Author + information
- Received August 1, 2016
- Revision received October 19, 2016
- Accepted October 19, 2016
- Published online October 31, 2016.
- S1936879816318416-888fc05ca31486ee93e65d994e1cad19Dean J. Kereiakes, MDa,∗ (, )
- S1936879816318416-3c931f0911a0c8cc9e4937169bfd1018Stephen G. Ellis, MDb,
- S1936879816318416-a10d7ef4220f0fc5fa512e3bdbe92444Takeshi Kimura, MDc,
- S1936879816318416-81ec716c3e2994c159635463dd33eef7Alexandre Abizaid, MD, PhDd,
- S1936879816318416-91fafb296b4c42eb3d01ebeb1a133de5Weiying Zhao, MD, PhDe,
- S1936879816318416-76960b5d4e1cde9220b1c86b0dce9387Susan Veldhof, RNe,
- S1936879816318416-d526d900ea475bbe642044b777921e9bMinh-Thien Vu, MSe,
- S1936879816318416-2fa25687a4ee18040baf2a2bf54214a7Zhen Zhang, PhDe,
- S1936879816318416-aea386ca6cabffb484999ae861402b24Yoshinobu Onuma, MD, PhDf,
- S1936879816318416-43ad1c7497faaaee24800893e14b0f0eBernard Chevalier, MDg,
- S1936879816318416-b8ad8cc3941b6a237c111f8185e844eaPatrick W. Serruys, MD, PhDh and
- S1936879816318416-92a59b709d30b344c0d1a5ba5fbb027cGregg W. Stone, MDi,j
- aThe Christ Hospital Heart and Vascular Center, The Lindner Research Center, Cincinnati, Ohio
- bCleveland Clinic, Cleveland, Ohio
- cDepartment of Cardiovascular Medicine, Kyoto University Graduate School of Medicine, Kyoto, Japan
- dInstitute Dante Pazzanese of Cardiology, São Paulo, Brazil
- eAbbott Vascular, Santa Clara, California
- fThorax Centre, Erasmus MC, Rotterdam, the Netherlands
- gInstitut Jacques Cartier, Massy, France
- hInternational Centre for Cardiovascular Health, Imperial College, London, United Kingdom
- iNew York Presbyterian Hospital, Columbia University Medical Center, New York, New York
- jCardiovascular Research Foundation, New York, New York
- ↵∗Reprint requests and correspondence:
Dr. Dean J. Kereiakes, The Christ Hospital Heart and Vascular Center/The Lindner Research Center, 2123 Auburn Avenue, Suite 424, Cincinnati, Ohio 45219.
Objectives The study sought to evaluate the efficacy and safety of the Absorb everolimus-eluting bioresorbable vascular scaffold (BVS) (Abbott Vascular, Abbott Park, Illinois) in patients with diabetes mellitus.
Background Randomized, controlled trials have demonstrated comparable clinical outcomes following percutaneous coronary intervention with either Absorb BVS or metallic Xience everolimus-eluting stent. However, these trials lack power required to provide reliable treatment effect estimates in this high-risk population.
Methods In a pre-specified, powered analysis, patients with diabetes who received ≥1 Absorb were pooled from the ABSORB II, III, and JAPAN randomized trials and from the single arm ABSORB EXTEND registry. The study composite primary endpoint was target lesion failure (TLF) at 1 year following Absorb BVS compared with a performance goal of 12.7%.
Results Among 754 diabetic patients included in analysis (27.3% insulin treated), the 1-year TLF rate was 8.3% (upper 1-sided 95% confidence limit: 10.1%; p = 0.0001 vs. performance goal). Scaffold thrombosis (definite or probable) was observed in 2.3% of patients. Multivariable regression identified older age, insulin treatment, and smaller pre-procedure reference vessel diameter as significant independent predictors of 1-year TLF.
Conclusions The Absorb diabetic substudy suggests efficacy and safety of the Absorb BVS for treatment of patients with diabetes mellitus.
The presence of diabetes mellitus remains a significant predictor of adverse clinical and angiographic outcomes following percutaneous coronary intervention (PCI) with contemporary drug-eluting stents (DES), with increased rates of myocardial infarction (MI), stent thrombosis, restenosis, and death (1–4). This poor prognosis in patients with diabetes has been ascribed to a greater level of vascular inflammation, the presence of a prothrombotic state and more complex clinical and angiographic features (5,6).
Among patients with diabetes undergoing PCI, both the severity of diabetes as reflected by the treatment required (insulin-providing medications vs. insulin-sensitizing medications) (7,8) as well as the level of glucose control (as reflected by glycated hemoglobin or fasting blood glucose levels) (9,10) have been correlated with periprocedural and late clinical outcomes. DES reduce angiographic as well as clinical restenosis (ischemia-driven target lesion and vessel revascularization) following PCI when compared with either bare-metal stents or balloon angioplasty in patients with or without diabetes (11,12). Although iterations in metallic DES including novel alloy composition, reduced strut thickness, and improved polymer biocompatibility or bioresorption have further improved outcomes compared with early generation DES (13), concerns regarding incomplete endothelialization, polymer hypersensitivity, neoatherosclerosis, and stent fracture persist (14–16). Indeed, beyond 1 year after implant, current metallic DES are associated with a 2% to 4% ongoing annual incidence of target lesion failure (TLF) events (composite occurrence of cardiac death, target vessel MI [TV-MI], and ischemia-driven target lesion revascularization), rates similar to that observed following either bare-metal stents or early generation DES (17,18). The occurrence of this phenomenon with all stents may be due to the presence of a metallic implant that mechanically distorts and constrains the vessel, thus preventing normalization of vasomotion, autoregulation, and adaptive remodeling (17,18).
Fully bioresorbable scaffolds provide mechanical support and drug-delivery functions similar to metallic DES early (within 6 to 12 months) following PCI, followed by complete resorption with recovery of more normal vascular structure and function, with the consequent potential for improving very late clinical outcomes (19). Recent randomized controlled clinical trials have demonstrated comparable 1-year clinical outcomes following PCI with the Absorb bioresorbable vascular scaffold (BVS) (Abbott Vascular, Abbott Park, Illinois) compared to the metallic Xience everolimus-eluting stent (EES) (Abbott Vascular) in patients with noncomplex, stable ischemic heart disease or stabilized acute coronary syndromes, and long-term follow-up is ongoing (20–22). However, subgroup analyses of patients with diabetes mellitus from these trials lack power required to provide reliable treatment effect estimates in this high-risk population. Thus, a pre-specified formal substudy was performed to evaluate the 1-year safety and effectiveness of Absorb BVS in patients with diabetes mellitus.
Design and population
The present study represents a pre-specified, powered analysis designed in concert with the U.S. Food and Drug Administration to support a U.S. diabetic indication for Absorb. The study cohort includes subjects with diabetes mellitus who were enrolled into the ABSORB II, ABSORB III, and ABSORB Japan randomized trials (20–22) plus the single arm, open-label ABSORB EXTEND registry (23). The design of each study has been described previously (20–23). Each trial included in this pooled analysis was conducted in accordance with the clinical investigational plan, the declaration of Helsinki, and applicable regulatory requirements. Institutional review boards or medical ethics committee approval for the protocols and informed consents were obtained prior to site and subject participation. Clinical endpoints were adjudicated by an independent, central clinical events committee, and study oversight was provided by an independent data safety monitoring board for each study. A summary of key study design characteristics as well as the number of subjects with diabetes stratified by diabetic treatment are shown in Online Table 1.
All subjects included in the analysis cohort had Absorb BVS implanted in at least 1 target lesion (as-treated population). For conformity of target lesion lengths across studies, subjects with lesion lengths >24 mm in ABSORB EXTEND and ABSORB II were excluded.
The powered primary endpoint for analysis is the incidence of TLF at 1 year in the Absorb BVS diabetic cohort. All endpoints in this analysis were defined the same as in the ABSORB III trial (22).
Patient level data from the 4 ABSORB studies were pooled into a common database. The powered primary endpoint of 1-year TLF rate of the pooled Absorb BVS diabetic cohort was tested against a pre-specified performance goal (PG). The analysis assumed the true 1-year TLF rate in the Absorb BVS diabetic cohort was 8.2%, which was derived from the historical Xience diabetic data from the SPIRIT IV trial (24) (Online Appendix). The PG of 12.7% includes the 8.2% TLF estimate plus a 4.5% noninferiority margin based on the putative placebo concept to preserve ≥50% of the treatment benefit for Xience versus bare-metal stents (25). Assuming a 1-sided alpha of 0.05 and 5% loss to follow-up at 1 year, we estimated a total of approximately 700 Absorb BVS treated patients with diabetes mellitus would provide >95% power.
Patients who were lost to follow-up in whom no known event had occurred were not included in the denominator for calculations of binary endpoints. Binomial Exact test was used to compare 1-year primary endpoint of TLF against the PG. Chi-square test or Fisher’s exact test (when Cochran’s rule is not met) was used for between group comparisons of endpoint events. Poolability across the 4 ABSORB studies was examined via chi-square test for the primary endpoint of TLF. In addition, a sensitivity analysis was also performed using both fixed and random effect meta-analysis for the primary endpoint. A multivariable Cox regression analysis of the primary endpoint of 1-year TLF was performed in the pooled Absorb BVS diabetic patients. Variables included in the model include age (5-year increment), gender (female vs. male), target vessel left anterior descending artery (yes vs. no), pre-procedure reference vessel diameter (RVD) (0.5 mm increment), lesion length (5 mm increment), insulin use (yes vs. no), lesion type (B2/C vs. A/B1), number of lesions treated (>2 vs. 1), and study (Absorb III vs. non-Absorb III patients). The graphical and numerical methods of Lin et al. (26) were used to assess the proportional hazards assumption. We used meta (version 4.3-2) in R version 3.2 (R Development Core Team, Vienna, Austria) to do the meta-analysis (27). All other statistical analyses were performed with the use of SAS software, version 9.2 (SAS Institute, Cary, North Carolina).
Patients and baseline characteristics
The analysis population was comprised of 754 patients with diabetes mellitus who were treated with at least 1 Absorb BVS in at least 1 target lesion. Baseline clinical, angiographic lesion characteristics and procedural data among these patients are shown in Table 1. At enrollment, 27.3% of subjects received insulin treatment and nearly 60% had glycated hemoglobin levels ≥7.0%. As expected from a global population in the pooled analysis, some geographic differences in the baseline patient demographic and risk factors were noted. In particular, the ABSORB III randomized controlled trial diabetic population had a higher risk profile compared with other trials. Of note, 18% of all treated lesions in this analysis had a baseline RVD of <2.25 mm by quantitative coronary angiography (QCA), and ∼60% had American Heart Association/American College of Cardiology type B2/C target lesion morphology (moderate to severe complexity). More than 70% of all scaffolds were post-dilated, and ∼7% of lesions were treated with overlapping devices. Adherence to dual antiplatelet therapy is shown in Online Table 2.
Outcomes at 1 year
The primary endpoint of TLF at 1-year occurred in 8.3% of diabetic patients treated with Absorb BVS (Figure 1), with an upper 1-sided 95% confidence limit (CL) (exact method) of 10.1%, well below the pre-specified PG of 12.7% (p for noninferiority = 0.0001). One-year TLF ranged from 4.4% to 10.9% by study (chi-square test for poolability p = 0.08). Sensitivity analyses using both fixed effect (Absorb 1-year TLF: 8.7%, upper 1-sided 95% CL: 10.6%; p for noninferiority = 0.0008) and random effect (Absorb 1-year TLF: 7.1%, upper 1-sided 95% CL: 10.5%; p for noninferiority = 0.006) meta-analysis models confirmed that 1-year TLF following Absorb BVS was significantly below the PG. Individual efficacy and safety outcomes to 1-year by trial and for the pooled diabetic cohort are shown in Table 2. Most outcomes including TLF, all MI, TV-MI, ischemia-driven target lesion and vessel revascularization, and scaffold thrombosis were significantly increased among diabetic patients who were receiving insulin treatment compared with those who were not (Table 2).
Multivariable Cox regression analysis identified older age, insulin treatment, and smaller pre-procedure RVD as significant independent predictors of 1-year TLF among subjects with diabetes mellitus (Figure 2) where the proportional hazards assumption was met for all variables included in the analysis. Clinical outcomes stratified by baseline QCA RVD <2.25 mm versus ≥2.25 mm demonstrates that adverse events were less frequent among diabetic subjects with RVD ≥2.25 mm (Figure 3).
Although comparisons of outcomes between Absorb BVS and Xience treated patients with diabetes are limited by lack of randomization as well as the absence of a Xience treatment arm in the ABSORB EXTEND study, both 1-year TLF and device thrombosis rates appear similar among patients appropriate for trial enrollment (baseline QCA RVD ≥2.25 mm) by device type (1-year TLF 6.6% [n = 40 of 606] vs. 6.5% [n = 17 of 261]; device thrombosis 1.3% [n = 8 of 603] vs. 0.4% [n = 1 of 259] for Absorb BVS vs. Xience, respectively).
The present pre-specified, prospective, pooled analysis is the largest outcome study of patients with diabetes mellitus treated with the Absorb BVS to date, and thus provides valuable insights into the efficacy and safety of this device in an important, increasingly prevalent high-risk subgroup. The major observations of this analysis include: 1) the powered primary endpoint of 1-year TLF following Absorb BVS in patients with diabetes was 8.3%, similar to the pre-specified TLF estimate of 8.2% and significantly less than the PG of 12.7%. In addition, achievement for the primary endpoint was confirmed in sensitivity analysis using formal meta-analysis. 2) Among patients with diabetes, the rates of TLF, and the TLF components of TV-MI and ischemia-driven target lesion revascularization were significantly increased among diabetic patients treated with insulin compared with those who were not. A similar observation was made for scaffold thrombosis. 3) Multivariable regression analysis identified older age, smaller target vessel RVD by QCA, and insulin treatment as independent predictors of 1-year TLF.
This study was designed to support label expansion of Absorb in the United States, and in this regard demonstrates efficacy and safety of Absorb BVS for the treatment of noncomplex stable ischemic heart disease and stabilized acute coronary syndromes in patients with diabetes. Our study also provides important insights as to which diabetic patients will have a more or less favorable 1-year prognosis after Absorb. As shown by the multivariable regression analysis, rates of 1-year TLF would be predictably lower in patients with diabetes who are younger, non–insulin treated and with larger baseline RVD. The higher TLF rate observed in the U.S. ABSORB III trial (10.9%) was likely due to more complex patients included in this trial. After adjusting for the other patient and lesion risk factors, ABSORB III was not an independent predictor of 1-year TLF in this pooled diabetic analysis.
The overall 2.3% 1-year rate of scaffold thrombosis observed in the present study is not surprising as both diabetes and small vessel size are well established risk factors for stent thrombosis (22,24,28), and around one-fifth of the diabetic patients had very small target vessels (QCA RVD <2.25 mm, roughly correlating to a visually estimated RVD of <2.5 mm). As in ABSORB III (19,29), baseline QCA RVD <2.25 mm was a powerful correlate of adverse outcomes, particularly TV-MI and scaffold thrombosis in the present pooled diabetic population. For diabetic patients with appropriately sized vessels (QCA RVD ≥2.25 mm), the scaffold thrombosis rate was lower (1.3%). Recent clinical experience suggests that an Absorb BVS specific deployment strategy that includes optimal target lesion preparation, fastidious scaffold to vessel sizing, and high-pressure post-dilation with appropriately sized (≥1:1 but <0.5 mm larger than scaffold) noncompliant balloons is effective in reducing BVS scaffold thrombosis (30). Interestingly, the incidence of post-dilation by trial in the present analysis ranged from 55.8% (ABSORB II) to 84.0% (ABSORB JAPAN), and was not clearly related to 1-year scaffold thrombosis rates of Absorb treated diabetic patients in these 2 trials (1.5% vs. 2.1%, respectively). This apparent lack of correlation likely reflects play of chance due to the low frequency occurrence for scaffold thrombosis as well as the limited number of patients with diabetes mellitus contributed by each of the individual trials.
In the larger portion of patients with diabetes mellitus who did not require insulin treatment (n = 548), the 1-year rate of scaffold thrombosis was 1.5%, similar to both the 1.5% rate observed for the overall Absorb patients enrolled into the ABSORB III trial (n = 1,322) (22) as well as the 1.4% rate observed in all Xience-treated patients with diabetes (n = 224) in the ABSORB III trial (31). These results are also consistent with a prior propensity-matched comparison of patients with diabetes mellitus treated with either the Absorb BVS or the Xience EES, which noted comparable rates of 1-year TLF and thrombosis between devices (32). It is noteworthy that in the present analysis, the smaller portion of patients who were insulin treated (27.3%) accounted for over 50% of the scaffold thrombosis events that occurred.
First, despite being the largest analysis of patients with diabetes treated with Absorb BVS to date, this study remains underpowered to precisely evaluate low frequency events such as scaffold thrombosis. Second, clinical outcomes and follow-up are limited to 1-year post PCI, a time frame when Absorb BVS resorption is incomplete. Third, the lack of randomized assignment of patients with diabetes to treatment with either Absorb BVS or EES precludes direct comparison of outcomes between the devices. Nevertheless, the powered primary endpoint of this study is not dependent on either a randomized (to EES) comparator group of patients with diabetes or comparison of device treatment by diabetic status. The study primary endpoint of 1-year TLF in Absorb BVS-treated patients with diabetes compared with a pre-specified PG was met with a high level of statistical significance, which was confirmed in sensitivity analysis. Furthermore, consistency of Absorb BVS treatment (compared with EES) was previously demonstrated in the large-scale ABSORB III trial regardless of diabetic status (22). Finally, it should be noted that for most investigators these studies reflect the first-time clinical use of Absorb BVS (compared with an extensive history with Xience). As a first experience with a novel device, the results in a diabetic cohort are encouraging, and one would expect that as with all new medical procedures, results will improve over time with increased operator experience.
In conclusion, this study suggests efficacy and safety of Absorb BVS in patients with diabetes mellitus particularly those with baseline RVD ≥2.25 mm. Although this work represents the largest clinical outcomes analysis to date of diabetic patients treated with Absorb BVS, larger-scale direct comparative trials of Absorb versus Xience with long-term follow-up are required to better define the relative outcomes between these devices in patients with diabetes mellitus.
WHAT IS KNOWN? Although patients with diabetes mellitus have worse clinical outcomes following percutaneous coronary revascularization, outcomes following BVS deployment in this high-risk population are not defined.
WHAT IS NEW? A prospective prespecified, powered analysis of 1-year TLF following Absorb BVS in patients with diabetes suggests efficacy and safety of this device with an observed TLF rate of 8.3% compared to a pre-specified PG of 12.7% (p for noninferiority = 0.0001).
WHAT IS NEXT? This study supports diabetic label expansion for Absorb BVS. Larger scale direct comparative trials (Absorb BVS vs. Xience) with long-term follow-up are required to better define the relative outcomes between these devices in patients with diabetes mellitus.
The authors thank all the clinical sites that participated in the ABSORB EXTEND, ABSORB II, ABSORB III, and ABSORB Japan trials. They also thank Jennifer M. Jones, Wai-Fung Cheong, and Stan Fink for their intellectual input in writing this article.
For expanded Methods and references sections as well as supplemental tables, please see the online version of this article.
The ABSORB EXTEND, ABSORB II, ABSORB III, and ABSORB Japan trials have been sponsored and funded by Abbott Vascular. Dr. Kereiakes has receieved research funding from Abbott Vascular and Boston Scientific; and has served as a consultant for Abbott Vascular, Boston Scientific, Svelte Medical Systems, Micell Technologies, and Sino Medical Sciences Technology. Drs. Ellis and Chevalier have served as consultants for Abbott Vascular. Dr. Kimura has served on the advisory board of and received research grant support from Abbott Vascular. Dr. Abizaid has received research grant support from Abbott Vascular. Dr. Zhao, Susan Veldhof, Minh-Thien Vu, and Dr. Zhang are full time employees of Abbott Vascular. Dr. Onuma has served on the advisory board of Abbott Vascular. Dr. Stone has served as consultant for Reva Corp. Dr. Serruys has reported that he has no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- bioresorbable vascular scaffold
- confidence limit
- drug-eluting stent(s)
- everolimus-eluting stent
- myocardial infarction
- percutaneous coronary intervention
- performance goal
- quantitative coronary angiography
- reference vessel diameter
- target lesion failure
- target vessel myocardial infarction
- Received August 1, 2016.
- Revision received October 19, 2016.
- Accepted October 19, 2016.
- American College of Cardiology Foundation
- Ritsinger V.,
- Saleh N.,
- Lagerqvist B.,
- Norhammar A.
- Singh M.,
- Gersh B.J.,
- McClelland R.L.,
- et al.
- Stone G.W.,
- Kedhi E.,
- Kereiakes D.J.,
- et al.
- Van Belle E.,
- Périé M.,
- Braune D.,
- et al.
- Moreno P.R.,
- Murcia A.M.,
- Palacios I.F.,
- et al.
- Corpus R.A.,
- George P.B.,
- House J.A.,
- et al.
- Hillegass W.B.,
- Patel M.R.,
- Klein L.W.,
- et al.
- Kang S.,
- Chae I.,
- Park J.,
- et al.
- Finn A.V.,
- Joner M.,
- Nakazawa G.,
- et al.
- Kan J.,
- Ge Z.,
- Zhang J.,
- et al.
- Otsuka F.,
- Vorpahl M.,
- Nakano M.,
- et al.
- Iqbal J.,
- Serruys P.W.,
- Silber S.,
- et al.
- Yamaji K.,
- Kimura T.,
- Morimoto T.,
- et al.
- Kereiakes D.J.,
- Onuma Y.,
- Serruys P.W.,
- Stone G.W.
- Serruys P.W.,
- Chevalier B.,
- Dudek D.,
- et al.
- Kimura T.,
- Kozuma K.,
- Tanabe K.,
- et al.
- Kereiakes D.J.,
- Cutlip D.E.,
- Applegate R.J.,
- et al.
- ↵Food and Drug Administration. Guidelines for Industry: Non-Inferiority Clinical Trials. March 2010. Available at: http://www.fda.gov/ucm/groups/fdagov-public/@fdagov-drugs-gen/documents/document/ucm202140.pdf. Accessed September 20, 2016.
- Lin D.,
- Wei L.J.,
- Ying Z.
- Schwarzer G.,
- Carpenter J.R.,
- Rücker G.
- Puricel S.,
- Cuculi F.,
- Weissner M.,
- et al.
- Absorb G.T.
- Muramatsu T.,
- Onuma Y.,
- van Geuns R.J.,
- et al.