Author + information
- Received August 25, 2016
- Revision received September 21, 2016
- Accepted September 25, 2016
- Published online December 19, 2016.
- Theodosios Bisdas, MD, PhDa,b,∗ (, )
- Matthias Borowski, PhDc,
- Konstantinos Stavroulakis, MDa,
- Giovanni Torsello, MDa,b,
- CRITISCH Collaborators
- aDepartment of Vascular Surgery, St. Franziskus Hospital GmbH, Muenster, Germany
- bDepartment of Vascular Surgery, University Clinic of Muenster, Muenster, Germany
- cInstitute of Biostatistics and Clinical Research, University of Muenster, Muenster, Germany
- ↵∗Reprint requests and correspondence:
Dr. Theodosios Bisdas, Department of Vascular Surgery, St. Franziskus Hospital GmbH, Hohenzollernring 72, Muenster D-48147, Germany.
Objectives The most effective first-line treatment between endovascular therapy and bypass surgery for patients with critical limb ischemia (CLI) is still not well defined. The primary aim of the interim analysis of CRITISCH (Registry of First-Line Treatments in Patients With Critical Limb Ischemia) was to compare both treatment options in a prospective confirmatory manner.
Background Only 1 randomized controlled trial between endovascular therapy and bypass surgery has been published yet. Several retrospective studies showed comparable outcomes between the 2 treatment strategies, but in the majority of them, current endovascular technologies have not been included.
Methods Between January 2013 and September 2014, 1,200 CLI patients (Rutherford 4 to 6) from 27 vascular centers were enrolled. The selection of the first-line treatment was left completely to the discretion of the responsible physician. The primary composite endpoint was amputation-free survival (AFS), that is, time to major amputation and/or death from any cause. A pre-specified interim analysis aimed at showing noninferiority of the endovascular therapy versus bypass surgery as to the hazard ratio (HR) of AFS (noninferiority bound = 1.33; interim α = 0.0058). Time-to-event analyses of major amputation, death, and the composite endpoint of reintervention and/or above-ankle amputation were also conducted.
Results Endovascular therapy was applied to 642 (54%) and bypass surgery to 284 (24%) patients. Median follow-up time was 12 months in both groups. One-year AFS was 75% and 72%, respectively. The noninferiority of endovascular therapy versus bypass surgery for AFS was confirmed (HR: 0.91; upper bound of 1-sided (1 − 0.0058) confidence interval [CI]: 1.29; p = 0.003). An impact of the treatment strategy on time until death (HR: 1.14; 95% CI: 0.80 to 1.63; p = 0.453), major amputation (HR: 0.86; 95% CI:0.56 to 1.30; p = 0.463), and reintervention and/or above-ankle amputation (HR: 0.89; 95% CI: 0.70 to 1.14; p = 0.348) was not observed.
Conclusions The interim analysis confirmed that when physicians are free to individualize therapy to CLI patients, the endovascular-first approach achieved a noninferior AFS rate compared with bypass surgery. (Registry of First-Line Treatments in Patients With Critical Limb Ischemia [CRITISCH]; NCT01877252)
The gold standard of treatment between bypass surgery and an endovascular approach in patients with critical limb ischemia (CLI) is still under debate. Bypass surgery is associated with encouraging patency- and amputation-free survival rate in the long run, especially in cases of autogenous vein conduits (1–3). On the other hand, recovery following endovascular therapy is faster than after bypass surgery, and the minimally invasive nature of the procedure prevents any additional wound or anesthesia-associated morbidity (up to 20%) (1–4). The BASIL (Bypass Versus Angioplasty for Severe Ischemia of the Leg) trial, which still remains the only published randomized controlled trial (RCT) in this field, showed comparable outcomes between the strategies (4), but favored bypass surgery for patients with a life expectancy >2 years (5).
Meanwhile, an evolution in the endovascular field was undertaken, and several RCTs confirmed the superiority of the current technologies to plain balloon angioplasty, which has been investigated in the BASIL trial (6). However, patients with CLI were the minority, if not an exclusion criterion, in those RCTs, and evidence concerning the effectiveness of these modalities in this challenging cohort is still missing. Moreover, there is a paucity of comparative data between the new technologies and bypass surgery (7). In this context, the results of the BASIL 2/3 and the BEST-CLI (Best Endovascular Versus Best Surgical Therapy in Patients With Critical Limb Ischemia) trials will surely provide strong evidence regarding this debate. However, due to delays and difficulties in recruitment, the first results are anticipated after 2019.
To inform the published reports as to the aforementioned issues, we conducted a prospective, multicenter registry of all first-line treatment strategies in patients with CLI (CRITISCH [Registry of First-Line Treatments in Patients With Critical Limb Ischemia]) (8). All available endovascular techniques at the time of patients’ recruitment were included and evaluated. In the framework of our pre-planned interim analysis, we compared the endovascular approach to bypass surgery as to amputation-free survival in a prospective confirmatory manner. Our initial hypothesis was that an endovascular approach is not inferior to conventional bypass surgery.
The study was performed in accordance with the principles in the Declaration of Helsinki. Ethical approval was obtained from all participating centers before patients were recruited. All patients gave written informed consent. This trial is registered (NCT01877252).
A detailed description of the study design has been previously published (8,9). Briefly, the CRITISCH registry is a prospective, multicenter registry assessing the current practice of all available treatments strategies in all-comers CLI patients treated in 27 vascular centers of excellence (Online Appendix). Two criteria were used for the definition of a center of excellence: 1) implemented CLI program with use of both endovascular and surgical techniques; and/or 2) selection of the treatment strategy after a vascular conference between the responsible disciplines (vascular surgeons/cardiologists/radiologists). At least 1 operator should be board certified (i.e., specialist) for the respective discipline. No further selection criteria for the interventionalist were used.
The inclusion criterion of this study was the presence of new onset CLI. CLI was defined as an ankle-brachial index <0.40 or ischemic rest pain, or both, with or without tissue loss in the presence of peripheral arterial disease (Rutherford classes 4 to 6/Fontaine III to IV) lasting >2 weeks. In case of a noncalculable index, ankle pressure or transcutaneous oxygen pressure was measured, as established at each center. Only 1 limb per patient was included. The modified PREVENT III (Edifoligide for the Prevention of Vein Graft Failure in Lower Extremity Bypass Surgery—Phase III Trial) score was used to address the frailty of the patients, due to its proven predictive role for short- and long-term outcomes after infrainguinal surgical revascularization (10) or endovascular therapy (11). The study excluded patients with acute limb ischemia (embolic or thrombotic), isolated aortoiliac interventions, vascular trauma of the index leg, known clotting disorders, and non-atherosclerotic vascular disease (e.g., vasculitis) (8).
The registry was validated via an audit at participating centers to monitor how accurately all patients and procedures were documented. The recruitment of the target cohort started in January 2013 and was completed in September 2014. The selected treatment strategy was left completely at the discretion of the treating physician following the principle of “best medical practice”; in case of CLI, where there are grey zones of medical practice for which evidence is conflicting or not existing, the case decisions had to rely on health professional experience, judgment, and the choice of patients (12).
There was no limitation concerning the selected treatment option. The different first-line treatment strategies were classified into 5 groups. Group I included patients undergoing all types of endovascular repair. Endovascular repair was considered any percutaneous endovascular attempt for guidewire traversal of lesions. In 6 cases, a lesion could not be crossed. In these cases, the endovascular approach remained the first-line treatment, and any further surgical revascularization was considered a reintervention. Group II consisted of patients undergoing primarily bypass surgery with all type of conduits. We included all types of conduits in order to demonstrate real-world practice as to this issue. Group III included patients undergoing common/deep femoral artery patchplasty, with or without a concomitant distal endovascular intervention. For this reason, these patients were not analyzed together with those of group II. Finally, group IV included patients treated conservatively, whereas group V subjects underwent primary major amputation.
The primary composite endpoint of the CRITISCH registry was amputation-free survival (AFS), that is, time until major amputation of the index limb and/or death from any cause. A major amputation was considered any above-ankle amputation. Secondary endpoints analyzed in this paper were time until death, time until amputation at the index limb, as well as time until major amputation and/or any reintervention at the index limb.
Study design and statistical analysis
A flexible adaptive-sequential study design with 1 interim analysis of the primary endpoint AFS was established. The statistical analysis was pre-specified before the recruitment started and was concretized in a statistical analysis plan before the interim analysis was conducted. Hence, the results of the primary efficacy analysis provide confirmatory evidence. Power calculations delivered a target cohort of 1,200 patients to be assessed for eligibility. The interim analysis was planned to be carried out after 250 events occurred. Because randomization of the treatment strategy was not feasible, we planned to establish a multivariable Cox regression model of AFS including the treatment strategy (endovascular treatment vs. bypass surgery) and further prognostic factors as covariates. The strategy of model building and the possible prognostic covariates had been pre-specified. By creating the multivariate Cox model, we were able to quantify the treatment effect of endovascular treatment versus bypass surgery by the (adjusted) hazard ratio (HR) of AFS. In order to confirm the noninferiority of endovascular treatment versus bypass surgery as to this HR, the 1-sided Wald test was applied. The noninferiority of endovascular treatment versus bypass surgery had been defined by a HR <1.33. In order to maintain a global significance level α = 2.5%, the local level α = 0.58% was pre-specified to be used in the interim analysis. The interim analysis was also carried out pursuing a kind of per-protocol approach. In this per-protocol analysis, 6 patients from group I were removed from analysis because the complete lesion could not be crossed, and therefore, surgical revascularization was applied. Further exploratory analyses were conducted in the course of the interim analysis, using common descriptive (median, quantiles, interquartile range [IQR], absolute and relative frequencies, Kaplan-Meier estimates) and inferential (Mann-Whitney test, Fisher test, chi-square test, log-rank test, Wald test, confidence intervals [CIs]) statistical methods. Statistical computations were performed using R version 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).
Endovascular treatment (group I) was applied to 642 (53%) and bypass surgery (group II) to 284 (24%; p < 0.001) patients. Table 1 summarizes the demographics and comorbidities of the patients, the Rutherford stages, and the anatomic characteristics of the lesions in each group. Table 2 provides an overview of the endovascular devices, as well as the type of bypass conduits used.
Median length of the in-hospital stay was 7 days (interquartile range [IQR]: 3 to 14 days) in group I and 15 days (IQR: 11 to 26 days) in group II (p < 0.001). In group I, 563 patients (88%) were discharged home compared with 216 patients (76%) in group 2 (p < 0.001). In-hospital mortality amounted to 1% (n = 7) in group I and 3% (n = 8) in group II (p = 0.085), whereas the major amputation rate was 3% (n = 20) in group I and 4% (n = 10) in group II (p = 0.841).
The median follow-up was approximately 1 year in both groups (group I 373 days, group II 365 days), and the AFS rate at 1 year was 75% in group I and 72% in group II. The Kaplan-Meier plot (Figure 1) and log-rank test (p = 0.994) did not indicate a difference in AFS between the groups.
The multivariate Cox regression model of AFS (Table 3) considered all variables listed in Table 1 as possible confounders of the treatment effect. After adjustment for several influential confounders, the model delivers a HR of AFS (comparing endovascular treatment vs. bypass surgery) of 0.91 with an estimated 2-sided 95% CI of 0.70 to 1.19. (Note that we did not find any systematic structures in the Schoenfeld residuals of each covariate over time, which is why we conclude that the assumption of proportional hazards is justifiable.)
The 1-sided (1 − α)-CI of the HR was derived from this multivariate Cox model (Wald test) with α = 0.0058 being the local significance level of the interim analysis. Because the p value was p = 0.0029 < α = 0.0058 or, equivalently, because the upper bound of the (1 − 0.0058)-CI was 1.29 and, thus, smaller than the pre-specified noninferiority bound of the HR of 1.33, we were able to confirm significant noninferiority of endovascular treatment versus bypass surgery as to the AFS at a 2.5% level of significance. In a kind of per-protocol analysis, 6 patients of group I were excluded from the full analysis set because the first-line endovascular approach failed and a bypass surgery was applied instead. This per-protocol analysis delivered similar results with an upper bound of the (1 − 0.0058)-CI of 1.30 and a p value of 0.003.
Because AFS is a composite endpoint, we also investigated the influence of the treatment strategy on the freedom from amputation and the overall survival by means of Kaplan-Meier estimation (Figures 2 and 3). At 12 months, freedom from amputation was 90% in group I and 85% in group II, and overall survival amounted to 81% in group I and 84% in group II. The log-rank tests indicated that there might be a difference between the groups as to freedom from amputation (p = 0.077) and overall survival (p = 0.036). However, after adjustment for relevant confounders by multivariate Cox regression analysis (Table 4), these differences could not be observed anymore, neither for time to amputation (HR: 0.86; 95% CI: 0.56 to 1.30; p = 0.463) nor for survival (HR: 1.14; 95% CI: 0.80 to 1.63; p = 0.453). The hazard of AFS is increased by an increase of the modified PREVENT III score (HR: 1.12), by the presence of chronic kidney disease (HR: 1.44) or diabetes mellitus (HR: 1.25), and the absence of statin use during follow-up (HR: 1.24) (Table 4). Considering limb loss, again chronic kidney disease (HR: 1.62), Rutherford stage 6 (HR: 2.27), and previous vascular intervention(s) (HR: 1.95) were identified as risk factors. Regarding overall survival, an increased modified PREVENT III score (HR: 1.15) and chronic kidney disease (HR:1 .58) were linked with an increased hazard.
An influence of the treatment strategy on the time until major amputation and/or reintervention could not be observed by Kaplan-Meier estimations (Figure 4) and log-rank test (p = 0.381). The rate of event-free survival at 12 months was 65% in group I and 62% in group II at 12 months. The numbers of events were 212 (33%) in group I and 98 (35%) in group II. Multivariate Cox models (Table 5) also delivered no indication that the treatment strategy influences the time until major amputation and/or reintervention. However, age (HR: 0.99), no run-off vessels (HR: 1.56), previous vascular intervention (HR: 1.70), and no statin administration during surveillance (HR: 1.33) were identified as risk factors for this endpoint.
In this study, we compared the outcomes between bypass surgery and an endovascular approach in patients with CLI in the framework of the pre-planned interim analysis. Adjustment for several confounders by means of multivariate Cox regressions was necessary, because noticeable differences between the groups were observed (Table 1). However, these differences could highlight a number of selection criteria of the physicians regarding each treatment strategy. Patients receiving endovascular treatment were older and frailer (higher PREVENT III score), they more frequently had chronic kidney disease and usually presented with ischemic lesions of the digits (Rutherford class 5). The angiography revealed a poorer run-off status, despite the moderate angiographic lesions (TASC [Trans-Atlantic Inter-Society Consensus] II A/B). As to this issue, TASC II classification is mainly a treatment suggestion and less an angiographic description of the hemodynamic impact of the lesion. On the other hand, bypass surgery was applied more commonly in patients having already 1 or more previous vascular interventions. This confirms the results of previous studies (13), which identified similar empiric selection criteria between the strategies and proves that decision making is still in accordance to Trans-Atlantic Inter-Society Consensus Document on Management of Peripheral Arterial Disease II guidelines (14); patients undergoing an endovascular approach are skewed to more severe clinical presentation and less severe anatomic lesions. Nonetheless, endovascular repair led to shorter median length of in-hospital stay and higher rates of ambulatory patients post-discharge.
It is interesting that physicians choose to apply more frequently the endovascular therapy, despite use of nephrotoxic contrast agent, as primary approach in patients with chronic kidney disease. Chronic kidney disease was a noticeable contributor to in-hospital mortality in our registry (8,15), and this was also confirmed by Willenberg et al. (16). Once advanced kidney disease is present, it may contribute to the progression of systemic endothelial dysfunction more than traditional cardiovascular risk factors do (17). More interestingly, a recent study showed that statins have no protective effect against cardiovascular morbidity in patients with end-stage renal disease (18). However, little is known as to the effectiveness of different treatment strategies in patients with chronic kidney disease and especially in those with end-stage renal disease, because these patients are excluded in the majority of RCTs. Ortmann et al. (19) showed that patients with CLI will benefit from revascularization compared to medical therapy alone at all stages of renal impairment. However, the best revascularization method in those patients remains unknown und requires further investigation, considering that 44% of the total cohort in our registry suffered from chronic kidney disease (7,15). Our multivariate Cox regression analyses revealed chronic kidney disease as a risk factor for the composite endpoint, limb loss, and overall mortality. Of note, a separate analysis of the impact of dialysis on the primary endpoint was not reasonable due to the small number of dialysis-dependent patients in both groups (Table 1).
Our initial hypothesis, that the endovascular repair was not inferior to bypass surgery could be confirmed in the framework of the planned interim analysis at a 2.5% level of significance. The AFS amounted to 75% in group I and 72% in group II at 1 year. Analyzing separately overall survival and freedom from amputation, both cohorts showed comparable outcomes (81% vs. 84% and 90% vs. 85%, respectively) as well. The BASIL trial showed also comparable effectiveness between angioplasty and bypass surgery although patients with life expectancy >2 years would benefit more from bypass surgery (4,5). However, the study design has been criticized, and the inclusion of only angioplasty cases in the endovascular arm makes the results of this trial less representative of the current practice. Our study indicates that for patients who do not meet the criteria of the RCTs, the bias introduced by allowing skilled physicians to practice medicine and to come to treatment decisions does not compromise patient outcomes.
Several retrospective studies have also compared endovascular therapy and bypass surgery in CLI patients, and a meta-analysis of them (January 1995 to August 2012) showed no difference in amputation-free survival and all-cause mortality (20). Abu Dabrh et al. (21) performed a systematic review of all comparative studies through October 2014 and again suggested that endovascular approaches and bypass surgery may have a similar effect on mortality and major amputation. As expected, a common finding of both meta-analyses was that the quality of evidence was low due to imprecision and heterogeneity between the studies. To our best knowledge, the included studies in both meta-analyses used either plain balloon angioplasty or bare-metal stenting (19,20).
In our study, different technologies of the endovascular armamentarium (Table 2) were assessed. In the superficial femoral artery (SFA), the majority of the patients were treated with bare-metal stents (41%). Despite the fact that few data as to the effectiveness of drug-coated balloons (DCBs) in the SFA in CLI patients existed at the time of recruitment, 18% of the patients in our study received DCB therapy in the SFA. Similar rates of DCB therapy were applied also in the popliteal artery (19%), where again, primary stenting was the leading therapy (33%). Likely, the length or severity of the complex femoropopliteal lesions in CLI patients might lead to subintimal recanalization requiring primary stenting. In this context, we are not able to provide more data about the exact location (p1 to p3 segment) of the stent as well, as its type (e.g., interwoven nitinol stent). Finally, as to the infrapopliteal vessels, the majority of the patients received plain balloon angioplasty (70%), and only 10% of patients underwent DCB angioplasty, probably due to the results of the InPact DEEP trial (Study of IN.PACT Amphirion™ Drug Eluting Balloon vs. Standard PTA for the Treatment of Below the Knee Critical Limb Ischemia) (22). Despite the encouraging 12-month outcomes of the XCELL trial (Study to Evaluate the Safety & Performance of the Xpert(TM) Stent in Treating Below-the-knee Lesions in Patients Undergoing Percutaneous Intervention for Chronic CLI) as to primary infrapopliteal nitinol stenting, 7% of the patients received this type of therapy (23). Important to mention is also that few physicians choose primary stenting with a drug-eluting stent in the SFA (2%) or popliteal artery (0.7%). By contrast, 3.3% of patients received drug-eluting stents in the tibial vessels and 8% of patients in the distal anastomosis after previous bypass. Possible explanations for the low implementation rate of the new technologies, namely, drug-coated devices and atherectomy in the femoropopliteal lesions are mainly the lack of evidence as to the effectiveness of these devices in CLI patients, the missing reimbursement at the time of recruitment, and the high procedural costs, especially in regard to drug-eluting stents and atherectomy (6).
We are not able to report on the exact selection criteria for each patient separately. These can be only extrapolated by comparing the baseline characteristics between the groups. The registry aimed at including consecutive patients based on the bedside clinical examination by the treating physician. The same person assessed and reported the angiographic and ultrasound findings as well as whether an endpoint was met. We chose AFS as primary endpoint to be in accordance with previous studies as to CLI; however, this can be criticized considering that a major adverse limb event, meanwhile, has been described as a superior outcome (24). We are not able to provide a number of details as to the type of previous interventions, the exact endovascular products and their manufacturers, the primary or secondary patency of the treated lesions or implanted bypass grafts, the time to wound healing, and the exact reason of death. In our study, we could not assess whether the quality or availability of the vein graft conduit was the reason to choose a prosthetic bypass graft or an endovascular approach as the first-line treatment strategy (25). Because randomization was not feasible, a multivariate Cox regression model was established in order to prevent biased results. Concrete instructions of model building were defined before the interim analysis was performed. Hence, we were able to show significant noninferiority of endovascular-first approach in comparison to bypass surgery as to the HR of AFS in a confirmatory manner. In this context, we performed a regression-adjusted analysis and not a matched pair analysis, due to the arguments given by Klein et al. (26). In particular, we wanted to use all the patient information, including also those patients for whom we could not find a match. Any comparison to the outcomes of the other first-line treatments (groups III to V) was not reasonable due to the high heterogeneity of the cohorts and the small sizes of the other groups, as reported elsewhere (8). Finally, a cost analysis was not done as part of this registry.
Regarding the primary endpoint, the interim analysis of the CRITISCH registry confirmed significant noninferiority of the current endovascular approach in comparison to bypass surgery as first-line treatment strategy. This study highlights that when physicians are free to individualize therapy for their CLI patients, they achieved encouraging outcomes with both therapies. Despite the fact that sicker and higher risk patients were included in the endovascular group, a low early mortality rate as well as a statistically noticeable shorter in-hospital stay compared with bypass surgery were observed. Because chronic kidney disease was the most striking risk factor for limb loss and death, this cohort requires further investigation as to the most effective treatment.
WHAT IS KNOWN? The gold standard of treatment between the endovascular-first and the bypass-first approaches in patients with CLI remains under debate with only 1 randomized controlled trial published more than a decade ago. Decision making is based mostly on local expertise, and little is known about the outcomes of physician-directed CLI therapy.
WHAT IS NEW? Our study shows in a confirmatory manner that for patients who do not meet the criteria of the randomized controlled trials, the bias introduced by allowing skilled physicians to individualize their therapy and make treatment decisions does not compromise patient outcomes for either endovascular therapy and bypass surgery.
WHAT IS NEXT? More evidence is needed to improve our knowledge about the treatment of choice in CLI patients with renal impairment, considering that this was the most striking risk factor for both amputation and death.
For a list of the CRITISCH collaborators, please see the online version of this article.
The German Society of Vascular Surgery and Medicine (Deutsche Gesellschaft für Gefäßchirurgie und Gefäßmedizin) sponsored this study. The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Bisdas and Borowski contributed equally to this work.
- Abbreviations and Acronyms
- amputation-free survival
- confidence interval
- critical limb ischemia
- drug-coated balloon
- hazard ratio
- randomized controlled trial
- superficial femoral artery
- Trans-Atlantic Inter-Society Consensus
- Received August 25, 2016.
- Revision received September 21, 2016.
- Accepted September 25, 2016.
- American College of Cardiology Foundation
- Twine C.P.,
- McLain A.D.
- Johnson W.C.,
- Lee K.K.
- Bradbury A.W.,
- Adam D.J.,
- Bell J.,
- et al.
- Katsanos K.,
- Geisler B.P.,
- Garner A.M.,
- et al.
- Teraa M.,
- Conte M.S.,
- Moll F.L.,
- et al.
- Bisdas T.,
- Stachmann A.,
- Weiss K.,
- et al.
- Beropoulis E.,
- Stavroulakis K.,
- Schwindt A.,
- et al.
- Soga Y.,
- Mii S.,
- Iida O.,
- et al.
- Meyer A.,
- Lang W.,
- Borowski M.,
- et al.
- Chong K.C.,
- Owens C.D.,
- Park M.,
- et al.
- Jones W.S.,
- Dolor R.J.,
- Hasselblad V.,
- et al.
- Zeller T.,
- Baumgartner I.,
- Scheinert D.,
- et al.
- Bisdas T.,
- Torsello G.,
- Stachmann A.,
- et al.