Article ID: CJ-19-1184
Background: Bioresorbable vascular scaffolds (BVS) are promising alternatives to metallic drug-eluting stents (DES) in percutaneous coronary interventions. Absorb BVS was comparable to XIENCE (DES) for patient- and device-oriented composite endpoints through 1 year post-procedure. Mid-term results showed increased rates of device-oriented events with Absorb. The objective of this study was to evaluate the long-term safety and effectiveness of Absorb BVS compared with XIENCE metallic DES when implanted in patients in Japan with de novo coronary artery lesions.
Methods and Results: ABSORB Japan randomized 400 patients into either Absorb (n=266) or XIENCE (n=134) treatment arm. Through 5-year follow-up, the composite endpoints of DMR (death, myocardial infarction [MI], and all revascularization), target vessel failure (TVF), major adverse cardiac events (MACE), target lesion failure (TLF), and cardiac death/all MI were evaluated. Individual endpoints included death, MI, coronary revascularization, and scaffold/stent thrombosis. There were no significant differences in the composite or individual endpoint outcomes between the Absorb and XIENCE arms through 5 years or between 3 and 5 years. Numerically lower TVF, MACE, and all MI rates were observed for the Absorb vs. XIENCE arm after 3 years. No scaffold/stent thrombosis was reported beyond 3 years. Post-procedure imaging subgroups showed comparable event rates.
Conclusions: Following resorption of the scaffold, between 3 and 5 years post-procedure, the Absorb BVS performed comparably to XIENCE in all patient- and device-oriented endpoints (ClinicalTrials.gov, #NCT01844284).
Bioresorbable vascular scaffolds (BVS) were developed with the aim of preventing the adverse events associated with the use of metallic drug-eluting stents (DES), such as very late stent thrombosis (ST), late restenosis, and neoatherosclerosis.1,2 Progressive resorption of the scaffold may help the vessel wall recover its normal physiological pulsatile strength, thereby reducing the long-term risk of adverse events.3,4 The Absorb BVS resorbs completely within 3 years of implantation,5,6 and at 5 years, there is evidence of restored vasomotion, endotheliummediated vasodilation, vessel remodeling, plaque regression, and the formation of stable neointima.7,8
Short-term randomized controlled trials (RCTs) up to 1 year have demonstrated noninferiority of Absorb BVS compared with XIENCE cobalt-chromium everolimus-eluting stent (CoCr-EES), with comparable rates of patient-oriented (all-cause death, all myocardial infarction [MI], or all revascularization) and device-oriented (target lesion failure [TLF]) composite endpoints.9 However, mid-term results showed increased rates of composite device-oriented adverse events and device thrombosis at 2 years,10,11 and 3 years12,13 of follow-up compared with EES. Scaffold undersizing was determined to be an independent predictor of very late scaffold thrombosis with BVS.14 Long-term follow-up to 4 years did not show improvement in TLF or device thrombosis rates for BVS between 3 and 4 years, and were likely related to scaffold sizing and suboptimal implantation technique.15 Currently, there is a lack of adequate evidence from RCTs to support the safety and performance of BVS vs. metallic stent beyond 3 years, and post-resorption of scaffold.
Herein, we report the 5-year follow-up results from the ABSORB Japan RCT to evaluate the long-term clinical outcomes of Absorb BVS in comparison with CoCr-EES following implantation in patients with ischemic heart disease (IHD).
ABSORB Japan was a prospective, multicenter, randomized, single-blind, active-controlled study with 400 patients randomized in a 2:1 ratio to treatment with BVS (Abbott Vascular, Santa Clara, CA, USA; n=266) or XIENCE (CoCr-EES, XIENCE PRIME/Xpedition; Abbott Vascular; n=134).
The purpose of the study was to evaluate the safety and effectiveness of the Absorb BVS in the treatment of patients with IHD caused by de novo native coronary artery lesions in comparison with XIENCE, an approved commercial metallic DES for clinical use in Japan.
This study was conducted in compliance with the ethical principles of the Declaration of Helsinki. The Absorb BVS was approved by the Ministry of Health, Labor and Welfare (MHLW) in Japan on November 2, 2016 and was also registered in ClinicalTrials.gov, #NCT01844284. Accordingly, on November 9, 2016, ABSORB Japan was reclassified as a post-marketing clinical trial after marketing approval. The study was then conducted per the standards for post-marketing surveillance and studies as required by the MHLW.
Enrollment occurred at 38 investigational sites after institutional review board approval. Patients with IHD, indicated for elective percutaneous coronary intervention (PCI) were included in the study. A complete list of the inclusion and exclusion criteria have been described previously.16
Patient Enrollment Patients were randomized into either the Absorb BVS arm or the XIENCE arm (2:1 ratio) and stratified by diabetic status and number of lesions to be treated. Patients and site personnel conducting follow-up clinical visits were blinded to treatment assignment.
Absorb BVS Implantation Procedure Treatment of up to 2 de novo lesions (single lesion/vessel or dual lesions/vessel), each located in different coronary arteries (right coronary artery, left anterior descending artery, or left circumflex artery) was allowed. Successful predilatation of the target lesion was mandatory. The stent sizes of Absorb BVS and XIENCE included: 2.5, 3.0, and 3.5 mm diameter, and 8,12, 18, and 28 mm length. Post-dilatation of lesion after scaffold/stent implantation was performed per operator’s discretion. All patients were maintained on thienopyridine for 12 months and on aspirin for an indefinite period.
Follow-up Period Patient enrollment for the study occurred from April 27, 2013 to December 27, 2013. Patients were evaluated post-procedure through 5 years at the following time points: post-procedure/in-hospital; 1, 6, 12, and 13 months; 2, 3, 4, and 5 years.
Clinical Endpoints at 5 Years Clinical component endpoints including death (cardiac/vascular/noncardiovascular), MI, target lesion revascularization (TLR), target vessel revascularization (TVR), all coronary revascularization, and scaffold/ST were based on Academic Research Consortium definitions.17 Periprocedural non-Q-wave MI was defined as post-procedural creatine kinase (CK)-MB >5-fold the upper limit of normal. Composite endpoints at 5 years included DMR (defined as a composite of all-cause death, all MI, and all revascularization), target vessel failure (TVF), major adverse cardiac events (MACE), TLF, cardiac death/all MI, and all death/all MI. A complete list of endpoints is provided elsewhere.16
Clinical Assessments An independent clinical events committee (CEC, The Baim Institute for Clinical Research [formerly Harvard Clinical Research Institute], Boston, MA, USA) was blinded while adjudicating clinical endpoints of all-cause death, MI, TLR/TVR, and ST. If ST was suspected, an angiographic core laboratory assessed the presence of thrombus. An independent safety oversight committee (The Baim Institute for Clinical Research) was blinded to periodically review the safety data throughout the study.
Statistical AnalysisThe clinical endpoint rates were evaluated in the intent-to-treat (ITT) population through 1,853 days to include events occurring through the end of the 5-year (+28 days) follow-up window. Binary variables were calculated as counts, percentages, and 95% confidence intervals. Pearson’s chi-squared test or Fisher’s exact test was used when appropriate. Continuous variables were summarized as mean±standard deviation or as median and interquartile range, wherever appropriate. Kaplan-Meier survival curves were constructed for time-to-event variables and were compared by the log-rank test. Statistical analyses were performed using SAS versions 9.2 and 9.3 (SAS Institute, Inc., Cary, NC, USA).
Role of Funding SourceThe sponsor was involved in study design, data collection, data analysis, data interpretation and writing of this report. The corresponding author had full access to the analyzed data in the study and accepts full responsibility for the integrity of the study and the decision to submit for publication.
The disposition of 400 randomized patients through the 5-year follow-up of the ITT population is shown in Figure 1. The ITT population included all subjects registered, regardless of whether they received the investigational device or not and included the baseline analysis. There were 12 patients in the Absorb arm, and 7 patients in the XIENCE arm who were lost to follow-up without DMR events, so they were excluded from the 5-year analysis.
Patient disposition flowchart through 5 years (ITT population). Numbers in parentheses are the analysis population. Numbers in each cell are the analysis population at each time point for clinical events. Patients who died or were withdrawn after DMR are still included in the analysis population. Only patients withdrawn without DMR are excluded from analysis. The “Other” reason for withdrawal at the 5-year follow up was refusal to sign the informed consent form. In-hospital: Day 7; 30 Day: Day 22; 6M: Day 151; 12M: Day 336; 2Y: Day 701; 3Y: Day 1,066; 4Y: Day 1,431; 5Y: Day 1,796 post procedure. DMR, composite of all-cause death, all myocardial infarction, and all revascularization; F/U, follow up; IFU, instructions for use; ITT, intent-to-treat; M, months; M/R, myocardial infarction or revascularization; w/o, without; Y, years.
The patients’ mean age was approximately 67 years, 88% had stable coronary artery disease, and approximately 36% were diabetic. No significant differences were found in patient demographics or baseline risk factors between the study arms (Table 1). Procedural results are described in Kimura et al.16
| % (n/N) | ABSORB (n=266) |
XIENCE (n=134) |
|---|---|---|
| Age,a years | 67.1±9.4 (266) | 67.3±9.6 (134) |
| Male | 78.9% (210/266) | 73.9% (99/134) |
| Body mass index,a kg/m2 | 24.01±3.03 (260) | 24.27±2.96 (130) |
| Current tobacco use | 19.9% (53/266) | 21.6% (29/134) |
| Hypertension | 78.2% (208/266) | 79.9% (107/134) |
| Requiring medication | 72.2% (192/266) | 73.1% (98/134) |
| Dyslipidemia | 82.0% (218/266) | 82.1% (110/134) |
| Requiring medication | 74.1% (197/266) | 79.1% (106/134) |
| Diabetes mellitus | 36.1% (96/266) | 35.8% (48/134) |
| Requiring medication | 31.2% (83/266) | 29.9% (40/134) |
| On insulin | 9.0% (24/266) | 8.2% (11/134) |
| HbA1c,a % | 6.23±1.06 (265) | 6.15±0.78 (133) |
| Prior coronary interventions | 36.1% (96/266) | 38.1% (51/134) |
| On target vessel | 3.4% (9/266) | 5.2% (7/134) |
| Prior MI | 16.0% (42/262) | 23.9% (32/134) |
| Family history of premature CAD | 6.5% (16/246) | 8.1% (10/124) |
| Current evidence of ischemia | ||
| Stable angina | 63.9% (170/266) | 65.7% (88/134) |
| Unstable angina | 9.8% (26/266) | 16.4% (22/134) |
| Silent ischemia | 26.3% (70/266) | 17.9% (24/134) |
aMean±SD. CAD, coronary artery disease; HbA1c, hemoglobin A1c; MI, myocardial infarction; N, total number of subjects.
Patients were required to take continuous dual antiplatelet therapy (DAPT) of thienopyridine (clopidogrel, ticlopidine, or prasugrel) and aspirin for at least 12 months post-procedure. At the initiation of the study, the majority (>93%) of the patients were taking DAPT. At 3-year follow-up, less than half (Absorb 106/265; XIENCE 50/134) of the patients remained on DAPT. Through 5 years, DAPT was continued in slightly over one-quarter of all patients (Absorb 79/265; XIENCE 36/134). Over 80% of patients were taking aspirin at 3 years (Absorb 237/266; XIENCE 116/134) and 5 years (Absorb 225/266; XIENCE 110/134) post-procedure.
Clinical Outcomes Through 5 YearsSummaries of all composite and component endpoints through 5 years, and between 3 and 5 years are provided in Table 2. No significant differences in event rates were detected between the Absorb and XIENCE arms between 0 and 5 years or 3 and 5 years. The cumulative 5-year rates for DMR, TVF, and TLF were numerically higher in the Absorb arm compared with the XIENCE arm, due to a higher number of events through the first 2 years.18 After 3 years, the event rates were similar between arms, with a slight trend for numerically lower rates of TVF, MACE, all MI in the Absorb arm compared with the XIENCE arm. The Kaplan-Meier estimates for TLF, ST, and ID-TLR are shown in Figure 2A–C. Landmark analysis at 1 year is demonstrated in Figure 3A–C. The majority of differences in clinical events were observed between 1 and 2 years.
| % (n/N) | ABSORB (n=266) | XIENCE (n=134) | P value |
|---|---|---|---|
| 3–5 years | |||
| Composite endpoints | |||
| DMR | 8.7% (21/242) | 8.9% (11/124) | 0.95a |
| TVF | 3.7% (9/242) | 4.8% (6/124) | 0.61a |
| MACE | 2.5% (6/242) | 3.2% (4/124) | 0.74b |
| TLF | 2.5% (6/242) | 2.4% (3/124) | 1.00b |
| Cardiac death/All MI | 1.2% (3/242) | 3.2% (4/124) | 0.23b |
| All-cause death/All MI | 3.3% (8/242) | 4.0% (5/124) | 0.77b |
| Individual endpoints | |||
| All-cause death | 3.3% (8/242) | 1.6% (2/124) | 0.50b |
| Cardiac death | 0.8% (2/242) | 0.8% (1/124) | 1.00b |
| All MI | 1.2% (3/242) | 2.4% (3/124) | 0.41b |
| TV-MI | 1.2% (3/242) | 1.6% (2/124) | 1.00b |
| QMI | 0.8% (2/242) | 0.0% (0/124) | 0.55b |
| NQMI | 0.8% (2/242) | 1.6% (2/124) | 0.61b |
| All revascularization | 6.2% (15/242) | 6.5% (8/124) | 0.92a |
| All TVR | 4.1% (10/242) | 4.0% (5/124) | 0.96b |
| ID-TVR | 3.3% (8/242) | 3.2% (4/124) | 1.00b |
| All TLR | 2.5% (6/242) | 1.6% (2/124) | 0.72b |
| ID-TLR | 1.7% (4/242) | 0.8% (1/124) | 0.67b |
| Scaffold/stent thrombosis | 0.0% (0/234) | 0.0% (0/122) | 1.00c |
| Definite | 0.0% (0/234) | 0.0% (0/122) | 1.00c |
| Probable | 0.0% (0/234) | 0.0% (0/122) | 1.00c |
| Cumulative events at 5 years | |||
| Composite endpoints | |||
| DMR | 29.1% (74/254) | 26.8% (34/127) | 0.63a |
| TVF | 16.1% (41/254) | 13.4% (17/127) | 0.48a |
| MACE | 11.8% (30/254) | 8.7% (11/127) | 0.35a |
| TLF | 11.0% (28/254) | 7.9% (10/127) | 0.33a |
| Cardiac death/All MI | 7.9% (20/254) | 6.3% (8/127) | 0.58a |
| All-cause death/All MI | 11.8% (30/254) | 7.9% (10/127) | 0.24a |
| Individual endpoints | |||
| All-cause death | 5.9% (15/254) | 3.1% (4/127) | 0.24b |
| Cardiac death | 1.2% (3/254) | 0.8% (1/127) | 1.00b |
| All MI | 7.5% (19/254) | 5.5% (7/127) | 0.47b |
| TV-MI | 6.7% (17/254) | 4.7% (6/127) | 0.45b |
| QMI | 3.1% (8/254) | 0.0% (0/127) | 0.06b |
| NQMI | 3.9% (10/254) | 4.7% (6/127) | 0.72b |
| All revascularization | 24.4% (62/254) | 22.8% (29/127) | 0.73a |
| All TVR | 15.0% (38/254) | 13.4% (17/127) | 0.68b |
| ID-TVR | 13.4% (34/254) | 10.2% (13/127) | 0.38b |
| All TLR | 10.2% (26/254) | 8.7% (11/127) | 0.62b |
| ID-TLR | 8.3% (21/254) | 4.7% (6/127) | 0.20b |
| Scaffold/stent thrombosis | 3.8% (9/237) | 1.6% (2/123) | 0.34c |
| Definite | 3.8% (9/237) | 0.8% (1/123) | 0.17c |
| Probable | 0.0% (0/237) | 0.8% (1/123) | 0.34c |
aChi-square test. bFisher’s exact test when Cochran’s rule not met. cFisher’s exact test. All P-values are two-tailed and not from prespecified hypothesis testing. Patients only counted once for each type of event in each time period. Revascularization includes TLR, TVR excluding TLR, and non-TVR. Denominators exclude patients who were truly lost to follow-up, defined as patients who were terminated through a given timepoint without any DMR event. Denominators used in the calculations of stent/scaffold thrombosis event rates were based on the analysis population and excluded patients who were lost to follow-up through 5 years without and stent/thrombosis event. DMR, composite endpoint of death, MI and revascularization; ID, ischemia driven; MI, myocardial infarction; NQMI, non-Q-wave MI; QMI, Q-wave MI; TLR, target lesion failure; TV, target vessel; TVR, target vessel revascularization.
Kaplan-Meier curves for the estimated cumulative incidence rates of (A) TLF, (B) scaffold/stent thrombosis, and (C) ID-TLR (per ARC definition) through 5 years (per ABSORB III protocol MI definition) with landmark analysis at 3 years. ARC, Academic Research Consortium; HR, hazard ratio; ID-TLR, Ischemia driven-target lesion revascularization; TLF, target lesion failure.
Kaplan-Meier curves for the estimated cumulative incidence rates of (A) TLF, (B) scaffold/stent thrombosis, and (C) ID-TLR (per ARC definition) through 5 years (per ABSORB III protocol MI definition) with Landmark analysis from 1 year. Abbreviations as in Figure 2.
At 5 years, the outcome rates for all-cause death, cardiac death, all MI, TV-MI, all revascularization, all TVR, all TLR, and scaffold/ST were similar in the Absorb and XIENCE arms (Table 2). Individual endpoint results between 3 and 5 years are described next.
Between 3 and 5 years, 8 deaths occurred in the Absorb arm: 6 noncardiac, and 2 cardiac. All deaths in the Absorb arm were deemed not related to the device by the investigators. In the same period, 2 deaths occurred in the XIENCE arm: 1 cardiac and 1 noncardiovascular. The cardiac death was deemed to be possibly related to the device by the investigator.
Beyond 3 years, there was no difference in all MI events between the Absorb and XIENCE arms. All MI events between 3 and 5 years happened in the target vessel (Table 2).
Numerically, there were more all TVR than all TLR cases, and most of the revascularizations were ischemia driven (ID). No significant difference in the incidence of all TVR, ID-TVR, all TLR, or ID-TLR was observed between the arms.
There were no incidences of definite/probable scaffold/ST in either arm beyond 3 years to 5-year follow-up.
There were no unanticipated adverse device effects in this study. Moreover, there were no new safety signals reported between 3 and 5 years.
Event Rates by Imaging SubgroupsAt the beginning of the study, patients were subrandomized into 1 of 3 intravascular imaging subgroups: OCT-1 group underwent optical coherence tomography (OCT) follow-up of the target lesion post-implantation, and at 2 and 3 years; OCT-2 group underwent OCT follow-up at 3 years; and the IVUS group underwent intravascular ultrasound (IVUS) follow-up post-implantation and at 3 years. The OCT-1 and -2 groups had 125 patients each (Absorb, n=83; XIENCE, n=42). The IVUS group comprised 150 patients, with 100 patients receiving the Absorb BVS and 50 patients receiving the XIENCE DES.
Kaplan-Meier estimated cumulative incidence rates through 5 years by imaging subgroup are presented in Figure 4A–C. There were no significant differences in TLF, scaffold/ST, and ID-TLR between the imaging subgroups through 5 years. However, a trend for slightly lower TLF and ID-TLR rates in the IVUS group, and higher scaffold/ST rates in the OCT-2 group was detected as compared with the respective other subgroups.
Kaplan-Meier curves for the imaging cohorts representing estimated cumulative incidence rates of (A) TLF, (B) scaffold/stent thrombosis, and (C) ID-TLR (per ARC definition) through 5 years (per ABSORB III protocol MI definition). IVUS, Intravascular ultrasound; OCT, optical coherence tomography. Other abbreviations as in Figure 2.
This study demonstrated 5-year clinical outcomes of Absorb BVS as compared with XIENCE stents in a randomized fashion. The rates for all-cause death, cardiac death, all MI, TV-MI, all revascularization, all TVR, all TLR, and scaffold/ST were similar in the Absorb and XIENCE arms at 5 years. No scaffold thrombosis occurred after 3 years in the Absorb BVS arm. There seems to be no signal of an excess of adverse events for BVS over CoCr-EES between 3 and 5 years, although this study was not powered for this clinical endpoint. This study also demonstrated for the first time comparisons of clinical outcomes among no periprocedural imaging, IVUS-guidance and OCT-guidance groups. A trend for slightly lower TLF and ID-TLR rates in the IVUS group, and higher scaffold/ST rates in the no imaging group was observed.
Although the primary endpoint at 1 year was met and approved in Japan in 2016, the Absorb BVS has already left the market because increases in scaffold thrombosis and TLR were demonstrated up to 3 years in many studies. Recently, 3-year imaging follow-up data of the current study were published.19 Longer-term data have been awaited to prove the benefit of the concept “leave nothing behind”. In this study, rates of target vessel MI and TVR in the Absorb group were identical to those in the XIENCE arm between 3 and 5 years. The Absorb BVS scaffold is made of poly L-lactic acid and is designed to be fully absorbed within approximately 3 years. After bioresorption, the vessel may be stabilized pathologically. Metallic stents, either bare metal or drug-eluting, have been proven in pathologic investigations to induce neoatherosclerosis at treated segments, which may increase late adverse events.2,20 Long-term follow-up data after 2nd-generation DES showed approximately 2% increase annually in TLF.21 Therefore, considerable numbers of adverse events will be anticipated for young patients with long life expectancy, and a completely bioresorbable scaffold may be the solution for this issue. The signal to be equivalent of adverse events in BVS over CoCr-EES seems promising for BVS to return to the field of PCI. However, improvement of the resorption profile and embedding in the vessel wall would be required for bioresorbable scaffolds to be involved in routine clinical practice. In fact, improved new scaffolds with thinner strut and equivalent mechanical integrity have been developed and are ready for clinical trials. Application for the superficial femoral artery is firstly going to be tested in a clinical trial.
Bioresorbable scaffolds are expected to recover physiological function after complete resorption. Recovery of vessel vasomotion was not shown at 3 years in the ABSORB II trial.22 It is reported that the vasomotor response to nitroglycerine after ABSORB implantation tended to increase between 2 and 5 years.23 The timing of the present study may be the proper moment to start demonstrating the benefit of BVS. It is regrettable that a vasomotion test at 5 years was not planned by the protocol. Posthoc investigation for vasomotor response may be endorsed in the future.
In this study, patients were subrandomized into 1 of 3 intravascular imaging subgroups at the time of enrollment. There were no significant differences in TLF, scaffold/ST, and ID-TLR between the imaging subgroups through 5 years due to insufficient numbers of subjects. Operators were not able to use any imaging devices for Absorb BVS implantation when assigned to the OCT-2 group. Although adequate technique (i.e., PSP [preparation, sizing, post-dilatation]) was not recommended during the enrollment period, the imaging arms (IVUS arm and OCT-1 group) tended to show better outcomes than the non-imaging arm. Because the scaffold was not visible by fluoroscopy, confirmation of apposition of struts was not possible with angiography. Incomplete apposition is a major cause of very late scaffold thrombosis.24 In this regard, imaging-guided PCI would be helpful to improve BVS outcomes. In Japan, in more than 80% of cases imaging is performed with IVUS or OCT. Outcomes for the Absorb BVS may be improved when the revised BVS comes onto the market, because appropriate techniques guided by imaging devices have been evolved. In fact, no scaffold thrombosis in 135 patients was reported from a post-marketing study, in which patients underwent BVS implantation with OCT guidance.25
Study LimitationsABSORB Japan was a modestly-sized trial and not powered for longer-term clinical endpoints. Results from meta-analysis of 4 randomized studies may be better to understand long-term safety.26 Scheduled angiographic follow-up (13 months, 2 years, 3 years) may have affected the ID-TLR rate (oculostenotic reflex). ABSORB Japan was initiated in 2013 (i.e., before the adoption of optimal techniques based on peer-to-peer experience). Although post-dilatation was performed in ∼80% of cases, the mean pressure applied was relatively low at 15.5 atm.
Neither very late ST nor scaffold thrombosis was observed in either arm of ABSORB Japan and TLF were low and comparable after 3 years, suggesting complete absorption and long-term safety of BVS.
We wish to acknowledge all investigators who participated in the ABSORB Japan Trial, and Ananya De of Criterion Edge for writing assistance.
The deidentified participant data will not be shared.
Abbott Vascular provided funding for this study.
This was an approval study reviewed by Ministry of Health, Labor and Welfare (MHLW) Reference no. Chiken number 12-301.
K.K., K.T., and T.K. were advisory board members of Abbott Vascular Japan. T.K. received research grant from Abbott Vascular Japan. H.K. and D.E. are full-time employees of Abbott Vascular. Others have nothing to disclose. Y.I. and T.K. are members of Circulation Journal ’ Editorial Team.
