedures. In this subgroup, SSGSV bypass was associated with reduced incidences of major amputation (HR 0.54; p=.025), MALE or death (HR 0.63; p<.001), MALE or postoperative death (MALE-POD:HR 0.5; p<.001), reintervention-amputation or death (RAD; HR 0.7; p<.001), and any reintervention (HR 0.66; p=.002) versus ENDO. AC bypass in the FP-only subgroup was also superior to Endo in the endpoints of MALE-POD (HR 0.59; p=.027) and any reintervention (HR 0.63; p=.012). While outcome trends were similar in the FP+IP and IP-only subgroups, they did not achieve significance with the exception of RAD (HR 0.81; p=.046) for SSGSV vs Endo in the combined FP+IP patients. We also examined the total number of target limb reinterventions experienced by patients over follow-up in the trial. Among the 621 patients treated with SSGSV bypass a total of 344 reinterventions were performed, yielding a rate of 23.8 reinterventions per 100 person-years. In the 236 patients treated with AC bypass, 147 reinterventions were performed for a rate of 34.8 per 100 person-years. Among the 923 patients treated with ENDO, 720 reinterventions were performed for a rate of 35.2 per 100 person-years. Cox proportional hazards models for selected endpoints identified other important predictors ( Table S4 ). ESRD was significantly associated with all-cause mortality, MALE-death, major amputation, and RAD (p<.001 for all). Diabetes was associated with major amputation (p=.01) and all-cause mortality (p=.04). Prior index limb revascularization was associated with RAD or any target limb reintervention (p=.05). Randomization strata (tissue loss vs rest pain, presence of significant IP disease versus none) was associated with all-cause mortality, MALE-death, major amputation, and RAD (p<.01 for all). When excluding patients who experienced early technical failure, there were 616 SSGSV, 223 AC, and 750 ENDO treatments. Superior outcomes for SSGSV in comparison to ENDO remained evident for the endpoints of MALE-death (HR 0.81, 95%CI: 0.69, 0.95; p=.01), major amputation (HR 0.73, 95% CI: 0.54, 0.99; p=.041), MALE-POD (HR 0.74, 95%CI: 0.59, 0.92; p=.008) and RAD (HR 0.83, 95% CI: 0.72, 0.95; p=.007;). There were no significant differences between AC bypass and ENDO subgroups once technical failures were excluded. Estimated KM event rates at 3 years for patients who experienced technical success are summarized in Table 4 (lower section) and survival curves are illustrated in Figure S3 .

Analysis of the as-treated outcomes from the BEST-CLI trial demonstrates that, among patients with infrainguinal CLTI who were deemed acceptable candidates for either OPEN or ENDO treatment strategies, those undergoing initial bypass with SSGSV experienced significantly fewer major amputations, major or any reinterventions, and no difference in all-cause mortality in comparison to those receiving initial ENDO procedures. These differences remained significant, albeit with attenuated effect size, when patients with early technical failure were excluded. These results should be viewed as supplementary to those reported from the primary trial intention-to-treat (ITT) analysis and provide further confirmation of those findings. They also justify the parallel trial design of BEST-CLI based on preoperative assessment of GSV quality, a determination readily made in clinical practice using noninvasive ultrasound vein mapping. Recent reports from propensity-matched cohort studies provide similar findings to BEST-CLI. Ricco et al reported on a multi-center retrospective study of 793 patients who underwent first time infrainguinal bypass or endovascular interventions for CLTI. 12 Propensity-matched analysis (N=472) demonstrated superior amputation-free survival and wound healing, and reduced MALE and major amputation, in those treated with open bypass (81% autogenous conduits). There were no differences in all-cause mortality or MACE between groups. Liu et al reported a single institution series of patients with CLTI (N=413) who underwent revascularization and had complete WIFI staging available. 13 Using inverse propensity weighting they identified improved limb salvage for those treated with autogenous vein bypass (29% of cohort) versus endovascular or non-autogenous bypass; notably, the benefit was greatest among patients presenting with advanced limb threat (WIFI stage 4). Zarrintan et al conducted a Medicare-linked Vascular Quality Initiative (VQI-VISION) analysis of patients who had undergone either open bypass or endovascular interventions for CLTI over a ten year period (2010–2019) employing a robust propensity score matching approach. 14 They subdivided open bypass by those employing GSV conduit versus alternative conduits. They found that open bypass with GSV was associated with reduced hazard for death, amputation, and combined amputation/death at both two and four years. In their study the improved outcomes for GSV bypass versus ENDO were evident across all three anatomic levels of disease (FP, IP, FP+IP). In contrast, a contemporary Medicare claims analysis found that CLTI patients in the Medicare population were older, had more comorbidities, and experienced higher mortality and MALE after bypass surgery than what was observed in the surgical arm of BEST-CLI. 15 Numerous published studies demonstrate the safety and superiority of SSGSV bypass over alternative bypass conduits in patients with CLTI. The results obtained in BEST-CLI align well with these prior reports. For example, Arvela et al reported an institutional series of 1,109 open bypass procedures among which 818 (74%) were constructed using SSGSV. 3 They observed primary patency rates of 74.4% and 67.1% at 1 and 3 years, and limb salvage rates of 88.9% and 86.9% respectively. The present results from BEST-CLI compare favorably, with a 1-year rate of any reintervention of 28.6% and major amputation in 8.1% among those treated with SSGSV. Further comparative outcomes by conduit types employed in BEST-CLI are beyond the scope of the present report and are a subject of ongoing analysis. It is also notable that among the patients enrolled in BEST CLI who were assessed preoperatively as likely to be able to undergo a SSGSV bypass (Cohort 1), 88% of those assigned to OPEN had their bypass completed using SSGSV. Conversely 20% of those assigned to OPEN in Cohort 2 were still able to receive a SSGSV bypass at time of surgery. This implies that the specificity of the preoperative assessment for GSV quality was high, but the sensitivity somewhat less so. Since the approach to preoperative vein mapping was not standardized in the trial, we are unable to shed further light on this. However experienced surgeons may explore GSV conduits deemed marginal on preoperative ultrasound and/or modify the inflow and outflow targets to achieve an SSGSV graft. The incidence, causes, and downstream consequences of early technical failure in BEST-CLI are the subject of a separate analysis. Much has been made of the technical failure rate reported in the trial, yet it appears largely concordant with data reported from other contemporary trials and registries particularly in the setting of complex anatomy. 12 , 16 , 17 Technical failure following ENDO interventions was an important driver of the observed differences in MALE in both the reported ITT results and the as-treated analysis. However, the present analysis demonstrates that SSGSV bypass remained superior to endovascular intervention across multiple efficacy endpoints even when those who experienced early technical failure were excluded. Superior outcomes with respect to need for repeat interventions and limb salvage speak to the greater durability and likely greater hemodynamic gain associated with SSGSV bypass. As the clinical indications underlying repeat interventions were not fully captured in the BEST-CLI database, it remains unclear to what extent ongoing or repeat CLTI symptoms were the primary drivers of these secondary procedures. We recently reported an ITT analysis from the trial focused on index limb reinterventions, which highlighted the superior durability associated with OPEN treatment assignment in the trial. 18 Another pre-planned ITT analysis reported on several novel secondary outcomes such as time to resolution of CLTI symptoms, recurrence of CLTI, and hemodynamic failure; all of these endpoints favored the OPEN treatment assignment within Cohort 1 of the trial. 19 In contrast to BEST-CLI, the recently reported BASIL-2 trial conducted primarily in the United Kingdom compared initial OPEN versus ENDO strategies in CLTI patients with infra-popliteal disease and found that amputation-free survival was better for those initially treated with ENDO. 17 The apparently conflicting results between BEST-CLI and the BASIL-2 trial have been the topic of much debate and discussion in the field. Important differences in the study populations, trial design, and healthcare systems have been enumerated and ongoing work between the study groups is looking to better clarify areas of concordance. 20 BEST-CLI demonstrated significant differences in limb-related outcomes, but not overall patient mortality, by treatment strategy; BASIL-2 essentially had the opposite findings. BASIL-2 has not reported an as-treated analysis to date. Differences in the surgical outcomes from the two trials are most relevant. In both the ITT and the as-treated analyses of the BEST-CLI trial, there were no significant differences in perioperative or long-term mortality between ENDO and OPEN strategies. Notably the perioperative mortality reported here for SSGSV bypass (1.3%) and even AC bypass (2.5%), is lower than for those assigned to open bypass with vein (6%) in BASIL-2. The BASIL-2 study reported a 53% mortality for the open bypass group at a median follow-up time of 40 months, which also appears substantially higher than the 31.4% 3-year mortality reported here for patients who received SSGSV bypass in the BEST-CLI trial. Importantly, BASIL-2 enrolled subjects who required treatment of IP disease, and the extent of FP disease among subjects in that trial is unclear. It is thus of interest that we found the greatest differences between SSGSV bypass and ENDO outcomes in BEST-CLI was in patients treated for FP-only disease (who would have been excluded from BASIL-2). A post-hoc analysis of angiographic patterns of disease in BEST-CLI is currently ongoing and will provide greater insight into anatomic determinants of treatment outcomes. Further work is needed to better unravel the confounders in these top-line comparisons between the two trials. This study has limitations. First, the as-treated analysis, by its nature, is subject to confounding and selection bias associated with any non-randomized comparison. We adjusted for multiple baseline variables to reduce such confounding. Intrinsic differences between subjects enrolled in the two randomization cohorts are reflected in the contrasts between the surgical conduit subgroups. A second major limitation is the lack of anatomic details on the disease patterns of patients in BEST-CLI, an area of ongoing post-hoc study. Third, BEST-CLI was a pragmatic trial and thus there was a lack of standardization of procedures within the randomized arms. Fourth, the protocol did not mandate surveillance imaging following index revascularizations, and as data from surveillance studies was not reliably collected, we are unable to ascertain primary or secondary patency rates in the trial. Finally, enrollment in the BEST-CLI trial was limited to patients judged by the investigators or CLI team to be eligible for both OPEN and ENDO treatment approaches, and in whom there was equipoise.

Among CLTI patients deemed suitable for either open bypass or endovascular intervention for infrainguinal disease, open bypass with SSGSV was associated with superior long-term clinical effectiveness and similar safety in comparison to ENDO. AC bypass also demonstrated lower rates of reintervention but was otherwise equivalent to ENDO. The improved results obtained with SSGSV bypass in comparison to ENDO were evident even when patients who experienced early technical failure were excluded. Patients with CLTI who are acceptable surgical candidates are recommended to undergo evaluation of GSV conduit availability to allow for optimal shared decision-making regarding their treatment options.

Effective revascularization is a fundamental pillar of limb preservation for patients with chronic limb-threatening ischemia (CLTI). Clinical success in CLTI is largely dependent on patient risk, severity of limb threat at presentation, and the anatomic complexity of arterial occlusive disease. 1 Open surgical bypass grafting (OPEN) is an established modality whose outcomes are strongly dependent on the quality of conduit employed, amongst other factors. For infrainguinal bypass, a single-segment great saphenous vein (SSGSV) of adequate caliber (i.e. >3 or 3.5 mm) provides the optimal patency and limb salvage rates. 2 , 3 Alternative bypass conduits (AC), which include spliced autologous vein grafts, prosthetic grafts, cryopreserved allografts, or composites of any of the above, may provide reasonable mid-term results in selected patients but are inferior to SSGSV in terms of patency and limb salvage. 4 Endovascular interventions (ENDO) have assumed a dominant role in the treatment of CLTI over the last two decades. ENDO techniques encompass a spectrum of plain balloon angioplasty (PBA), atherectomy, bare metal or covered stents, or drug-eluting devices (drug-coated balloons, drug-eluting stents). ENDO procedures are minimally invasive but may be limited by immediate technical failure as well as high rates of restenosis in complex patterns of disease. While recent randomized controlled trials (RCT) demonstrate that re-interventions are significantly reduced when drug-eluting devices are employed versus PBA for femoro-popliteal artery disease, few patients with CLTI are represented in these trials. 5 In addition, patients with CLTI commonly present with infra-popliteal (IP) or multi-level disease 6 and only recently has any drug-eluting device for IP disease gained regulatory approval in the United States * . Thus, in contrast to open surgery, the optimal ENDO strategy for clinical success in CLTI remains highly variable and operator-dependent. The Best Endovascular versus Best Surgical Therapy in Patients with CLTI (BEST-CLI) trial was a large, multi-center, pragmatic, randomized, controlled trial (RCT) designed to compare the effectiveness of OPEN versus ENDO revascularization among patients with CLTI due to infrainguinal peripheral artery disease, who were deemed acceptable candidates for either treatment. 7 , 8 The trial was designed as two parallel RCT, based on a-priori hypotheses related to the differential performance of bypass conduits. Specifically, Cohort 1 included patients deemed to have an acceptable SSGSV based on preoperative evaluation, with the hypothesis that bypass with SSGSV would be superior to ENDO. In contrast, the hypothesis for Cohort 2 (patients lacking adequate SSGSV) was that ENDO would outperform AC bypass. The specific strategy for “best” OPEN or ENDO approach in each patient was left to the treating investigator. Randomization in the trial was stratified by clinical indication (ischemic rest pain vs. tissue loss) and by the presence of significant infra-popliteal (IP) disease. The primary endpoint was freedom from a major adverse limb event (MALE) or death from any cause. Secondary endpoints included components of the primary endpoint, major (above ankle) limb amputation, all-cause mortality, and the composite outcome of target limb reintervention, amputation, or death (RAD). The primary trial results analyzed by intention-to-treat have been reported, demonstrating superiority of OPEN over ENDO in cohort 1, and no significant difference in cohort 2. Immediate technical failure (ITF) was higher in the ENDO arms of both cohorts (15.1%, 18.6% respectively), leading some to question whether the selection of patients, expertise or bias of the trial investigators may reduce the applicability of the findings. Herein we report a planned analysis of the key clinical outcomes in BEST-CLI, based on the initial treatment received. We hypothesized that open bypass using a SSGSV graft would provide superior clinical effectiveness to ENDO in all limb-related endpoints, regardless of the immediate technical outcome. We examined the outcomes of all patients as-treated (both cohorts) across the three groups- SSGSV bypass, AC bypass, and ENDO, and then after excluding subjects who experienced immediate technical failure.